PrimeSim™ HSPICE® User Guide:

Signal Integrity Modeling and Analysis

Version W-2024.09, September 2024
 Feedback

Contents
New in This Release . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
Related Products, Publications, and Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Customer Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Statement on Inclusivity and Diversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Part 1: Signal Integrity Analyses

1. Introduction to Signal Integrity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Getting Started on Signal Integrity Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Signal Integrity Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Analog Side of Digital Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
System Design Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Simulating Circuits with IBIS Models in PrimeSim HSPICE . . . . . . . . . . . . . . . . . . . 23
Using the B-element to Instantiate Individual Buffers . . . . . . . . . . . . . . . . . . . . .24
Using the IBIS Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
Determining Output and Input Pins in the IBIS Component . . . . . . . . . . . . .26
Power and Ground Pins in the IBIS Component . . . . . . . . . . . . . . . . . . . . . 27
Package and Pin Parasitics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Interconnect Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Power Supply of the IBIS Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2. Linear Network Parameter Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

LIN Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Identifying Ports with the P-element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
Using a Hierarchy Delimiter to Narrow the AMI Parameter Assignment . . . . . . 38
Using the P-element for Mixed-Mode Measurement . . . . . . . . . . . . . . . . . . . . . 38
.LIN Input Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
.LIN Output Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
.PRINT and .PROBE Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Hybrid Parameter Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

2
 Feedback

Contents

Multi-Port Scattering (S) Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Two-Port Transfer and Noise Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Equivalent Input Noise Voltage and Current . . . . . . . . . . . . . . . . . . . . . . . . 44
Equivalent Noise Resistance and Conductance . . . . . . . . . . . . . . . . . . . . . 45
Noise Correlation Impedance and Admittance . . . . . . . . . . . . . . . . . . . . . . 45
Optimum Matching for Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Noise Figure and Minimum Noise Figure . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Associated Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Output Format for Group Delay in .sc* Files . . . . . . . . . . . . . . . . . . . . . . . . 46
Output Format for Two-Port Noise Parameters in *.sc# Files . . . . . . . . . . . 46
Noise Parameters in 2-Port and N-Port Networks . . . . . . . . . . . . . . . . . . . . . . . 47
Hybrid (H) Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Group Delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Summary of Printing and Probing Network Parameters . . . . . . . . . . . . . . . . . . .50
Specifying the Analysis Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Port Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Port Numbers with More Than One Digit . . . . . . . . . . . . . . . . . . . . . . . . . . .51
Network Parameter Data Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51
Mixed-Mode Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
SNLIN Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Multitone Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
SNLIN Input Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
SNLIN Output Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
.PRINT and .PROBE Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .54
Output to Data Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Additional Measurements From .LIN Command . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Impedance Characterizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Stability Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .57
Gain Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Matching for Optimal Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
VSWR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
ZIN(i) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
YIN(i) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .59
K_STABILITY_FACTOR (Rollett Stability Factor) . . . . . . . . . . . . . . . . . . . . 59
MU_STABILITY_FACTOR (Edwards-Sinsky Stability Factor) . . . . . . . . . . . 59
Maximum Available Power Gain - G_MAX . . . . . . . . . . . . . . . . . . . . . . . . . 60
Maximum Stable Gain - G_MSG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Maximum Unilateral Transducer Power Gain - G_TUMAX . . . . . . . . . . . . . 60
Unilateral Power Gain - GU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Simultaneous Conjugate Match for G_MAX . . . . . . . . . . . . . . . . . . . . . . . . 61
Equivalent Input Noise Voltage and Current - IN2, VN2, RHON . . . . . . . . . 62

3
 Feedback

Contents

Equivalent Noise Resistance and Conductance - RN, GN . . . . . . . . . . . . . 62

Noise Correlation Impedance and Admittance - ZCOR, YCOR . . . . . . . . . .63
Optimum Matching for Noise - ZOPT, YOPT, GAMMA_OPT . . . . . . . . . . . .63
Noise Figure and Noise Figure Minimum - NF, NFMIN . . . . . . . . . . . . . . . . 63
Associated Gain - G_As . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .64
Example - Extracting Bandpass Filter S-parameters . . . . . . . . . . . . . . . . . . . . . . . . 64
Extracting Mixed-Mode Scattering (S) Parameters . . . . . . . . . . . . . . . . . . . . . . . . . .67
Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Defaults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Output File Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Two-Port Parameter Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .69
Output Format and Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Features Supported . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Prerequisites and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Reported Statistics for the Performance Log . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Errors and Warnings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71
Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier . . . . . . . . . . . . . . 71
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

3. Statistical Eye Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

Understanding StatEye Analysis in PrimeSim HSPICE . . . . . . . . . . . . . . . . . . . . . . 81
Statistical Eye Analysis Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .82
Port Element Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83
.PRINT and .PROBE Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Output Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .85
.MEASURE Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Understanding the Worst Pattern Measurement Result . . . . . . . . . . . . . . . . . . . 89
Understanding n-UI Accumulated Clock Period Jitter Measurements (RMS/
PP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Sample Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Saving and Reusing Initial Transient Results . . . . . . . . . . . . . . . . . . . . . . . . . . .92
Examples - Statistical Analysis Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Example – StateEye Crosstalk using W Element . . . . . . . . . . . . . . . . . . . . . . . 94
Determining Number of Bits Simulated in Statistical and Bit-by-bit Modes . . . . . . . .96
Statistical Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .96
Bit-by-bit Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

4
 Feedback

Contents

Pulse Superposition and Edge Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

Edge Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Statistical Eye Full Transient Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Example - Controlling Edges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Example - Run StatEye with Precomputed Transient Waveform . . . . . . . . . . . 100
Separate Rise Time and Fall Time Settings . . . . . . . . . . . . . . . . . . . . . . . . . . .101
Multi-Edge Responses for Accurate Modeling of Nonlinearity . . . . . . . . . . . . . 101
Capturing Steady State Noise for Power Integrity . . . . . . . . . . . . . . . . . . . . . . . . . 102
Example - Create SSN Eye Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Example - ADD SSN Eye Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Duty Cycle Distortion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .107
Capturing DCD Effect on System Clock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Periodic and Random Jitter Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Periodic Jitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Random Jitter Noise Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .111
Specifying Random Jitter by Peak-to-Peak Value with BER . . . . . . . . . . . . . . 113
Specifying Custom Jitter Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Port Element Syntax for Jitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
N-Cycle Jitter Modeling for Data and Clock Signals . . . . . . . . . . . . . . . . . . . . . . . .116
Pre-Emphasis and De-Emphasis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Syntax for P-Element and Transient V and I Sources . . . . . . . . . . . . . . . . . . . 122
Usage Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .124
Example 1: Pre-Emphasis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Example 2: De-Emphasis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .125
Built-In DFE Function for Rx Port Element in StatEye Analysis . . . . . . . . . . . . . . . 126
Perturbation Analysis with StatEye . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Unfolding Statistical Eye Diagrams to Waveforms . . . . . . . . . . . . . . . . . . . . . . . . . 130
Multiple Data Rate Eye Diagram Analysis with StatEye . . . . . . . . . . . . . . . . . . . . . 133
Accelerating StatEye’s Initial Transient with Process Distribution and Multi-
Threading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .133
Known Limitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Troubleshooting Output Warnings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .134
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

4. Time Domain Reflectometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .136

5
 Feedback

Contents

Optimizing TDR Packaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

Performing a TDR Simulation in PrimeSim HSPICE . . . . . . . . . . . . . . . . . . . . . . . 137
Basic TDR Impedance Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Example 1: Short Circuit Termination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Example 2: Open Circuit Termination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .139
Example 3: Mismatched Load Termination . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Example 4: Shunt Capacitance Discontinuity . . . . . . . . . . . . . . . . . . . . . . . . . .140
Example 5: Series Inductance Discontinuity . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Testbench Netlist Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .141
Input and Output Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Selecting the Risetime of the Incident Wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Simulating the Example DUTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Troubleshooting: TDR Zo Result Starts at Negative Value . . . . . . . . . . . . . . . . . . .145
Differential TDR Example Netlist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

Part 2: Signal Integrity Modeling

5. S-parameter Modeling Using the S-element . . . . . . . . . . . . . . . . . . . . . . . . . . . .149

S-parameter Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Notifications and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Using the Scattering Parameter Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
S Model Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
S-element Data File Model Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Mixed-Mode S-parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Relating Voltage and Current Waves to Nodal Waves . . . . . . . . . . . . . . . . . . . 161
Characterizing Differential Data Transfer Systems . . . . . . . . . . . . . . . . . . . . . .162
Deriving a Simpler Set of Voltage and Current Pairs . . . . . . . . . . . . . . . . . . . . 163
Using the Mixed-Mode S-parameters (S-element) . . . . . . . . . . . . . . . . . . . . . . 164
Accelerating S-Element Time Domain Performance with Recursive Convolution . .166
Multithreading Acceleration for S-element on Linux . . . . . . . . . . . . . . . . . . . . . 168
Ensuring Causality in the Rational Function Model . . . . . . . . . . . . . . . . . . . . . 169
Rational Function Matrix (.rfm) File Format . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Multiport Noise Model for Passive Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

6
 Feedback

Contents

Input Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172

Output Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
S-element Noise Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .173
Two-port Noise Parameter Support in Touchstone Files . . . . . . . . . . . . . . . . . 173
Input Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Output Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Notifications and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
S-parameter Standalone Manipulation Utility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
SPutil Program Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Invoking the Utility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
SPutil Runset Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
SPUTIL Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .186
Status Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Causality Checking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Using Interpolation for Causality Checking . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
Preconditioning S-parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Group Delay Handler in Time Domain Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
S-Model Data Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
Data Smoothing Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
S-Model Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
Predicting an Initial Value for FMAX in S-Element Models . . . . . . . . . . . . . . . . . . . 192
De-Embedding S-parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Small-Signal Parameter Data Frequency Table Model . . . . . . . . . . . . . . . . . . . . . . 195
SP Model Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
Four Valid Forms of the SP Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .197
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .199
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

6. W- and T-elements Transmission-Line Modeling . . . . . . . . . . . . . . . . . . . . . . . . 203

Overview of W-element Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
W-element Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Using the W-element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .205
Input Model 1: W-element, RLGC Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206

7
 Feedback

Contents

Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Sample Netlist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Specifying the RLGC Model in an External File . . . . . . . . . . . . . . . . . . . . 210
Input Model 2: U-element, RLGC Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Using RLGC Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Sample Netlist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
Input Model 3: Built-in Field-Solver Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Input Model 4: Frequency-Dependent Tabular Model . . . . . . . . . . . . . . . . . . . 215
Notation Used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .216
Table Model Card Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .216
Examples: 4-Conductor Tx Line and RLGC Model List . . . . . . . . . . . . . . . 217
Introducing Causality Check for W-element RLGC Table Model . . . . . . . . 219
Input Model 5: S Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
S-Model Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
S-Model Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
Input for W-element Passive Noise Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Input Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .223
Output Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Controlling W-element Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Setting Frequency Range of Interest for Greater Accuracy . . . . . . . . . . . . . . . 225
Setting .OPTION RISETIME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .225
Using DELAYOPT Keyword for Higher Frequency Ranges . . . . . . . . . . . . 226
Using DCACC Keyword for Lower Frequency Ranges . . . . . . . . . . . . . . . 226
Using Time-Step Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Static Time-Step Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Dynamic Time-Step Control using WACC . . . . . . . . . . . . . . . . . . . . . . . . . 227
Understanding Frequency-dependent Matrices and Transmission Line Theory . . . 228
Equations and Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Frequency-Dependent Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Introducing the Complex Dielectric Loss Model . . . . . . . . . . . . . . . . . . . . .231
Fitting Procedure Triggered by INCLUDEGDIMAG Keyword . . . . . . . . . . 233
Example: Complex Dielectric Loss Model . . . . . . . . . . . . . . . . . . . . . . . . . 234
Determining Matrix Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
Using the PRINTZO Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
Defining Frequency-dependent Impedance in Mixed Mode Printing . . . . . 237
File Description for *.wzo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Wave Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .241
Propagating a Voltage Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Handling Line-to-Line Junctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Overview of T-element Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .246

8
 Feedback

Contents

Using T-element for Predicting Simple Tx Line Effects . . . . . . . . . . . . . . . . . . . . . .246

T-element Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
Invoking Ideal T-element Transmission Lines . . . . . . . . . . . . . . . . . . . . . . . . . 247
Ideal Transmission Line Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
Selecting Ideal or Lossy Transmission Line Elements . . . . . . . . . . . . . . . .249
Lossless Voltage and Current Propagation . . . . . . . . . . . . . . . . . . . . . . . . 253
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255

7. Using the Field Solver to Extract Tx Line Parameters . . . . . . . . . . . . . . . . . . . 257

Using the Field Solver Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Filament Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Modeling Geometries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
Solver Limitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
Field-Solver-Related Netlist Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
Field Solver Model Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Using the Field Solver to Extract an RLGC Tabular Model . . . . . . . . . . . . . . . 262
Syntax for Extracting RLGC Tabular Model . . . . . . . . . . . . . . . . . . . . . . . . 262
COMPUTE_TABLE/COMPUTETABLE Keyword . . . . . . . . . . . . . . . . . . . . 263
Default . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .263
RLGC Tabular Model Sample Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
Accounting for Surface Roughness Effect in W-element . . . . . . . . . . . . . . . . . 265
Syntax for Scaling RS Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
Syntax for Taking RMS Surface Roughness of Conductor Materials . . . . . 266
Syntax for Choosing Surface Roughness Method . . . . . . . . . . . . . . . . . . . 266
Example: Transmission Line Surface Roughness . . . . . . . . . . . . . . . . . . . 266
Accelerating the W-element Field Solver Using Multicore Technology . . . . . . . . . . 267
Visualizing Cross-Sectional Geometric Information . . . . . . . . . . . . . . . . . . . . . . . . 268
Field Solver Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Example 1: Cylindrical Conductor Above a Ground Plane . . . . . . . . . . . . . . . .270
Example 2: Stratified Dielectric Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Example 3: Two Traces Between Two Ground Planes . . . . . . . . . . . . . . . . . . 274
Example 4: Using Field Solver with Monte Carlo Analysis . . . . . . . . . . . . . . . .275
Example 5: Modeling Coaxial and Shielded Twin-Lead Lines Using the Polar
Field Solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Coax Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Shielded Twin-Lead Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278

8. Using IBIS Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280

9
 Feedback

Contents

Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Verifying IBIS Files with the Golden Parser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Specifying Models for IBIS Buffers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Troubleshooting Signal Propagation Issues/Overclocking . . . . . . . . . . . . . . . . . . . 283
Multiple Rising and Falling Waveforms Support . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Differential Pins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Differential and Pseudo-Differential Buffers . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
Buffers in Subcircuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
Netlist Example with Output Buffer, Transmission Line, and Input Buffer . . . . . . . . 289
Using the IBIS Component Command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
How .IBIS Creates Buffers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
Required Keywords . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .291
file=’file_name’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
component=’component_name’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
Optional Keywords . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
pkgfile=’pkg_file_name’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
[Model Selector] Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Other Optional Keywords . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Component Calls for SPICE or Verilog-A Formatted Pins . . . . . . . . . . . . . . . . 294
Component Calls for SPICE or Verilog-A Formatted [External Circuit] . . . . . . .295
Buffer Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
Buffer Power ON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
Buffer Power OFF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Using IBIS Package Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .303
Accessing Nets inside a Package Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Using IBIS Board-Level Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
.EDB and .IBIS Command Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
Circuit Topology Created by the .EBD and .IBIS Commands . . . . . . . . . . . . . .307
B-element Naming Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Circuit Topology Created with SPICE or Verilog-A Formatted Pins . . . . . . . . . 310
SPICE or Verilog-A Formatted B-element Naming Rules . . . . . . . . . . . . . 310
IBIS Board-Level Component Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
Using IBIS Interconnect Modeling (ICM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318

10
 Feedback

Contents

9. Using IBIS-AMI Equalizer Models with StatEye . . . . . . . . . . . . . . . . . . . . . . . . . 319

IBIS-AMI Model Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
Requirements for Using IBIS-AMI Models with PrimeSim HSPICE . . . . . . . . . 321
Using the Standalone AMI Object Checking Program ’amicheck’ . . . . . . . . . . . . . 321
amicheck Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
Windows OS ’amicheck’ Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
amicheck Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
How AMI Works in PrimeSim HSPICE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
Input Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
Saving AMI Pulse/Step Responses and Filters . . . . . . . . . . . . . . . . . . . . . . . . 325
Visualizing Eye Diagram Output of an IBIS-AMI RX Model . . . . . . . . . . . . . . . 325
Usage Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .326
StatEye Analysis with AMI Using Demo Example . . . . . . . . . . . . . . . . . . . . . . . . . 326
Parametric Control over AMI File for Quoted Expressions . . . . . . . . . . . . . . . . . . . 328
Backchannel Communication Interface for IBIS-AMI’s Link Training Protocol . . . . 331
Input Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
BIC_ID Naming Convention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
Using an IBIS-AMI Mid-Channel Repeater Model . . . . . . . . . . . . . . . . . . . . . . . . . 332
Mid-Channel Repeater Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .333
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334

Part 3: Feed-Forward Equalization with the PrimeSim HSPICE Tool

10. Introduction to Feed-Forward Equalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336

FFE Modeling With Verilog-A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
Built-in FFE Model Using Port Emphasis Parameters . . . . . . . . . . . . . . . . . . . . . . 337
E-element Delay Based Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .338

Appendices

A. Simulating Circuits with Signetics Drivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . .340

11
 Feedback

Contents

Simulating Circuits with Signetics Drivers—Examples . . . . . . . . . . . . . . . . . . . . . . 340

B. Simulating Circuits with Xilinx FPGAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344

Introduction to Simulating Circuits with Xilinx FPGAs . . . . . . . . . . . . . . . . . . . . . . .344
Syntax for IOB (xil_iob) and IOB4 (xil_iob4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .344
Ground-Bounce Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
Coupled Line Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .348

12
Copyright and Proprietary Information Notice
© 2024 Synopsys, Inc. This Synopsys software and all associated documentation are proprietary to Synopsys, Inc.
and may only be used pursuant to the terms and conditions of a written license agreement with Synopsys, Inc. All
other use, reproduction, modification, or distribution of the Synopsys software or the associated documentation is
strictly prohibited.
Destination Control Statement
All technical data contained in this publication is subject to the export control laws of the United States of America.
Disclosure to nationals of other countries contrary to United States law is prohibited. It is the reader’s responsibility to
determine the applicable regulations and to comply with them.
Disclaimer
SYNOPSYS, INC., AND ITS LICENSORS MAKE NO WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
WITH REGARD TO THIS MATERIAL, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
Trademarks
Synopsys and certain Synopsys product names are trademarks of Synopsys, as set forth at
https://www.synopsys.com/company/legal/trademarks-brands.html.
All other product or company names may be trademarks of their respective owners.
Free and Open-Source Licensing Notices
If applicable, Free and Open-Source Software (FOSS) licensing notices are available in the product installation.
Third-Party Links
Any links to third-party websites included in this document are for your convenience only. Synopsys does not endorse
and is not responsible for such websites and their practices, including privacy practices, availability, and content.
www.synopsys.com

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 13

W-2024.09
 Feedback

About the PrimeSim HSPICE User Guide: Signal

Integrity Modeling and Analysis
The PrimeSim HSPICE User Guide: Signal Integrity Modeling and Analysis describes how
to use the Synopsys PrimeSim HSPICE® tool to achieve and maintain signal integrity in
your chip design.
This preface includes the following sections:
• New in This Release
• Related Products, Publications, and Trademarks
• Conventions
• Customer Support
• Statement on Inclusivity and Diversity

New in This Release

Information about new features, enhancements, and changes, known limitations, and
resolved Synopsys Technical Action Requests (STARs) is available in the PrimeSim
HSPICE Release Notes on the SolvNetPlus site.

Related Products, Publications, and Trademarks

For additional information about the PrimeSim HSPICE simulator, see the documentation
on the Synopsys SolvNetPlus support site at the following address:
https://solvnetplus.synopsys.com
You might also want to see the documentation for the following related Synopsys products:
PrimeWave™ Design Environment

Conventions
The following conventions are used in Synopsys documentation.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 14

W-2024.09
 Feedback
About the PrimeSim HSPICE User Guide: Signal Integrity Modeling and Analysis
Customer Support

Convention Description

Courier Indicates syntax, such as write_file.

Courier italic Indicates a user-defined value in syntax, such as

write_file design_list

Courier bold Indicates user input—text you type verbatim—in examples, such
as
prompt> write_file top

Purple • Within an example, indicates information of special interest.

• Within a command-syntax section, indicates a default, such as
include_enclosing = true | false

[] Denotes optional arguments in syntax, such as

write_file [-format fmt]

... Indicates that arguments can be repeated as many times as

needed, such as
pin1 pin2 ... pinN.

| Indicates a choice among alternatives, such as

low | medium | high

\ Indicates a continuation of a command line.

/ Indicates levels of directory structure.

Bold Indicates a graphical user interface (GUI) element that has an

action associated with it.

Edit > Copy Indicates a path to a menu command, such as opening the Edit
menu and choosing Copy.

Ctrl+C Indicates a keyboard combination, such as holding down the Ctrl

key and pressing C.

Customer Support
Customer support is available through SolvNetPlus.

Accessing SolvNetPlus
The SolvNetPlus site includes a knowledge base of technical articles and answers to
frequently asked questions about Synopsys tools. The SolvNetPlus site also gives you
access to a wide range of Synopsys online services including software downloads,
documentation, and technical support.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 15

W-2024.09
 Feedback
About the PrimeSim HSPICE User Guide: Signal Integrity Modeling and Analysis
Statement on Inclusivity and Diversity

To access the SolvNetPlus site, go to the following address:

https://solvnetplus.synopsys.com
If prompted, enter your user name and password. If you do not have a Synopsys user
name and password, follow the instructions to sign up for an account.
If you need help using the SolvNetPlus site, click REGISTRATION HELP in the top-right
menu bar.

Contacting Customer Support

To contact Customer Support, go to https://solvnetplus.synopsys.com.

Statement on Inclusivity and Diversity

Synopsys is committed to creating an inclusive environment where every employee,
customer, and partner feels welcomed. We are reviewing and removing exclusionary
language from our products and supporting customer-facing collateral. Our effort also
includes internal initiatives to remove biased language from our engineering and working
environment, including terms that are embedded in our software and IPs. At the same
time, we are working to ensure that our web content and software applications are usable
to people of varying abilities. You may still find examples of non-inclusive language in our
software or documentation as our IPs implement industry-standard specifications that are
currently under review to remove exclusionary language.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 16

W-2024.09
 Feedback

Part 1: Signal Integrity Analyses

Describes some of the factors and various analyses that can affect signal integrity in your
design.

Part 1 focuses on signal integrity analyses and consists of the following chapters:
• Chapter 1, Introduction to Signal Integrity
• Chapter 2, Linear Network Parameter Analysis
• Chapter 3, Statistical Eye Analysis
• Chapter 4, Time Domain Reflectometry

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 17

W-2024.09
 Feedback

1
Introduction to Signal Integrity

Describes some of the factors that can affect signal integrity in your design.

The performance of an IC design does not only depends on how many million transistors
a vendor fits on a single chip. With tighter packaging space and increasing clock
frequencies, packaging issues and system-level performance issues (such as crosstalk
and transmission lines) become increasingly significant. At the same time, the popularity of
multichip packages and increased I/O counts is forcing package design to become more
like chip design.
This user guide consists of three parts and a set of appendices:

Part Content

Part 1, Signal Integrity Analyses Chapter 1, Introduction to Signal Integrity

Chapter 2, Linear Network Parameter Analysis

Chapter 3, Statistical Eye Analysis

Chapter 4, Time Domain Reflectometry

Part 2, Signal Integrity Modeling Chapter 5, S-parameter Modeling Using the

S-element

Chapter 6, W- and T-elements

Transmission-Line Modeling

Chapter 7, Using the Field Solver to Extract Tx

Line Parameters

Chapter 8, Using IBIS Files

Chapter 9, Using IBIS-AMI Equalizer Models

with StatEye

Part 3, Feed-Forward Equalization with the Chapter 10, Introduction to Feed-Forward

PrimeSim HSPICE Tool Equalization

Appendixes Appendix A, Simulating Circuits with Signetics

Drivers

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 18

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity

Part Content

Appendix B, Simulating Circuits with Xilinx

FPGAs

PrimeSim HSPICE ships hundreds of examples for your use. Find signal integrity-related
demo files under Signal Integrity Examples, S-parameter Examples, IBIS Examples,
Transmission (W-element) Line Examples, and Transmission Lines Examples, in the
PrimeSim HSPICE User Guide: Demonstration Netlists.
To copy and paste proven syntax, use the demonstration files shipped with your
installation of PrimeSim HSPICE (see Listing of Demonstration Input Files in the PrimeSim
HSPICE User Guide: Demonstration Netlists). Attempting to copy and paste from the
manual or help documentation may present unexpected results as the text used in
formatting may include hidden characters and white space for visual clarity.
Note:
The measurement system in this manual always refers to MKS units (meter,
kilogram, second measurement), unless otherwise stated. PrimeSim HSPICE
expects length and width units of meters. But PrimeSim HSPICE does directly
support units of “mil” (0.001 inch, 25.4e-06 meters) as input. For example, you
can specify a transmission line with a length of 0.4 inches:
T1 IN 0 OUT 0 Z0=50 f=1meg L=400mil

To get the results you expect, use caution when mixing units of measure. For
reference, here are other “m” units. Do not confuse Mega, which you express
as “meg” or “x”, with “m” (mili):
1m = 1e-3 (mili)
1meg = 1x = 1e6 (mega)
1u = 1e-6 (micro)
The following sections discuss these topics:
• Getting Started on Signal Integrity Simulations
• Simulating Circuits with IBIS Models in PrimeSim HSPICE

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 19

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Getting Started on Signal Integrity Simulations

Getting Started on Signal Integrity Simulations

In a signal integrity simulation, you must model the following components:
• Driver cell including parasitic pin capacitances and package lead inductances
• Transmission lines
• A receiver cell with parasitic pin capacitances and package lead inductances
• Terminations or other electrical elements on the line
Model the transmission line as closely as possible—that is, to maintain the integrity of
the simulation, include all electrical elements exactly as they appear on the backplane or
printed circuit board.
printed circuit boards
You can use readily-available I/O drivers from ASIC vendors and the PrimeSim HSPICE
device models advanced lossy transmission lines to simulate the electrical behavior of
the board interconnect, bus, or backplane. You also can analyze the transmission line
behavior under various conditions.
transmission linessimulationinterconnectanalyzing
In addition, PrimeSim HSPICE preserves the necessary electrical characteristics with full
transistor-level library circuits.
PrimeSim HSPICE can simulate systems by using:
• System-level behavior, such as local component temperature and independent models,
to accurately predict electrical behavior.
systems, simulating
• Automatic inclusion of library components by using the SEARCH option.
• Lossy transmission line models that:
◦ Support common-mode simulation.
◦ Include ground-plane reactance.
◦ Include resistive loss of conductor and ground plane.
◦ Allow multiple signal conductors.
◦ Require minimum CPU computation time.

Signal Integrity Problems

Table 1 lists some of the signal integrity problems that might cause failures in high-speed
designs.
noisesourcesdelayscausesdelaysproblems and solutionsnoiseproblems and solutionsdesignhigh speed, problems and solutions
PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 20
W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Getting Started on Signal Integrity Simulations

Table 1 High-Speed Design Problems and Solutions

Signal Integrity Causes Solution
Problem

Noise: delta I (current) Multiple simultaneously-switching Adjust or evaluate location, size, and
drivers; high-speed devices create value of decoupling capacitors.
larger delta I

Noise: coupled Closely-spaced parallel traces Establish design rules for lengths of
(crosstalk) parallel lines.

Noise: reflective Impedance mismatch Reduce the number of connectors and

select proper impedance connectors.

Delay: path length Poor placement and routing; too Choose MCM or other high-density
many or too few layers; chip pitch packaging technology.

Propagation speed Dielectric medium Choose the dielectric with the lowest
dielectric constant.

Delay: rise time Resistive loss and impedance Adjust width, thickness, and length of
degradation mismatch line.

The following sections discuss these topics:

• Analog Side of Digital Logic
• System Design Issues

Analog Side of Digital Logic

Circuit simulation of a digital system becomes necessary when the analog characteristics
analog, circuit simulation of a digital system

of the digital signals become electrically important. The integrity of the digital quality of the
signals require careful circuit analysis.
The roadblocks to successful high-speed digital designs are noise and signal delays.
Digital noise can originate from several sources. The fundamental digital noise sources
are:
noisesources
• Line termination noise - Additional voltage reflected from the load back to the driver
caused by an impedance mismatch. Digital output buffers do not accurately control the
output impedance. Most buffers have different rising and falling edge impedances.
• Ground bounce noise - Generated where leadframes or other circuit wires cannot form
into transmission lines. The resulting inductance creates an induced voltage in the
ground circuit, supply circuit, and output driver circuit. Ground bounce noise lowers the
noise margins for the rest of the system.
• Coupled line noise - Induced from lines that are physically adjacent. This noise is
generally more severe for data lines that are next to clock lines.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 21

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Getting Started on Signal Integrity Simulations

As system cycle times approach the speed of electromagnetic signal propagation for the
printed circuit board, consideration of the line length becomes critical. The system noises
and line delays interact with the electrical characteristics of the gates and might require
circuit-level simulation.

System Design Issues

Exceeding the noise quota might not cause a system to fail. Maximum noise becomes a
problem only when PrimeSim HSPICE accepts a digital input. If a digital system engineer
can decouple the system, PrimeSim HSPICE can accept a much higher level of noise.
noisequota
Some common methods that a digital system engineer can use to decouple a system
include:
decoupling methods
• Multiple ground and power planes on the printed circuit board (PCB), multichip module
(MCM), and pin grid array (PGA)
• Separating signal traces with ground traces
• Decoupling capacitors
• Series resistors on output buffer drivers
• Twisted-pair line driving
In present systems designs, you must select the best packaging methods at three levels:
• PCB
• MCM
• PGA
Extra ground and power planes are often necessary to lower the supply inductance and to
provide decoupling:
• Decoupling capacitors must have very low internal inductance to be effective for high-
speed designs
• Newer designs frequently use series resistance in the output drivers to lower circuit
ringing
• Critical high-speed driver applications use twisted differential-pair transmission lines
A system engineer must determine how to partition the logic. The propagation speed of
signals on a printed circuit board is about 6 in/ns. As digital designs become faster, wiring
interconnects become a factor in how you partition a logic.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 22

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Simulating Circuits with IBIS Models in PrimeSim HSPICE

The critical wiring systems are:

• IC-level wiring
• Package wiring for SIPs, DIPs, PGAs, and MCMs
• Printed circuit-board wiring
• Backplane and connector wiring
• Long lines–power, coax, or twisted pair
If you use ASIC or custom integrated circuits as part of your system logic partitioning
strategy, you must make decisions about integrated circuit level wiring. The more-familiar
decisions involve selecting packages and arranging packages on a printed circuit board.
Large systems generally have a central backplane, which becomes the primary challenge
at the system partition level.
designpartitioningtransmission lineseffects
Use the following equation to estimate wire length when transmission line effects become
noticeable:

critical length=(rise time)*velocity/8 wire

For example, if rise time is 1 ns and board velocity is 6 in./ns, then distortion becomes
noticeable when wire length is 3/4 in. The PrimeSim HSPICE circuit simulator contains
a field solver to extract full loss transmission line models and the linear analysis (.LIN)
feature can extract S-parameter models for packages or complex interconnects.

Simulating Circuits with IBIS Models in PrimeSim HSPICE

There are two ways to instantiate IBIS buffers in your PrimeSim HSPICE testbench. You
can instantiate buffers one at a time with the B-element or you can instantiate all the
buffers for a given component using the .IBIS component card. With the IBIS component,
package parasitics can be annotated automatically. With the B-element, these must be
added as PrimeSim HSPICE circuit elements if desired. For a complete discussion of the
IBIS model components, see Chapter 8, Using IBIS Files.
The following sections discuss these topics:
• Using the B-element to Instantiate Individual Buffers
• Using the IBIS Component
• Power Supply of the IBIS Buffer

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 23

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Simulating Circuits with IBIS Models in PrimeSim HSPICE

Using the B-element to Instantiate Individual Buffers

If you decide to select and instantiate individual buffers using the B-element syntax, you
create the node list and choose what to name the nodes. The list of required nodes for
a buffer depends on the buffer type. IBIS models are supported according to the current
PrimeSim HSPICE-supported IBIS standard. For a complete list of supported buffers, see
Buffer Types in PrimeSim Continuum User Guide: Elements. For illustration, here are six
common types:
• Input buffer:
B_INPUT nd_pc nd_gc nd_in nd_out_of_in

• Output buffer:
B_OUTPUT nd_pu nd_pd nd_out nd_in [nd_pc nd_gc]

• Input ECL Buffer:

B_INPUT_ECL nd_pc nd_gc nd_in nd_out_of_in

• Output ECL Buffer:

B_OUTPUT_ECL nd_pu nd_out nd_in [nd_pc nd_gc]

• Tri-state buffer:
B_3STATE nd_pu nd_pd nd_out nd_in nd_en [nd_pc nd_gc]

• Input/Output buffer:
B_IO nd_pu nd_pd nd_out nd_in nd_en nd_out_of_in [nd_pc nd_gc]

In the preceding examples, the required nodes are listed first and the optional nodes are
in brackets ([ ]). The position in the list, not the name, selects the node function. You can
name the nodes using any valid PrimeSim HSPICE node name. In these examples, pu
and pd are pullup and pulldown; pc and gc are power clamp and ground clamp; nd simply
stands for node.
The following is a sample B-element instantiation of an output (driver) buffer:
b_out1 nd_pu nd_pd out1 in1
+ file = 'at16245.ibs'
+ model = 'AT16245_OUT'

This is the minimum syntax to add a B-element. Note that both the file and model names
are case-sensitive. There are numerous additional options, many of which are shared with
the IBIS component.
For more details, see Specifying Required and Optional Common Keywords in PrimeSim
Continuum User Guide: Elements.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 24

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Simulating Circuits with IBIS Models in PrimeSim HSPICE

Using the IBIS Component

To use the .IBIS component card, your IBIS file must have a [Component] section. When
you use the IBIS component, PrimeSim HSPICE automatically creates node names for
all the buffer pins in the component. These node names are created by concatenating the
component name, the pin name, and the pin’s function. For node naming, the component
name is the user-supplied name given to the .IBIS card—not the component name.
For example, if the root of derived node name is
.IBIS ddr256

The output node of an output buffer (driver) might look like:

ddr256_q1_o

where ddr256 is the component name, q1 is the pin name, and _o denotes the driver
output.
PrimeSim HSPICE adds underscore symbols (_) in between for readability. You do not
have to create these nodes yourself in the netlist. They exist logically, and you can make
connections to them simply by using the derived name.
Just as in the B-element node list, the nodes that are created for each buffer depend on
the buffer type. For example, a tri-state buffer generates an enable node, where a simple
input or output buffer does not.
An easy way to get started with the IBIS component is to create a skeleton netlist with only
the IBIS component and a simple analysis. When you probe all signals, you see a list of all
the known nodes for each buffer in your waveform tool.
Here is an example of the minimum syntax to instantiate an IBIS component and generate
a plotfile:
* Instantiate an IBIS component to see the logical nodes created
.ibis ddr256
+ file ='hyb256_400.ibs'
+ component ='hyb25d256'
.tran 1n 5n
.option post probe dcstep=1
.probe v(ddr*)
.end

Since all logical nodes of the IBIS component start with ddr256_, using .probe
v(ddr256_*) outputs all the nodes created by the IBIS component instantiation. The
dcstep option prevents PrimeSim HSPICE from printing numerous warnings about nodes
with no path to ground.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 25

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Simulating Circuits with IBIS Models in PrimeSim HSPICE

The following sections describe these topics:

• Determining Output and Input Pins in the IBIS Component
• Power and Ground Pins in the IBIS Component
• Package and Pin Parasitics
• Interconnect Model

Determining Output and Input Pins in the IBIS Component

How you connect the nodes of a buffer instantiated by an IBIS component depends on if
it is an input, output or I/O function. There is a special pin for I/O buffers named “outofin”.
When an IO buffer is in receive mode, the input is external to the component and sits on
your SI path. The output of that input buffer (or “outofin”) is inside the component. If you
probe the outofin node, you see it is an ideal, behavioral digital waveform. The regular
input buffer does not have a separate “outofin” node. Instead, the “_i” node acts as outofin.

Buffer Type Derived Logical Node Name Example

Input buffers - receivers

The input (receiving) node of an input buffer: 'cname'_'pin_name' ddr256_q1

The output node of an input buffer: cname'_'pin_name'_i ddr256_q1_i

Output buffers - drivers

Input node of an output buffer: 'cname'_'pin_name'_i ddr256_q1_i

Output node of an output buffer: 'cname'_'pin_name' ddr256_q1

Die side output node, if package model used: 'cname'_'pin_name'_o ddr256_q1_o

IO buffers - input/output

Input of an IO buffer in input mode: 'cname'_'pin_name' ddr256_q1

Output of an IO buffer in input mode: 'cname'_'pin_name'_outofin ddr256_q1_outofin

Input of an IO buffer in output mode: 'cname'_'pin_name'_i ddr256_q1_i

Output node of an IO buffer in output mode: 'cname'_'pin_name' ddr256_q1

Die side output node of an IO buffer in output mode, if 'cname'_'pin_name'_o ddr256_q1_o

package model used:

Buffer Node Names General Guidelines

Die side output node, if package model used: 'cname'_'pin_name'_o

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 26

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Simulating Circuits with IBIS Models in PrimeSim HSPICE

Buffer Node Names General Guidelines

Output node or input node for input buffers: 'cname'_'pin_name'

Enable node for buffers with enable function: 'cname'_'pin_name'_en

Input node or outofin node for input buffers: 'cname'_'pin_name'_i

Outofin node for buffers other than input buffers: 'cname'_'pin_name'_outofin

Power and Ground Pins in the IBIS Component

You may wish to externally power your IBIS buffer in certain situations, such as to simulate
Simultaneous Switching Noise, or “ground bounce”. Nodes of power and ground pins are
created in the “logical” netlist but no buffer is created for them. The pins can also have
specific pin or general package parasitics associated with them such as the other IO pins
in the component.

Package and Pin Parasitics

The IBIS component gives you the option to automatically annotate the general R_pkg,
L_pkg and C_pkg parasitics in the [Package] section, or the pin specific R_pin, L_pin and
C_pin parasitics in the [Pin] section, or the complicated package model declared in the
[Package Model] section. Please refer the keyword package of the .IBIS command.

Interconnect Model
IBIS 7.0 supports broadband interconnect models describing connections between the
pins of a component, die pads, and its I/O buffers. The interconnect models define the
interconnections to either IBIS-ISS subcircuits or n-port networks described by Touchstone
files. PrimeSim HSPICE automatically creates X and S elements with proper connection
based on the data in the IBIS file.
Note:
If IBIS-ISS subcircuits are used in the component, then the files containing
the subcircuits should be explicitly included using the .INC command in the
PrimeSim HSPICE netlist.
For an IBIS file with version 7.0, if interconnect models are used in an IBIS component,
then any package data in ([Pin], [Package], or [Package Model] are ignored. For backward
compatibility, you can use the package keyword of the .IBIS command. For more
information, see .IBIS in the PrimeSim HSPICE Reference Manual: Commands and
Control Options.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 27

W-2024.09
 Feedback
Chapter 1: Introduction to Signal Integrity
Simulating Circuits with IBIS Models in PrimeSim HSPICE

Power Supply of the IBIS Buffer

For the B-element: The physical pull-up and pull-down nodes are, by default, connected
to supply and ground by the power=on directive. If you physically connect these nodes to
an external voltage source and ground, be sure to set power=off in your netlist.
For the IBIS component logical netlist: If there is [Pin Mapping] for this component in
the IBIS file, then physical pull-up and pull-down nodes are connected automatically to
power and ground pins of the component. You should externally power your buffer from
such pins manually. So power=off is the default in this situation.
If there is no [Pin Mapping] for this component in the IBIS file, then, like the B-element
card, the physical pull-up and pull-down nodes are, by default, connected to the supply
and ground by the power=on directive. You need not add an external voltage source for
them.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 28

W-2024.09
 Feedback

2
Linear Network Parameter Analysis

Describes how to perform an AC sweep to extract small-signal linear network parameters

in PrimeSim HSPICE and advanced analog simulations.

The chapter covers LIN analysis, RF measurements from .LIN command, extracting
mixed-mode S (scattering) and noise parameters, and additional measurements.
PrimeSim HSPICE ships hundreds of examples for your use; see Listing of Demonstration
Input Files in the PrimeSim HSPICE User Guide: Demonstration Netlists for paths to the
demo files.
For descriptions of individual commands referenced in this chapter, see the PrimeSim™
Continuum Reference Manual: Commands and Control Options.
These topics are covered in the following sections:
• LIN Analysis
• Additional Measurements From .LIN Command
• Example - Extracting Bandpass Filter S-parameters
• Extracting Mixed-Mode Scattering (S) Parameters
• Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier
• References

LIN Analysis
The .LIN command extracts noise and linear transfer parameters for a general multi-port
network.
When used with the .AC command, .LIN makes available a broad set of linear port-wise
.LINcommandmulti-port scattering parametersextracting noise, linear transfer parameters

measurements:
• Multi-port scattering [S] parameters
• Noise parameters

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 29

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

• Stability factors
• Gain factors
• Matching coefficients
The LIN analysis is similar to basic small-signal, swept-frequency .AC analysis, but it also
automatically calculates a series of noise and small-signal transfer parameters between
the terminals identified using port (P) elements.
PrimeSim HSPICE can output the result of group delay extraction and two-port noise
analysis to either a *.sc file, a Touchstone 1.0/2.0 file, or a CITI file.
The .PRINT/.PROBE/.MEAS output syntax for .LIN supports S/Y/Z/H (hybrid) parameters,
group delay, noise parameters, stability factors, gain factors, and matching coefficients.

Figure 1 Basic circuit in LIN analysis

I1 I2

+ Circuit +
Z01 P1 V1 under V2 P2 Z02
- test -

The following sections discuss these topics:

• Identifying Ports with the P-element
• Using the P-element for Mixed-Mode Measurement
• .LIN Input Syntax
• .LIN Output Syntax
• Multi-Port Scattering (S) Parameters
• Two-Port Transfer and Noise Calculations
• Noise Parameters in 2-Port and N-Port Networks
• Hybrid (H) Parameters
• Group Delay
• Summary of Printing and Probing Network Parameters

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 30

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Identifying Ports with the P-element

The .LIN command computes the S- (scattering), Y- (admittance), Z- (impedance), and H-
.LINscattering, admittance, impedance parameters

(hybrid) parameters directly based on the location of the port (P) elements in your circuit,
and the specified values for their reference impedances.
The port element identifies the ports used in LIN analysis.
Syntax

Pxxx node1 node2 [node3 node4 node5] port=portindex

+ [Z0=val] $Port Impedance
+ [DC mag] [AC mag phase] [HBAC mag phase] $Voltage/Power Info
+ [HB mag phase harm tone modharm modtone]
+ [transient_waveform]
+ [ENCODE=DW8B10B [RD_INIT=0|1]]
+ [ENCODE=DW64b66b [TXC=txcdata]]
+ [ENCODE=DW128b130b|DW128b132b [HEAD=DATA|]]CONTROL
+ [TRANFORHB=[0|1]] [DCOPEN=[0|1]]
+ [power=[0|1|2|W|dbm]] $Power Switch
+ [RDC=val] [RAC=val] $ Source Impedance Overrides
+ [RHBAC=val] [RHB=val] [RTRAN=val]
+ [Emphasis_Level=val] [Emphasis_Time=val] $ Emphasis
+ $ **** DFE ****
+ [DFE_tap=n1 n2 … nN DFE_coeff=val1, val2, … valN]

+ [DFE_TAP_OPT=N]
+ [DFE_VTH=val]
+ [DFE_CLK=0|1]
+ [DCD=val] [DCD_TYPE=0|1|2|3] $ Duty Cycle Distortion
+ [PJ=val] [PJ_TYPE=0|1|2|3|4|5|6] $ Period Jitter
+ [RJ=val1, [val2, ... valN]] [VN=val] $ Rand Jitter Stateye
+ [AMI_OBJ=filename] [AMI_PARAM=filename] [AMI_CDR=0|1]$AMI
+ [AMI_CORNER=idx]$AMI
+ [MODULATION=PAM4]
+ [PAM4_MAPPING=string]
+ [IBIS_AMI_FILE=filename IBIS_AMI_MODEL=modelname]
+ [AMI_PARAM_xx = string|quoted string]
+ [IBIS_AMI_USE_CCOMP = 0|1]
+ [IBIS_AMI_CCOMP_ORDER = 1|2]
+ [IBIS_AMI_VRANGE_MODE = 1|2]
+ [IBIS_AMI_USE_RAMP = 0|1]
+ [IBIS_AMI_USE_VRANGE = 0|1]
+ [IBIS_AMI_USE_ZSRC = 0|1]

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 31

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Parameter Description

node2 [node3 node4 The port element can have either two, or three, or five terminals.
node5] The three-terminal model can be used for differential ports (see Using
the Port Element for Mixed-Mode Measurement).
You can also add a five-terminal model (see Using the Five Terminal Port
Element) mainly to take a four-port S-parameter block for the IBIS-AMI's
"Ts4file Analog Buffer Models" (BIRD 158).

port=portindex The port index. Numbered sequentially beginning with 1 with no shared
port indexes.

z0=val (or Zo=val) Port impedance (Ohms). (Default: 50). Sets port characteristic
impedance used for LIN analysis, sets port termination impedance for
other analyses, and also sets source impedance when you use the port
element as a signal source.

DC mag DC voltage or power source value.

AC mag phase AC voltage or power source value.

HBAC mag phase (PrimeSim HSPICE advanced analog analyses) HBAC voltage or power
source value.

HB mag phase harm tone (PrimeSim HSPICE advanced analog analyses) HB voltage, current,
modharm modtone or power source value. Allows multiple HB specifications with different
harm, tone, modharm, and modtone values.
• phase is in degrees.
• harm and tone are indices corresponding to the tones specified in
the .HB statement. Indexing starts at 1 (corresponding to the first
harmonic of a tone).
• modtone and modharm specify sources for multi-tone simulation.
A source specifies a tone and a harmonic, and up to 1 offset tone
and harmonic (modtone for tones and modharm for harmonics).
Then it describes the signal as: V(or I)=mag*cos(2*pi*
(harm*tone+modharm*modtone)*t + phase)

transient_waveform (Transient analysis) Voltage or power source waveform. Any one of

waveforms: AM, EXP, PULSE, PWL, SFFM, SIN, PAT, or PRBS. Does
not allow multiple transient descriptions.

ENCODE=DW8b10b Keyword to specify 8b/10b encoding.

RD_INIT=0|1 Initial value of Running Disparity. Running Disparity.1 is the name of the
one-bit memory that recalls the bias of the last unbalanced code word.
• 0: Specifies that a Running Disparity value of zero is synonymous
with negative Running Disparity (-).
• 1: Specifies that a Running Disparity value of one is synonymous with
positive Running Disparity (+).

ENCODE=DW64b66b Keyword to specify 64b/66b encoding.

Note:
DW64b66b encoding is valid only for the PAT transient_waveform
type.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 32

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Parameter Description

TXC=txcdata The input 64-bit pattern signal will be divided into 8 blocks with 8 bits
each; txcdata is an 8-bit string with only 0 and 1s. It indicates the block
type (data block or control block) for each 8-bit block.
• 0: Specifies that the corresponding input block is a data block.
• 1: Specifies that the corresponding input block is a control block.
For example: TXC=10000000 indicates that the first block in the input
pattern signal (bits 1 to 8) is a control block, and blocks 2, 3, 4, 5, 6, 7,
and 8 (that is, 9-16, 17-24, 25-32, 33-40, 41-48, 49-56, and 57-64 bits
respectively) are data blocks.

ENCODE=DW128b130b | Keyword to specify 128b/130b or 128b/132b encoding.

DW128b132b
Note:
DW128b130b and DW128b132b encoding are valid only for the PAT
transient_waveform type.

HEAD = DATA|CONTROL Indicates whether the pattern input signal is a data signal or control
signal. Default is DATA.
• DATA: Specifies that the 128-bit pattern input signal is a DATA signal.
• CONTROL: Specifies that the 128-bit pattern input signal is a
CONTROL signal.

TRANFORHB=[0|1] PrimeSim HSPICE advanced analog analyses

• 0 (default): Ignores the transient description if you specify an HB
value or a DC value. If you do not provide a DC or HB value and
TRANFORHB=0, then HB analysis treats the source as a DC source,
and the DC source value is the time=0 value.
• 1: HB analysis uses the transient description if its value is VMRF, SIN,
PULSE, PWL, or LFSR. If the type is a non-repeating PWL source,
the simulation treats the time=infinity value as a DC analysis source
value. For example, the simulation treats the following statement as a
DC source with value=1 for HB analysis:
v1 1 0 PWL (0 0 1n 1 1u 1) + TRANFORHB=1
In contrast, the following statement is a 0V DC source:
v1 1 0 PWL (0 0 1n 1 1u 1) + TRANFORHB=0
The simulation treats the following statement as a periodic source
with a 1u period that uses PWL values:
v1 1 0 PWL (0 0 1n 1 0.999u 1 1u 0) R + TRANFORHB=1
To override the global TRANFORHB option, explicitly set TRANFORHB for a
voltage or current source.

DCOPEN Switch for open DC connection if you have not set DC mag.
• 0 (default): P-element behaves as an impedance termination
• 1: P-element behaves as an open circuit in DC operating point
analysis. LIN analysis mainly uses DCOPEN=1 so the P-element
does not affect the self-biasing device under test by opening the
termination at the operating point.

RDC=val (DC analysis) Series resistance (overrides z0).

RAC=val (AC analysis) Series resistance (overrides z0).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 33

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Parameter Description

RHBAC=val (PrimeSim HSPICE advanced analog HBAC analysis) Series resistance

(overrides z0).

RHB=val (PrimeSim HSPICE advanced analog HB analysis) Series resistance

(overrides z0).

RTRAN=val (Transient analysis) Series resistance (overrides z0).

power=[0 | 1 | 2 | W | Power switch. Causes treatment of the signal amplitude quantities

dbm] specified with the port element to as power levels instead of voltage
levels. The port element then becomes a power source, realized as a
voltage source in series with the port impedance (Z0). The simulation
calculates the voltage source value internally as that is necessary to
realize the appropriate RMS available power. For transient analysis, this
is only appropriate for DC and SIN sources.
• When 0 (default), power entry disabled.
• When 1 or W, power given in units of Watts.
• When 2 or dBm, power given in units of dBm (dB relative to 1 mW).

Emphasis_Level Aids in .STATEYE analysis pre-emphasis and de-emphasis.

Emphasis_Time Aids in .STATEYE analysis pre-emphasis and de-emphasis.

DFE_tap Integer number(s) to specify relative feedback tap location.

DFE_coeff Real number(s) to specify corresponding feedback coefficient.

Note:
If DFE_tap is specified, DFE_coeff is required.

DFE_TAP_OPT Integer number to specify the number of DFE taps or coefficients.

If DFE_TAP_OPT is specified and the following analysis conditions are
met, the tool automatically extracts the optimized DFE coefficients and
applies them to the DFE operation:
• The STATEYE MODE keyword is set to EDGE or STAT.
• The P-element MODULATION=PAM4 keyword is not set.
Note:
• The tool ignores this keyword and prints a warning message in
the following situations:
◦ If you specify the number of DFE taps as less than or equal to
0
◦ If the analysis conditions are not met (see the preceding list)
• If DFE_TAP_OPT is specified, the tool ignores the DFE_tap and
DFE_coeff keywords.

DFE_vth A real number to specify logical threshold of the system. Default is 0.

DFE_CLK Works with CLOCK_PORT and CLOCK_MODE=2 keywords in .STATEYE

command. When DFE_CLK=1, StatEye passes clock ticks at CLOCK_PORT
to the built-in DFE model which will use the given clock ticks in slicer
operation.
The default is 0.
See also Applying External Clock to the Built-In DFE Model.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 34

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Parameter Description

Note:
For more information on DFE function, see Built-In DFE Function for Rx Port Element in StatEye
Analysis.

DCD Aids in .STATEYE analysis; specifies peak percentage of the duty cycle
distortion (DCD). Default is 0.

DCD_TYPE Aids in .STATEYE analysis; specifies variation type. Default type for
non-zero DCD is 1 (constant).
• 0: No DCD
• 1: Constant DCD
• 2: Uncorrelated triangular DCD variation
• 3: Uncorrelated sinusoidal DCD variation

PJ Aids in .STATEYE analysis; specifies periodic jitter (voltage) magnitude.

Default is 0.

PJ_TYPE Aids in .STATEYE analysis. Specifies an array of integer numbers

for periodic jitter variation type. Default type for nonzero PJ is 3
(uncorrelated sinusoidal voltage variation):
• 0: No periodic jitter
• 1: Constant voltage shift
• 2: Uncorrelated triangular voltage variation
• 3: Uncorrelated sinusoidal voltage variation
• 4: Constant time shift
• 5: Uncorrelated triangular time variation
• 6: Uncorrelated sinusoidal time variation
Note:
• Triangular assumes random change between 0 and specified
value with equal probability.
• Sinusoidal assumes the probability of the jitter amount has
sinusoidal distribution.
• Types 4, 5, and 6 are deterministic jitters.

RJ An array of the real numbers to specify the standard deviation of the

Gaussian random jitter. The array must be in the order of the port
element index. By default, the simulation does not add random jitter.

VN (Voltage noise) Aids in .STATEYE analysis. Specify the standard

deviation of the Gaussian random voltage noise. By default, the
simulation does not add random voltage noise.

AMI_OBJ Aids in .STATEYE analysis; specifies an Algorithmic Modeling Interface

(AMI) shared object (typically, lib*.so for UNIX, and *.dll for Windows)

AMI_PARAM Aids in .STATEYE analysis; specifies an AMI parameter file (*.ami)

AMI_CDR Aids in .STATEYE analysis. Switch to use AMI_GetWave's clock data

recovery (CDR) output in eye diagram generation. Default is 0 (disabled
state).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 35

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Parameter Description

AMI_CORNER Specifies an index number. The IBIS keyword "...(Corner val1 val2 ...)"
defines the corner values in the IBIS-AMI file specified by AMI_PARAM.
When you use AMI_CORNER, idx is used to choose the idxth value
defined by the "...(Corner ...)" keyword.
For example, if AMI_CORNER = 2 is specified in a port element and the
associated AMI parameter file has the following:
(My_AMI_param1 ... (Corner 1 2 3))
(My_AMI_param2 ... (Corner 4 5 6))
(My_AMI_param3 ... (Corner 7 8 9))
Then, 2, 5 and 8 are chosen for My_AMI_param{1,2 and 3} respectively.

MODULATION=PAM4 Activates pulse amplitude modulation (PAM) 4 signaling when the source
has bit pattern (LFSR/PAT) input. By default, the source element uses
the conventional two logic level signaling.

PAM4_MAPPING=string Specifies mapping from a pair of bits to four-level symbols.

PAM4_Mapping takes a four-character string argument, a combination
of '0' '1' '2' and '3' which represent pairs of binary symbols "00", "01",
"10", and "11" respectively. The first, second, third, and fourth character
positions in the argument string correspond to v0 (lowest), v1, v2, and
v3 (highest) voltage levels, respectively. For example, PAM4_MAPPING =
"0132" maps "00", "01", "11", and "10" to v0, v1, v2 and v3, respectively.
Default value for PAM4_MAPPING (when MODULATION=PAM4) is "0132".

IBIS_AMI_FILE=filename Arguments for direct IBIS (*.ibs) file call.

IBIS_AMI_MODEL=modeln PrimeSim HSPICE retrieves the following information based on the
ame specified pair of file and model names:
• AMI object and parameter file names for the appropriate computer
platform (Linux or Windows)
• Parasitic information such as C_comp
• Input source information such as source impedance, rise/fall time, and
voltage range
• AMI_CORNER=n argument which currently is used to specify the corner
index of the AMI parameter (*.ami) file. This corner index will also be
applied in retrieving these quantities from the IBIS (*.ibs) file.
A port element applies these parasitic and input source properties in
all the PrimeSim HSPICE simulations. AMI information is used in the
StatEye simulations.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 36

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Parameter Description

AMI_PARAM_xx = string | Argument for direct AMI parameter control from the PrimeSim HSPICE
quoted string netlist.
• PrimeSim HSPICE extracts "xx" as the target name, then assigns
the argument string when the name is found in the AMI parameter
(*.ami) file. For example, when AMI_PARAM_Ignore_Bits=1000 is
specified, PrimeSim HSPICE assigns 1000 to Ignore_Bits in the
AMI parameter file.
• When the argument value is quoted using single quotation character,
PrimeSim HSPICE first evaluates the string as an expression and
assigns the result to the AMI parameter file.
• You can have multiple declarations of this argument.
• Since the keywords in both the PrimeSim HSPICE netlist and the
AMI parameter file are case insensitive, parameter name matching
will be carried out in a case-insensitive manner. However, PrimeSim
HSPICE preserves the case of the given argument in the assignment.

IBIS_AMI_USE_CCOMP = Specifies whether or not to use Ccomp value in the *.ibs file for port
0|1 element’s load capacitance in IBIS-AMI simulation. Default is 1.

IBIS_AMI_CCOMP_ORDER = Selects the AMI_CORNER indexing scheme for C_comp when C_comp
1|2 value is selected up from *.ibs file for IBIS-AMI simulation.

IBIS_AMI_VRANGE_MODE = Specifies how to use the VRANGE value in the *.ibs file in IBIS-AMI
1|2 simulation.
• 1 (default): Uses VRANGE as differential voltage. The incident source
will have (-VRANGE to VRANGE) voltage swing.
• 2: Uses VRANGE as peak-to-peak voltage. The incident source will
have (-0.5*VRANGE to 0.5*VRANGE) voltage swing.

IBIS_AMI_USE_RAMP = 0|1 Specifies whether or not to use the RAMP information in the *.ibs file
for incident’s rise/fall time in IBIS-AMI simulation. Default is 1.

IBIS_AMI_USE_VRANGE = Specifies whether or not to use the VRANGE information in the *.ibs
0|1 file for incident’s voltage swing magnitude in IBIS-AMI simulation. Default
is 1.

IBIS_AMI_USE_ZSRC = 0|1 Specifies whether or not to use linearlized pull-up/pull-down information

in the *.ibs file for port element’s impedance in IBIS-AMI simulation.
Default is 1.

A full description of the port element and its syntax is found in Port Element in the
PrimeSim Continuum User Guide: Elements.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 37

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Using a Hierarchy Delimiter to Narrow the AMI Parameter

Assignment
When using IBIS-AMI models with AMI parameter file (*.ami), which contains multiple
parameters with the same leaf names under the different hierarchy such as:
(EQ1
(Mode (Usage In) (Type Integer) (List 2 0 1) (Default 2)
(Description "Mode for equalizer 1 : 0=Off, 1=Fixed, 2=Adapt")
)
)
(EQ2
(Mode (Usage In) (Type Integer) (List 2 0 1) (Default 2)
(Description "Mode for equalizer 2 : 0=Off, 1=Fixed, 2=Adapt")
)
)

The AMI_PARAM_xx keyword with parameter name “xx” assigns the value to all the
parameters with that name. For example,
P1 ..
+ AMI_PARAM_Mode = 1

results in the AMI parameter string as follows:

(EQ1(Mode 1))(EQ2(Mode 1))

To selectively assign the parameter value, use the hierarchy delimiter ‘/’ in the
AMI_PARAM_xx keyword so as to separately identify the same leaf name, Mode from
different hierarchy, EQ1 and EQ2. For example:
P1 ..
+ AMI_PARAM_EQ1/MODE = 1
results in
(EQ1(Mode 1))(EQ2(Mode 2))

P1 ..
+ AMI_PARAM_EQ2/MODE = 1
results in
(EQ1(Mode 2))(EQ2(Mode 1))

Note:
In this example, if no assignment is assigned from port element, then default
value 2 is used.

Using the P-element for Mixed-Mode Measurement

You can use a port element with three terminals as the port element for measuring the
mixed mode S-parameters. Except for the number of external terminals, the syntax of the

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 38

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

port element remains the same. The LIN analysis function internally sets the necessary
.LINport element and mixed mode measurement

drive mode (common/differential) of these mixed mode port elements. For analyses other
than the LIN analysis (such as DC, AC, TRAN, and so on), the mixed-mode P-element
acts as a differential driver that drives positive nodes with half of their specified voltage
and the negative nodes with a negated half of the specified voltage. Figure 2 shows the
block diagram of the mixed mode port element.

Figure 2 Mixed mode port element

P1 (port element)

n1+
V+
Z0

Z0
V-
n1-

n1_ref
P1 n1+ n1- n1_ref Z 0 =50

The port element can also be used as a signal source with a built-in reference impedance.
For further information on its use as a signal source, see Sources and Stimuli in the
PrimeSim Continuum User Guide: Elements.

.LIN Input Syntax

.LIN [sparcalc=1|0 [type=s|y] [modelname = modelname]]
+ [filename = filename]
+ [format=selem|citi|touchstone | touchstone2]
+ [noisecalc=1|0] [gdcalc=1|0]
+ [mixedmode2port=dd|dc|ds|cd|cc|cs|sd|sc|ss]
+ [dataformat=ri|ma|db] [FREQDIGIT=x] [SPARDIGIT=x]
+ [listfreq=(frequencies|none|all|freq1 freq2...)]
+ [listcount=num] [listfloor=val] [listsources=1|0|yes|no]
+ COMMENT_FILE={_}filename{_}]
+ [PORTS = P1[, P2 … Pn]]

The FREQDIGIT and SPARDIGIT keyword values allow you to extend the precision number
of digits for Touchstone 1.0/2.0, CITI, and .sc# files. The default is 6 for both keywords.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 39

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

For argument descriptions, see the .LIN command in the PrimeSim HSPICE Reference
Manual: Commands and Control Options.

.LIN Output Syntax

This section describes the syntax for the .PRINT and .PROBE statements used for LIN
.LINoutput syntax

analysis.
The following sections discuss these topics:
• .PRINT and .PROBE Statements
• Examples
• Hybrid Parameter Calculations

.PRINT and .PROBE Statements

.PRINT AC Xmn | Xmn(TYPE) | X(m,n) |X(m,n)(TYPE) |Hmn|
+ Hmn(TYPE) |H(m,n) | H(m,n)(TYPE) | LINPARAM |
+ LINPARAM(TYPE)
.PROBE AC Xmn | Xmn(TYPE) | X(m,n) | X(m,n)(TYPE) | Hmn |
+ Hmn(TYPE) |H(m,n) | H(m,n)(TYPE) | LINPARAM |
+ LINPARAM(TYPE)

Argument Description

Xmn, X(m,n) One of these parameter types:

• S (scattering parameters)
• Y (admittance parameters)
• Z (impedance parameters)
mn refers to a pair of port numbers.

Hmn, H(m,n) Complex hybrid (H-) parameters. mn refers to a pair of port numbers, where m and n
are positive integer numbers. If m,n=0 or m,n>2, PrimeSim HSPICE issues a warning
and ignores the output request.
• To calculate a one-port H parameter, you must specify at least one port (P) element.
• To calculate a two-port H parameter, you must specify two or more port (P)
elements.
For additional information, see Hybrid Parameter Calculations.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 40

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Argument Description

LINPARAM Two-port noise parameters:

noise parametersparameterstwo-port noise

• NFMIN (noise figure minimum)

• NF (Noise figure)
• VN2 (Equivalent input noise voltage squared
• IN2 (Equivalent input noise current squared)
• RHON (Correlation coefficient between input noise voltage and input noise current)
• RN (Noise equivalent resistance)
• GN (Noise equivalent conductance)
• ZCOR (Noise correlation impedance)
• YCOR (Noise correlation admittance)
• ZOPT (Optimum source impedance for noise)
• YOPT (Optimum source admittance for noise)
• GAMMA_OPT (source reflection coefficient that achieves the minimum noise figure)
• ZOPT (source impedance that achieves minimum noise)
• RN (noise equivalent resistance)
• K_STABILITY_FACTOR (Rollett stability factor)
• MU_STABILITY_FACTOR (Edwards & Sinsky stability factor)
• G_MAX (maximum available/operating power gain)
• G_MSG (Maximum stable gain)
• G_TUMAX (Maximum unilateral transducer power gain)
• G_U (Unilateral power gain)
• G_MAX_GAMMA1 (source reflection coefficient that achieves maximum available
power gain)
• G_MAX_GAMMA2 (load reflection coefficient that achieves maximum operating power
gain)
• G_MAX_Z1=Source impedance needed to realize G_MAX (complex, Ohms)
• G_MAX_Z2=Load impedance needed to realize G_MAX (complex, Ohms)
• G_MAX_Y1=Source admittance needed to realize G_MAX (complex, Siemens)
• G_MAX_Y2=Load admittance needed to realize G_MAX (complex, Siemens)
• G_AS (associate gain—maximum gain at the minimum noise figure)
• VSWR(n) (voltage standing-wave ratio at the n port)
• G_MSG (maximum stable gain)
• G_TUMAX (maximum unilateral transducer power gain)
• G_U (unilateral power gain)

TYPE Data type definitions:

definitionsdata typesdata type definitions

• R=Real
• I=Imaginary
• M=Magnitude
• P=PD=Phase in degrees
• PR=Phase in radians
• DB=decibels

Examples
.print AC S11 S21(DB) S(2,3)(D) S(2,1)(I)
.print AC NFMIN GAMMA_OPT G_AS
.probe AC RN G_MAX ZOPT Y(3,1)(M) Y31(P)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 41

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Hybrid Parameter Calculations

The hybrid parameters are transformed from S-parameters:
hybrid parameter calculations
• For a one-port circuit, the calculation is:

Equation 1

• For a two-port circuit, the calculation is:

Equation 2

Equation 3

Equation 4

Equation 5

For networks with more than two ports when computing the 1,2 H index parameters,
PrimeSim HSPICE assumes that ports numbered 3 and above terminate in their port
reference impedance (z0). The above two-port calculations therefore remain appropriate
because S11, S12, S21, and S22 remain valid, and simulation can ignore higher order S-
parameters.

Multi-Port Scattering (S) Parameters

S-parameters represent the ratio of incident and scattered (or forward and reflected)
normalized voltage waves. Figure 3 shows a two-port network.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 42

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Figure 3 Two-port network

I1 I2

+ Two-Port +
Port=1 Z01 V1 Network V2 Z02 Port=2
- -

The following equations define the incident (forward) waves for this two-port network:

The following equations define the scattered (reflected) waves for this two-port network:

The following equations define the S-parameters:

Each S-parameter is a complex number, which can represent gain, isolation, or a reflection
coefficient.
Example
The following examples show how you can represent a gain, isolation, or reflection
coefficient:
.PRINT AC S11(DB) $ Input return loss
.PRINT AC S21(DB) $ Gain
.PRINT AC S12(DB) $ Isolation
.PRINT AC S22(DB) $ Output return loss

Two-Port Transfer and Noise Calculations

Two-port noise analysis is a linear AC noise analysis method that determines the noise
figure of a linear two-port for an arbitrary source impedance.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 43

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Several output parameter measurements are specific to two-port networks. The LIN
analysis supports two-port calculations for 3 or more ports if port=1 is the input and port=2
the output. All other ports terminate in their characteristic impedance. This is equivalent
to operating on the two-port [S] submatrix extracted from the multi-port [S] matrix. This
occurs for both signal and noise calculations. A warning appears if N<2 and you specified
two-port quantities.
Noise and signal port-wise calculations do not require that port elements use a ground
reference. You can therefore measure fully-differential circuits.
.LIN command generates a set of noise parameters. The analysis assumes a noise model
consisting of:
• A shunt current noise source, called In, at the input of a noiseless two-port linear
network
• A series voltage noise source, called Vn, at the input of a noiseless two-port linear
network
• A source with impedance, called Zs, that drives this two-port network
• The two-port network drives a noiseless load, called Zl
The following sections discuss these topics:
• Equivalent Input Noise Voltage and Current
• Equivalent Noise Resistance and Conductance
• Noise Correlation Impedance and Admittance
• Optimum Matching for Noise
• Noise Figure and Minimum Noise Figure
• Associated Gain
• Output Format for Group Delay in .sc* Files
• Output Format for Two-Port Noise Parameters in *.sc# Files

Equivalent Input Noise Voltage and Current

For each analysis frequency, PrimeSim HSPICE computes a noise equivalent circuit for
a linear two-port. The noise equivalent circuit calculation results in an equivalent noise
voltage and current, and their correlation coefficient.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 44

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

2
• VN2: Equivalent input noise voltage squared (Real, V )
2
• IN2: Equivalent input noise current squared (Real, A )
• RHON: Correlation coefficient between the input noise voltage and the input noise
current (complex, unitless)

Equivalent Noise Resistance and Conductance

These measurements are the equivalent resistance and conductance, which generate
the equivalent noise voltage and current values at a temperature of T=290K in a 1Hz
bandwidth.
• RN: Noise equivalent resistance (Real, Ohms)
• GN: Noise equivalent conductance (Real, Siemens)

Noise Correlation Impedance and Admittance

These measurements represent the equivalent impedance and admittance that you can
insert at the input of the noise equivalent circuit to account for the correlation between the
equivalent noise voltage and the current values.
• ZCOR: Noise correlation impedance (Complex, Ohms)
• YCOR: Noise correlation admittance (Complex, Siemens)

Optimum Matching for Noise

These measurements represent the optimum impedance, admittance, and reflection
coefficient value that result in the best noise performance (minimum noise figure).
• ZOPT: Optimum source impedance for noise (Complex, Ohms)
• YOPT: Optimum source admittance for noise (Complex, Siemens)
• GAMMA_OPT: Optimum source reflection coefficient (Complex, unitless)
Because ZOPT and YOPT can commonly take on infinite values when computing optimum
noise conditions, calculations for optimum noise loading reflect the GAMMA_OPT coefficient.

Noise Figure and Minimum Noise Figure

Noise figure represents the ratio of the SNR (signal to noise ratio) at the input to the SNR
at the output. You can set the input source impedance to ZOPT to obtain the minimum
noise figure.
• NFMIN: Minimum noise figure (source at ZOPT) (real, unitless, power ratio)
• NF: Noise figure (value obtained with source impedance at Zc[1]) (real, unitless, power
ratio)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 45

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Associated Gain
This measurement assumes that the input impedance matches the minimum noise figure,
and the output matches the maximum gain.
G_AS is the associated gain—maximum power gain at NFMIN (real, power ratio)

Output Format for Group Delay in .sc* Files

All of the S/Y/Z/H parameters support a group delay calculation. The output syntax
of .PRINT and .PROBE statements for group delay is:

Xmn(T) | Xmn(TD) | X(m,n)(T) | X(m,n)(TD)

• X=S, Y, Z, or H
• m, n=port number (1 or 2 for H parameter)
The output of group delay matrices in .sc* files lets PrimeSim HSPICE directly read back
the group delay information. The tabulated data uses the regular PrimeSim HSPICE model
syntax with the SP keyword:
*| group delay parameters
.MODEL SMODEL_GD SP N=2 SPACING=POI INTERPOLATION=LINEAR
+ MATRIX=NONSYMMETRIC VALTYPE=REAL
+ DATA=3
+ 1e+08
+ 0 5e-09
+ 5e-09 0
+ {...data...}

model name is the model name of the S-parameters, plus _GD.

GROUPDELAY=[0|1] in the top line indicates group delay data:

*| N=2 DATA=3 NOISE=0 GROUPDELAY=1
*| NumOfBlock=1 NumOfParam=0

Output Format for Two-Port Noise Parameters in *.sc# Files

Output of two-port noise parameter data in *.sc# files shows the tabulated data with the
following quantities in the following order:
*| 2-port noise parameters
*| frequency Fmin[dB] GammaOpt(M) GammaOpt(P) RN/Z0
*| {...data...}

In this syntax:
• Fmin[dB]=minimum noise figure (dB)
• GammaOpt(M)=magnitude of the reflection coefficient needed to realize Fmin

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 46

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

• GammaOpt(P)=phase (degrees) of the reflection coefficient needed to realize Fmin

• RN/Z0=normalized noise resistance
Both GammaOpt and RN/Z0 values are normalized with respect to the characteristic
impedance of the port=1 element (that is, Z01).

Noise Parameters in 2-Port and N-Port Networks

You can use the LIN analysis to compute the equivalent two-port noise parameters
.LINnoise parameters

or multi-port noise parameters for a network. The noisecalc=1 option automatically

calculates the following equivalent circuit values.

Figure 4 Noise equivalent circuit

Vn
+-

In Two-Port
Port=1 Network Port=2

• Vn is the equivalent input-referred noise voltage source.

• In is the equivalent input-referred noise current source.
• InVn is their correlation.
The noisecalc=1 option activates N-port noise analysis. Invoking an N-port noise
analysis produces an *.nc# file to output the N-port noise correlation matrix in the
following format, which is similar to a *.sc# file.
*| N=2 DATA=1 NOISE=0 GROUPDELAY=0 COMPLEX_DATAFORMAT=RI
*| NumOfBlock=2 NumOfParam=1 ParamNames=res
*| ParamSweep res:50 100
*|-------------------------------------
*| res=50
.MODEL NFQMODEL SP N=2 SPACING=POI INTERPOLATION=LINEAR
MATRIX=NONSYMMETRIC
+ DATA=1
+ 100
+ 0.457141 0 -0.457141 0
+ -0.457141 0 0.457141 0
*|-------------------------------------
*| res=100
.MODEL NFQMODEL1 SP N=2 SPACING=POI INTERPOLATION=LINEAR
MATRIX=NONSYMMETRIC

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 47

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

+ DATA=1
+ 100
+ 0.514284 0 -0.514284 0
+ -0.514284 0 0.514284 0

The noise analysis data is output in the .lis and .ac* files in the following format:
Frequency = 100.000 Hz
NF (mag) = 1.0+1.02857
NF (dB) = 3.07189
Element Contribution to NF (mag) (dB)
r1 TOT 1.02857 122.327m

All the 2-port noise parameters are deduced and output in the .lis file with
.PRINT/.PROBE commands.

PrimeSim HSPICE can output the result of .LIN noise analysis to a *.sc, Touchstone
1.0/2.0, or CITI file. Note this limitation for Touchstone files: because Touchstone files
currently provide only two-port noise parameters, this type of noise model only supports
two-port noise analysis for both passive and active systems.

Hybrid (H) Parameters

LIN analysis can calculate the complex two-port H (hybrid) parameter of a multi-terminal
.LINhybrid parameters

network.
The H parameters of a two-port network relate the voltages and currents at input and
output ports:

In the preceding equations:

• Hybrid matrix

• V1=Voltage at input port

• I2=Current at output port
• V2=Voltage at output port
• I1=Current at input port
You can add the hybrid H parameter matrixes of two networks to describe networks that
are in series at their input and in parallel at their output.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 48

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

.LIN command calculates H parameters based on the scattering parameters of the

networks. LIN analysis can extract one-port and two-port network H parameters. For
networks with more than two ports, .LIN command assumes that the ports numbered 3
and above terminate in their port characteristic impedance (Zc[i], i>2).

Group Delay
Group delay measures the transit time of a signal through a network versus frequency.
It reduces the linear portion of the phase response to a constant value, and transforms
the deviations from linear phase into deviations from constant group delay (which causes
phase distortion in communications systems). The average delay represents the average
signal transit time through a network system.
PrimeSim HSPICE can output the result of .LIN group delay measurement to a .sc*,
Touchstone 1.0/2.0, or CITIfile.
Group delay is a function of frequency:

where:
• gd = Group delay at the f frequency,
• phase = phase response at the f frequency
• w = radians frequency
All complex S, Y, Z, and H parameters support a group delay calculation.
Syntax

Xmn(T) | Xmn(TD) | X(m,n)(T) | X(m,n)(TD)

X=S, Y, Z, or H (parameters)
m,n=port index (1 or 2 for H-parameters)

The results of the group delay calculation are scalar real numbers in units of seconds.
For .LIN, group delay values are a function of frequency. The calculation is:

Differentiating the complex logarithm with respect to omega results in:

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 49

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

The group delay is the negative derivative of the phase. Simulation can compute it from
the imaginary component of the derivative w.r.t. frequency of the measurement:

Summary of Printing and Probing Network Parameters

When used in combination with P (port) elements and the .AC command, .LIN makes
available network parameters such as noise, S, H and Y/Z. You can save these network
parameters values in a plot file with .PROBE and view them with a waveform tool, or use
.PRINT to record the values in your listing file in a tabular format.

The following sections discuss these topics:

• Specifying the Analysis Type
• Port Reference
• Port Numbers with More Than One Digit
• Network Parameter Data Types
• Mixed-Mode Ports

Specifying the Analysis Type

Since you may have more than one analysis in your netlist (AC/DC/TRAN), when using
.PRINT or .PROBE you have the option to specify which analysis to wish to reference. If
the only analysis you are performing is the .AC combined with .LIN to extract network
parameters, you can omit this and therefore the following are equivalent:
.PROBE AC S21(DB) S12(DB)
.PROBE S21(DB) S12(DB)

Port Reference
The numbers in the port reference (e.g., S21, S12 above) refer to the PORT= number in
your netlist. In this single-ended example, P_IN is port 1 and P_OUT is port 2.
P_IN IN 0 Z0=50 PORT=1 AC=1
P_OUT OUT 0 Z0=50 PORT=2

So, in the print command:

.PRINT AC S21(DB)

You are printing the insertion loss (or gain) observed at port 2 from port 1.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 50

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
LIN Analysis

Port Numbers with More Than One Digit

PrimeSim HSPICE can understand S12, S34, S91, etc., but port numbers with more than
one digit (e.g., > 9) require putting parenthesis around the port reference and separating
them with a comma. For example:
.PROBE S(12,2)(DB)

Network Parameter Data Types

The previous examples reference the DB data type. If you omit the data type argument,
the default is actually magnitude. For example:
.PRINT S11

is equivalent to:
.PRINT S11(M)

Both will return the magnitude of the return loss. These are the available data type
definitions:
• R = Real
• I = Imaginary
• M = Magnitude (default)
• P=PD = Phase in degrees
• PR = Phase in radians
• DB = decibels

Mixed-Mode Ports
You can also specify whether to reference the common, mixed or cross-mode network
parameters in your .PRINT or .PROBE statements. By default, PrimeSim HSPICE
determines the mode of the network parameters from the port configuration. For example,
in this mixed mode (differential) port configuration:
P_IN IN_P IN_N 0 Z0=50 PORT=1 AC=1
P_OUT OUT_P OUT_N 0 Z0=50 PORT=2

The following statements will return the same values:

.PROBE S21(DB)
.PROBE SDD21(DB)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 51

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
SNLIN Analysis

But you can explicitly specify the mode of the network parameters to reference as follows:
• SDD = differential mode S parameters
• SCC = common mode S parameters
• SDC or SCD = cross mode S parameters

SNLIN Analysis
The .SNLIN command invokes the SNLIN analysis. This analysis is used for:
• Frequency translation scattering parameter extraction for a periodically-driven circuits
that exhibits frequency translation effects, such as mixers.
• Noise parameter extraction for a periodically-driven circuits that exhibits frequency
translation effects.
Frequency translate S parameter is used to describe N-port circuits that exhibits frequency
translation, such as mixers. It is a small-signal analysis like .LIN, except that the circuit
is first linearized about a periodically varying operating point, instead of a simple DC
operating point. In this manner, it is possible to compute S parameters between circuit
ports that convert signals from one frequency band to another. Frequency translate S
parameter describes the capability of a periodically linear time varying systems to shift
signals in frequency. Similar to the S parameters of a linear-time-varying system, it is
defined as below:

The incident and reflected waves are defined as:

Parameter Description

Tone, refers to to the fundamental frequency

n Harmonic index

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 52

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
SNLIN Analysis

Parameter Description

i Port index

The input wave at the frequency

for the i'th port.

The reflected wave at the frequency

for the i'th port.

The fourier coefficient at the frequency

, of the voltage at port i.

The fourier coefficient at the frequency

, of the voltage at port i.

The reference impedance at port i.

th th
S[i,j] S parameter from the j port to the i port.

Multitone Analysis
For a multitone analysis, the calculation for multitone a, b vector definition is:

SNLIN Input Syntax

.SNLIN
+ [freq_sweep]
+ [NOISECALC=1|0|on|off]
+ [FILENAME=file_name]
+ [DATAFORMAT=ri|ma|db|]
+ [MIXEDMODE2PORT=dd|cc|cd|dc|sd|sc|cs|ds]]

For argument descriptions, see the .LIN command in the PrimeSim Continuum Reference
Manual: Commands and Control Options.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 53

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
SNLIN Analysis

SNLIN Output Syntax

This section describes the syntax for the .PRINT and .PROBE statements used for SNLIN
.LINoutput syntax

analysis.
.SNLIN outputs to the *.sl0 and *.printsl0 files.

The following sections discuss these topics:

• .PRINT and .PROBE Statements
• Output to Data Files
• Examples

.PRINT and .PROBE Statements

.print SNLIN Smn|Smn(TYPE) | S(m,n) |S(m,n)(TYPE)|
.probe SNLIN Smn | Smn(TYPE) | S(m,n) | S(m,n)(TYPE) |
.print SNLIN SXYmn|SXYmn(TYPE) | SXY(m,n) |SXY(m,n)(TYPE)|
.probe SNLIN SXYmn|SXYmn(TYPE) | SXY(m,n) |SXY(m,n)(TYPE)|
.print SNLIN NF | SSNF | DSNF
.probe SNLIN NF | SSNF | DSNF

Argument Description

m and n m and n refers to port number index. Both of them must be with in the range 0 < m, n <
port_number.

X and Y X and Y are used for mixed mode S parameter output. It can be any one of the
following:
• D - differential
• C - common
• S - single end

TYPE • R = Real
• I = Imaginary
• M = Magnitude
• P = PD = Phase in degrees
• D = DB = Decibels
• DBM = Decibels per milli

NF and SSNF Refers to the single side band noise figure as a function of the input frequency band
points.

DSNF Refers to the double side band noise figure as a function of the input frequency band
points.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 54

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
SNLIN Analysis

Output to Data Files

The extracted S parameters can also be the output to data files. The file format can only
be the Touchstone file. For the Touchstone file output, the frequency translation flag,
and the port harmonic indices must be mentioned in the header to declare the translated
frequency band. The following is an example.
! NPort=2 DATA=4 NOISE=0 GROUPDELAY=0
! NumOfBlock=1 NumOfParam=0 FrequencyTranslate=1
!
! PORT1 = [p1_h1 p1_h2 ... p1_hNn ... +/-1]
!
! PORT2 = [p2_h1 p2_h2 ... p2_hNn ... +/-1]
! ...
! PORTm = [pm_h1 pm_h2 ... pm_hNn ... +/-1]
! SS_TONE = i

# TONES = [F1 F2 ... Fn]

# Hz S MA R 50.0000
.......

Frequency translates S parameter data with this port harmonic combination.

# TONES = [F1 F2 ... Fn] ! exist only when large signal tone sweep
# Hz S MA R 50.0000
.....

Frequency translate S parameter data with this harmonic combination.

If mixed mode S parameters are extracted, the mixed-mode information is be specified in
the file header at the very beginning, such as:
! NPort=2 DATA=1000 NOISE=0 GROUPDELAY=0
! NumOfBlock=1 NumOfParam=0 FrequencyTranslate=1
! MIXEDMODE=1 DATATYPE=D1D2C1C2
!
! S(1,1) = SDD(1,1)
! S(1,2) = SDD(1,2)
! S(1,3) = SDC(1,1)
! S(1,4) = SDC(1,2)
! S(2,1) = SDD(2,1)
! S(2,2) = SDD(2,2)
! S(2,3) = SDC(2,1)
! S(2,4) = SDC(2,2)
! S(3,1) = SCD(1,1)
! S(3,2) = SCD(1,2)
! S(3,3) = SCC(1,1)
! S(3,4) = SCC(1,2)
! S(4,1) = SCD(2,1)
! S(4,2) = SCD(2,2)
! S(4,3) = SCC(2,1)
! S(4,4) = SCC(2,2)
!

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 55

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
SNLIN Analysis

! PORT1 = [p1_h1 p1_h2 ... p1_hNn +/-1]

! PORT2 = [p2_h1 p2_h2 ... p2_hNn +/-1]
! ...
! PORTm = [pm_h1 pm_h2 ... pm_hNn +/-1]
! SS_TONE = i

# TONES = [F1 F2 ... Fn]

# Hz S MA R 50.0000

Frequency translate S parameter data with this harmonic combination.

# TONES = [F1 F2 ... Fn] ! exist only when large signal
tone sweep
# Hz S MA R 50.0000

Frequency translate S parameter data with this harmonic combination.

Examples
This example does single-tone analysis, with frequency translation:
• Extract two-port S parameters transferred from RF (1G-del_f) to IF (del_f). The LO
signal is specified by normal voltage source Vlo.
• The frequency on port 1 is in RF band:1G-del_f, the frequency on port 2 is in IF band:
del_f; del_f is swept from 0 to 100meg.

• The results are output to the ex1.s2p file.

p1 RFin gnd port = 1 SNLIN =(1, -1)
p2 IFout gnd port = 2 SNLIN=(0,1)
Vlo Loin gnd 1.0 cos(1.0 0.5 1g)
.SN tones=1G nharm=5

.SNLIN
+ lin 5 0 100meg
+ noisecalc = off
+ filename = ex1
+ dataformat = ma

The following example is for singletone analysis, with frequency translation:

• Extract two-port S parameters transferred from RF (1G-del_f) to IF (del_f). The LO
signal is specified by normal voltage source Vlo.
• The frequency component is picked at port 1 is del_f; the frequency component
picked at port 2 is 1GHz*2-del_f; the frequency component picked at port 3 is 1GHz
+del_f; del_f is swept from 0 to 100meg.
• SNNOISE calculation is required.
• The results are output to the ex2.s3p file.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 56

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Additional Measurements From .LIN Command

p1 1 0 port=1 SNLIN=(0, 1)
p2 2 0 port=2 SNLIN=(2,-1)
p3 3 0 port=3 SNLIN=(1,1)

.PARAM smfreq=100meg
.SN tones=1G nharms=2

.SNLIN
+ lin 5 0 100meg
+ noisecalc=on
+ filename=ex2

Additional Measurements From .LIN Command

In addition to S, Y, Z, and H parameters, a LIN analysis can include other output
measurements discussed in the following sections:
• Impedance Characterizations
• Stability Measurements
• Gain Measurements
• Matching for Optimal Gain

Impedance Characterizations
• VSWR(i) = Voltage standing wave ratio at port i (real, unit-less)
• ZIN(i) = Input impedance at port i (complex, Ohms)
• YIN(i) = Input admittance at port i (complex, Siemens)

Stability Measurements
• K_STABILITY_FACTOR = Rollett stability factor (real, unit-less)
• MU_STABILITY_FACTOR = Edwards & Sinsky stability factor (real, unit-less)

Gain Measurements
• G_MAX = Maximum available/operating power gain (real, power ratio)
• G_MSG = Maximum stable gain (real, power ratio)
• G_TUMAX = Maximum unilateral transducer power gain (real, power ratio)
• G_U = Unilateral power gain (real, power ratio)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 57

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Additional Measurements From .LIN Command

Matching for Optimal Gain

• G_MAX_GAMMA1 = Source reflection coefficient needed to realize G_MAX (complex, unit-
less)
• G_MAX_GAMMA2 = Load reflection coefficient needed to realize G_MAX (complex, unit-
less)
• G_MAX_Z1 = Source impedance needed to realize G_MAX (complex, Ohms)
• G_MAX_Z2 = Load impedance needed to realize G_MAX (complex, Ohms)
• G_MAX_Y1 = Source admittance needed to realize G_MAX (complex, Siemens)
• G_MAX_Y2 = Load admittance needed to realize G_MAX (complex, Siemens)
The following sections discuss these topics:
• VSWR
• ZIN(i)
• YIN(i)
• K_STABILITY_FACTOR (Rollett Stability Factor)
• MU_STABILITY_FACTOR (Edwards-Sinsky Stability Factor)
• Maximum Available Power Gain - G_MAX
• Maximum Stable Gain - G_MSG
• Maximum Unilateral Transducer Power Gain - G_TUMAX
• Unilateral Power Gain - GU
• Simultaneous Conjugate Match for G_MAX
• Equivalent Input Noise Voltage and Current - IN2, VN2, RHON
• Equivalent Noise Resistance and Conductance - RN, GN
• Noise Correlation Impedance and Admittance - ZCOR, YCOR
• Optimum Matching for Noise - ZOPT, YOPT, GAMMA_OPT
• Noise Figure and Noise Figure Minimum - NF, NFMIN
• Associated Gain - G_As

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 58

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Additional Measurements From .LIN Command

VSWR
The Voltage Standing Wave Ratio represents the ratio of maximum to minimum voltages
along a standing wave pattern due to a port’s impedance mismatch. All ports other than
the port of interest terminate in their characteristic impedances. VSWR is a real number
related to that port’s scattering parameter:

ZIN(i)
The Input Impedance at the i port is the complex impedance into a port with all other
ports terminated in their appropriate characteristic impedance. It is related to that port’s
scattering parameter:

YIN(i)
The Input Admittance at the i port is the complex admittance into a port with all other
ports terminated in their appropriate characteristic impedance. It is related to that port’s
scattering parameter:

K_STABILITY_FACTOR (Rollett Stability Factor)

The Rollett stability factor is:

Δ determines the two-port S matrix calculated from this equation:

An amplifier where K>1 is unconditionally stable at the selected frequency.

MU_STABILITY_FACTOR (Edwards-Sinsky Stability Factor)

The following equation defines the Edwards-Sinsky stability factor.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 59

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Additional Measurements From .LIN Command

An amplifier with μ>1 is considered unconditionally stable at the specified frequency.

Maximum Available Power Gain - G_MAX

This is the gain value that can be realized if the two-port is simultaneously conjugate-
matched at both input and output (with no additional feedback):

K is the Rollett stability factor. Special cases of G_MAX are handled in the following manner:
2
• If |S12|=0 and (|S11|=1 or |S22|=1), G_MAX=|S21|
• If |S12|=0 and |S11|≠1 and |S22|≠1, G_MAX=G_TUMAX
• If |S12|≠0 and K≤ 1, G_MAX=G_MSG
When values for K≤ 1, the Maximum Available Power Gain is undefined, and PrimeSim
HSPICE returns the Maximum Stable Gain.

Maximum Stable Gain - G_MSG

For a two-port that is conditionally stable (K<1), the following equation calculates the
maximum stable gain:

To achieve this gain, resistively load the unstable two-port so that K=1, and then
simultaneously conjugately match the input and output ports. G_MSG is therefore equivalent
to G_MAX with K=1. In terms of admittance parameters:

MSG is equivalent to the Maximum Available Power Gain if K=1.

Maximum Unilateral Transducer Power Gain - G_TUMAX

This is the highest possible gain that a two-port with no feedback (that is, S12=0) can
achieve.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 60

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Additional Measurements From .LIN Command

Unilateral Power Gain - GU

This is the highest gain that the active two-port can ever achieve by embedding in a
matching network that includes feedback. The frequency where the unilateral gain
becomes unity defines the boundary between an active and a passive circuit. The
frequency is usually referred to as fmax, the maximum frequency of oscillation.
To realize this gain, PrimeSim HSPICE neutralizes the feedback of the two-port, and
simultaneously conjugate-matches both input and output:

Simultaneous Conjugate Match for G_MAX

A simultaneous conjugate match is required at the source and load to realize the Gmax
gain value. The source reflection coefficient at the input must be:

The load reflection coefficient (G_MAX_GAMMA_2) is:

In the preceding equation:

You can obtain useful solutions only when:

These equations also imply that K>1.

PrimeSim HSPICE derives calculations for the related impedances and admittances from
the preceding values.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 61

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Additional Measurements From .LIN Command

For G_MAX_Z1:

For G_MAX_Z2:

For G_MAX_Y1:

For G_MAX_Y2:

Equivalent Input Noise Voltage and Current - IN2, VN2, RHON

For each analysis frequency, PrimeSim HSPICE computes a noise-equivalent circuit
for a linear two-port. The noise analysis assumes that all ports terminate in noise-
less resistances. For circuits with more than two ports, ports identified as 3 and above
terminate, and the analysis considers only ports 1 and 2. The noise-equivalent circuit
calculation results in an equivalent noise voltage and current, and their correlation
coefficient. These values are:

Equivalent Noise Resistance and Conductance - RN, GN

These measurements are the equivalent resistance and conductance that would generate
the equivalent noise voltage and current values at a temperature of in a 1 Hz
bandwidth (that is, ).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 62

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Additional Measurements From .LIN Command

Noise Correlation Impedance and Admittance - ZCOR, YCOR

These measurements represent the equivalent impedance and admittance that you can
insert at the input of the noise-equivalent circuit to account for the correlation between the
equivalent noise voltage and the current values.

Optimum Matching for Noise - ZOPT, YOPT, GAMMA_OPT

The equivalent input noise sources and their correlation make it possible to compute the
impedance, admittance, and reflection coefficient values that, if presented at the input of
the noisy two-port, result in the best noise performance. These values are:

Noise Figure and Noise Figure Minimum - NF, NFMIN

If you set the input source impedance to ZOPT, the two-port operates with the minimum
Noise Figure. The definition of the Noise Figure (F) is unusual because it involves the
available gain of the two-port, not its transducer gain. You can express it in the following
form:

• Ga is the available power gain.

• Na is the available noise power at the output of the two-port (due solely to the two-
port’s noise and not to the input impedance).
• k is Boltzmann’s constant.
• T0 is the 290 Kelvin reference temperature.
The NMIN minimum noise figure value is computed as:

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 63

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Example - Extracting Bandpass Filter S-parameters

where NFMIN≥1. For input source impedance values other than ZOPT, the Noise Figure
varies as a function of the input source reflection coefficient, according to:

The PrimeSim HSPICE Noise Figure measurement (NF) returns the noise figure value if
the input terminates in the port characteristic impedance (that is, ). This value is:

Associated Gain - G_As

PrimeSim HSPICE also includes a measurement named Associated Gain, which assumes
that the in-out impedance is matched for the minimum noise figure (that is,
), while the output is matched for the maximum gain.

In the preceding equation:

Example - Extracting Bandpass Filter S-parameters

This example uses linear analysis to extract the S-parameters in the Touchstone 1.0/2.0
format for a bandpass filter. The input impedance of the filter is also measured. The circuit
for this example is fbplin.sp, which is located in the directory $installdir/demo/
hspice/sparams.
*band pass filter
c1 in 2 3.166pf
l1 2 3 203nh
c2 3 0 3.76pf
c3 3 4 1.75pf
c4 4 0 9.1pf
l2 4 0 36.81nh
c5 4 5 1.07pf
c6 5 0 3.13pf
l3 5 6 233.17nh
c7 6 7 5.92pf
c8 7 0 4.51pf
c9 7 8 1.568pf
c10 8 0 8.866pf
l4 8 0 35.71nh

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 64

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Example - Extracting Bandpass Filter S-parameters

c11 8 9 2.06pf
c12 9 0 4.3pf
l5 9 10 200.97nh
c13 10 out 2.97pf

To extract the S-parameters, port elements are inserted at the input and output of the filter.
p1 in 0 port=1
p2 out 0 port=2

Set up the linear analysis to extract the S-parameters of the filter. The center frequency of
the filter is 250 MHz.
.ac lin 250 200meg 300meg
.lin format=touchstone

Probe the magnitude and phase of the input return loss (S11), the insertion loss (S21) and
the input impedance (zin).
.probe ac s11(db) s11(p) s21(db) s21(p)
.probe ac zin(1)(m) zin(1)(p)

The frequency in the passband where the input impedance is 50 ohm and the phase in the
passband where the input impedance is 50 ohm can be measured.
.meas ac cross50 when zin(1)=50 td=230meg
.meas ac phase50 find zin(1)(p) when zin(1)(m)=50 td=230meg

Results can be seen in Figure 5 and Figure 6.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 65

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Example - Extracting Bandpass Filter S-parameters

Figure 5 Input return loss magnitude and phase

Figure 6 Input impedance magnitude and phase

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 66

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Extracting Mixed-Mode Scattering (S) Parameters

Extracting Mixed-Mode Scattering (S) Parameters

The LIN analysis includes a keyword for extracting mixed-mode scattering (S) parameters.

Syntax
.LIN … [ mixedmode2port= dd|dc|ds|cd|cc|cs|sd|sc|ss ]

The following keywords in the .PRINT and .PROBE statements specify the elements in the
mixed mode S-parameter matrices:

Sxy|Yxy|Zxy nm (t)

Argument Description

x,y One of the following:

• D (differential)
• C (common)
• S (single-ended)
For example:
• SCC = Common mode S-parameters
• SDC or SCD = Cross mode S-parameters
If you omit x,y, then PrimeSim HSPICE uses the value set for the mixedmode2port.

Scc Common-mode S-parameters

Scd and Sdc Mode-conversion or cross-mode S-parameters

m, n port number

type One of the following:

• DB: Magnitude in decibels
• I: Imaginary part
• M: Magnitude (default)
• P: Phase in degree
• R: Real part

The following sections discuss these topics:

• Defaults
• Output File Formats
• Two-Port Parameter Measurement
• Output Format and Description
• Features Supported

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 67

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Extracting Mixed-Mode Scattering (S) Parameters

• Prerequisites and Limitations

• Reported Statistics for the Performance Log
• Errors and Warnings

Defaults
Availability and default value for the mixedmode2port keyword depends on the port
configuration.

Example 1
p1=p2=single

where:
• Available: ss
• Default: ss

Example 2
p1=p2=balanced

where:
• Available: dd,cd,dc,cc
• Default: dd

Example 3
p1=balanced p2=single

where:
• Available: ds,cs
• Default: ds

Example 4
p1=single p2=balanced

where:
• Available: sd,sc
• Default: sd

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 68

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Extracting Mixed-Mode Scattering (S) Parameters

Output File Formats

An .sc# file format for the mixed mode:
• The S-element model has additional keywords, such as mixedmode and datatype, if
the netlist includes one or more balanced ports.
• The mixedmode2port keyword prints in the header line.
• The other S-element keywords also appear in the header lines.
Touchstone 1.0/2.0 format for the mixed mode:
The following lines for data mapping are added to the head of the Touchstone output file if
the netlist includes one or more balanced ports.
! S11=SDD11
! S12=SDD12
! S13=SDC11
! S14=SDC12

Two-Port Parameter Measurement

Two-port parameter measurement function takes the first 2 ports, then reads the
corresponding parameter with the drive condition specified by the mixedmode2port
keyword.

Output Format and Description

File Type Description

*.ac# Output from both the .PROBE and .PRINT statements.

*.printac# Output from the .PRINT statement. Available in PrimeSim HSPICE when .option
LIS_NEW is set.

*.sc# The extracted S-parameters/2-port noise parameters are written to a *.sc# file by
using the S-element format. If you want to simulate the S element, you can reference
the *.sc# file in your netlist.

*.s#p* Touchstone 1.0/2.0 file

*.citi# CITI file

*.nc# File to output noise matrix. Generated by .lin noisecalc=1

* N=numOfPorts DATA=numOfFreq NOISE=[0|1] GROUPDELAY=[0|1]

* NumOfBlock=numOfSweepBlocks NumOfParam=numOfSweptParameters
* MIXEDMODE=[0|1] DATATYPE=mixedModeDataTypeString

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 69

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Extracting Mixed-Mode Scattering (S) Parameters

.MODEL mname S
+ N=numOfPorts FQMODEL=SFQMODEL TYPE=S Z0=*** *** ...
.MODEL SFQMODEL SP N=numberOfPorts SPACING=POI
INTERPOLATION=LINEAR MATRIX=NONSYMMETRIC
+ DATA= numberOfData
+ freq1
+ s11real s11imag s12real s12imag ... s1Nreal s1Nimag
...
+ sN1real sN1imag ... sNNreal sNNimag
...
...
+ freqNumberOfData
+ s11real s11imag s12real s12imag ...s1Nreal s1Nimag
...
+ sN1real sN1imag ... sNNreal sNNimag

* 2-port noise parameter

* frequency Fmin [dB] GammaOpt(M) GammaOpt(P) RN/Z0
* 0.10000E+09 0. 1.0000 0. 1.0281
* ...

The 2-port noise section starts with “*” so that you can include this file in your PrimeSim
HSPICE input netlists.

Features Supported
LIN analysis in PrimeSim HSPICE supports the following features:
.LINsupported features
• Automatic calculation of bias-dependent S, Y, and H parameters. No additional sources
required.
• Calculation of noise parameters (triggered by options).
• Calculation of group delay matrices (triggered by options).
In addition, PrimeSim HSPICE supports all existing PrimeSim HSPICE models for noise
analysis; PrimeSim HSPICE view port 1 as the input and port 2 as the output.

Prerequisites and Limitations

The following prerequisites and limitations apply to LIN analysis in PrimeSim HSPICE:
• Requires one .LIN statement to specify calculation options.
• Requires one .AC statement to specify frequency sweep and parameter sweep.
• Requires at least one P element, numbered from port 1 to N.
• For noise analysis, PrimeSim HSPICE views port 1 as the input and port 2 as the
output.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 70

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier

Reported Statistics for the Performance Log

• Simulation time
◦ DC op time
◦ Total simulation time
• Memory used
◦ Total memory

Errors and Warnings

• If the circuit contains fewer than two P-elements and noisecalc=1, then the 2-port
noise calculation is skipped.
• If the circuit contains fewer than two P-elements, you cannot use the .PRINT, .PROBE,
or .MEAS commands with any two-port noise or gain parameters.
• If the circuit contains more than two P-elements, all two-port parameters are computed.
By default, port=1 is the input and port=2 is the output. All other ports terminate in
their reference impedances.

Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier

The .LIN command simplifies the calculation of linear multi-port transfer parameters and
noise parameters. In the LIN analysis, Port (P) elements are used to specify port numbers
and their characteristic impedances. The analysis automatically computes the frequency-
dependent complex transfer coefficients between ports. The result is a convenient means
to get scattering parameters, noise parameters, stability parameters, and gain coefficients.
The .LIN command renders obsolete the .NET command. By default, the output from
the .LIN command is saved in the *.sc0 file format that can, in turn, be referenced as a
model file for the S-parameter element.
To set up a linear transfer parameter analysis, the PrimeSim HSPICE input netlist must
contain the following:
• An.AC command to activate small-signal AC analysis, and to specify a frequency
sweep. Also, use the .AC command to specify any other parameter sweeps of interest.
• A .LIN command with the .AC command to activate small-signal linear transfer
analysis. The .AC command specifies the base frequency sweep for the LIN analysis.
The LIN analysis automatically performs multiple AC and NOISE analyses as needed
to compute all complex signal transfer parameters.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 71

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier

• The necessary number of port (P) elements, numbered sequentially beginning with
one to define the terminals of the multi-port network. For example, a two-port circuit
must contain two port elements with one listed as port=1 and the other as port=2. The
port elements define the ordering for the output quantities from the .LIN command (for
example, the terminals for port=1 are used for S11, Y11, and Z11 measurements).
Much of the LIN analysis is automated so the PrimeSim HSPICE input netlist often does
not require the following:
• AC signal sources. The .LIN command computes transfer parameters between the
ports with no additional AC sources needed.
• DC sources. You can analyze a purely passive circuit without adding sources of any
kind.
The following tutorial example shows how to set up a LIN analysis for an NMOS low noise
amplifier circuit. This netlist is shipped with PrimeSim HSPICE distribution as gsmlna.sp
and is available in the following directory: $installdir/demo/hspice/rf_examples/.
Compare this analysis to the use of the .HB command on the same circuit. See Example
2 — Using HB Analysis for a Low Noise Amp in the PrimeSim HSPICE User Guide:
Advanced Analog Simulation and Analysis.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 72

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier

Figure 7 Schematic showing instantiation for low noise amplifier

** NMOS 0.25um Cascode LNA for GSM applications

** setup for s-parameter and noise parameter measurements
**
.temp 27
.options post
.param Vdd=2.3
.global gnd
**

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 73

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier

** Cascode LNA tuned for operation near 1 GHz

**
M1 _n4 _n3 _n5 _n5 CMOSN l=0.25u w=7.5u
+ as=15p ad=15p ps=19u pd=19u m=80
M2 _n6 _n1 _n4 _n4 CMOSN l=0.25u w=7.5u
+ as=15p ad=15p ps=19u pd=19u m=80
M3 rfo _n6 gnd gnd CMOSN l=0.25u w=7.5u
+ as=15p ad=15p ps=19u pd=19u m=40
r1 _vdd _n6 400
l1 _n5 gnd l=0.9nH
l2 rfin _n3 l=13nH
vvb _n1 gnd dc=1.19 $ bias for common base device
vvdd _vdd gnd dc=Vdd
rfb rfo _n6 120 $ feedback
**
** 50 Ohm input port (incl. bias), 255 Ohm output port.
**
P1 rfin gnd port=1 z0=50 dc = 0.595 $ input port includes DC bias
P2 rfo _vdd port=2 z0=255 $ port doubles as pull-up resistor
**
** Measure s-parameters and noise parameters
**
.AC DEC 50 100MEG 5G
.LIN noisecalc=1 sparcalc=1 format=touchstone
.PRINT S11(DB) S21(DB) S12(DB) S22(DB) NFMIN
**
.PROBE NFMIN G_AS K_STABILITY_FACTOR
** Approximate parameters for TSMC 0.25 Process (MOSIS run T17B)
**
.MODEL CMOSN NMOS ( LEVEL = 49
3.1 TNOM = 27 TOX = 5.8E-9
+XJ = 1E-7 NCH = 2.3549E17 VTH0 = 0.3819327
+K1 = 0.477867 K2 = 2.422759E-3 K3 = 1E-3
+K3B = 2.1606637 W0 = 1E-7 NLX = 1.57986E-7
+DVT0W = 0 DVT1W = 0 DVT2W = 0
+DVT0 = 0.5334651 DVT1 = 0.7186877 DVT2 = -0.5
+U0 = 289.1720829 UA = -1.300598E-9 UB = 2.3082E-18
+UC = 2.841618E-11 VSAT = 1.482651E5 A0 = 1.6856991
+AGS = 0.2874763 B0 = -1.833193E-8 B1 = -1E-7
+KETA = -2.395348E-3 A1 = 0 A2 = 0.4177975
+RDSW = 178.7751373 PRWG = 0.3774172 PRWB = -0.2
+WR = 1 WINT = 0 LINT = 1.88839E-8
+XL = 3E-8 XW = -4E-8 DWG = -1.2139E-8
+DWB = 4.613042E-9 VOFF = -0.0981658 NFACTOR = 1.2032376
+CIT = 0 CDSC = 2.4E-4 CDSCD = 0
+CDSCB = 0 ETA0 = 5.128492E-3 ETAB = 6.18609E-4
+DSUB = 0.0463218 PCLM = 1.91946 PDIBLC1 = 1
+PDIBLC2 = 4.422611E-3 PDIBLCB = -0.1 DROUT = 0.9817908
+PSCBE1 = 7.982649E10 PSCBE2 = 5.200359E-10 PVAG = 9.31443E-3
+DELTA = 0.01 RSH = 3.7 MOBMOD = 1
+PRT = 0 UTE = -1.5 KT1 = -0.11
+KT1L = 0 KT2 = 0.022 UA1 = 4.31E-9
+UB1 = -7.61E-18 UC1 = -5.6E-11 AT = 3.3E4
+WL = 0 WLN = 1 WW = 0
+WWN = 1 WWL = 0 LL = 0
+LLN = 1 LW = 0 LWN = 1
+LWL = 0 CAPMOD = 2 XPART = 0.5
+CGDO = 5.62E-10 CGSO = 5.62E-10 CGBO = 1E-12
+CJ = 1.641005E-3 PB = 0.99 MJ = 0.4453094
+CJSW = 4.179682E-10 PBSW = 0.99 MJSW = 0.3413857
+CJSWG = 3.29E-10 PBSWG = 0.99 MJSWG = 0.3413857

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 74

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier

+CF = 0 PVTH0 = -8.385037E-3 PRDSW = -10

+PK2 = 2.650965E-3 WKETA = 7.293869E-3 LKETA = -6.070E-3)
*
.END

A LIN analysis also includes the following:

• .LIN command:
.LIN noisecalc=1 sparcalc=1 format=touchstone

This invokes a LIN analysis and activates noise calculations and S- parameter output
files in Touchstone format.
• Two port elements:
P1 rfin gnd port=1 z0=50 dc=0.595

Specifies that an input port is assumed between terminals rfin and ground, that it is
has a 50 ohm termination, and it has a built-in DC bias of 0.595 V. The output (second)
port is:
P2 rfo _vdd port=2 z0=255

This syntax specifies that the output port is between terminals rfo and _vdd, and is
being used as a pull up resistor with impedance of 255 ohm.
• A .PRINT command for plotting the output S-parameters in dB and the noise figure
minimum.
To run this netlist, type the following command:
hspicerf gsmlna.sp

This produces three output files, named gsmlna.ac0, gsmlna.s2p, and

gsmlna.printac0, containing the AC analysis results, S-parameter and noise parameter
results, and the requested PRINT data.
To view the output:
1. Type wv at the prompt to invoke WaveView.
2. Click the Open Waveform icon and select the gmslna.s2p file from the Open:
Waveform Files dialog box.
3. Select Smith Chart from the Panel section of the Waveform tab to open a blank Smith
chart.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 75

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier

4. Select the S(1,1) and S(2,2) signals in the signal browser. Drag and drop the selected
signals in the Smith Chart.
5. Select Polar Plot from the Panel section of the Waveform tab to open a blank Polar
chart.
6. Select the S(2,1) signal in the signal browser. Drag and drop the signal in the Polar
chart to plot the complex gain of the LNA.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 76

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
Tutorial Example: LIN Analysis for a NMOS Low Noise Amplifier

7. Click the Open Waveform icon and select the gmslna.ac0 file from the Open:
Waveform Files dialog box.
8. Click the WaveView icon in the Add section of the Waveform tab to open a new
waveview.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 77

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
References

9. Select the g_as (associated gain), k_stability_factor (Rollet stability factor) and nfmin
(the noise figure minimum) signals in the signal browser. Drag and drop the selected
signals in the new waveview.

References
[1] Goyal, Ravender. “S-Parameter Output From SPICE Program”, MSN & CT, February
1988, pp. 63 and 66.
[2] Robert J. Weber “Introduction to Microwave Circuits”, IEEE Press.
[3] Behzad Razavi, “Design of Analog CMOS Integrated Circuits”, McGraw Hill.
[4] Reinhold Ludwig, Pavel Bretchko, “RF Circuit Design Theory and Applications”.
[5] G.D. Vendelin, Design of Amplifiers and Oscillators by the S-Parameter Method, John
Wiley & Sons, 1982.
[6] R.S. Carson, High-Frequency Amplifiers, 2nd Edition, John Wiley & Sons, 1982.
[7] G. Gonzalez, Microwave Transistor Amplifiers: Analysis and Design, 2nd Edition,
Prentice-Hall, 1997.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 78

W-2024.09
 Feedback
Chapter 2: Linear Network Parameter Analysis
References

[8] M.L. Edwards and J.H. Sinsky, “A single stability parameter for linear 2-port networks,”
IEEE 1992 MTT-S Symposium Digest, pages 885-888.
[9] H. Rothe and W. Dahlke, “Theory of noisy fourpoles”, Proc. IRE, volume 44, pages
811-818, June 1956.
[10] David E. Bockeman, “Combined Differential and Common-Mode Scattering
Parameters: Theory and Simulation,” IEEE trans. on MTT Volume 43, Number 7, Jul.
1995.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 79

W-2024.09
 Feedback

3
Statistical Eye Analysis

Describes statistical eye diagram analysis by using the .STATEYE command.

Analysis of high-speed serial interfaces requires processing millions of bits of data

making analysis by traditional analysis tools computationally expensive. Eye diagrams
aid extensively in the evaluation of high-speed serial interfaces and as a fundamental
performance metric for high-speed serial interfaces in the bit error rate (BER). Statistical
statistical eye analysis.STATEYE commandbit error rate (BER)BERStatEye

eye diagram techniques enable you to quickly and accurately evaluate eye diagrams and
BER.
PrimeSim HSPICE includes several full demonstration examples for your use:
• For paths to the StatEye demo files, see Signal Integrity Examples in the PrimeSim
HSPICE User Guide: Demonstration Netlists for paths to the StatEye demo files.
• For a demonstration of Stateye analysis, go to https://www.synopsys.com/verification/
ams-verification/hspice.html and click PrimeSim HSPICE TIPS & TRICKS to load the
Eye Diagrams (PrimeSim HSPICETips & Tricks #2) mini webinar.
• For a discussion of the StatEye feature working with IBIS-AMI, see Chapter 9, Using
IBIS-AMI Equalizer Models with StatEye.
• For a mini webinar demo of StatEye with IBIS-AMI, see https://www.synopsys.com/
verification/ams-verification/hspice.html > PrimeSim HSPICE TIPS & TRICKS > IBIS-
AMI (PrimeSim HSPICE Tips & Tricks #3).
The following sections discuss these topics:
• Understanding StatEye Analysis in PrimeSim HSPICE
• Statistical Eye Analysis Setup
• Determining Number of Bits Simulated in Statistical and Bit-by-bit Modes
• Edge Control
• Capturing Steady State Noise for Power Integrity
• Duty Cycle Distortion
• Capturing DCD Effect on System Clock

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 80

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Understanding StatEye Analysis in PrimeSim HSPICE

• Periodic and Random Jitter Effect

• N-Cycle Jitter Modeling for Data and Clock Signals
• Pre-Emphasis and De-Emphasis
• Built-In DFE Function for Rx Port Element in StatEye Analysis
• Perturbation Analysis with StatEye
• Unfolding Statistical Eye Diagrams to Waveforms
• Multiple Data Rate Eye Diagram Analysis with StatEye
• Accelerating StatEye’s Initial Transient with Process Distribution and Multi-Threading
• Known Limitation
• Troubleshooting Output Warnings
• References

Understanding StatEye Analysis in PrimeSim HSPICE

StatEye analysis in PrimeSim HSPICE represents eye diagrams by using-probability-
density function (PDF). When a probability of a target waveform's crossing
between voltage and is given using nonnegative, continuous function as :

Equation 6

where, is defined as a probability density function of with the unit of one

over interval. In PrimeSim HSPICE .StatEye, [1/mV] is used for the unit of eye diagram
probably density function. Here, satisfies:

Equation 7

Then, the BER can be evaluated as an accumulation of probability density from the center
of the eye up to the target voltage point as:

Equation 8

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 81

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

where, is the center of the eye diagram. Assuming a given bit stream’s even probability
for each symbol, the boundary conditions of the BER are given as:

Equation 9

Statistical Eye Analysis Setup

The .STATEYE command performs pulse/step response examinations before the statistical
computation and performs this initial time domain analysis as a common transient
analysis. Therefore, you can apply all the .print/.PROBE/.MEASURE commands with the
tran keyword.

Use the port element to designate the incident (input) and probe (output) ports for the
system under analysis. You can specify ports can as single-ended or mixed-mode. You
stateyeanalysis setup

can apply random jitter to each incident and probe point in the system.
Each incident port acts as random bit pattern source with specified voltage magnitude. .
Probe ports function as observation points where you can define .PRINT, .PROBE, and
.MEASURE statements.

To perform the statistical eye diagram analysis, use the analysis command .STATEYE.
The following sections discuss these topics:
• Port Element Configuration
• .PRINT and .PROBE Syntax
• Output Data
• .MEASURE Syntax
• Understanding the Worst Pattern Measurement Result
• Understanding n-UI Accumulated Clock Period Jitter Measurements (RMS/PP)
• Sample Measurements
• Output
• Saving and Reusing Initial Transient Results
• Examples - Statistical Analysis Setup
• Example – StateEye Crosstalk using W Element

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 82

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

Port Element Configuration

Incident ports act as pure random bit stream generators in .STATEYE analysis unless you
specify either LFSR or PAT source keywords. These keywords force the incident port to act
as a specific bit pattern generator. If an incident port element does not have a time-domain
voltage magnitude specification, PrimeSim HSPICE uses the default values, V_high=1.0,
V_low=-1.0.

You can print or probe the following statistical eye diagram analysis:
• Eye diagram at a port
• Bit error rate map at a port
• Bathtub curve of the bit error rate at a port for specified voltage
• Bathtub curve of the bit error rate at a port for a specified time
• Emphasis (see Pre-Emphasis and De-Emphasis)
• Duty cycle distortion (see Duty Cycle Distortion)
• Perjitter analysis (see Periodic and Random Jitter Effect)
For P-element syntax, refer to Identifying Ports with the P-element.
If you define port elements, but do not specify them as incident or probe ports, PrimeSim
HSPICE assumes them to be DC voltage sources with no time dependence.

.PRINT and .PROBE Syntax

.print stateye eye(port_idx)
.print stateye eyeBW(port_idx)
.print stateye eyeC(port_idx)
.print stateye eyeV(port_idx,v)
.print stateye eyeT(port_idx,t)
.print stateye ber(port_idx)
.print stateye berC(port_idx)
.print stateye bathtubV(port_idx,v)
.print stateye bathtubT(port_idx,t)
.print stateye eyeUF(port_idx, bit_idx, bit_idx2)

.probe stateye eye(port_idx)

.probe stateye eyeBW(port_idx)
.probe stateye eyeC(port_idx)
.probe stateye eyeV(port_idx,v)
.probe stateye eyeT(port_idx,t)
.probe stateye ber(port_idx)
.probe stateye berC(port_idx <,BER_value>)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 83

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

.probe stateye bathtubV(port_idx,v)

.probe stateye bathtubT(port_idx,t)
.probe stateye eyeUF(port_idx, bit_idx, bit_idx2)

where:
• eye(port_idx) specifies the port to output the eye diagram.
• eyeBW(port_idx)specifies the port to output the eye diagram as black and white data.
• eyeC(port_idx) specifies the port to output the eye diagram as contour data (see
Figure 8).
• eyeV(port_idx,v)specifies the port to output as a cross-section of the eye diagram at
specified port at a specified voltage.
• eyeT(port_idx, t) specifies cross-section of the eye diagram at specified port at a
specified time.
• ber(port_idx) specifies the port to output the bit error rate.
• berC(port_idx, ber_value)) specifies the port to output the bit error rate as contour
data (see Figure 8). When BER_value is specified, the output contains single contour
at a given BER level. When no BER_value is specified, multiple contour levels are
included. See also BERC_Levels keyword description for more information about
contour levels.
• bathtubV(port_idx, v) specifies the port and voltage to output a bathtub curve.
• bathtubT(port_idx, t) specifies the port and time to output a bathtub curve.
• eyeUF(port_idx, bit_idx, bit_idx2)specifies the port and bit index to output
an unfolded time domain waveform (see Unfolding Statistical Eye Diagrams to
Waveforms). When one bit index is specified, StatEye outputs a time domain waveform
around the specified bit position. When two bit indices are specified, StatEye outputs a
bit stream waveform from the first index to the second index. In this case, you can use
“-1” for either or both indices where -1 in the first argument means the beginning of the
simulation and -1 in the second argument means the end of the simulation.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 84

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

Figure 8 (Left) eyeC(port_idx) and (igrht) berC(port_idx) contour waveforms

Output Data
This section discusses output for .PROBE, .PRINT, and.MEASURE commands and initial
transient results.
Probing Output
The .probe stateye command generates netlist.stet# and netlist.stev# for
following purposes:
• netlist.stet#: eye(t,v), eyeBW(t,v), eyeV(t), ber(t,v) and bathtubV(t)
• netlist.stev#: eyeT(v) and bathtubT(v)
Printing Output
The .print stateye command generates the data directory netlist.printSte#/. Then
the command stores a convenient gnuplot script for each .print stateye target with
necessary data files. For example:
• The .print stateye eye(2) statement creates the netlist.printSte0/eye_2.plt
script file.
• The .print stateye bathtubV(2,0.1) creates the netlist.printSte0/
bathtub_2_0.1000e+00.plt script file.

On the gnuplot command shell, you can load these scripts directly to display the target
data. For example:
% gnuplot
gnuplot> load "printSte0/eye_2.plt"
gnuplot> load "netlist.printSte0/bathtub_2_0.1000e+00.plt"

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 85

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

Measurement Output
The .measure stateye command generates the netlist.mste# file and stores the
measurement results.
StatEye Initial Transient Results
Storage of probe and measurement results takes place in the netlist.pNeM.tr# files,
where:
• N: Incident port index
• M: Edge index
• #: Alter index
Typically, netlist.p1e1.tr0 is the probe result file for the pulse response of the first
incident in single edge mode.

.MEASURE Syntax
You can measure the following statistical eye diagram analysis results:
eyetypemeasure syntax, StatEyeStateyeeyetype measurements
• Vertical eye opening at a specified time
• Horizontal eye opening, RMS, and peak-to-peak jitter at a specified voltage
• A bit pattern that produces the worst eye opening at a specified time in a high or low
state (See Figure 9.):
• Crosstalk measurement
• Accumulated period jitter, RMS, and peak-to-peak with the lowest target probability
specified as a tolerance (default is tol=1e-20)

.measure stateye result_name Veye port_idx [time=val]

+ [tol=val] [eyetype=1|2]
.measure stateye result_name Veye_AVG port_idx
+ [tol=val] [eyetype=1|2]
.measure stateye result_name Heye port_idx [volt=val]
+ [tol=val][eyetype=1|2]
.measure stateye result_name WorstBits port_idx [time=val]
+ [state=low|high]
.measure StatEye result_name Eye port# time=val volt=val
.measure StatEye result_name BER port# time=val volt=val
.measure stateye result_name rmsjitter port_idx [volt=val]
+ [eyetype=1|2]
.measure stateye result_name ppjitter port_idx [volt=val]
+ [tol=val] [eyetype=1|2]
.measure stateye result_name VBOX port_idx time=val
+ [tol=val] [eyetype=1|2]

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 86

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

.measure stateye result_name HBOX port_idx volt=val

+ [tol=val] [eyetype=1|2]
.measure stateye result_name Vboundary port_idx [time=val]
+ [tol=val] [eyetype=1|2]
.measure stateye result_name VREF port_idx
+ [tol=val] [eyetype=1|2]
.measure stateye result_name CROSSTALK probe_port_idx
.measure stateye result_name TREF port_idx
+ [tol=val] [eyetype=1|2]
.measure stateye result_name RMSaccPerJitter clk_idx <nUI = n>
.measure stateye result_name PPaccPerJitter clk_idx <nUI = n> <tol=value>

Figure 9 Measurements show vertical and horizontal eyes, worst bit pattern, and voltage
with time eye

where:
• Veye port_idx [time=val] [tol=val] specifies the measurement port, time, and
tolerance for the vertical eye opening. (For more on tol, see Vertical/Horizontal Eye-
Opening Tolerance.) If you do not specify time, PrimeSim HSPICE measures the
maximum eye height and reports the time of the maximum eye opening in the listing
file. eyetype=1 (default) specifies a PDF-based eye opening measurement. eyetype=2
specifies a BER- based eye opening measurement.
• Veye_AVG port_idx[tol=val] specifies the measurement port and tolerance for
the average eye opening. (For more on tol, see Vertical/Horizontal Eye-Opening
Tolerance.) This measurement computes the average vertical eye opening over the
time interval for which eye height is nonzero.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 87

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

• Heye port_idx [volt=val] [tol=val] specifies the measurement port, voltage,

and tolerance for the horizontal eye opening. (For more on tol, see Vertical/Horizontal
Eye-Opening Tolerance.) If you do not specify voltage, StatEye measures the
maximum eye width, and PrimeSim HSPICE reports the voltage position in the listing
file.
• WorstBits port_idx time=val [state=low|high] specifies the port, time, and
state of the bit pattern that produces the worst eye opening. The default values of
tolerance (tol) and logic state (state) are 1e-20 and high, respectively. If you do
not specify time, StatEye measures the worst bit pattern at the maximum vertical eye
opening point, and PrimeSim HSPICE reports the measured time point in the listing file.
• Eye and BER measurements pick up pinpoint probability at time and voltage points that
you specify.
• rmsjitter/ppjitter port_idx [volt=val] specifies the measurement port and
voltage for the RMS or peak-to-peak jitter. If you do not specify voltage, PrimeSim
HSPICE measures jitter at voltage=0. tol=val specifies the tolerance for the peak-
to-peak jitter measurement. eyetype=1 (default) specifies a PDF-based eye opening
measurement. eyetype=2 specifies a BER-based eye opening measurement.
• VBOX measures the vertical eye opening at a given time from the reference time.
The .STATEYE specifies how to retrieve the reference time for VBOX and WorstBits
measurements.
• HBOX measurement first captures the voltage at which the horizontal maximum eye
opening is given as reference. Then HBOX measures horizontal eye opening at given
voltage from the reference voltage.
• Vboundary port_idx[time=val][tol=val] specifies the measurement port, time,
and tolerance for the vertical eye boundary where the maximum possible voltage swing
is above probability tolerance. If you do not specify time, PrimeSim HSPICE measures
the maximum eye boundary and reports the time at which the maximum value is given.
eyetype=1 (default) specifies a PDF-based eye boundary measurement. eyetype=2
specifies a bit error rate-BER-based eye boundary measurement.
• VREF measurement returns the reference voltage level at which the eye diagram gives
the maximum horizontal eye opening.
• CROSSTALK probe_port_idx lists the magnitude of the crosstalk noise from each
incident port observed at the specified probe_port_idx.
• TREF measurement returns the reference time at which the eye diagram gives the
maximum vertical eye opening.
Vertical/Horizontal Eye-Opening Tolerance
Tolerance for eye width/height measurements means that the eye is considered open
where PDF/BER values are below the tolerance value, where the default is tol=1e-20.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 88

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

Without random jitter, changes to the tolerance value have little effect on the eye-opening
result. Confirm this by viewing vertical/horizontal PDF/BER cross sections (eyeT, eyeV,
bathtubT, and bathtabV). Then, if you add random jitter to the eye, you can see the
usefulness of the tol=val control since random Gaussian distribution is unbounded. Thus
PDF/BER never reach zero.
Worst Bits Output Format
In the worst bits output, the number within parenthesis is the incident port index which
becomes important when you have multiple data lanes that include the effect of crosstalk.
In case of a multiple data lane system, the worst bits result that looks something like:
(1)00111111100(2)11110111101

This means that you will get the worst possible eye edge with this bit combination at the
two incident inputs.
In cases that have different bit lengths, the worst bits result that looks something like:
(1)01111(3)1111111111111101

The shorter bit sequence is left padded with don’t cares (-):
Incident port 1 bit sequence: -----------01111
Incident port 3 bit sequence: 1111111111111101

The values of the preceding bits do not matter and will still produce the worst bits.

Understanding the Worst Pattern Measurement Result

The worst pattern catcher results are stored in both the log (*.lis) file and measurement
(*.mste0) file. Following is the example of the result pattern set.
(1:241)10101001011110011111 (2:36359)00000100101010010110

The worst patterns set above consists of two N patterns where N is the number of
incidents including main signal and bits from crosstalk aggressors. Each pattern contains

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 89

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

header part and body (pattern) part. The header part contains 1, 2 or 3 numbers
depending on the StatEye analysis set up as follows:
1. When using the multilevel (PAM-N) logic, the header format contains 3 numbers:
(port_index : bit_index : symbol_index)

where,
◦ port_index is the input port index.
◦ bit_index is the bit position of the right most bit of the worst pattern.
◦ symbol_index is the symbol position corresponding to the bit position.
For example, Figure 10 indicates that, the signal from port-1 produces the worst
waveform situation at 40286-th bits, which means the 20143-th symbol (PAM-4 in this
case). And the worst pattern displays the pattern towards the worst point. Here, the
right most bit represents the 40286-th bit.

Figure 10 Using the multilevel (Pam-N) logic

2. When using system clock-based DCD (DCD_mode=2), the header contains 2

numbers:
(port_index : bit_index)

where,
◦ port_index is the input port index
◦ bit_index is the bit position of the right most bit of the worst pattern.

Figure 11 Using system clock-based DCD

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 90

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

In DCD_mode=2, narrow and wide clock cycles toggle. When positive DCD is given to
the incident port, the bits in odd indices are in narrow cycle and bits in even indices are
in wide cycle.

Figure 12 Narrow and Wide Cycles

3. Other than these conditions, the header contains 1 number where

port_index is the input port index.

Understanding n-UI Accumulated Clock Period Jitter

Measurements (RMS/PP)
The root mean square (RMS) and peak-to-peak (PP) values are measured from the
following period jitter distribution:

where,
is the nUI accumulated period jitter.
and are the ideal period and the period at the p-th bit/symbol position.

is the number of jitter accumulation.

is the total number of bits/symbols.
For peak-to-peak jitter, the lowest target probability can be specified as a tolerance where
the default is tol=1e-20.
Note:
This measurement is available on clock output port in numerical sequence
mode.
See the Clock_Port and Clock_Mode keyword in the .STATEYE command reference.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 91

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

Sample Measurements
.StatEye …
+ WORST_WAVEFORM=1
* 1 : worst waveform into binary (tr0) file
* 2: worst waveform into text (.txt) file
.measure stateye wst_h WorstBits 3 time=0e-12 state=high
.measure stateye wst_l WorstBits 3 time=0e-12 state=low
.measure stateye wst2_h WorstBits 3 time=70e-12 state=high
.measure stateye wst2_l WorstBits 3 time=70e-12 state=low

Output
****** Measured values for the netlist ***********************
worst_h= (1)000000000000000000010(2)000011111111111110101
worst_l= (1)100000111111011111101(2)110000110000010000010
worst2_h= (1)000000000000000000010(2)000011011111111110110
worst2_l= (1)100000111111011111101(2)110000010000010000001
% ls my_netlist.steuf0/
worst3_0.000000e+00_h.tr0 worst3_7.000000e-11_h.tr0
worst3_0.000000e+00_l.tr0 worst3_7.000000e-11_l.tr0

Figure 13 Output

Saving and Reusing Initial Transient Results

StatEye analysis shows superior performance in generating eye diagrams over
conventional transient analysis. However, when a system has a large number of circuit
elements and/or has many incident ports, the computational time for initial pulse/step
response examinations becomes significant. Therefore, it is useful to be able to save
and load the initial transient analysis data. This capability is especially useful when
running multiple StatEye analyses without changing network configuration, but changing
noise properties and/or post process (.PRINT/.PROBE/.MEASURE) targets. Using the data

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 92

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

loading function you can import externally generated pulse/step responses. The .StatEye
command takes the Save_tr and Load_tr keywords for such purposes.
For the save/load functionality, the name of data storage directory is netlist.save#/ (#:
alter number). In the storage directory, StatEye saves/seeks the following files for each
purpose:
• incident#.txt stores pulse responses (for Edge=1)
• incident#_r.txt stores rising step responses (for Edges=2)
• incident#_f.txt stores falling step responses (for Edges=2)
• incident#_eN.txt, N is 1-4 for Edges=4
• incident#_eN.txt, N is 1-8 for Edges=8
where # is the index number (PORT=n) of the incident port element. Save/Load data
files are in simple table data format. The first column is always the “time” entry then
probe# (# index number (PORT=n) of the probe port element) entries follow in the order of
probe_port keyword specification in the .StatEye statement.

Example (stateye.save0/incident1.txt)
time probe1 probe2
0.0000000000e+00 -4.9945800254e-01 0.0000000000e+00
1.0000000000e-12 -4.9945800254e-01 2.7317300241e-17
2.0000000000e-12 -4.9945800254e-01 7.4348539204e-17
3.0000000000e-12 -4.9945800254e-01 1.3213235524e-16
...

Examples - Statistical Analysis Setup

Circuit Example:
* StatEye example circuit
* Define parameters
.param _BW=0.05n
.param trf = '_BW*0.01'
.param opfreq = '1.0/_BW'
* define ports
* Port P1 in the input or incident port
* Port P2 is the output or probe port
P1 in 0 port=1 LFSR (1 -1 0 'trf' 'trf''opfreq' 1 [5,2] )
P2 out 0 port=2
* Channel or devices to be simulated
X1 in out CONN
* Stateye command
.stateye T='_BW' trf='0.1*_BW' incident_port=1 probe_port=2
* Simulation options
.options post accurate

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 93

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

* Subcircuit definition
.subckt CONN _in _out
Rtest _in _out 50 RS=1e-3 $ skin effect resistor
Cin _in 0 0.3p
Cout _out 0 0.3p
.ends
* print and probe statements to output stateye results
* output eye diagram in color and black and white at the probe port
* output bit error rate (ber) at the probe port
.probe stateye eye(2) ber(2) eyeBW(2)
.print stateye eye(2) ber(2) eyeBW(2)
* print and probe statements to output bathtub curves
* Output bathtub curve at a time point and bathtub curve at a voltage
point
.probe stateye bathtubT(2,1e-11) bathtubV(2,0.0)
.print stateye bathtubT(2,1e-11) bathtubV(2,0.0)
* Print and probe statements to output the eye size
* Output eye cross section at a time point and at a voltage point
.probe stateye eyeV(2,0.0) eyeT(2,1.0e-11)
.print stateye eyeV(2,0.0) eyeT(2,1.0e-11)
* Measurement statements
* measure the vertical eye size at a time point
* measure the horizontal eye size at a voltage point
* measure the worst bits at a time point
.measure stateye veye1 Veye 2 time=1e-11 tol=1e-10
.measure stateye heye1 Heye 2 volt=0.0 tol=1e-10
.measure stateye worst1 WorstBits 2 time=10p
.end

The PrimeSim HSPICE demonstration example, stateye_ex1.sp, is in

$installdir/demo/hspice/si.

Example – StateEye Crosstalk using W Element

Circuit Example:
* superposing LTI responses

P1 in1 0 port=1
+ LFSR (1 -1 0 '0.01*_BR' '0.01*_BR''1.0/_BR' 1 [13,5,2] )
P2 in2 0 port=2
+ LFSR (1 -1 0 '0.01*_BR' '0.01*_BR''1.0/_BR' 1 [15,14,6,1] )
P3 in3 0 port=3
+ LFSR (1 -1 0 '0.01*_BR' '0.01*_BR''1.0/_BR' 1 [14,13,2,1] )

P4 out1 0 port=4
P5 out2 0 port=5
P6 out3 0 port=6

.param _BR=0.1n

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 94

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Statistical Eye Analysis Setup

.stateye T='_BR' trf='0.01*_BR'

+ incident_port= 1 2 3
+ probe_port= 4
+ save_tr=ascii

.opt post accurate

.probe stateye eye(4)
.measure stateye xtalk4 crosstalk 4

W1 in1 in2 in3 gnd out1 out2 out3 gnd FSmodel=line3 N=3 l=0.1
+ delayopt=3
+ includegdimag=yes

.material diel dielectric er=4.3, losstangent=0.02

.material copper metal conductivity=57.6meg

.shape rect rectangle width=400e-6

+height=40e-6
+nw=20 nh=20

.layerstack stack1
+layer=(copper,10e-6)
+layer=(diel,200e-6)

.fsoptions opt1 printdata=yes computegd=yes computers=yes

.model line3 W Modeltype=fieldsolver,

+layerstack=stack1,
+fsoptions=opt1,
+Rlgcfile=line2.rlgc,
+conductor=(shape=rect,origin=(0,100e-6),material=copper)
+conductor=(shape=rect,origin=(500e-6,100e-6),material=copper)
+conductor=(shape=rect,origin=(1000e-6,100e-6),material=copper)

.end

Ports 1, 2, and 3 are incident ports. Ports 4, 5, and 6 are termination ports; and port 4 is
designated as probe port in the StatEye analysis.
The crosstalk measurement on port 4 is written in the .mste0 file and lists the magnitude
of the crosstalk noise from each.
The PrimeSim HSPICE demonstration example, xt_meas.sp, is in $installdir/demo/
hspice/si/stat_xtalk.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 95

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Determining Number of Bits Simulated in Statistical and Bit-by-bit Modes

Determining Number of Bits Simulated in Statistical and Bit-by-bit

Modes
This section covers the following topics
• Statistical Mode
• Bit-by-bit Mode
• Pulse Superposition and Edge Mode

Statistical Mode
The number of bits simulated is directly related to the length of the pulse response. The
pulse response length is reported in the log file (.lis). For example if the length reported
8.5
is 8.5T where T is the length of the UI, the number of bits simulated is 2 = 362 bits. To
increase the number of bits, use keyword ISI_DEPTH (Iter-Symbol Interference Depth)
in the .STATEYE command. Specifying ISI_DEPTH=20, the number of bits simulated is
20
2 =1048576.
Note:
By default, PrimeSim HSPICE determines the length of the time window (8.5T)
based on settling of the pulse response within a certain tolerance.
For example:
When ISI_DEPTH=5, all possible bit patterns simulated are:
00000
00001
00010
00011
.
.
11100
11101
11110
11111

The STATEYE analysis produces the output of all possible patterns above and generates
corresponding eye diagram. This can be thought of as a sequence of these bits
combination.
ISI_DPETH
The length of the sequence above becomes 2 =32 bits
The .STATEYE TRAN_INIT option limits the initial transient to 60 UIs. This would limit the
ISI_DEPTH in case ISIS_DEPTH is specified higher than TRAN_INIT.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 96

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Determining Number of Bits Simulated in Statistical and Bit-by-bit Modes

Note:
The first all zero does not practically occur in PRBS (LFSR cannot have
seed=0). But statistically, all zero and ones are possible patterns.

Bit-by-bit Mode
When port elements are specified as LFSR or PAT sources, by default .STATEYE analysis
is run in bit-by-bit mode. The length of the simulated pattern determines the number of bits
simulated.
In the STATEYE transient mode (mode=tran), the number of bits is set by default to
1000 (MAX_BITS=1000). When edge mode (mode=edge) is used the number of bits
21
is 2 = 2097152 (approximately 2million bits). MAX_BITS and PATTERN_REPEAT can be
increased if more bits are desired. For more information, see keywords PATTERN_REPEAT
and MAX_BITS in the .STATEYE section of the PrimeSim Continuum Reference Manual:
Commands and Control Options.
Switching from bit-by-bit mode to statistical mode can be achieved by adding
MODE=STATISTICAL in the STATEYE analysis.

Pulse Superposition and Edge Mode

In .STATEYE edge mode, the edge superposition method is used. The pulse response
is generated based on the superposition of multiple edges. If specified, jitter with its
corresponding variance is then added to the pulse. The STATEYE analysis convolves
the distribution of the jitter on top of the ISI based PDF. Same treatment of jitter is
implemented in the STATEYE transient mode.

Figure 14 Edge superposition method

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 97

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Edge Control

Edge Control
Statistical eye analysis method generates eye diagrams by superposing single bit
StatEyeedge control

responses based on symbols in a given bit stream. While the statistical approach
significantly reduces the computational cost for eye diagram generation compared to the
conventional transient-based approach, it may not be accurate when the target system is
strongly nonlinear.
Typical accuracy degradation occurs when a nonlinear buffer is imbalanced and the
negative pulse response becomes different from a positive one. In these cases, an
eye diagram becomes virtually nonsymmetric (Figure 15). As a result, the conventional
statistical method is unable to capture an eye diagram with great accuracy due to
imbalanced buffer.

Figure 15 (Left) Vertically nonsymmetric eye diagram due to an imbalanced buffer;

generated by transient analysis with a pseudo random bit stream (PRBS) (right)
Conventional .STATEYE result with imbalanced buffer

To overcome the difficulty, you can use the Edge keyword in the .STATEYE command.
Here you can represent a pulse response, P(t), by using the rising edge step, R(t), and
falling edge step, F(t), responses with T as the unit interval (width of the input pulse):

Here, since nonlinear buffers usually have circuit-bias dependencies, you must specify
these edge response acquisitions at desired bias points. Then the statistical process
must correctly account for these bias points. Because of the double edge-based statistical
computations, the .STATEYE command realizes higher levels of consistency with
conventional transient-based results (Figure 16).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 98

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Edge Control

Figure 16 Comparison between .STATEYE and transient-based eye diagrams. On left

is the eye diagram of .STATEYE with double edge mode; the right shows the conventional
transient-based eye diagram

The following sections present these edge-related topics:

• Statistical Eye Full Transient Mode
• Example - Controlling Edges
• Example - Run StatEye with Precomputed Transient Waveform
• Separate Rise Time and Fall Time Settings
• Multi-Edge Responses for Accurate Modeling of Nonlinearity

Statistical Eye Full Transient Mode

Statistical eye diagram analysis algorithm consider a systems' inter-symbol interference
(ISI) and other linear memory effects. However, it is difficult to consider buffers nonlinear
memory effects since superposition theory gradually breaks down as the amount of
nonlinearity increases. By using multiple edge patterns, the StatEye accuracy can be
increased dramatically. However, ideally, there should be a method to fully verify a
nonlinear system that is faster than traditional transient analysis. The StatEye full transient
mode will run a transient analysis and generate the PDF eye diagrams, and BER eye
diagrams.The full transient mode uses all the available StatEye features such as random
(RJ) and deterministic (DJ) and equalization using pre- and de-emphasis. The full transient
mode is compatible with transistor level, IBIS, and IBIS-AMI models.
By Specifying MODE=TRAN, StatEye uses full transient analysis mode instead of edge
mode. The EDGE keyword will be ignored when full transient analysis mode is specified.
Transient Probe Data Information about .STATEYE in Transient Mode (mode=tran)
When .STATEYE analysis is running in transient mode (with mode=tran), .probe
statements with tran notation will be stored in netlist.e0.tr# (#: alter number) file. And

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 99

W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Edge Control

when IBIS-AMI filtering objects are attached to incident and/or probe port, transient probe
data in netlist.e0.tr# mean:
1. Probe data at incident terminals are waveforms filtered by transmitter side AMI model
(output from AMI).
2. Probe data at probe terminals are waveforms to be filtered by receiver side AMI model
(input to AMI).
To see the time domain waveforms filtered by receiver side AMI object, use:

.probe stateye eyeUF(M, N)

Where M is the port index, and N is the bit position.

Example - Controlling Edges

.param BW=0.2n
.param trf = 'BW*0.01'
.param opfreq = '1.0/BW'
.param vdd = '3.0'

.STATEYE T='BW' TRF='trf' incident_port=1 probe_port=2

+ V_HIGH='vdd' V_LOW=0.0
+ edge=2

Example - Run StatEye with Precomputed Transient Waveform

Circuit Example:
* transient

.opt post probe accurate wdf=1

.param UI=0.1n
.param trf = 'UI*0.1'
.param opfreq = '1.0/UI'
.param tstop='500*UI'

*** dummy channel ***

P1 in 0 port=1
P2 out 0 port=2

.stateye
+ T='UI' trf='trf'
+ incident_port=1 probe_port=1 2
+ mode=tran
+ STREAM_FILE='tran_wdf.tr0'
+ STREAM_SIGNAL='v(inp,inn)' 'v(outp,outn)'
+ ignore_bits=100

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 100
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Edge Control

.probe stateye eye(1) eye(2)

.end

The PrimeSim HSPICE demonstration example, ste_wdf.sp, is in $installdir/demo/

hspice/si/tran_stateye.

Separate Rise Time and Fall Time Settings

In double edge mode (Edge=2), StatEye can handle vertically asymmetric eye diagrams by
using TR/TF keywords within the .STATEYE command and the existing TRF keyword. You
must specify either TR/TF or TRF in the ..STATEYE command. If you specify both TR/TF
and TRF, the last keyword takes priority.

Multi-Edge Responses for Accurate Modeling of Nonlinearity

PrimeSim HSPICE StatEye performs initial time domain analyses with predetermined
edge patterns for each EDGE mode for each incident port (shown in Table 2). Then,
StatEye applies the best-matched pattern response in the statistical process.
Table 2 Edge value and resulting pattern
Edge Values Edge Patterns

1 (single pulse) 01000…

2 (rise/fall steps) 01111…, 1000…

4 0111…, 1000…, 01000…, 011000…

8 0111…, 1000…, 01000…, 011000...,

0111000…, 01111000…, 010111…, 010100…

For a system with highly nonlinear memory effect in drivers and receiver circuits, Figure 17
shows a comparison of resulting eye diagrams from transient analysis versus StatEye
with Edge=1, 2, 4, and 8. As the number of edge patterns increases, you can model higher
nonlinearity accurately.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 101
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Capturing Steady State Noise for Power Integrity

Figure 17 Edge mode comparisons with transient analysis-based eye diagram (top)

Capturing Steady State Noise for Power Integrity

When the power supply for a system has voltage noise, the noise causes jitter on data
transfer on the channel. Such jitter is called power supply induced jitter (PSIJ). There are
multiple causes of the PSIJ, such as clock effect on the voltage regulator, simultaneous
switching noise (SSN) and so on. Many of these are aligned with system clock period.
Since these PSIJ effects appear independently of input data pattern, PrimeSim HSPICE
treats this as steady state noise.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 102
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Capturing Steady State Noise for Power Integrity

Figure 18 System with steady state noise

When input data independent steady state noise appears, StatEye’s edge superposition
algorithm requires special handling of such noise components; otherwise effect may
be over counted because the same noise appears on the top of all the edge response
waveforms as shown in Figure 19. As a result, eye diagram may become overly
pessimistic as shown in Figure .

Figure 19 Steady state noise over counted in edge response superposition

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 103
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Capturing Steady State Noise for Power Integrity

To accurately account the steady state noise, StatEye introduces the steady_state
keyword. With this keyword, StatEye precharacterizes the steady state noise, removes the
noise effect from edge responses, and then performs edge superposition.

Figure 20 Analysis flow with the steady_state keyword

Finally, precharacterized steady state noise convolves to obtain the overall system eye
diagram accurately as shown in Figure 20.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 104
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Capturing Steady State Noise for Power Integrity

• When running StatEye with steady_state=1, the tool performs the noise
characterizations on logical low state.
• When running StatEye with steady_state=2, the tool performs the noise
characterization on low and high states.

Example - Create SSN Eye Diagram

Circuit Example:
.param pvdd=0.8 $Typical
.param f=1333.33 $signal frequency in MHz
.param vol =0 $ siganl output low voltage
.param voh=pvdd $ signal output hight voltage $$we want to add noise to
this.
.param tper='(1/f)*1e-6'
.param tr=40p tf=40p
.param dbit='tper/2' tdq='3*tper/4'

.opt post probe

.stateye
+ T='tper/2' trf='tr' $ UI, rise/fall time
+ incident_port=1 3 $ incident ports
+ probe_port=5 $ probe port
+ load_tr=ascii $ reuse edge response
*+ max_pattern = 1000000 $ pattern length
+ max_pattern = 1000 $ pattern length
*+ max_pattern = 320 $ pattern length
+ unfold_length = 500 $ length of bits for eyeUF
+ align_eye=sync $ eye alignment
+ XTALK_TYPE=DDP $ crosstalk type
+ VD_RANGE=1.6 $ vertical display range

.probe stateye eye(5)

.probe stateye eyeUF(5,0)
.probe stateye eyeT(5,200p)

* incident/probe ports
P1 1 0 port=1 pat (voh vol tper tr tr dbit sw1)
P2 2 0 port=2
P3 3 0 port=3 LFSR (pvdd 0 tdq 'tr' 'tr''1/dbit' 1 [20,14,5,2]
P4 4 0 port=4 LFSR (0 pvdd tdq 'tr' 'tr''1/dbit' 1 [20,14,5,2]
P5 5 0 port=5
P6 6 0 port=6
P7 7 0 port=7
P8 8 0 port=8

* clock pattern
.pat sw=
b0101010101010101010101010101010101010101010101010101010101010101
R=100000

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 105
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Capturing Steady State Noise for Power Integrity

.pat start= b00

.pat sw1=start sw

.end

Reusing edge responses from initial transient analysis on the original PDN circuit, StatEye
generates edge superposition-based eye diagram (PDF) of the simultaneous switching
noise and writes it in the make_ssn.stev0 file. The probability distribution matures as
you increase max_pattern in .StatEye command. Also, make_ssn.steuf0/ directory
contains reproduced SSN (time domain) waveform.
The PrimeSim HSPICE demonstration example, make_ssn.sp, is in $installdir/demo/
hspice/si/ddr_sso/MakeSSN.

Example - ADD SSN Eye Diagram

Circuit Example:
.param pvdd=0.8 $Typical
.param f=1333.33 $signal frequency in MHz
.param vol =0 $ siganl output low voltage
.param voh=pvdd $ signal output hight voltage $$we want to add noise to
this.
.param tper='(1/f)*1e-6'
.param tr=40p tf=40p
.param dbit='tper/2' tdq='3*tper/4'

.opt post probe

.stateye
+ T='tper/2' trf='tr' $ UI, rise/fall time
+ incident_port=1 3 4 $ incident ports
+ probe_port=7 $ probe port
+ load_tr=ascii $ reuse edge response
*+ max_pattern = 320 $ pattern length
*+ max_pattern = 1000 $ pattern length
+ max_pattern = 5000 $ pattern length
*+ max_pattern = 1000000 $ pattern length
+ unfold_length = 500 $ length of bits for eyeUF
+ align_eye=sync $ eye alignment
+ XTALK_TYPE=DDP $ crosstalk type
+ VD_RANGE = 1.6 $ vertical display range
+ edges = 8 $ use 8 edge patterns
+ initial_low = 'tper' $ start edge response w/ low
+ TD_IN = '10*tper' $ initial time for edge response
+ t_resolution = 400 $ horizontal resolution
+ v_resolution = 400 $ vertical resolution

.probe stateye eye(7)

.meas stateye vref7 Vref 7 tol=1e-16
.meas stateye hbox7 hbox 7 volt = 60e-3 tol=1e-16

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 106
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Duty Cycle Distortion

* clock pattern
.pat sw=
b0101010101010101010101010101010101010101010101010101010101010101
R=100000
.pat start= b00
.pat sw1=start sw

* incident/probe ports
Pvdqsd ins 0 port=1 Z0=0.0 pat (voh vol tper tr tr dbit sw1)
Pvdqd indq1 0 port=3 Z0=0.0 LFSR (pvdd 0 tdq 'tr' 'tr''1/dbit' 1
[14,5,2]
Pdqdv indq2 0 port=4 Z0=0.0 LFSR (0 pvdd tdq 'tr' 'tr''1/dbit' 1
[14,5,2]

P6 2 0 port=6
P7 3 0 port=7 $ w/o SSN
P8 4 0 port=8

.alter add SSN on victim

P7 3 0 port=7
+ jitter_file='SSN_jitter.txt'

.end

Reusing edge responses from initial transient analysis on the original nonlinear netlist,
StatEye performs edge superposition based eye diagram generation. Then second alter
adds SSN jitter on victim output.
add_ssn.stet0 : without SSN
add_ssn.stet1 : with SSN

The PrimeSim HSPICE demonstration example, add_ssn.sp, is in $installdir/demo/

hspice/si/ddr_sso/AddSSN.

Duty Cycle Distortion

StatEye analysis with the double-edge mode is capable of evaluating duty cycle distortion
StatEyeduty cycle distortionStatEyeperiodic jitter effectStatEyerandom noise effect

(DCD) effect. DCD occurs when transmitter drivers show unwanted fluctuation in their
logical threshold. When the logical threshold value shifts higher than its ideal value, a
transmitter's output has a longer low state than expected (Figure 21).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 107
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Duty Cycle Distortion

Figure 21 Illustration of the DCD effect with 10% distortion

DCD typically displays the eye diagrams. StatEye analysis with the double-edge mode
(Edge=2) is capable of evaluating this effect.

Figure 22 DCD_TYPES

(0) No DCD (1) 10% constant DCD

(2) 10% triangular DCD variation (3) 10% sinusoidal DCD variation

Port Element Syntax for Duty Cycle Distortion

An incident port element takes DCD and DCD_TYPE keywords to specify the duty cycle
distortion. See DCD Example.

Pi n1 n2 port=1 + [DCD=value] [DCD_TYPE=0|1|2|3]

DCD Aids in .STATEYE analysis; specifies peak percentage of the duty cycle distortion
(DCD). Default is 0.

DCD_TYPE Aids in .STATEYE analysis; specifies variation type. Default type for non-zero
DCD is 1 (constant).
• 0: No DCD
• 1: Constant DCD
• 2: Uncorrelated triangular DCD variation
• 3: Uncorrelated sinusoidal DCD variation

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 108
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Capturing DCD Effect on System Clock

DCD Example
* StatEye w/ 10% DCD
.param _BW=0.05n
.param trf = '_BW*0.01'
.param opfreq = '1.0/_BW'

P1 in 0 port=1
+ DCD=10 DCD_TYPE=1 $ adds 10% constant DCD
P2 out 0 port=2
X1 in out CONN

.stateye T='_BW' trf='0.1*_BW' incident_port=1 probe_port=2

+ edges=2
.opt post accurate
.probe stateye eye(2)
.print stateye eye(2)

.subckt CONN _in _out

Rtest _in _out 50 RS=1e-3 $ skin effect resistor
Cin _in 0 0.3p
Cout _out 0 0.3p
.ends
.end

Capturing DCD Effect on System Clock

When a clock that contains duty cycle distortion (DCD) is fed to the output buffer
which drives the channel, that also causes distortion in the channel output as shown in
Figure 23.

Figure 23 System with DCD-ed clock

st nd rd
Under such condition, primary bit (1 2 3 …) sequence is in narrower cycle and
nd th th
secondary bit (2 , 4 , 6 …) sequence is in wider cycle or vice versa depending on the
sign of DCD on the system clock.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 109
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Periodic and Random Jitter Effect

Figure 24 Primary and secondary bit sequences under DCD-ed clock

To capture this effect, StatEye analysis with the designated keyword DCD_mode=2
generates probability density function (PDF) eye diagrams for primary and secondary
sequence separately as shown in Figure 24 . And these two PDFs will be merged. This
method prevents inter-sequence crosstalk from being miscounted; therefore resulting eye
diagrams as well as worst bit pattern measurements accurately capture the system clock
DCD effect. The DCD_mode=2 supports constant DCD only.

Figure 25 Eye diagram under DCD-ed system clock

Periodic and Random Jitter Effect

The following sections present these topics:
• Periodic Jitter
• Random Jitter Noise Effect
• Specifying Random Jitter by Peak-to-Peak Value with BER
• Specifying Custom Jitter Distribution
• Port Element Syntax for Jitter

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 110
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Periodic and Random Jitter Effect

Periodic Jitter
Periodic jitter is an effect of unwanted periodic voltage/time interferences added to the
target signal. For example, an electromagnetic field from a power supply line adds a low
frequency sinusoidal voltage variation to the target signal. When a periodic interference
appears at an incident port, the shaping of the jitter is due to channel characteristics
before appearing at probing points (see Figure 26). StatEye analysis can evaluate this
type of effect. Incident and/or port elements take PJ and PJ_TYPE keywords to specify
periodic jitter.

Figure 26 StatEye with periodic jitter at transmitter (Tx) side

(0) No perjitter (1) 100-mV constant shift at Tx

(2) 100-mV pp uncorrelated triangular Tx-PJ (3) 100-mV pp uncorrelated sinusoidal Tx-PJ

Random Jitter Noise Effect

StatEye can also display random noise effects (see Figure 27) by adding RJ and VN
keywords to port elements. For descriptions of keywords, see Port Element Syntax for
Jitter.
StatEye analysis incident and probe ports include random Gaussian jitter when you use
the [RJ=value] keyword. In Equation 10, the RJ value specifies the root mean square
(RMS) value of the distribution (known as the standard deviation, , of the following
Gaussian distribution:

Equation 10

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 111
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Periodic and Random Jitter Effect

The VN keyword adds vertical uncertainty by specifying the standard deviation of the
Gaussian random voltage noise.

Figure 27 StatEye with random noise effects

Without random noise

Adding 1-p sec random jitter

to the probe port

Adding 5-mV voltage noise

to the probe port

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 112
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Periodic and Random Jitter Effect

Specifying Random Jitter by Peak-to-Peak Value with BER

Another convenient way to specify the random noise distribution is to use the peak-to-peak
value associated with the specified BER. Since BER is a cumulative error probability of
outside of specific “peak value” as:

Equation 11

Then, the RMS jitter value can relate to the RJPP with a specific BER by using the inverse
of complementary error function as:

Equation 12

For example, PrimeSim HSPICE allows [RJPP=value] and [RJPP_BER=value] to specify

the following:

Equation 13

Note:
An RMS Jitter specification overrides a peak-to-peak jitter specification.

Specifying Custom Jitter Distribution

The StatEye analysis convolutes custom jitter distribution described in an external data
file. A port element statement specifies the file with JITTER_FILE=filename keyword. The
jitter file can specify either time- (horizontal) wise or voltage- (vertical) wise distribution.
In the first row of the data file, “time” or “volt” can be used as x-axis variable. The second
variable name must be “PDF” or “probability.”

Time wise jitter Voltage wise jitter

Data file name jit_t.txt Jit_v.txt

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 113
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Periodic and Random Jitter Effect

Data file contents time PDF volt PDF

-8e-12 0.00443185 -0.08 3.64553e-09
-7.6e-12 0.00791545 -0.064 1.49655e-05
-7.2e-12 0.013583 -0.06 8.02981e-05
-6.8e-12 0.0223945 -0.056 0.000367141
-6.4e-12 0.0354746 -0.052 0.00143045
-6e-12 0.053991 -0.048 0.00474927
-5.6e-12 0.0789502 -0.044 0.0134367
-5.2e-12 0.110921 -0.04 0.0323946
-4.8e-12 0.149727 -0.032 0.116512
-4.4e-12 0.194186 -0.028 0.173817
-4e-12 0.241971 -0.024 0.220972
-3.6e-12 0.289692 -0.02 0.239419
-3.2e-12 0.333225 -0.016 0.221207
-2.8e-12 0.36827 -0.012 0.174769
-2.4e-12 0.391043 -0.008 0.119678
-2e-12 0.398942 -0.004 0.0755103
-1.6e-12 0.391043 4.16334e-17 0.053991
-1.2e-12 0.36827 0.004 0.0578051
-8e-13 0.333225 0.008 0.0824237
-4e-13 0.289692 0.012 0.117307
-1.31266e-27 0.241971 0.016 0.147675
4e-13 0.194186 0.02 0.159657
8e-13 0.149727 0.024 0.147323
1.2e-12 0.110921 0.028 0.115879
1.6e-12 0.0789502 0.032 0.0776747
2e-12 0.053991 0.036 0.0443684
2.8e-12 0.0223945 0.04 0.0215964
3.2e-12 0.013583 0.044 0.00895781
3.6e-12 0.00791545 0.048 0.00316618
4e-12 0.00443185 0.052 0.000953635
4.4e-12 0.00238409 0.056 0.000244761
4.8e-12 0.00123222 0.06 5.35321e-05
7.6e-12 3.9613e-06 0.076 2.47305e-08

P-element statement Px n1 n2 … Px n1 n2 …
+ JITTER_FILE=jit_t.txt + JITTER_FILE=jit_v.txt

Result (w/o jitter) w/ time wise jitter w/ voltage wise jitter

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 114
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Periodic and Random Jitter Effect

Port Element Syntax for Jitter

Pi n1 n2 port=1
+ [PJ=[val1 .. valN.] ] [PJ_TYPE=[0|1|2|3|4|5|6]]
+ [RJ=[val1 ... valN]] [RJ_MEAN=[val1...valN] ] [VN=val]
+ [RJPP=val RJPP_BER=val]
+ [VNPP=val VNPP_BER=val]
+ [JITTER_FILE=filename]

Syntax Description

RJ Random jitter aids in .STATEYE analysis. Specifies an array of the real numbers for the standard
deviation of the Gaussian random jitter. By default, StatEye adds no random timing jitter. RJ=val
overrides RJPP and RJPP_BER keywords.

RJ_MEAN Specifies an array of the real numbers for the mean values of Gaussian random jitter specified by
the RJ keyword. Default mean value for nonzero RJ is zero.

PJ Aids in .STATEYE analysis. Specifies an array of real numbers for peak-to-peak values of periodic
jitter in voltage or time. Default is 0.

PJ_TYPE Aids in .STATEYE analysis. Specifies an array of integer number for periodic jitter variation type.
Default type for nonzero PJ is 3 (uncorrelated sinusoidal voltage variation).
• 0: No periodic jitter
• 1: Constant voltage shift
• 2: Uncorrelated triangular voltage variation
• 3: Uncorrelated sinusoidal voltage variation
• 4: Constant time shift
• 5: Uncorrelated triangular time variation
• 6: Uncorrelated sinusoidal time variation

VN Random voltage noise aids in .STATEYE analysis. Specifies an array of real numbers for the
standard deviation of the Gaussian random noise. By default, StatEye adds no random noise.

RJPP Peak-to-peak jitter

RJPP_BER BER for the peak-to-peak jitter specification

VNPP Peak-to-peak jitter with voltage noise (VN)

VNPP_BER BER for the peak-to-peak jitter specification with voltage noise (VN)

JITTER_FILE Specifies custom jitter file name

Periodic Jitter Example

* StatEye w/ 100mV PJ at Tx
.param _BW=0.05n
.param trf = '_BW*0.01'
.param opfreq = '1.0/_BW'

P1 in 0 port=1

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 115
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
N-Cycle Jitter Modeling for Data and Clock Signals

+ PJ=1e-1 PJ_TYPE=2
P2 out 0 port=2
X1 in out CONN

.stateye T='_BW' trf='0.1*_BW' incident_port=1 probe_port=2

.opt post accurate
.probe stateye eye(2)
.print stateye eye(2)

.subckt CONN _in _out

Rtest _in _out 50 RS=1e-3 $ skin effect resistor
Cin _in 0 0.3p
Cout _out 0 0.3p
.ends
.end

N-Cycle Jitter Modeling for Data and Clock Signals

In the /hspice/StatEye_clk_skew/nUI_sj.sp directory, use the following setting in the
sj.sp and n_sj.sp files:
• UI=1n
• Sinusoidal jitter (SJ) with freq=1e8 (1/10UI) and phase=90 degrees
• Long 4.5UI skew to achieve maximum DQ and DQS jitter difference (jitter freq=1e8)
1. The netlists settings:
sj.sp: Adds SJ to DQ and DQS. Alter-0 has DQ/DQS skew within a UI
(CLK_skew=0.5*UI) and alter-1 has 4.5*UI skew
n_sj.sp: Same as netlist above but without jitter to compare with injected jitter

2. Use WaveView to load the presimulated results session:

%wv -x view_eye.sx

First WaveView (alter_0): port-3 shows no jitter, port-4 (bottom) has jitter. This is
because DQ and DQS have same in-phase jitter canceling out.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 116
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
N-Cycle Jitter Modeling for Data and Clock Signals

Figure 28 First WaveView (alter_0)

In WaveView (alter_1), port-3 shows 2x the jitter of port-4. This is because the DQ and
DQS jitter is out of phase (4.5*CLK_skew).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 117
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
N-Cycle Jitter Modeling for Data and Clock Signals

Figure 29 WaveView (alter_1)

3. Viewing the unfolded eyes waveforms of port 4: eyeUF(4,0):

% wv -x view_clk.sx

◦ Top: skew=0.5UI w/ and w/o jitter

◦ Bottom: skew=4.5UI w/ and w/o jitter
Since jitter sequence is applied from the beginning of the sequence, jitter amount of the
two yellow circles match, and orange circles match. However, yellow versus orange do
not match.
Therefore, jitter on clock at each timing is different between skew=0.5*UI and
skew=4.5*UI.
4. Jitter relationship between DQ and DQS, using unfolded waveforms from port-3 and
port-4: alter-0 (0.5UI skew):
% wv -x view_skew0.5.sx

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 118
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
N-Cycle Jitter Modeling for Data and Clock Signals

Top: DQ (w/ and w/0 jitter. Bottom: DQS (w/ w/o jitter)
Since skew is within a UI, you can see the same jitter at each edge transition between
DQ and DQS. Therefore, if you use the DQS as trigger, the jitter is canceled.

Figure 30 Jitter relationship between DQ and DQS

alter-1 (4.5UI skew):

%wv -x view_skew4.5.sx

Top: DQ (w/ and w/o jitter ) bottom DQS (w/ and w/o jitter)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 119
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Pre-Emphasis and De-Emphasis

Since skew is within more than UI, the different jitter is observed at each edge
transition between DQ and DQS.

Figure 31 Edge transition between DQ and DQS

Pre-Emphasis and De-Emphasis

Eye diagrams enable you to estimate and evaluate high-speed serial data links. As the
data rate increases, however, the design goal of sufficient eye opening becomes difficult
to achieve. In over-10 gigabit-per-second (bps) designs, the loss of high frequency
StatEyepre-emphasis,

components in the rising and falling edges of the digital bit streams becomes significant.
de-emphasisStatEyeunfolding waveforms

To compensate for the loss, preshaping the incident waveform to increase the amount
of high frequency components becomes essential. You can use techniques called
pre-emphasis and de-emphasis for this purpose. Pre-emphasis adds high frequency
components to the rising and falling edges of the incident waveforms as shown in purple in
Figure 32. The de-emphasis voltage swing from the waveform portions specifies time after
the rising and falling edges. The red waveform in Figure 32 illustrates that the rising and
falling edges are also de-emphasized.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 120
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Pre-Emphasis and De-Emphasis

Figure 32 Emphasized pulse sources: (green) original waveform; (purple) 10%

pre-emphasized pulse; (red) 10% de-emphasized pulse

De-emphasized pulse

You can achieve clearer eye openings by using these emphasis techniques for the same
system with the same digital symbol voltage level as shown in Figure 33.

Figure 33 The effect of incident pre-emphasis: (left) original eye diagram; (right) eye diagram
with additional 30% of pre-emphasis to the incident port

Figure 34 shows how the amount of pre- or de-emphasis represents a fraction of 1. For
example, an emphasis_level = -.5 (or 50%) would correspond to 6-dB transmitter de-
emphasis.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 121
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Pre-Emphasis and De-Emphasis

Figure 34 Effect of emphasis on a pulse source

To add pre-emphasis and de-emphasis, apply the keywords Emphasis_Level and

Emphasis_Time to the port-element, the transient voltage and the current sources.

The following section discuss these topics:

• Syntax for P-Element and Transient V and I Sources
• Usage Examples

Syntax for P-Element and Transient V and I Sources

Port Element Syntax

Pxxx node1 node2 [node3 node4 node5] port=portindex

+ [Z0=val] $Port Impedance
+ [DC mag] [AC mag phase] [HBAC mag phase] $Voltage/Power Info
+ [HB mag phase harm tone modharm modtone]
+ [transient_waveform]
+ [ENCODE=DW8B10B [RD_INIT=0|1]]
+ [ENCODE=DW64b66b [TXC=txcdata]]
+ [ENCODE=DW128b130b|DW128b132b [HEAD=DATA|CONTROL]]
+ [TRANFORHB=[0|1]] [DCOPEN=[0|1]]
+ [power=[0|1|2|W|dbm]] $Power Switch
+ [POWERSIN=[0|1|2|W|dbm]]
+ [POWERHB=[0|1|2|W|dbm]]
+ [POWERHBAC=[0|1|2|W|dbm]]
+ [POWERSNAC=[0|1|2|W|dbm]]
+ [RDC=val] [RAC=val] $ Source Impedance Overrides
+ [RHBAC=val] [RHB=val] [RTRAN=val]

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 122
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Pre-Emphasis and De-Emphasis

+ [Emphasis_Level=val] [Emphasis_Time=val] $ Emphasis

+ $ **** DFE ****
+ [DFE_tap=n1 n2 … nN DFE_coeff=val1, val2, … valN]
+ [DFE_VTH=val]
+ [DFE_CLK=0|1]
+ [DCD=val] [DCD_TYPE=0|1|2|3] $ Duty Cycle Distortion
+ [PJ=val] [PJ_TYPE=0|1|2|3|4|5|6] $ Period Jitter
+ [RJ=val1, [val2, ... valN]] [VN=val] $ Rand Jitter Stateye
+ [IBIS_AMI_FILE=filename IBIS_AMI_MODEL=modelname]
+ [AMI_OBJ=filename] [AMI_PARAM=filename] [AMI_CDR=0|1]$AMI
+ [AMI_CORNER=idx]$AMI
+ [AMI_Remove_DC_Offset = 0|1]
+ [AMI_Recover_DC_Offset = 0|1]
+ [MODULATION=PAM4]
+ [PAM4_MAPPING=string]
+ [AMI_PARAM_xx = string|quoted string]
+ [IBIS_AMI_VRANGE_MODE = 1|2]
+ [IBIS_AMI_USE_RAMP = 0|1]
+ [IBIS_AMI_USE_VRANGE = 0|1]
+ [IBIS_AMI_USE_ZSRC = 0|1]

Pulse Voltage and Current Source Syntax

Vxxx n+ n- PU[LSE] [(]v1 v2 [td [tr [tf [pw [per]]]]] [)]

+ [Emphasis_Level=val] [Emphasis_Time=val]
Ixxx n+ n- PU[LSE] [(]v1 v2 [td [tr [tf [pw [per]]]]] [)]
+ [Emphasis_Level=val] [Emphasis_Time=val]

SIN Voltage and Current Source Syntax

Vxxx n+ n- SIN [(] vo va [freq [td [q [j]]]] [)]

+ [PERJITTER=val] [SEED=val]] [Emphasis_Level=val]
+ [Emphasis_Time=val]
Ixxx n+ n- SIN [(] vo va [freq [td [q [j]]]] [)]
+ [PERJITTER=val] [SEED=val] [Emphasis_Level=val]
+ [Emphasis_Time=val]

PWL Voltage and Current Source Syntax

Vxxx n+ n- PWL [(] t1 v1 [t2 v2 t3 v3…] [R [=repeat]]

+ [TD=delay] [)] [Emphasis_Level=val]
+ [Emphasis_Time=val]
Ixxx n+ n- PWL [(] t1 v1 [t2 v2 t3 v3…] [R [=repeat]]+ [TD=delay]
[)][Emphasis_Level=val]
+ [Emphasis_Time=val]

Exponential Voltage and Current Source Syntax

Vxxx n+ n- EXP [(] v1 v2 [td1 [t1 [td2 [t2]]]] [)]

+ [Emphasis_Level=val] [Emphasis_Time=val]
Ixxx n+ n- EXP [(] v1 v2 [td1 [t1 [td2 [t2]]]] [)]
+ [Emphasis_Level=val] [Emphasis_Time=val]

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 123
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Pre-Emphasis and De-Emphasis

PAT Voltage and Current Source Syntax

Vxxx n+ n- PAT [(] vhi vlo td tr tf tsample data [RB=val]

+ [R=repeat] [)] [Emphasis_Level=value]
+ [Emphasis_Time=val]
Ixxx n+ n- PAT [(] vhi vlo td tr tf tsample data [RB=val]
+ [R=repeat] [)] [Emphasis_Level=val]
+ [Emphasis_Time=val]

Pseudo-Random Bit Source Syntax

Vxxx n+ n- LFSR [(] vlow vhigh tdelay trise tfall rate seed
+ [[] taps []] [rout=val] [)] [Emphasis_Level=val]
+ [Emphasis_Time=val]
Ixxx n+ n- LFSR [(] vlow vhigh tdelay trise tfall rate seed
+ [[] taps []] [rout=val] [)][Emphasis_Level=val]
+ [Emphasis_Time=val]

Single Frequency FM Voltage and Current Source Syntax

Vxxx n+ n- SFFM [(] vo va [fc [mdi [fs]]] [)]

+ [Emphasis_Level=val] [Emphasis_Time=val]
Ixxx n+ n- SFFM [(] vo va [fc [mdi [fs]]] [)]
+ [Emphasis_Level=val] [Emphasis_Time=val]

Single Frequency AM Voltage and Current Source Syntax

Vxxx n+ n- AM [(] sa oc fm fc [td] [)]

+ [Emphasis_Level=val] [Emphasis_Time=val]
Ixxx n+ n- AM [(] sa oc fm fc [td] [)]
+ [Emphasis_Level=val] [Emphasis_Time=val]

Usage Examples
• Example 1: Pre-Emphasis
• Example 2: De-Emphasis

Example 1: Pre-Emphasis
In this example, the voltage source Ve adds to the
original pulse waveform to emphasize the rising and falling edges of the original pulse
(Figure 35).
.param Vmag = 1.0
.param T = 3.33n
.param trf = 'T/20'
.param td = 4.5n

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 124
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Pre-Emphasis and De-Emphasis

Ve e 0 Emphasis_Level =0.1 Emphasis_Time ='T/4'

+ pulse (0 mag td trf trf 'T/2-trf' T)

Figure 35 Pre-emphasis added by specifying positive Emphasis_Level

Example 2: De-Emphasis
In this example, the voltage source Vde applies de-emphasis by adding
to the original waveform. A negative
Emphasis_Level specifies the de-emphasis. Figure 36 shows the pulse waveform after
you reduce Emphasis_Time in magnitude.
.param Vmag = 1.0
.param T = 3.33n
.param trf = 'T/20'
.param td = 4.5n

Vde de 0 Emphasis_Level =-0.1 Emphasis_Time ='T/4'

+ pulse (0 mag td trf trf 'T/2-trf' T)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 125
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Built-In DFE Function for Rx Port Element in StatEye Analysis

Figure 36 De-emphasis applied by specifying negative Emphasis_Level

Built-In DFE Function for Rx Port Element in StatEye Analysis

Figure 37 shows a block diagram of the decision feedback equalizer (DFE). The system
applies the waveform filter represented in Equation 14 to equalize the output bit stream
waveform.

Figure 37 DFE block diagram

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 126
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Built-In DFE Function for Rx Port Element in StatEye Analysis

Equation 14

Where:
• xk is the DFE input voltage
• yk is the DFE output voltage
• d is the feedback coefficient
• k is the sample index in UI (bit position)
• m is the relative bit position to be fed back and
• vth is the logical threshold of the system.
An example netlist is provided in the PrimeSim HSPICE demo package in ../demo/
hspice/si/SNPS_AMI/biDFE.sp. This example uses the AMI model which represents a
transmitter-side feed-forward equalizer and a built-in DFE as the receiver-side equalizer.
The following figure compares eye diagrams with and without the DFE effect.

Figure 38 Eye diagrams without and with DFE

Without DFE With DFE

Applying External Clock to the Built-In DFE Model

The DDR5 next generation memory interface can use equalizer(s) in transmitter and/
or receiver circuitries. For receiver, DFE will be deployed. Since the DDR system uses
designated clock lane (DQS), the DFE implementation for DDR simulation will take the
external clock signal into account in its slicer process.
For DFE’s external clock input, set DFE_CLK=1 for the probe side port element.
The demo netlist in $installdir/demo/hspice/si/BiDFE/clk_fwd.sp provides an
example of this feature.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 127
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Built-In DFE Function for Rx Port Element in StatEye Analysis

.StatEye ....
+ incident_port=1, 2
+ probe_port=3, 4
+ Clock_Port=(4, 0.0)
+ Clock_Mode =2

In the preceding statements in the example, the StatEye command sets up one data (DQ)
lane and one clock (DQS) lane. The Clock_Port command specifies port-4 as the probing
point of the DQS signal with trigger at 0V. And Clock_Mode=2 sets the clock jitter injection
as a numerical perturbation sequence. Then, the following statements add timing variation
(PJ) with different frequency to port-1 (DQ input) and port-2 (DQS input).
P1 DQ0_n 0 port=1
+ pj = 10p
+ pj_type=6
+ pj_freq=1e8
+ LFSR (0 1 0 _TRF _TRF _BPS 1 [21,1] )

* clock : 90 degree phase shifted

P2 DQS_n 0 port=2
+ pj = 10p
+ pj_type=6
+ pj_freq=1e6
+ PAT (0.5 -0.5 '0.5*_UI' _TRF _TRF _UI b01010101 R=1000000)

Finally, to add built-in DFE at receiver side of the DQ lane, the following statements are
used:
Pdq0_f dq0_f 0 port=3
+ DFE_TAP=1 2 DFE_COEFF=0.02 0.01 + DFE_VTH=0.16
+ DFE_CLK=1

Figure 39 Results of the StatEye simulation in clk_fwd.sp

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 128
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Perturbation Analysis with StatEye

The results show that a non-ideal clock signal is considered in the DFE slicer.

Perturbation Analysis with StatEye

With the PTB_DATA keyword in StatEye analysis, PrimeSim HSPICE loads a perturbed
set of edge responses. When the PTB_DATA keyword is specified, PrimeSim HSPICE
computes a perturbation transfer function at each voltage by numerical difference between
nonperturbed and perturbed data. Then StatEye applies vertical fluctuation on the
resulting PDF eye diagram. The contents and file format of the perturb data directory are
the same as regular edge response data generated by StatEye analysis with the SAVE_TR
keyword. Therefore, you can simply run one StatEye analysis with an arbitrary parameter
set to obtain the perturbed edge response data.
The following example shows the supply voltage fluctuation effect on the eye diagram.
In the netlist, the initial run computes an eye diagram with a nominal supply voltage.
Then the first .alter block changes the supply voltage to 90 percent of the nominal level
and saves the initial edge responses under the perturb.save1/ directory. Finally, the
second .alter block sets a nominal supply voltage again with PTB_DATA=perturb.save1
to perform perturbation analysis using the perturbed edge response generated in the
previous alter.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 129
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Unfolding Statistical Eye Diagrams to Waveforms

Example Netlist: perturb.sp Initial Run With Nominal VDD

* VDD Perturbation Analysis

.param BW=0.2n
.param trf = 'BW*0.01'
.param opfreq = '1.0/BW'
.param vdd_nom = '3.0'
.param vdd = vdd_nom

P1 in 0 port=1
P2 out 0 port=2

.include 'buff.inc'
x1 in m vdd buffer
x2 m mm vdd buffer First alter with 90-percent VDD
x3 mm out connector

.opt post accurate

.StatEye T='BW' TRF='trf'
+ incident_port=1 probe_port=2
+ V_HIGH='vdd' V_LOW=0.0
+ edges=2 save_tr=ascii
.probe stateye eye(2)

.alter $ 90% vdd

.param vdd = 'vdd_nom*0.9'

.alter $ perturb analysis

.param vdd = 'vdd_nom'
Second alter with perturbation analysis
.StatEye T='BW' TRF='trf'
+ incident_port=1 probe_port=2
+ V_HIGH='vdd' V_LOW=0.0
+ edges=2
+ PTB_DATA=perturb.save1
* load alter-1 result
* as perturbation data .end

Unfolding Statistical Eye Diagrams to Waveforms

The following describes the method for.STATEYE analysis to unfold resulting eye diagrams
to output partial time domain waveforms when you provide a specific pattern to the
StatEyeunfolding waveforms

incident port.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 130
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Unfolding Statistical Eye Diagrams to Waveforms

Under this condition, .STATEYE allows the following .PROBE and .PRINT commands to
extract partial time domain waveforms:
.PROBE eyeUF(port_index, bit_index)
.PRINT eyeUF(port_index, bit_index)

PrimeSim HSPICE extracts partial time domain waveforms about the bit stream
location, bit_index, from the resulting eye diagram at a probing port when you specify
port_index. The length of the partial time domain waveform is 100 bits. When -1 is
specified as the bit_index, PrimeSim HSPICE outputs the time domain waveform of the
entire bit sequence.
Note:
• Unfolding of eye diagrams applies only to pattern-specific eye diagram
generation.
• When using -1 as the bit_index, it is strongly recommended to use the
MAX_PATTERN keyword to limit the length of bits, to avoid a very large output
file.
Output
The .PROBE command outputs a binary time domain data file readable to WaveView. The
unfolded time domain files are stored under netlist.steuf#. The naming convention of
the binary data file is:
netlist.steuf#/portN_M

The .PRINT command outputs the text time domain data file under the
netlist.printSte0 directory. The naming convention of the text data file is:
Netlist.printSte#/data_v_t_M_

Also, the .PRINT command produces a script file which you can upload to the gnuplot
program:
Netlist.printSte#/eyeUF_N_M

where #, M, and N are the alter index number, probe port index number, and bit index
number, respectively.
Example: Unfolding StateEye Diagram
An example netlist is provided in the PrimeSim HSPICE demo package in demo/hspice/
si/stateye_unfold.sp. The example PrimeSim HSPICE netlist performs .STATEYE
analysis with an unfolding time domain waveform around the 500th bit (Figure 40).
*
* StatEye Unfolding Example
*

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 131
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Unfolding Statistical Eye Diagrams to Waveforms

.param BW=0.2n
.param trf = 'BW*0.01'
.param opfreq = '1.0/BW'
.param vdd = '3.0'
P1 in 0 port=1 LFSR 0 'vdd' 0 'trf' 'trf''opfreq' 1 [14,9,5,2]
P2 out 0 port=2
x1 in m vdd INV
x2 m mm vdd INV
x3 mm out CONN
.opt post accurate
.StatEye T='BW' TRF='trf' incident_port=1 probe_port=2
+ V_HIGH='vdd' V_LOW=0.0
+ edges=2
.print stateye eye(2) eyeUF(2,500)
.probe stateye eye(2) eyeUF(2,500)
.model nch nmos level=54
.model pch pmos level=54
Vd vdd 0 'vdd'
.subckt inv in out vdd
mn1 out in 0 0 nch l=0.25u w=30u
mp1 out in vdd vdd pch l=0.25u w=60u
.ends
.subckt CONN _in _out
Rtest _in _out 50 RS=1e-3 $ skin effect resistor
Cin _in 0 0.3p
Cout _out 0 0.3p
.ends
.end

Figure 40 StatEye result with unfolded waveform around 500th bit

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 132
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Multiple Data Rate Eye Diagram Analysis with StatEye

Multiple Data Rate Eye Diagram Analysis with StatEye

StatEye allows you to inject incident signals with different data rates using the optional
T_Factor keyword.

T_factor = n1 [, n2, ... nN] multiplies the unit interval of the incident bit stream at
each port.
Example:
.param _UI=1n
.param _TRF='0.05*_UI'

.stateye
+ T = '_UI'
+ TRF = '_TRF'
+ incident_port= 1 2
+ probe_port= 3 4
+ T_Factor= 1 2 $ UI x2 for 2nd incident
+ save_tr=ascii
+ TD_nUI=4
+ align_eye=sync

This example injects the bit stream with 1 ns unit interval at port 1 and the bit stream with
2 ns unit interval at port 2.
Note:
TD_nUI uses the unscaled unit interval.

Accelerating StatEye’s Initial Transient with Process Distribution

and Multi-Threading
StatEye deploys distributed processing (DP) for the initial edge response transient. When
the target is a highly complex nonlinear system with crosstalk couplings that require
multiple incidents with multiple edge responses, applying process distribution (with the

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 133
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
Known Limitation

-dp command-line argument) in conjunction with multi-threading (or MT, with the -mt
command-line argument) can significantly reduce the overall runtime.
For more information on invoking PrimeSim HSPICE with DP, see Running Multiple
Simulations in the PrimeSim HSPICE User Guide: Basic Simulation and Analysis.
For more information on invoking PrimeSim HSPICE with MT, see Running Multithreaded
Simulations in the PrimeSim HSPICE User Guide: Basic Simulation and Analysis.

Known Limitation
.StatEye assumes an equivalent to a bit stream response that you can obtain by
superposition of single bit responses. This assumption may not apply when you have
strongly nonlinear components in the system under test.

Troubleshooting Output Warnings

When running a StatEye test case, you may see the following messages:
Warning: (line 60) port 1 is not specified as .stateye probe_port
Warning: [STAT_EYE error] invalid probe port index for stateye
measurement

For StatEye analysis, .PRINT, .PROBE, and .MEASURE statements are taken at the probe
port. The statements must match the "probe_port" defined in the .stateye statement.
For example if the .STATEYE command is:
.stateye T='BW' trf='0.1*BW' incident_port=1 probe_port=2

You should identify the probe port in your output statement. The following example shows
valid .PRINT, .PROBE, and .MEASURE statements for this analysis:
.print stateye eye(2) * eye diagram of specified port
.probe stateye eyeV(2,V) * cross-section of eye at V
.print stateye bathtubT(2,T) * port and time for bathtub curve
.probe stateye ber(2) * output BER plot for port
.measure stateye res_name Veye 2 * maximum vertical eye measurement

PrimeSim HSPICE issues no warnings when the port listed in the output statement
matches the probe_port specified in the .STATEYE command.

References
[1] IBIS Version 5.0 Specification:http://eda.org/pub/ibis/ver5.0/ver5_0.txt.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 134
W-2024.09
 Feedback
Chapter 3: Statistical Eye Analysis
References

[2] IBIS (I/O Buffer Information Specification) home page: http://www.eigroup.org/ibis/

default.htm

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 135
W-2024.09
 Feedback

4
Time Domain Reflectometry

This chapter discusses using digitized time domain reflectometry (TDR) in conjunction with
PrimeSim HSPICE to select design components.

The following sections discuss these topics:

• Optimizing TDR Packaging
• Performing a TDR Simulation in PrimeSim HSPICE
• Basic TDR Impedance Analysis
• Testbench Netlist Overview
• Input and Output Ports
• Selecting the Risetime of the Incident Wave
• Simulating the Example DUTs
• Troubleshooting: TDR Zo Result Starts at Negative Value
• Differential TDR Example Netlist
• References

Optimizing TDR Packaging

The packaging plays an important role in determining the overall speed, cost, and
reliability of a system. With small feature sizes and high levels of integration, a significant
portion of the total delay is the time required for a signal to travel between chips.
Time domain reflectometry (TDR) is the closest measurement to actual digital component
functions. It provides a transient display of the impedance versus time for pulse behavior.
timedomain reflectometryTDR (time domain reflectometry)impedanceversus timetimeversus impedance
PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 136
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Performing a TDR Simulation in PrimeSim HSPICE

Performing a TDR Simulation in PrimeSim HSPICE

When performing a signal integrity analysis, a bit stream is often the first stimulus that
you use to validate the system. If you obtain poor results, it can be useful to first perform
a TDR simulation on either the system or the individual models to find unexpected
impedance discontinuities or mismatches. A TDR simulation measures the reflections that
result from an ideal step edge traveling through a transmission medium.
The following is an example PrimeSim HSPICE testbench to demonstrate TDR
simulations with a variety of classic discontinuities. The included testbench is completely
functional and uses single-ended lossless transmission lines and lumped elements
to demonstrate the analysis but you can be easily adapt use of the more complex W-
element or S-parameter models. The last section of this description supplies a differential
equivalent example netlist.
PrimeSim HSPICE provides a number of methods to model transmission lines and media.
These include:
• Lumped models with R-, L-, and C-elements connected by lossless (T-element)
transmission lines
• S-parameter data (S-element), either from instrument measurements or extracted
during previous PrimeSim HSPICE simulations
• The frequency-dependent lossy W-element. The W-element has several different
modeling options:
◦ The PrimeSim HSPICE field-solver based on materials and geometries
◦ RLGC per unit length matrices
◦ S-parameter data called by the W-element SMODEL parameter
◦ A network of behavioral sources
A TDR simulation measures the reflections that result from an ideal step edge traveling
through a transmission medium. This allows you to analyze impedance discontinuities and
mismatches. The medium that you study is often an interconnect topology, and could be
a combination of circuit board traces, cables, connectors and even the wire and bonding
between an IC and its package. The following uses an example PrimeSim HSPICE
testbench to demonstrate TDR simulations with a variety of classic discontinuities.
An interconnect simulation often contains a combination of these methods. Singularly or
collectively you can refer to them as the Device Under Test (DUT). It is common for the
manufacturer to provide the DUT model, or portions of it (as in the case of a connector) or
it is the output by a signal-integrity analysis tool which creates a model by analyzing the
packaging or layout data.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 137
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Basic TDR Impedance Analysis

Basic TDR Impedance Analysis

The resulting waveform of the TDR simulation is the combination of the incident wave and
reflections that occur when the step edge encounters impedance variations.
This example netlist uses the following names and equations:
• Vincident: An ideal copy of the step edge source you use as a reference
• Vmeasured: Voltage at the input to the DUT which is the combination of incident and
reflected waves
• Vreflect: The reflected portion only which you obtain by subtracting the incident portion
from the measured voltage
• Impedance (Zo): Calculated from the above parameters:

Use this calculation to graph the response of the transmission line in unit of characteristic
impedance which is often more useful than plotting the voltage of the TDR measured
wave. In at least one case, however, the open circuit termination is more useful to plot the
measured voltage. Note that due to the voltage divider effect, the measured voltage is 1/2
of the incident wave.
The following examples demonstrate classic impedance mismatches and discontinuities:
• Example 1: Short Circuit Termination
• Example 2: Open Circuit Termination
• Example 3: Mismatched Load Termination
• Example 4: Shunt Capacitance Discontinuity
• Example 5: Series Inductance Discontinuity

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 138
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Basic TDR Impedance Analysis

Example 1: Short Circuit Termination

Figure 41 Short circuit termination

Tp

Example 2: Open Circuit Termination

Figure 42 Open circuit termination

Tp

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 139
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Basic TDR Impedance Analysis

Example 3: Mismatched Load Termination

Figure 43 Mismatched load termination

Z0 >

ZI<>Z0

<

Example 4: Shunt Capacitance Discontinuity

Figure 44 Shunt capacitance discontinuity

Z0 Z0

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 140
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Testbench Netlist Overview

Example 5: Series Inductance Discontinuity

Figure 45 Series inductance discontinuity

Z0 Z0

Testbench Netlist Overview

In creating a PrimeSim HSPICE TDR test bench, this example uses the “P” (port) element
for both the input sources and output terminations because it can:
• Represent either single-ended or differential signals, and
• Provide both a source voltage and an impedance value in a single construct
The first example simulates a single-ended TDR example. Additional syntax for a
differential testbench and DUT appears in Differential TDR Example Netlist.
Four main sections comprise the testbench:
1. The port that represents the source voltage and impedance for the incident wave
2. The Device Under Test (DUT)
3. An output termination port
4. A reference port and matching termination port
The reference port functions to create an ideal version of the incident pulse which is
separate from the DUT. The tool uses this port to calculate the reflected portion of the TDR
measurement.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 141
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Input and Output Ports

Input and Output Ports

The P-element is the basis for the input and output ports. As you see above, you can
visualize it as an impedance and a voltage source in series:

Figure 46 PrimeSim HSPICE port element

P1 input1 0 z0=50 port=1 pulse (0 1.5 100p 40p 40p)

In the example shown in Figure 46, a port named “P1” connects between “input1” and
ground and has a characteristic impedance of 50 ohm. It supplies a single 40 ps rising
edge to input1 after a 100 ps delay. If you omit the voltage source argument, the port
serves as a simple termination. Each port must have a unique sequential numerical
designation in the “port=” argument. You also use the P-element in the LIN analysis to
extract S-parameters.
For more information on port elements, see Port Element in the PrimeSim Continuum User
Guide: Elements.

Selecting the Risetime of the Incident Wave

One way to determine the risetime of the incident wave is using the knee frequency.
To guarantee reliable operation of a digital system, you should develop and verify the
circuit design for frequencies below the knee frequency. Most energy in digital pulses
concentrates below the knee; the behavior of a circuit at the knee frequency determines its
ability to process a step edge. The knee frequency for any digital signal relates to the rise
and fall time of its digital edges rather than its clock rate.
You can calculate the risetime based the desired knee frequency and this calculation
depends on the basis for the measurement: a 10-90% or 20-80% measurement.
• For a 10-90% measurement, Trise = .5/Fknee
• For a 20-80% measurement, Trise = .35/Fknee
For example, the risetime measured at 20-80% needed for a 10 GHz knee frequency:

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 142
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Simulating the Example DUTs

In a PrimeSim HSPICE trapezoidal pulse source function, specify the risetime from
0-100%.
So, for a 20-80% risetime, multiply the desired risetime by 1.67 to obtain the “tr” parameter
and for a 10-90% measured risetime, multiply the desired risetime by 1.25. For example:

Simulating the Example DUTs

The sample netlist includes three separate example DUTs:
1. An inductive discontinuity
2. A capacitive shunt discontinuity
3. A transmission line followed by a shunt capacitance discontinuity, an impedance
mismatch in the second transmission line and a series inductance
The design in Figure 47 shows a combination of behaviors. An example netlist for
simulation is provided following Figure 47 and the results are shown in Figure 48.

Figure 47 Testbench and DUT

The example testbench below, “An example PrimeSim HSPICE TDR Netlist with DUT3
Uncommented”, is completely functional and uses simple lossless transmission lines
and lumped elements to demonstrate the analysis, but it could just as easily use more
complex W-element or S-parameter models. A differential example follows at the end of
this section, and contains appropriate modifications to the ports, DUTs and equations.
An Example PrimeSim HSPICE TDR Netlist with DUT3 Uncommented
** Single-ended TDR example **
*
.opt post probe runlvl=5

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 143
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Simulating the Example DUTs

.tran 0.1n 20n

*
.param zref=50.0 vlo=0 vhi=1 td=1n tr=.2n tf=.2n pw=50n per=200n
*
* P1 is the input port for the incident wave. It provides
* both an impedance and a pulse source
*
Psource dut_in 0 z0=zref port=1 pulse(vlo vhi td tr tf)
*
* DUT1 - An inductive discontinuity
*T1 dut_in 0 node1 0 Z0=50 td=1n
*L1 node1 node2 5e-9
*T2 node2 0 dut_out 0 Z0=50 td=1n
*
* DUT2 - A capacitive shunt discontinuity
*T1 dut_in 0 node1 0 Z0=50 td=1n
*C1 node1 0 1e-12
*T2 node1 0 dut_out 0 Z0=50 td=1n
*
* DUT3 - Series inductance - shunt capacitance
* with impedance mismatch in the second transmission
* line
T1 dut_in 0 node1 0 Z0=50 td=1n
C1 node1 0 2e-12
T2 node1 0 node2 0 Z0=75 td=1n
L1 node2 dut_out 5e-9
*
* Pterm is the output port of the DUT and provides termination
*
Pterm dut_out 0 z0=50 port=2
*
* Pref and Prefterm are a reference source and termination to use
* as the ideal incident wave in the reflection calculation
*
Pref vref 0 z0=zref port=3 pulse(vlo vhi td tr tf)
Prefterm vref 0 z0=zref port=4
*
* Calculations of the reflected wave and impedance
*
.probe tran vincident = par('v(vref)')
.probe tran vmeasured = par('v(dut_in)')
.probe tran vreflect = par('vmeasured-vincident')
.probe tran Z0 = par('zref*(vincident+vreflect)/(vincident-vreflect)')
*
.end

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 144
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Troubleshooting: TDR Zo Result Starts at Negative Value

Procedural Notes:
1. Increase the default RUNLVL from 3 to 5 to observe the subtleties in the discontinuities
of the waveforms.
2. Alternately, you can specify: .option runlvl=6 delmax=10p. Since the pulse is
a single event (non-repeating), you can omit the pw (pulse width) and per (period)
arguments, although a warning message is reported. Since delmax forces the
maximum time step length, it may sacrifice the efficiency of the dynamic time
step control. Therefore, unless you have a specific need for dense time points, do
not specify delmax. Uncomment only one DUT at a time to observe the example
discontinuities and impedance mismatches.

Figure 48 Result of the DUT3 Analysis

Troubleshooting: TDR Zo Result Starts at Negative Value

You may see TDR Zo results start at negative 50 ohm (single-ended) or negative 100 ohm
(differential). This is due to a change in the precision of expression evaluation in PrimeSim
HSPICE. You can safely ignore this, since the TDR result following the incident wave is the
same. If you would rather see it to start at zero, add a very small value (such as 1e-20) to

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 145
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
Differential TDR Example Netlist

the denominator in the Zo equation. Using the formula in Basic TDR Impedance Analysis,
modify the equation as follows:
.probe tran Zo = par('zref*(vincident+vreflect)/
(vincident-vreflect+1e-20)')

The Zo result starts at zero ohm and can be shared without the need to explain the initial
starting impedance.

Differential TDR Example Netlist

** Differential TDR example **
*
.opt post probe runlvl=5
.tran 0.1n 20n
.param zref=50.0 vlo=0 vhi=1 td=1n tr=.2n tf=.2n
* P1 is the differential input port for the incident wave.
* It provides both an impedance and a pulse source
*
Psource dut_inp dut_inn 0 z0=zref port=1 pulse(vlo vhi td tr tf)
*
* DUT1 - An inductive discontinuity
* T1P dut_inp 0 node1p 0 Z0=50 td=1n
* T1N dut_inn 0 node1n 0 Z0=50 td=1n
* L1 node1p node2p 5e-9
* L2 node1n node2n 5e-9
* T2P node2p 0 dut_outp 0 Z0=50 td=1n
* T2N node2n 0 dut_outn 0 Z0=50 td=1n
*
* DUT2 - A capacitive shunt discontinuity
* T1P dut_inp 0 node1p 0 Z0=50 td=1n
* T1N dut_inn 0 node1n 0 Z0=50 td=1n
* C1 node1p 0 1e-12
* C2 node1n 0 1e-12
* T2P node1p 0 dut_outp 0 Z0=50 td=1n
* T2N node1n 0 dut_outn 0 Z0=50 td=1n
*
* DUT3 - A shunt capacitance followed by an impedance mismatch
* and a series inductance
T1P dut_inp 0 node1p 0 Z0=50 td=1n
T1N dut_inn 0 node1n 0 Z0=50 td=1n
C1 node1p 0 2e-12
C2 node1n 0 2e-12
T2P node1p 0 node2p 0 Z0=75 td=1n
T2N node1n 0 node2n 0 Z0=75 td=1n
L1 node2p dut_outp 5e-9
L2 node2n dut_outn 5e-9
*
* Pterm is the dut_output port of the DUT and provides termination
*
Pterm dut_outp dut_outn 0 z0=50 port=2

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 146
W-2024.09
 Feedback
Chapter 4: Time Domain Reflectometry
References

*
* Pref and Prefterm are a reference source and termination to use
* as the ideal incident wave in the reflection calculation
*
Pref refp refn 0 z0=zref port=3 pulse(vlo vhi td tr tf)
Prefterm refp refn 0 z0=zref port=4
*
* Calculations of the reflected wave and impedance
*
.probe tran vincident = par('v(refp)-v(refn)')
.probe tran vmeasured = par('v(dut_inp)-v(dut_inn)')
.probe tran vreflect = par('vmeasured-vincident')
.probe tran Zdiff
= par('2*zref*(vincident+vreflect)/(vincident-vreflect)')
*
.end

References
TDR Impedance Measurements: A Foundation for Signal Integrity, Copyright © 2001,
Tektronix, Inc.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 147
W-2024.09
 Feedback

Part 2: Signal Integrity Modeling

Describes the signal integrity modeling.

Part 2 focuses on signal integrity modeling and consists of the following chapters:
• Chapter 5, S-parameter Modeling Using the S-element
• Chapter 6, W- and T-elements Transmission-Line Modeling
• Chapter 7, Using the Field Solver to Extract Tx Line Parameters
• Chapter 8, Using IBIS Files
• Chapter 9, Using IBIS-AMI Equalizer Models with StatEye

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 148
W-2024.09
 Feedback

5
S-parameter Modeling Using the S-element

Describes the S-parameter, S-parameter modeling, and other topics related to the S-
element.

PrimeSim HSPICE provides numerous examples. For paths to S-parameter-related demo

files, see S-parameter Examples in the PrimeSim HSPICE User Guide: Demonstration
Netlists.
You can use the S-element to describe a multi-terminal network within most PrimeSim
networkmulti-terminalmulti-terminal network

HSPICE and RF analyses. (The exception is Shooting-Newton SN analysis.) For more

information about using the S-element (S-parameter) for signal integrity analysis, see S-
element (Scattering Parameter Data) in the PrimeSim Continuum User Guide: Elements.
The following sections discuss these topics:
• S-parameter Model
• Using the Scattering Parameter Element
• S Model Syntax
• S-element Data File Model Examples
• Mixed-Mode S-parameters
• Accelerating S-Element Time Domain Performance with Recursive Convolution
• Multiport Noise Model for Passive Systems
• S-element Noise Model
• S-parameter Standalone Manipulation Utility
• Preconditioning S-parameters
• Group Delay Handler in Time Domain Analysis
• S-Model Data Smoothing
• Predicting an Initial Value for FMAX in S-Element Models
• De-Embedding S-parameters

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 149
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Model

• Small-Signal Parameter Data Frequency Table Model

• References

S-parameter Model
You can use small-signal parameters at the network terminals to characterize linear or
S-parameter modelmodelsS-parameter

nonlinear networks that have sufficiently small signals. (For details, see Chapter 2, Linear
Network Parameter Analysis.) After you set the parameters, you can simulate the block
in any external circuit. Designers of high frequency circuits widely use S-parameters to
characterize a linear network.
In multiport networks, S-parameters ( ) take the following definition:

Equation 15

In Equation 15, is an incident wave vector, and is a reflected wave vector, defined as
follows:

Equation 16

Equation 17

Equation 16 and Equation 17 use the following definitions:

• is the forward voltage vector.

• is the backward voltage vector.

• is the forward current vector.

• is the backward current vector.

• is the characteristic impedance matrix of the reference system.

• is the characteristic admittance matrix.
• and satisfy the relationship
The S-parameters are frequency-dependent. When all ports terminate with impedance-
matching without a voltage/current source, the forward wave becomes zero. This is
because there is no reflection if the ports have no voltage/current source.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 150
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Using the Scattering Parameter Element

Notifications and Limitations

Because the S-element supports two types of noise models, the priority is:
• For multiport (N≠2) S-elements, PrimeSim HSPICE only considers passive noise
models in noise analysis. If NOISE=0, the tool considers the system as noiseless.
• For two-port S-elements, if a Touchstone file provides two-port noise parameters, those
two-port noise parameters generate the noise model. If the file does not provide two-
port noise and NOISE=1, a passive noise model is used. Otherwise, the tool considers
the system as noiseless.
• PrimeSim HSPICE does not support bias-dependent S-parameters.

Using the Scattering Parameter Element

The S-element gives you a convenient way to describe a multi-terminal network. You can
use the S-element with the generic frequency-domain model (.MODEL SP). You also can
use the S-element with data files that describe frequency-varying behavior of a network
and provide discrete frequency-dependent data such as a Touchstone 1.0/2.0 file and a
CITI file.
In particular, the S-parameter in the S-element represents the generalized scattering
parameter (S) for a multi-terminal network.
The S-parameter and the Y-parameter satisfy the following relationship:

Equation 18

where is the characteristic admittance matrix of the reference system. The following
formula relates to the characteristic impedance matrix:

Equation 19

Similarly, you can convert the Y-parameter to the S-parameter as follows:

Equation 20

For detailed information on the S-element syntax and examples, see S-element
(Scattering Parameter Data) in PrimeSim Continuum User Guide: Elements.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 151
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S Model Syntax

S Model Syntax
Use the following syntax to describe specific S models:
S model syntaxsyntaxS model
.MODEL Smodel_name S [N=dimension]
+ [TSTONEFILE=filename | CITIFILE=filename |
+ RFMFILE=file_name.rfm | BNPFILE=filename]
+ [TYPE=s|y] [Z0=value| vector_value]
+ [FBASE=base_frequency] [FMAX=maximum_frequency]
+ [INTERPOLATION=STEP|LINEAR|SPLINE|HYBRID]+ [INTDATTYP=[RI|MA|DBA]]
+ [HIGHPASS=0|1|2|3|4] [LOWPASS=0|1|2|3]
+ [DELAYHANDLE=1|2|0|ON|OFF] [DELAYFREQ=val]
+ [MIXEDMODE=0|1]
+ [DATATYPE=data_string] [XLINELENGTH=val]
+ [NoiPassiveChk [1|0]
+ [SMOOTH=val] [SMOOTHPTS=val]
+ [RATIONAL_FUNC=[0|1] [RATIONAL_FUNC_REUSE=0|1|2]
+ [RATIONAL_FUNC_FOR_AC=0|1]
+ [RATIONAL_FUNC_WINDOWING=[0|1]]
+ [PASSIVE=[0|1] [PASSIVE_TOL=val] [COLSUM_LIMIT=val]
+ [ENFORCE_PASSIVE=0|1] [STAMP=AUTO|Y|S|YSTS|SSTS|DEEMBED]
+ [PRECFAC=val] FQMODEL=sp_model_name
+ [NOISE=[1|0]] [DTEMP=value]
+ [CACHE_DIR=dir_str]
+ [CAUSALITY=[0|1]]
+ [CAUSALITY_FMAX_SCALE = val]
+ [WRITE_CAUSALITY_ENFORCED_FILE = [0|1]]
+ [WRITE_FITTED_FILE = 1|0]
+ [CAUSALITY_ENFORCED_FILE_NAME = file_name]

Keyword Description

Smodel_name Name of the S model.

S Specifies that the model type is an S model.

N S model dimension, which is equal to the terminal number of an S-element and excludes
the reference node.

TSTONEFILE Specifies the name of a Touchstone file v 1.0/2.0. Data contains frequency-dependent
array of matrixes. Touchstone v1.x files must follow the .s#p file extension rule, where #
represents the dimension of the network.
You can specify the Touchstone file by specifying the filename within single quotes or by
using string parameters.
Example of using str( ) to specify a Touchstone file:
.subckt sparam n1 n2 tsfile=str('ss_ts.s2p') S1 n1 n2 0
mname=s_model .model s_model S TSTONEFILE=str(tsfile) .ends x1 A B
sparam tsfile=str('ss_ts.s2p') ...
For details, see Touchstone® File Format Specification by the IBIS Open Forum
(https://ibis.org/).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 152
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S Model Syntax

Keyword Description

CITIFILE Specifies the name of the CITI file, which is a data file that contains frequency-dependent
data. Note that string parameters are supported for calling a CITIFILE. For details, see
Using Instruments with ADS by Agilent Technologies (http://www.agilent.com).

RFMFILE Specifies S-element rational function (RFM) file. See Accelerating S-Element Time
Domain Performance with Recursive Convolution.

BNPFILE Specifies Broadband Network Parameter (BNP) file (proprietary). Note that when
using the BNP file, there is no need for INTERPOLATION or LOW_PASS keywords. For
HIGH_PASS, other than all options are HIGH_PASS=4 supported.
Note:
If you get a warning such as BNP file read failure at f=xx, this means that
the BNP API is returning a read failure. It is likely that this BNP file does not cover the
f=xx frequency point.

TYPE Parameter type:

• S: (scattering) (default)
• Y: (admittance)

Z0 (or Zo) Characteristic impedance value of the reference line (frequency-independent). For
multi-terminal lines (N>1), PrimeSim HSPICE assumes that the characteristic impedance
matrix of the reference lines are diagonal, and their diagonal values are set to Z0. You
also can set a vector value for nonuniform diagonal values. Use Z0 to specify more
general types of a reference-line system. The default is 50.

FBASE Base frequency used for transient analysis. PrimeSim HSPICE uses this value as the
base frequency point for IFFT:
• If FBASE is not set, PrimeSim HSPICE uses a reciprocal of the transient period as the
base frequency.
• If FBASE is set to a value smaller than the reciprocal value of transient period, transient
analysis performs circular convolution by using the reciprocal value of FBASE as a base
period.

FMAX Maximum frequency for transient analysis. Used as the maximum frequency point for
IFFT. See Predicting an Initial Value for FMAX in S-Element Models.

INTERPOLATION The interpolation method:

• STEP: piecewise step.
• SPLINE: b-spline curve fit.
• LINEAR: piecewise linear (default).
• HYBRID: PrimeSim HSPICE combines different interpolation/extrapolation methods,
and switches automatically between them to get the best accuracy. If needed, it
also does causality correction down to DC. It is most useful for the S-parameters
showing local resonances, and provides the proper interpolation and low-frequency
extrapolation method for each entry of the S matrix, which shows different behaviors.
For best accuracy, low frequency examples should be provided.

INTDATTYP Data type for the interpolation of the complex data. It's used only with linear and spline
interpolation.
• RI: real-imaginary based interpolation.
• DBA: dB-angle based interpolation.
• MA: magnitude-angle based interpolation (default).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 153
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S Model Syntax

Keyword Description

HIGHPASS Specifies high-frequency extrapolation:

• 0: Use zero in Y dimension (open circuit).
• 1: Use the given spectrum for IFFT process. No extrapolation is done. This is only
available for S-type-based model.
• 2: Use linear extrapolation with the highest two points.
• 3: Apply window function (default).
• 4: Estimates average derivatives of the phase and magnitude from highest 10% of
sampling points. Extrapolation is performed using the highest sampling point and these
derivatives.
This option overrides EXTRAPOLATION in .MODEL SP.

LOWPASS Method to extrapolate lower frequency points:

• 0: Cut off.
• 1: Make use of the S-matrix at the magnitude of the lowest given frequency point; Set
the magnitude value of each entry as the element of DC matrix. The sign of each value
is determined by the real part of the extrapolated value at DC point. (default)
• 2: Perform linear extrapolation using the magnitude of the lowest two points.
• 3: Perform rational function approximation based on low end frequency extrapolation.
The LOWPASS option overrides EXTRAPOLATION in .MODEL SP.

DELAYHANDLE Extracts the system delay before constructing the system impulse response. This might
improve transient accuracy when the system does have delay, such as a transmission
line system, and RATIONAL_FUNC=0 is used.
When RATIONAL_FUNC=1 (default) is used, it is recommended to not specify the
DELAYHANDLE keyword. The S-element automatically detects and handles the
propagation delay.
If DELAYHANDLE is 2, the S-element extracts the propagation delay according to the model
property over the whole frequency band, instead of calculating group delay upon single
frequency point as that when DELAYHANDLE=1. This option is now only available when the
rational_func=1.
If DELAYHANDLE is 2, the S-element extracts the propagation delay according to the model
property over the whole frequency band, instead of calculating group delay upon single
frequency point as that when DELAYHANDLE=1. This option is now only available when
rational_func=1.
If you set DELAYHANDLE=OFF (or 0), but DELAYFREQ is not zero, PrimeSim HSPICE
simulates the S-element in the delay mode.
For Y-parameter models, DELAYHANDLE is not supported and will be ignored.
See also Group Delay Handler in Time Domain Analysis.

DELAYFREQ Delay frequency for transmission-line type parameters. The default is FMAX. If the
DELAYHANDLE is set to OFF, but DELAYFREQ is nonzero, PrimeSim HSPICE still simulates
the S-element in the delay mode.

MIXEDMODE Sets to 1 if the parameters are represented in the mixed mode.

DATATYPE A string used to determine the order of the indices of the mixed-signal incident or reflected
vector. The string must be an array of a letter and a number (Xn) where:
• X = D to indicate a differential term
• X = C to indicate a common term
• X = S to indicate a single (grounded) term
• n = the port number

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 154
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S Model Syntax

Keyword Description

XLINELENGTH The line length of the transmission line system where the S-parameters are extracted.
This keyword is required only when the S Model is used in a W-element.

NoiPassiveChk Checks S-parameter for passivity in noise analysis (only):

• 1 (default): Checks for passivity; if it fails at any frequency, thermal noise is turned off
for the specific frequency point.
• 0: Disables the passivity checker; thermal noise is always turned on.

SMOOTH An integer value to choose one of following methods:

• 0: No smoothing (default)
• 1: Mean
• 2: Median
• 3: 2nd order polynomial fit
• 4: 4th order polynomial fit
See S-Model Data Smoothing.

SMOOTHPTS An integer value to specify width of the smoothing window on each side of the target
point. In total, 2*x +1 point is taken at each point calculation.

RATIONAL_FUNC • 0: Performs the same as conventional S-element. FBASE/FMAX-based linear

convolution is performed.
• 1 (default): Performs rational function approximation then recursive convolution; also
handles noncausal S-parameters.
• 2: Performs rational function approximation to both time and frequency domain
simulation.
When RATIONAL_FUNC=1 is used, it is recommended to not specify the DELAYHANDLE
keyword. S-element automatically detects and handles the propagation delay.
See also .OPTION S_RATIONAL_FUNC in the PrimeSim HSPICE Reference Manual:
Commands and Control Options.

RATIONAL_FUNC_ REUSE • 0: Discard previously extracted rational function data and re-run the rational function
approximation.
• 1: Reuse rational function data if available.
• 2: (default) Reuse rational function data if available and make no change in parameter
source file (time stamp), FBASE, FMAX, HIGHPASS, LOWPASS, and passivity enforcement
configurations; otherwise, rerun the rational function approximation.

RATIONAL_FUNC_FOR_AC When used in conjunction with RATIONAL_FUNC=1, RATIONAL_FUNC_FOR_AC = 1,

causes the S-element to use the extracted rational function in frequency domain
evaluations.
Default is 0.

RATIONAL_FUNC_WINDOW When you specify this argument with a value 1, to an S-parameter model, a window
ING function is applied that will smoothly roll off the propagation toward the maximum
frequency when generating the rational function. The default maximum frequency is the
highest sampling point in the given S-parameter data.
When FMAX is specified, the value of FMAX is used for the maximum frequency.
When you specify this argument with a value 0 (default), no windowing function is applied.
To apply windowing globally to all S-parameter models in the design, see .OPTION
S_RATIONAL_FUNC_WINDOWING in the PrimeSim HSPICE Reference Manual:
Commands and Control Options.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 155
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S Model Syntax

Keyword Description

PASSIVE Activates the passive checker to help debug the passive models. The default is 0 for the
S-element where 0=deactivate and 1=activate. Using the tolerance value specified
by PASSIVE_TOL keyword, the eigenvalues of matrix (I-S*S'), ev[i], will be checked.
• If any frequency point violates RE(ev[i]) > -(TOL*0.1), PrimeSim HSPICE issues
a warning containing a list of violating frequencies with an “E” flag. Also, the checker
verifies potential passivity violations by checking the summation of each S-parameter
matrix column.
• If Sum > COLSUM_LIMIT, PrimeSim HSPICE issues a warning containing a list of
those violating frequencies with a “C” flag.

PASSIVE_TOL Tolerance for eigenvalue checking activated by the PASSIVE keyword. Default value is
0.01.

COLSUM_LIMIT Maximum value allowed for S-parameter matrix column summation. The limiting value will
be used for passivity checker and passivity enforcement. Default value is 1.0

ENFORCE_ PASSIVE With the ENFORCE_PASSIVE=1 keyword, the S-element checks passivity of all the given
frequency sampling points. Once passivity violations are found, the S-element seeks a
minimum amount of loss property which restores passivity of all the violated points. Then
the S-element adds the loss to all the given frequency points.
Note:
Passivity enforcement function generates cache data and stores it in a
<model_name>.penf file. When the cache file exists, the PrimeSim HSPICE
S-element reuses passivity enforcement result without re-enforcement.

STAMP • AUTO (default): Automatically selects an optimized stamping method

• Y: Conventional admittance based stamp.
• S: Scattering parameter based stamp (Note 1).
• YSTS: Admittance parameter based state space stamp (Note 2).
• SSTS: Scattering parameter based state space stamp (Note 2).
• DEEMBED: Produces a negated stamp to de-embed given an S-parameter block from
the adjacent DUT connected in series (see De-Embedding S-parameters).
Note:
Use STAMP=SSTS when running Shooting Newton (SN) based analyses using the
PrimeSim HSPICE tool.

PRECFAC In almost all cases, you do not need to specify a value for this parameter. This parameter
specifies the precondition factor keyword used for the precondition process of the
S-parameter. A precondition is used to avoid an infinite admittance matrix. The default is
0.75, which is good for most cases. See also, Preconditioning S-parameters.

FQMODEL Specifies the name of the frequency model file behavior of the S-,Y-, or Z-
parameters. The .MODEL statement of sp type defines the frequency-dependent matrices
array.

NOISE Activates thermal noise.

• 1 (default): element generates thermal noise.
• 0: element is considered noiseless.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 156
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S Model Syntax

Keyword Description

DTEMP Temperature difference between the element and the circuit, expressed in ×C. The default
is 0.0.
Element temperature is calculated as:
T = Element temperature (°K)
= 273.15 (°K) + circuit temperature (°C)
+ DTEMP (°C)
Where circuit temperature is specified using either the .TEMP statement, or by sweeping
the global TEMP variable in .DC, .AC, or .TRAN statements.
When a .TEMP statement or TEMP variable is not used, the circuit temperature is set
by .OPTION TNOM, which defaults to 25°C unless you use .OPTION SPICE, which raises
the default to 27°C.

CACHE_DIR Specifies the directory to write (for reuse) the generated cache files, such as rational
function model generation.

CAUSALITY Turns on or off the causality enforcement algorithm for the IFFT model:
• 0 (default): OFF **
• 1: ON **
When causality enforcement is turned on, the HIGHPASS and FMAX keyword of S element
or model is ignored.
• For PrimeSim HSPICE: default is 0.
• For PrimeSim: default is 1.

CAUSALITY_FMAX_SCALE A real value val, which is used to manually set FMAX=val * Fmax_given in causality
enforcement algorithm, where Fmax_given is the highest frequency value in the S model.
Default is 3. Only valid when CAUSALITY=1.

WRITE_CAUSALITY_ENFO Turns on or off the causality enforced result output to a Touchstone format file.
RCED_FILE • 0 (default): do not output
• 1: output
Only valid when CAUSALITY=1

WRITE_FITTED_FILE Writes the rational function fitted s-parameters file when the RATIONAL_FUNC or
S_RATIONAL_FUNC commands are used. Default is 1.

CAUSALITY_ENFORCED_F Touchstone file name for the causality enforced result output.
ILE_NAME • Only valid when CAUSALITY=1.
When file name is not specified, while WRITE_CAUSALITY_ENFORCED_FILE=1, the default
file name is constructed by S model name, such as model_name_enforced.s#p.

The FQMODEL, TSTONEFILE, CITIFILE, and RFMFILE parameters describe the frequency-
varying behavior of a network. Only specify one of the parameters in an S model card.
If more than one method is declared, only the first one is used, and PrimeSim HSPICE
issues a warning message.
For full example demo files of the S model usage, see S-parameter Examples in the
PrimeSim HSPICE User Guide: Demonstration Netlists.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 157
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-element Data File Model Examples

S-element Data File Model Examples

The S model statement samples shown in Example 1 and Example 2 generate the same
S-elementdata file model examples

results.
Example 1
S model statement code example:
s1 n1 n2 n3 n_ref mname=smodel
.model smodel s n=3 z0=50 fbase=25e6 fmax=1e9
s1 n1 n2 n3 n_ref z0=50 fbase=25e6 fmax=1e9

Example 2
In this example, the S model statement has the characteristic impedance equal to 100
instead of the 50 as defined in smodel. The impedance changes because the parameters
defined in the S element statement have higher priority than the parameters defined in the
S model statement:
s1 n1 n2 n3 n_ref mname=smodel z0=100
.model smodel s n=3 z0=50 fbase=25e6 fmax=1e9

Example 3
In this example, tstonefile and citifile are both declared in smodel. PrimeSim
HSPICE accepts tstonefile, ignores citifile, and issues a warning message.
It is illegal to define a tstonefile and CITIfile smodel in the same statement. This
prevents conflicts in the frequency-varying behavior description of the network. From the
tstonefile file extension .s3p, you can tell that the network has three ports:
s1 n1 n2 n3 n_ref mname=smodel
.model smodel s tstonefile=exp1.s3p
citifile=exp1.citi0

Example 4
In this example, is declared both in the S-element statement and the S model statement.
s1 n1 n2 n3 n_ref mname=smodel
.model smodel s n=3

Example 5
This example shows a generic S-parameter statement using port elements. For
information on port elements, see Identifying Ports with the P-element.
**S-parameter example
.OPTION post
.probe v(n2)
P1 n1 0 port=1 Z0=50 ac=1v PULSE 0v 5v 5n 0.5n 0.5n 25n

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 158
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-element Data File Model Examples

P2 n2 0 port=2 Z0=50
.ac lin 500 1Hz 30MegHz
.tran 0.1ns 10ns
* reference node is set
S1 n1 n2 0 mname=s_model
* S parameter
.model s_model S TSTONEFILE = ss_ts.s2p
Rt1 n2 0 50
.end

Example 6
This example shows the option line and noise parameters of a Touchstone file:
!
! touchstone file example
!
# Hz S MA R 50.0000
0.00000 0.637187 180.000 0.355136 0.00000
0.355136 0.00000 0.637187 180.000
......
! # HZ S DB R 50.0000
! 0.00000 -3.91466 180.000 -8.99211 0.00000
! -8.99211 0.00000 -3.91466 180.000
! ......
!
!# Hz S RI R 50.0000
! 0.00000 -0.637187 0.00000 0.355136 0.00000
! 0.355136 0.00000 -0.637187 0.00000
! ......
!
! 2-port noise parameter
! frequency[Hz] Nfmin[dB] GammaOpt(M) GammaOpt(P) RN/Z0
0.0000 0.29166 0.98916 180.00 0.11055E-03
0.52632E+08 6.2395 0.59071 -163.50 0.32868
0.10526E+09 7.7898 0.44537 175.26 0.56586
! ......
! end of file

Example 7
This example shows an S-parameter statement using port elements and the referenced
CITI file. For information on port elements, see Identifying Ports with the P-element.
**S-parameter
.OPTION post
.probe v(n2)
P1 n1 0 port=1 Z0=50 ac=1v PULSE 0v 5v 5n 0.5n 0.5n 25n
P2 n2 0 port=2 Z0=50
.ac lin 500 1Hz 30MegHz
.tran 0.1ns 10ns
*reference node is set
*S1 n1 n2 0 mname=s_model

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 159
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Mixed-Mode S-parameters

* use default reference node

S1 n1 n2 mname=s_model
* S parameter
.model s_model S CITIFILE = ss_citi.citi Z0=50
Rt1 n2 0 50
.end

Mixed-Mode S-parameters
Mixed-mode refers to a combination of differential and common mode characteristics
mixed-mode S-parameterS-parametermixed mode

in PrimeSim HSPICE linear network (LIN) analysis by using the S-element. It is useful
to understand mixed-mode S-parameters since they present an overview of the wiring
channels. The S-parameters also provide a view of inherent behaviors of differential and
common-mode signal propagation characteristics.
PrimeSim HSPICE accepts both conventional single-ended S-parameters and mixed-
mode S-parameters. In PrimeSim HSPICE, since the mixed-mode S-parameters undergo
conversion to the single-ended S-parameters to fit with ground-referenced nodal analyses,
there is no difference in simulation results between single-ended S-parameters and the
equivalent mixed-mode representations of them.

Figure 49 Node indexing convention of the ground referenced (single ended) S-parameter

Sxxx n1 n2 n3 n4 [nref] mname=xxx

n1 n2
Line A
n3 n4
Line B

• Nodes 1 and 3 are the ports for one end of the transmission-line pair.
• Nodes 2 and 4 are the ports for the opposite end of the transmission-line pair.
The following sections discuss these topics:
• Relating Voltage and Current Waves to Nodal Waves
• Characterizing Differential Data Transfer Systems
• Deriving a Simpler Set of Voltage and Current Pairs
• Using the Mixed-Mode S-parameters (S-element)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 160
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Mixed-Mode S-parameters

Relating Voltage and Current Waves to Nodal Waves

Figure 50 and Equation 21 through Equation 24 include common and differential mode
voltage,

voltage and current waves, relating them to nodal waves. Although you can apply mixed-
current waves-nodal waves

mode data propagation to an arbitrary number of pairs of transmission lines, Figure 50

shows using a single pair model.
This figure also shows a schematic of symmetric coupled pair transmission lines
commonly used for the differential data transfer system.

Figure 50 Schematic of symmetric coupled-pair transmission line

port 1 port 2
i1 i2
Line A
V1 V2

i3 i4
Line B
V3 V4

Solving the telegrapher’s equation, you can represent nodal voltage and current waves of
the data transfer system as:

Equation 21

Equation 22

Equation 23

Equation 24

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 161
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Mixed-Mode S-parameters

where:
• is the propagation constant for even mode waves.
• is the propagation constant for odd mode waves.
• is the characteristic impedance for even mode waves.
• is the characteristic impedance for odd mode waves.
• and represent phasor coefficients for the forward propagating modes.

• and represent phasor coefficients for the backward propagating modes.

Each voltage and current pair at each node represents a single propagating signal
wave that references to the ground potential. This type of expression is nodal wave
representation.

Characterizing Differential Data Transfer Systems

Equation 25 through Equation 28 use differential and common mode waves to
differential, common mode wavesvoltage, current pairs

characterize differential data transfer systems. The difference of the nodal wave defines
the voltage and current of the differential wave:

Equation 25

Equation 26

Common mode voltage and current definitions are:

Equation 27

Equation 28

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 162
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Mixed-Mode S-parameters

Deriving a Simpler Set of Voltage and Current Pairs

In the following equations, substituting Equation 15, Equation 16 and Equation 17 into
Equation 15 derives a simpler set of voltage and current pairs:

Equation 29

Equation 30

Equation 31

Equation 32

You also can relate characteristic impedances of each mode to the even and odd mode
characteristic impedances:

and

Having defined a generalized parameter power, you can then define differential normalized
waves:

and

and

Similarly, you can define common mode normalized waves as:

and

and

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 163
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Mixed-Mode S-parameters

You can then specify S-parameters for mixed-mode waves as ratios of these waves as
shown in Equation 33:

Equation 33

where:
• is the differential-mode S-parameter.

• is the common-mode S-parameter.

• and represent the mode-conversion or cross-mode S-parameters.

Based on these definitions, you can linearly transform nodal wave (standard) S-
parameters and mixed mode S-parameters:

The M transformation matrix is:

Using the Mixed-Mode S-parameters (S-element)

The S-element can recognize and parse the mixed-mode S-parameters when the keyword
mixed-mode S-parametersS-parametermixed-mode

mixedmode=1 is used. Any keywords besides mixedmode and datatype remain the same.
Use the following syntax for a mixed-mode S-parameter:

Sxxx p1+ [p1-] p2+ [p2-] p3+ [p3-]...[n_ref] mname=Smodel

.MODEL Smodel S ...
[+ mixedmode=[0 1]]
[+ datatype=XiYjZk...]

Keyword Description

pn+, pn- Positive and negative terminals of the port n, respectively. The port numbers must be in
an increasing order corresponding to the S matrices notation.
• If the port is in mixed-mode (balanced), it requires that both positive and negative
terminal names are in series.
• If the port is single-ended, mixed-mode requires only one terminal name.

mixedmode When mixedmode=1, PrimeSim HSPICE passes the information to the S-element that
the S-parameters are in mixed-mode. The default is 0 (standard mode).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 164
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Mixed-Mode S-parameters

Keyword Description

datatype A string that determines the order of indices of the incident or reflected vectors (a and
b) in Equation 25. The string must be an array of pairs that consists of a letter and a
number (for example, Xn), where X =
• D or d to indicate differential term.
• C or c to indicate common term.
• S, s, G, or g to indicate single (grounded) term and n = port number.

The definition datatype = D1D2C1C2 is the default for a 2-balanced port network and
specifies the nodal relationship of the following equation:

Equation 34

where:
• is the incident wave goes into positive terminal of the port 1.

• is the incident wave goes into negative terminal of the port 1.

• is the incident wave goes into positive terminal of the port 2.

• is the incident wave goes into negative terminal of the port 2.

You also can derive the nodal relationship of the reflection wave in the same way.
PrimeSim HSPICE assigns nodes from the given S-matrices to the S-element in the order
of . For example, incident and reflected waves at the positive terminal of the
appear at the first node of the S-element.

The definition datatype=D1C1S2 specifies the nodal relationship of the following equation:

Equation 35

The default is datatype=D1D2...DnC1C2...Cn, which is available for systems with mixed-

mode (balanced) ports only.
Mixed-Mode S-parameter Netlist Examples
Example 1: Differential Transmission Line Pair
For an example netlist of a differential transmission line pair, go to the following directory:
$installdir/demo/hspice/sparam/mixedmode_s.sp

Example 2: Differential Amplifier

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 165
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Accelerating S-Element Time Domain Performance with Recursive Convolution

For an example netlist of a differential amplifier, go to the following directory:

$installdir/demo/hspice/sparam/diffamp_s.sp

Accelerating S-Element Time Domain Performance with

Recursive Convolution
By default, the recursive convolution method is enabled by using the S-element keyword
RATIONAL_FUNC. The original S-parameters are fit to their equivalent rational functions by
using the original sampling points. This avoids interpolation or extrapolation. PrimeSim
HSPICE saves the rational function to a *.yrf file to use recursive convolution during the
simulation. By default, PrimeSim HSPICE reuses saved rational functions in subsequent
simulations. See also Multithreading Acceleration for S-element on Linux, Ensuring
Causality in the Rational Function Model, and Rational Function Matrix (.rfm) File Format.
When RATIONAL_FUNC is disabled, the S-parameter evaluation method uses inverse FFT.
N
This method requires equal-spaced sampling with 2 points. If the S-parameter frequency
points do not match the IFFT array sampling, PrimeSim HSPICE uses interpolation and/or
extrapolation. This means that the sampling point location for the IFFT depends heavily on
the combination of the FBASE and FMAX values. Accuracy also is strongly dependent on the
value of FMAX.
Since the recursive convolution method is independent of FMAX or FBASE values,
this method is useful when S-parameters are nonpassive or there are many ports.
The S-element RATIONAL_FUNC model captures long and multiple delay components
automatically without requiring the explicit DELAYHANDLE keyword. The recommended
configuration for using the S-element is to set RATIONAL_FUNC=1 and remove all other
control keywords, unless there is a specific requirement.
The existence of decoupling capacitors in the S-parameters are automatically detected.
The decoupling capacitors are de-embedded and separate sets of rational functions are
generated before a recursive convolution is performed.
Both low impedance networks such as power distribution networks (PDN) and networks
where the insertion loss is close to an open circuit can effectively be evaluated using
rational functions.
The recommended configuration when using recursive convolution is to remove all other
control keywords, unless you have a special need.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 166
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Accelerating S-Element Time Domain Performance with Recursive Convolution

The convolution integral is commonly used to handle frequency-dependent transfer

convolution integralrecursive convolutionS-elementrecursive convolution

characteristics. To get a system response at time , the convolution integral can be done
as shown in Equation 36:

Equation 36

where are input at the , system response function in the time domain
and output at , respectively. As observed in Equation 36, the convolution integral is
computationally expensive, especially if becomes a long transient simulation due to
an increasing time window for each time point evaluation. The conventional S-element
obtains by applying IFFT to the original system function in the frequency domain and
performs a discrete linear convolution integral, while Equation 36 is continuous.
On the other hand, the frequency domain transfer function, can be described as:

Equation 37

Converting Equation 37 to the time domain gives the exponential decay function shown in
Equation 38:

Equation 38

The computational cost of the convolution integral at time point can be reduced using
the convolution result at a previous time point (recursive convolution). Since recursive
convolution only requires numerical integration from a previous time point to the current
time point, it saves computational time and storage for input signal history. Recursive
convolution can be formulated only when the system response can be represented in
certain forms of rational functions, as shown in Equation 39:

Equation 39

The PrimeSim HSPICE S-element generates a rational function matrix based on a given
function and performs recursive convolution. Once the rational function is generated, the
S-element stores the intermediate data for reuse in the following form: MODEL_NAME.yrf.
When RATIONAL_FUNC_REUSE=1, the S-element seeks an available data file and reuses it
without running a redundant rational function generation process.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 167
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Accelerating S-Element Time Domain Performance with Recursive Convolution

PrimeSim HSPICE also accepts rational function data input as external input. The input file
syntax is described in the following section, Rational Function Matrix (.rfm) File Format.
PrimeSim HSPICE accepts S- or preconditioned Y- parameter matrices as expressions
with pairs of poles and residues. In cases of frequency-dependent scattering parameters,
S( ), or preconditioned admittance parameter, Y'( ) can be represented as rational function
matrix components as:

Equation 40

Equation 41

The following sections discuss these topics:

• Multithreading Acceleration for S-element on Linux
• Ensuring Causality in the Rational Function Model
• Rational Function Matrix (.rfm) File Format

Multithreading Acceleration for S-element on Linux

One of the benefits of using the rational function model is that since the function can be
modeled independent of the transient simulation configuration, generated rational function
data can be reused in subsequent simulation runs. To reduce the computational time for
this process, the S-element takes the number of threads specified on the command-line
invocation of -mt as the maximum number of threads to be used. See Running Multi-
threading (MT) and Distributed Processing (DP) Concurrently in the PrimeSim HSPICE
User Guide: Basic Simulation and Analysis. The actual number of threads to be used can
be smaller depending on the size of the target S-parameters.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 168
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Accelerating S-Element Time Domain Performance with Recursive Convolution

Figure 51 Wall clock time for the rational function generation of a 131 port S-parameter
block (Linux system with 8 x 2666 MHz Intel® Xeon®)

Ensuring Causality in the Rational Function Model

When using the rational function model, PrimeSim HSPICE enforces the causal behavior.
If RATIONAL_FUNC is set to 1, the original transfer function is approximated as a summation
of the partial rational functions which can be proved to be causal:

Therefore, the resulting function must be causal.

In the rational function generation process, potential unstable poles located in the
righthand side of the complex plane are filtered out automatically.

Rational Function Matrix (.rfm) File Format

In addition to the rational function matrix (*.yrf) file discussed in the previous section,
PrimeSim HSPICE provides syntax for users or third parties to create an ASCII
representation of the rational function matrix. The resulting *.rfm file can then be read by
the S-element via the S-model RFMFILE=file_name.rfm keyword.
The *.rfm file (text file) interface allows you to input transfer function (matrix) in the form
of rational function directly instead of using s-parameters. If you have an extraction tool
that creates a rational function model directly, using it may reduce numerical errors rather
than let the tool convert the function into s-parameters. PrimeSim HSPICE then extracts
the rational function from there.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 169
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Accelerating S-Element Time Domain Performance with Recursive Convolution

The *.rfm file is divided into two parts:

• The header consists of keywords and setup information for the entire system. This
section (first five lines in the following example) contains information about the data
that follows, such as number of ports, matrix type, preconditioning factor, and reference
impedance.
• The data field consists of rational function coefficients of each matrix component. Each
matrix component begins with a BEGIN keyword and ends with the END keyword.
Version 200600
NPORT 2
MATRIX_TYPE Y
PRECFAC 0.75
Z0 50

BEGIN 1 1
CONST 0.0
C 0.0
DELAY 0.0
BEGIN_REAL 2
3.50774e+07 -4.54754e-05
2.37196e+08 -0.00327245
BEGIN_COMPLEX 2
3.81668e+08 3.74508e+08 0.00583496 -2.54387
6.88144e+08 2.08242e+08 6.66955e-06 -2.78498
END

A single line can only contain single pairs of pole and residue. Therefore, two numbers
must appear in a line for a real pole and four numbers must appear in a line for a complex
pole. A single complex pole represents a complex conjugate pair of poles. An *.rfm file
does not need to include all the matrix components. If certain terms are not found, the S-
element regards these terms as ones with no propagation. The comment special character
is an exclamation point. Lines that begin with '!' are ignored.
An RFM keyword (with no whitespace) is always the first word on the new line. The
following table lists available keywords.

Keyword Description

VERSION n Version number.

NPORT n Number of ports.

MATRIX_TYPE [S|Y|Z] Currently, S and Y are supported.

SYMMETRIC Indicates symmetric matrix. Only a single declaration must appear in the
data field for transposing of pair of nondiagonal matrix components.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 170
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Multiport Noise Model for Passive Systems

Keyword Description

Z0 val(s) or Zo val(s) Reference impedance of ports. Real number impedance only. When a single
value is specified, the value is applied to all the ports. A vector of values with
the size of the number of port also can be specified. A single line can only
contain single number.

PRECFAC val Preconditioning factor; must be between 0.5 and 1.0 (0.5 < < 1.0).

BEGIN row col Beginning of a matrix component specified by row and column. The row and
column must be 1-based index of the matrix component.

CONST val Constant term of the rational function “B” term of Equation 39; if not
specified, CONST=0.

C val Reactive term of the rational function “C” term of Equation 39; if not
specified, C=0.

DELAY val Propagation delay from port[col] to port[row]. Must be zero or a positive
number. If not specified DELAY=0.

BEGIN_REAL n Pairs of real poles and residues follow. Following each line must contain a
real pole and a real residue in this order. If BEGIN_REAL is not specified, no
real pole is constructed. Other keywords must appear before BEGIN_REAL.

BEGIN_COMPLEX n Pairs of complex pole and residue follows. Following each line must contain
a real part and an imaginary part of pole, real and imaginary part of residue
in this order. A single complex pole and residue pair represents a conjugate
pair of poles. If BEGIN_COMPLEX is not specified, no complex pole is
constructed. Other keywords must appear before BEGIN_COMPLEX.

END End of the matrix component.

Multiport Noise Model for Passive Systems

Multiport passive and lossy circuits, such as transmission lines and package parasitics,
multiport passive and lossy circuits, thermal noisethermal noise

can exhibit considerable thermal noise. The passive noise model is used to present
such thermal noise for the S-element representing such circuits. The S-element passive
noise model supports normal, two-port and multiport noise analysis (.NOISE=1) and .LIN
noisecalc=1 for two-port and .LIN noisecalc=2 for N-port.

The following sections discuss these topics:

• Input Interface
• Output Interface

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 171
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Multiport Noise Model for Passive Systems

Input Interface
To trigger a passive multiport noise model, the NOISE and DTEMP keywords in an S-
element statement are used:

Sxxx n1...nN
+ ...
+ [NOISE=[1|0]] [DTEMP=value]

See S Model Syntax for the description of the NOISE and DTEMP keywords.
When NOISE=1, PrimeSim HSPICE generates a N×N noise-current correlation matrix from
the N×N S-parameters according to Twiss' Theorem. The result can be stamped into a
PrimeSim HSPICE noise analysis as N-correlated noise current sources: ji (i=1~N) as
shown Equation 42:

Equation 42

where:
The noise-current correlation matrix represents the frequency-dependent statistical
relationship between N noise current sources, ji (i=1~N), shown in Figure 52.

Figure 52 Frequency-dependent statistical relationship between N noise current sources

Original System Transformed System

Port j
Port i

...
Port i

Port j

... ji jj

Port 2 Port N–1 Port 2 Port N–1

Lossy Passive j2
N-Port Lossless Passive jN–1
Port 1 Port N N-Port System
Port 1 Port N

j1 jN

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 172
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-element Noise Model

Output Interface
PrimeSim HSPICE creates a .lis output list file that shows the results of a noise analysis
just as any other noisy elements. The format is as follows:
**** s element squared noise voltages (sq v/hz)

element 0:s1
N(i,j) data
r(N(i,j)) data
... i,j = 1~N ...
total data

where:
• N(i,j) = contribution of jijj* to the output port.
• r(N(i,j)) = transimpedance of ji to the output port.
• total = contribution of total noise voltage of the S-element to the output port.

S-element Noise Model

This section describes how the S-element supports two-port noise parameters and
multiport passive noise models.
The following sections discuss these topics:
• Two-port Noise Parameter Support in Touchstone Files
• Input Interface
• Output Interface
• Notifications and Limitations

Two-port Noise Parameter Support in Touchstone Files

The S-element can read in two-port noise parameter data from Touchstone data files, and
two-port noise parameter data

then transform the raw data into a form used for .NOISE and.LIN noisecalc=1[or 2]
analysis.
For example, you can represent a two-port with an S-element and then perform a noise
analysis (or any other analysis). The S-element noise model supports normal and two-port
(.NOISE and .LIN noisecalc=1). See Noise Parameters in 2-Port and N-Port Networks.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 173
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-element Noise Model

Note:
Because Touchstone files currently provide only two-port noise parameters, this
type of noise model only supports two-port S-parameter noise analysis for both
passive and active systems.

Input Interface
The frequency-dependent two-port noise parameters are provided in a network description
block of a Touchstone data file following the S-parameter data block.
The noise parameter data typically is organized by using the following syntax:

frequency[Hz] Nfmin[dB] GammaOpt(M) GammaOpt(P) RN/Z0

{ ...data... }

where:
• frequency = frequency in units.
• Nfmin[dB] = minimum noise figure (in dB).
• GammaOpt(M) = magnitude of reflection coefficient needed to realize Fmin.
• GammaOpt(P) = phase (in degrees) of reflection coefficient needed to realize Fmin.
• RN/Z0 = normalized noise resistance.
• ! = indicates a comment line.
For example:
! 2-port noise parameter
! frequency[Hz] Nfmin[dB] GammaOpt(M) GammaOpt(P) RN/Z0
0.0000 0.29166 0.98916 180.00 0.11055E-03
0.52632E+08 6.2395 0.59071 -163.50 0.32868
0.10526E+09 7.7898 0.44537 175.26 0.56586

Both GammaOpt and RN/Z0 values are normalized for characteristic impedance, Z0,
specified in the header of the Touchstone data file. PrimeSim HSPICE reads this raw data
and converts it to a coefficient of the noise-current correlation matrix. This matrix can be
stamped into a PrimeSim HSPICE noise analysis as two correlated noise current sources:
j1 and j2, as shown here:

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 174
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-element Noise Model

The noise-current correlation matrix represents the frequency-dependent statistical

relationship between two noise current sources, j1 and j2, as illustrated in Figure 53.

Figure 53 Frequency-dependent statistical relationship between two noise current sources

Original System Transformed System

Noisy System Noiseless System

j1 j2
S-element S-element

Output Interface
PrimeSim HSPICE creates a .lis output list file that shows the results of a noise analysis
just as any other noisy elements. The format is:
**** s element squared noise voltages (sq v/hz)
element 0:s1
N11 data
r(N11) data
N12 data
r(N12) data
N21 data
r(N21) data
N22 data
r(N22) data
total data

where:
• N11 = contribution of j1 to the output port
• r(N11) = transimpedance of j1 to the output port
• N12 = contribution of j1j2* to the output port
• r(N12) = transimpedance of j1 to the output port
• N21 = contribution of j2j1* to the output port
• r(N21) = transimpedance of j2 to the output port
• N22 = contribution of j2 to the output port
• r(N22) = transimpedance of j2 to the output port
• total = contribution of total noise voltage of the S-element to the output port

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 175
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Notifications and Limitations

• Because Touchstone files currently provide only two-port noise parameters, this type
of noise model only supports two-port S-parameter noise analysis for both passive and
active systems.
• If your Touchstone file includes square brackets in a z0 definition, PrimeSim HSPICE
does not support the square brackets. Acceptable syntax is to list the z0 values without
any brackets.
For example:
.model s_par s tstonefile='tsn.s4p'
+ z0=50 50 50 50

For readability, parentheses can be used.

.model s_par s tstonefile='tsn.s4p'
+ z0=(50 50 50 50)

S-parameter Standalone Manipulation Utility

PrimeSim HSPICE provides several capabilities to manipulate S-parameters including
accepting multiple file formats, inter-/extrapolation schemes, data smoothing, rational
function approximation, and so on. These capabilities become available when an S-
element is specified in a PrimeSim HSPICE circuit netlist and passed to PrimeSim
HSPICE. The following sections describe a standalone executable (sputil), which is
available on Linux and Windows. This standalone utility program is called SPutil. This
utility allows you to access these S-element features without invoking a PrimeSim HSPICE
simulation and additional utility functions. Also, this utility may be used as a tool to predict
and verify the behavior of given S-parameters in advance of PrimeSim HSPICE simulation
runs.
Note:
You can use sputil to manipulate the S-parameters that you plan to use in
your W-element. (Remember that the length of the transmission lines that you
have extracted S-parameters for must be specified when you use them in a W-
element.)
• SPutil Program Features
• Invoking the Utility
• SPutil Runset Format
• Commands

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 176
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

• SPUTIL Functions
• Status Messages
• Causality Checking
• Using Interpolation for Causality Checking

SPutil Program Features

SPutil performs many of the functions which are built into the PrimeSim HSPICE
S-element. In addition, you can add utility functions on demand. The following list
categorizes the current set of SPutil functions:
• Read in and combine s/y parameter files
• Interpolate and/or extrapolate necessary output data points
• Passivity check and enforcement
• Rational function approximation
• Output manipulated S-parameter data
• Format conversion (S/Y/Z, Touchstone-v1,v2, CITI file)
• Re-sampling with interpolation, extrapolation, or windowing with multiple sweep blocks
• Causality checking
• Change reference impedance
• Partial view and zero padding
• Port swapping, reordering, and picking
• Connection checking

Invoking the Utility

Note:
The SPutil program requires that you check out a PrimeSim HSPICE license.
Run a set of S-parameter manipulation commands in the input file by typing:
sputil runset.in

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 177
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

SPutil Runset Format

The SPutil runset file must contain a list of commands and arguments. The following
example shows the basic structure of a SPutil runset. Command names must be
compatible with PrimeSim HSPICE S-element/model keywords and other PrimeSim
HSPICE netlist keywords. SPutil directly calls the PrimeSim HSPICE S-element object
library to obtain identical behavior to the corresponding control keyword available in the
PrimeSim HSPICE S-element.
This example combines three S-parameter files, ts1.s4p, ts2.s4p and citi1.citi, into
one Touchstone (version 1) file, and writes results to the output.s4p file with a combined
list of frequency points specified by the FREQ_SWEEP keyword:
NPORT 4
TSTONEFILE ts1.s4p
TSTONEFILE ts2.s4p
CITIFILE citi1.citi
INTERPOLATION LINEAR
HIGHPASS 3
LOWPASS 1
TSTONE_OUT output.s4p
FREQ_SWEEP DEC 10 1.0 10e9

Note:
1. Accepts both "!" and "*" as comment characters in the runset file.
2. Allows the first line to be a blank line in the runset file.
3. Ignores spaces or extra line feeds at the end of the runset file.
In cases where you want to connect capacitors and subsequently reduce the S-parameter
ports, you can use the MAKEDOTLIN command. An example command set looks like this:
NPORT 4
TSTONEFILE dc.s4p
DATAFORMAT MA
MAKE_DOT_LIN example.sp

This command set creates an example.sp file, which is modified to remove ports 3 and
4 and add capacitors for nets connected to ports 3 and 4. Running PrimeSim HSPICE on
the modified file gives the reduced 2 port S-parameter file.

Commands
A command keyword must always be the first word on the new line. The command
keyword is always one continuous word without embedded spaces.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 178
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Table 3 Current SPutil Command Set

Keyword Description

S-element keywords. See also S-element Syntax in PrimeSim Continuum User Guide: Elements for full keyword
descriptions.

NPORT n Number of ports (optional when the input file is in the

Touchstone version 1 (*.sNp) format).
Note:
Multiple declarations of different NPORT keywords are
invalid.

MNAME modelname Model name used for output.

TYPE [S|Y|Z] Parameter type. Currently, only Y and S are supported.

Default is S.

TSTONEFILE filename Specifies a Touchstone file (version 1.0/2.0).

TSTONE_HEADER filename Specifies the name of the file which contains the
necessary comments for the Touchstone file header.
SPUTIL will copy all the lines before the index line (that
is, the line beginning with "#") from the specified file to
the output Touchstone file.

CITIFILE filename Specifies a CITI file.

BNPFILE filename Specifies Broadband Network Parameter (BNP) file.

FBASE val FBASE value for S-element.

FMAX val FMAX value for S-element.

INTERPOLATION [STEP |LINEAR|SPLINE|HYBRID] INTERPOLATION setting for S-element.

INTDATTYP [RI|DBA|MA] INTDATTYP setting for S-element.

HIGHPASS n HIGHPASS setting for S-element.

LOWPASS n LOWPASS setting for S-element.

DELAYHANDLE val DELAYHANDLE setting for S-element.

DELAYFREQ val DELAYFREQ value for S-element.

RATIONAL_FUNC n Rational function setting for S-element.

RATIONAL_FUNC_FOR_AC n Accesses to the original s-parameter data.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 179
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Table 3 Current SPutil Command Set (Continued)

Keyword Description

PASSIVE [0|1] Activates the passive checker to help debug passive

models. The default is 0 for the S-element where
0=deactivate and 1=activate. Using the tolerance value
specified by PASSIVE_TOL keyword, PrimeSim HSPICE
checks the eigenvalues of matrix (I-S*S'), ev[i].
• If any frequency point violates RE(ev[i]) > -(TOL*0.1),
PrimeSim HSPICE issues a warning containing a
list of violating frequencies with an “E” flag. Also,
the checker verifies potential passivity violations by
checking the summation of each S-parameter matrix
column.
• If Sum > COLSUM_LIMIT, SPutil issues a warning
containing a list of those violating frequencies with a
“C” flag.

PASSIVE_TOL val Tolerance for eigenvalue checking activated by PASSIVE

keyword. Default value is 0.01.

COLSUM_LIMIT val Maximum value allowed for S-parameter matrix column

summation. The limiting value will be used for passivity
checker and passivity enforcement. Default value is 1.0

ENFORCE_PASSIVE [0|1] When enforce_passive 1 is set, SPutil enforces

PASSIVE setting for the S-parameter. Default is 0 (off).

PRECFAC Preconditioning Factor value for S-element. Nonzero

precfac is used only when y-parameter output is
requested with S-parameter input.

Control Keywords

TSTONE_OUT tstone_name Output Touchstone v. 1.0 file name.

If this file name is not provided in causality enforcement,
an error message is displayed.

TSTONE2_OUT Specifies Touchstone v. 2.0 output file.

CAUSAL CHECK This command invokes causality checks (see Causality

causalitycommand

Checking). If the command CAUSAL is not followed by an

argument, SPUTIL signals an error.
Every time this command is invoked, a *.lis file
includes a list of the worst causal violations for any port
combination that is affected. If a port combination is not
affected, it does not appear in that file.
Additionally, a Touchstone RI file brings in its real part
the value of its Global Difference Measure (GDM), while
in the imaginary part it gives a Boolean value according
to the threshold entered in CAUSAL_TOL.

CAUSAL_TOL value This command places as target for causality a value for
the threshold. The value starts at 0.0 (causal). If no value
is given, the threshold is set to its default value: 0.1.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 180
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Table 3 Current SPutil Command Set (Continued)

Keyword Description

CITIFILE_OUT citifile_name Output CITI file name.

SELM_OUT selm_file_name Output sc# file name.

OUT_PORT_BEGIN n When partial S-parameter matrix output is required,

specifies the beginning port index number in the output
data. Default value is 1.

OUT_PORT_END n When partial S-parameter matrix output is required,

specifies the ending port index number in the output
data. Default is NPORT.

ZREF_OUT value | values Specifies the reference impedance value(s) of the

output.
A single value of the argument sets the given impedance
to all the ports.
If a vector (of size the same as the number of ports)
is given, each impedance value is assigned to the
corresponding port.

SPARDIGIT n Sets the numerical precision (the number of digits) for

the S-parameter output.

FREQDIGIT n Sets the numerical precision (the number of digits) for

the frequency output.

DATAFORMAT Specifies the format of the output data file:

• RI: real-imaginary
• MA: magnitude-phase (default format for Touchstone
files)
• DB: DB(magnitude)-phase

FREQ_SWEEP hspice_sweep Specifies frequency points in output data file. You can
specify any of ORG, LIN, DEC, OCT, or POI. ORG takes
frequency points from the input data file, while others
specify various frequency sweeps with nsteps, start,
and stop values using the following syntax for each
type:
• LIN nsteps start stop
• DEC nsteps start stop
• OCT nsteps start stop
• POI nsteps freq_values
• ORG [fmin fmax]

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 181
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Table 3 Current SPutil Command Set (Continued)

Keyword Description

FREQ_SWEEP with the ORG keyword picks up frequency

points given in the input data. The optional fmin and
fmax values, to the ORG keyword, limit the frequency
range to be picked up.
If the FREQ_SWEEP keyword is not specified, then
PrimeSim HSPICE chooses output frequency points
as the combination of the input frequency point. For
example: If the first input file has f=1 2 3 4 and second
input file has f=5 6 7, then the output frequency points
(by default, with no FREQ_SWEEP) are f=1 2 3 4 5 6 7.

SYMMETRY_CHECK [0|1] Default: 0. When SYMMETRY_CHECK 1 is specified,

the keyword checks symmetry of the S-parameter
matrix at each frequency point (especially useful
in evaluating differential signaling systems where
preserving symmetric S-parameter matrices is key).

COLUMN_SUM [0|1] Default: 0. When column_sum 1 is specified, the

keyword computes the summation of each column
of the S-parameter matrix at each frequency point. A
summation of each column indicates whether or not
corresponding incidents are amplified according to:

See Note on the COLUMN_SUM Keyword.

Results are printed in an input_filename.colsum
text file which can be uploaded to the waveform viewer
as a Text/PWL data file. The output file contains real,
imaginary, and magnitude as functions of frequency
for each column summation.

TO_MIXEDMODE [0|1] When set to 1. S-parameters will be converted to mixed

mode assuming each pair of adjacent terminals forms
one differential port. For example, the first and the
second ports of the original S-parameter block will be
the positive and negative terminals of the first differential
port in the resulting mixed mode S-parameter block,
respectively.

SWAP_PORT p1 p2 This command swaps the ordering of the two specified

ports of the S-parameters.

REORDER_PORT io2bus | bus2io | port_indices Swaps the port ordering method used for the
S-parameters between I/O pair ordering and bus
ordering conventions.
• io2bus changes from I/O pair ordering (i1 o1 i2 o2...)
to bus ordering (i1 i2... o1 o2...).
• bus2io changes from bus ordering (i1 i2... o1 o2...) to
I/O pair ordering (i1 o1 i2 o2...).
• port_indices, which is an array of integers,
specifies port indices for custom port reorder.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 182
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Table 3 Current SPutil Command Set (Continued)

Keyword Description

PICK_PORT indices Takes an array of S-parameter port indices and outputs

only the S-parameters of the specified ports in the
specified order.
The argument can contain the TO keyword in between
two indices. The TO keyword represents all the ports
from the first index to the next index in incremental order.
For example, 2 TO 5 is interpreted as 2 3 4 5.
This command allows you to create a subset of the
target system.

REDUCE_W_OPEN array_of_indices Specifies an array of port indices that are set to open
termination, when performing port reduction using the
PICK_PORT command.

REDUCE_W_SHORT array_of_indices Specifies an array of port indices that are set short to
the ground/supply termination, when performing port
reduction using the PICK_PORT command.

Note:
• REDUCE_W_OPEN and REDUCE_W_SHORT arguments accept the To keyword to represent an array of
incremental indices from the left index to the right index. For example, REDUCE_W_OPEN 4 7 to 10 is
equivalent to REDUCE_W_OPEN 4 7 8 9 10.
• REDUCE_W_OPEN and REDUCE_W_SHORT can be specified multiple times. For example, REDUCE_W_SHORT 1
2 and REDUCE_W_SHORT 3 4 are equivalent to REDUCE_W_SHORT 1 2 3 4.
• When performing port reduction using the PICK_PORT command, for ports not specified for either open or
short termination, SPUTIL terminates them with the reference impedance.

SCALE_FREQ scale_factor Every frequency point in the output file is scaled by this
factor.

CHECK_CONNECTION port_index | all Prints all the connections from the specified
port_index, in the order of strength. If all is specified,
the connection list will be generated for all ports. SPUTIL
recognizes a connection between a port n and a port m,
when the magnitude of Smn (port n to port m direction)
exceeds CONNECTION_TOL at any frequency point.

CONNECTION_TOL value Specifies the tolerance for the CHECK_CONNECTION

command. The default value of this tolerance is 0.1.

XLINELENGTH value Specifies the line length of the transmission line system
from which the S parameter model is extracted. This
must be exactly the same as the line length of the
extracted system.
For information on usage, see RLGC Model Output from
S-parameters.

WRITEXLINEPARAMS = prefix Specifies the prefix of the output RLGC file name.
For information on usage, see RLGC Model Output from
S-parameters.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 183
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Table 3 Current SPutil Command Set (Continued)

Keyword Description

WINDOWING target Fstart Fend [tolerance This command applies a function to roll off the
[tolerance_fstart tolerance_fend]] magnitude terms from the initial magnitude value at
the Fstart frequency to a value of zero at the Fend
frequency.
• The value of target can be one of the following:
• all: All terms.
• minor: Only minor terms where the maximum
magnitude is below the tolerance.
• major: Only major terms where the maximum
magnitude is above the tolerance.
• diag: Diagonal terms only.
• offdiag: Off diagonal terms only.
tolerance_fstart and tolerance_fend specify the range of
frequency for tolerance checking when target is specified
as major or minor.

SCALE_PARAM target scale_value [tolerance This command applies a scaling factor, scale_value, to
[tolerance_fstart tolerance_fend]] target parameters.
For a description of target, tolerance, tolerance_fstart,
and tolerance_fend arguments, see the WINDOWING
command description.

ADD_DELAY target delay_value [tolerance This command adds a propagation delay, delay_value,
[tolerance_fstart tolerance_fend]] to target parameters.
For a description of target, tolerance, tolerance_fstart,
and tolerance_fend arguments, see the WINDOWING
command description.

MAKE_DOT_LIN filename.sp [start_index] Creates a PrimeSim HSPICE .LIN netlist,

[prefix_string] filename.sp. The netlist uses an S-element with the
given S-parameter file. All the S-element terminals will
be terminated with port elements.
If the ZREF_OUT keyword is specified, port elements use
the specified reference impedance value(s); otherwise,
the value(s) specified in the S-parameter file will be
used.
Optionally, the starting port index can be specified by
the start_index keyword. If prefix_string is given,
the string will be added to the node name as well as the
S-element name.
When the given S-parameter file contains sampling
frequency information, the resulting netlist will
contain an .AC command with all the given frequency
points, so that the AC analysis will not need to use
interpolation/extrapolation.

! (comment) Lines beginning with “!” or “*” are ignored.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 184
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Note on the COLUMN_SUM Keyword

Given incident vector , reflected vector , the S-parameter matrix relates vector
and as:
.
Therefore, each term of the vector expands as:

As a result, in the output, vector, the energy originated from the term, becomes:

Therefore, indicates that is amplified.

RLGC Model Output from S-parameters

The SPutil command set includes two keywords XLINELENGTH and WRITEXLINEPARAMS
to output RLGC model information from an S-parameter model.
A SPutil script example to output RLGC model information using these two keywords:
***********sputil.sp***************************
NPORT 4
TSTONEFILE test.s4p
XLINELENGTH 0.0127
WRITEXLINEPARAMS = sputil_test
*************************************************

The following five files are output:

• $prefix_RLGC.inc: The RLGC model information specification.
• $prefix_R.inc: The frequency-dependent R-parameter file.
• $prefix_L.inc: The frequency-dependent L-parameter file.
• $prefix_G.inc: The frequency dependent G-parameter file.
• $prefix_C.inc: The frequency dependent C-parameter file.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 185
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

where $prefix is sputil_test for the preceding example.

Note:
Currently, SPutil only supports S-parameters with nodemap=i1 i2...iN o1
o2...oN. That is, the input ports are all specified before the output ports.

SPUTIL Functions
The following are the SPUTIL functions:
• write_fitted_file
Use the write_fitted_file function to generate an S-parameter file from the rational
function file file.srf. The generated S-parameter file is then used to compare to the
original S-parameter file from which the rational function is generated.
TSTONEFILE <input_sparam_file>
rational_func=1
write_fitted_file 1

• write_bbs_file
Use the write_bbs_file function to generate a broadband SPICE model from an S-
parameter file. This function is used with rational_func=1 for a broadband SPICE
model.
TSTONEFILE <input_sparam_file>
rational_func 1
write_bbs_file 1

• rfmfile
Use the rfmfile function to manipulate a rational function matrix file (.rfm) in SPUTIL.
The runset generates an S-parameters file from the RFM file. Frequency sweep is
optional and used to specify a given frequency range in the output file.
Note:
Currently broadband SPICE model output is supported in S-domain rational
function model only.
RFMFILE <input.rfm>
FREQ_SWEEP <sweep_type>
TSTONE_OUT <output_sparam_file>

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 186
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

Status Messages
Status messages are written in the command terminal as an SPutil runset proceeds. For
example, when the following runset is executed:
NPORT 4
TSTONEFILE good_tx.s4p
PASSIVE 1
TSTONE_OUT output.s4p
FREQ_SWEEP DEC 10 1e3 1e9

The following types of messages might be returned:

% sputil test.in
Reading Touchstone File [good_tx.s4p]...

Read Touchstone File [good_tx.s4p]

s-params: 801 pts 300000Hz - 8.5e+09Hz

no noise parameter read

Performing passivity check...

Warning: s-model [good_tx.s4p] passivity violation detected...

Writing Touchstone File [output.s4p]... done
See "test.lis" for results and statistics.

Causality Checking
• The SPUTIL's causality checker checks the causality of the given frequency points in
the raw S-parameter.
• The IFFT causality enforcement during simulation in the PrimeSim HSPICE and
PrimeSim HSPICE tool is to ensure causality when preparing data for the IFFT
algorithm throughout the extrapolated frequency points. The SPUTIL causality checker
is used to check the quality of the raw S-parameters prior to causality enforcement
during simulation.
To check the sampled data included in a scattering-parameters Touchstone-file, the input
file should look like:
TSTONEFILE rlc128p_7169_0_1e11.s128p
CAUSAL CHECK
CAUSAL_TOL 0.953496955

In this example, the input file name represents a 128 port-network (whose data has been
sampled 7169 times for a frequency range that goes from 0 Hz to 1e11 Hz). Causality
checking is asked using a GDM better than 0.953496955.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 187
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-parameter Standalone Manipulation Utility

There are two output files. One of them is a *.lis file that displays a short summary as in
the following example:
Parsing rlc128p_7169_0_1e11.in ...
Reading TSTONEFILE rlc128p_7169_0_1e11.s128p......
Touchstone File [rlc128p_7169_0_1e11.s128p] s-params: 7169 pts 0Hz -
1e+11Hz
No noise parameter read.
s-parameters: 7169 pts 0 - 1e+11
no noise parameter read
The three worst e-average Global Difference Measures (1.0 >= eGDM >= 0.0)
for 8192 points are listed below.
Port combination [58,122], its average eGDM is: 0.987236
Port combination [122,58], its average eGDM is: 0.987236
Port combination [3,68], its average eGDM is: 0.988165
Average eGDM for the input Scattering Matrix: 0.999882
Note: 1.0 >= eGDM > 0.65 is good. 0.65 >= eGDM > 0.2 is fair.
0.20 >= eGDM is poor.
More details can be seen by plotting rlc128p_7169_0_1e11REP.s128p
SPUTIL concluded

The other output file is an RI Touchstone file where the suffix .sNp is replaced by REP.sNp
(with N being the number of ports). The file holds, for each frequency point, the GDM value
in the corresponding real part and a causality boolean in the imaginary part. That is, one
for each GDM is equal to or less than the threshold given in CAUSAL_TOL, or zero when the
GDM is greater than that threshold.
For the given example, the name of the second output file is
rlc128p_7169_0_1e11REP.s4p.

Using Interpolation for Causality Checking

When the SPUTIL CAUSAL CHECK command is used, interpolation might be required. The
causality checking uses FFT. So, the number of frequency points must be an even power
of 2.
The interpolation is defined as follows:
• Number of frequency points after the interpolation:

• Frequency spacing:

In the preceding example, the number of frequency points, k, is 7169 and the maximum
frequency, fmax, is 100GHz. The number of frequency points is interpolated to 8192

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 188
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Preconditioning S-parameters

evenly-spaced points. The number of frequency points used in the causality check global
difference measurements are printed to the *.lis file.

Preconditioning S-parameters
Certain S-parameters, such as series inductor (2-port), show a singularity when converting
S- to Y-parameters. To avoid this singularity, the S-element adds series resistance
to precondition S matrices:

Equation 43

• is the reference impedance vector.

• is the preconditioning factor.

To compensate for this modification, the S-element adds a negative resistor ( ) to
the modified nodal analysis (NMA) matrix in actual circuit compensation. To specify this
preconditioning factor, use the PREFAC keyword in the S model statement. The default
preconditioning factor is 0.75.
bufferIBIS buffers
Figure 54 Preconditioning S-parameters

S kRref S
Preconditioning

S’

S to Y

-kRref Y’
Y’
NMA stamp

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 189
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Group Delay Handler in Time Domain Analysis

Group Delay Handler in Time Domain Analysis

The S-element accepts a constant group delay matrix in time-domain analysis. You also
can express a weak dependence of the delay matrix on the frequency as a combination of
the constant delay matrix and the phase shift value at each frequency point. To activate or
deactivate this delay handler, specify the DELAYHANDLE keyword in the S model statement.
The delay matrix is a constant matrix, which PrimeSim HSPICE extracts using finite
difference calculation at selected target frequency points. PrimeSim HSPICE obtains the
delay matrix component as:

Equation 44

• is the target frequency, which you can set using DELAYFREQ. The default target
frequency is the maximum frequency point.
• is the phase of .

After time domain analysis obtains the group delay matrix, Equation 45 eliminates the
delay amount from the frequency domain system-transfer function:

Equation 45

The convolution process uses Equation 46 to calculate the delay:

Equation 46

S-Model Data Smoothing

Four smoothing functions are provided for the S-model. Each of these is available for the
S modeldata smoothingdata smoothing, S- and W-element

S-element and W-element. Scattering parameters are frequently given from measurement
instruments such as vector network analyzers (VNA). In measurements, there are many
causes of noise injection such as calibration failure, electromagnetic interference (EMI)
and so on, especially in high frequency range. For such cases, several data smoothing
functions are available to the S-parameter data reader to restore the original noiseless
data.
Data smoothing alters the original data to suppress unwanted noise. Therefore, if you are
confident of the accuracy of the original data, data smoothing is not recommended.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 190
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
S-Model Data Smoothing

Data Smoothing Methods

Each smoothed data at ith point S’i is given as a function of original data Si and its
data smoothing, S-modelS-modeldata smoothing

neighbors as:

Four functions for data smoothing are provided:

• Mean: take the average value of

• Median: take the value situated in the middle of:

• 2nd order polynomial fit: perform least square fitting of:

with 2nd order polynomial, and then compute the value at ith frequency.
• 4th order polynomial fit: perform least square fitting of:
with 4th order polynomial, and then compute the value at ith frequency.
The following sections discuss these topics:
• S-Model Syntax
• Description
• Example

S-Model Syntax
.model model_name S ....
+ [SMOOTH=val] [SMOOTHPTS=val]

Default 0
See S Model Syntax for the description of the SMOOTH and SMOOTHPTS keywords.

Description
Each smoothing function has different characteristics. You should observe the original data
on the waveform viewer when determining the smoothing filter configuration. Typically, the
average function has a strong ability of smoothing, but it may lose the necessary bumps
in data if they are narrow. Since both the average and median (SMOOTH=1 or 2) takes an
intermediate value of the points within the specified window, if these reference points have
a wide range of phase difference due to sparse frequency sampling, the smoothing result
may lose accuracy.
The Median filter is effective if there are sharp and high noise spikes. These spikes
are eliminated by the median filter without changing the offset level. Polynomial fittings
are relatively weak in data smoothing, but they preserve narrow bumps. Typically, for

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 191
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Predicting an Initial Value for FMAX in S-Element Models

transmission line type S-parameters, polynomial fittings are effective since sinusoidal
curves (many narrow bumps) are expected in real and imaginary versus frequency plots
due to constant propagation delay.
The 2nd order and 4th order polynomial smoothing methods preserve the overall
waveform trend better than the former two methods. However, they are relatively weak in
strong, narrow range noises.

Example
The plot on the left side of Figure 55 shows the original measurement data for
the propagation term of the differential pair of transmission lines. With SMOOTH=3
SMOOTHPTS=5, 2nd polynomial fitting with 11 points (5 points from each side in addition to
the target point) is applied. The plot on the right side shows smoothed data.

Figure 55 Using the smoothing keywords on S Model data

Predicting an Initial Value for FMAX in S-Element Models

When selecting a starting point for the FMAX parameter in your S-parameter, it is important
to set FMAX high enough to account for the fastest edges and higher order harmonics in
the input waveforms. There are two methods to determine a starting point for setting FMAX.
Both these methods require performing a transient analysis and are only meant to provide
an initial value. Always check your results to ensure you are getting the accuracy you
need. Also, setting FMAX without having enough data present in your S-parameter data file
may result in extrapolation errors.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 192
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
De-Embedding S-parameters

Method 1: Based on Risetime using the Knee Frequency

This method is handy for TDR-type simulations where the incident wave has only one
rising or falling edge. Most energy in digital pulses concentrates below the knee frequency.
The behavior of a circuit at the knee frequency determines its processing of a step
edge. The knee frequency for any digital signal is related to the rise and fall time of
its digital edges, but not its clock rate. If you want to pass a certain rise time with little
degradation, you need the medium it propagates through to be about two times the knee
frequency. The knee frequency can be calculated based on a 10–90% or 20–80% risetime
measurement. For 10–90%, FKNEE = (.35/Trise). For 20–80%, FKNEE = (.5/Trise). For
example, the FMAX needed for a 25-ps risetime measured at 10–90% of the rising edge is
.

Method 2: Using FFT

You can use the waveform calculator in WaveView to check the frequency and eleventh
harmonic.
In WaveView:
1. Select the data signal.
2. Open the Waveform Calculator.
3. Paste the waveform into the calculator with the middle mouse button.
4. Click the WAVE button, and select FFT.
5. Modify the number of points and start/stop times if desired. Click OK.
6. Click the Graph X button to plot the FFT.
Note:
1. PrimeSim HSPICE usually selects a suitable value of FBASE that provides a
good tradeoff between the number of sampling points and performance, so
allow FBASE to default unless you are not seeing the resolution and accuracy
you require.
N
2. Because FFT requires an array with size of 2 , PrimeSim HSPICE may
select the highest frequency to be sampled and report a different (higher)
FMAX value than given as the FMAX in the netlist.

De-Embedding S-parameters
The S-element in PrimeSim HSPICE provides an interface to use measured or extracted
S-parameters. When obtaining S-parameters by measurement, the measured S-
parameters may contain characteristics of unwanted peripherals such as probes,

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 193
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
De-Embedding S-parameters

connectors, and so on, in addition to the target device under test (DUT) characteristics.
If the characteristics of these unwanted peripherals are known as S-parameters, you can
use PrimeSim HSPICE to de-embed them through common text-based data file formats
such as Touchstone, CITIfile, and the netlist model format, *.sc# by using the existing LIN
analysis and S-elements with the following stamping scheme:

Figure 56 (Left) The measured S-parameter model may include the characteristics of
fixtures and the DUT itself. (Right) When the characteristics of fixtures are given, software
de-embedding extracts pure DUT characteristics.

S-Model Syntax

S1 1 2 3 4 5 6 7 8 mname=model_name …
+ STAMP=DEEMBED

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 194
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Small-Signal Parameter Data Frequency Table Model

or:

.model model_name S …
+ STAMP=DEEMBED

The STAMP=DEEMBED setting produces a negated stamp to de-embed a given S-parameter

block from the adjacent DUT connected in series. For example, the following netlist
removes 50-ohm resistance represented by the S1 element from the DUT of 100-ohm
resistor, R1. As a result, the LIN analysis produces S-parameters of 50-ohm series resistor.
P1 1 0 port=1 ac=1
P2 2 0 port=2
* DUT (w/ unwanted 50 ohm)
R1 m 2 100
* de-embed 50 ohm s-parameters
S1 1 m mname=smodel stamp=DEEMBED
.opt post
.ac POI 4 0.0 1.0 1e8 1e9
.lin format=touchstone
.MODEL Smodel S
+ N=2 FQMODEL=SFQMODEL TYPE=S Z0=50
.MODEL SFQMODEL SP N=2 SPACING=POI MATRIX=NONSYMMETRIC
+ DATA=1
+ 0
+ 0.3333333333 0 0.6666666667 0
+ 0.6666666667 0 0.3333333333 0
+
.end

Note:
Since negated stamp causes noncausal behavior in time domain analysis,
STAMP=DEEMBED is only supported in frequency domain analyses. For time
domain analyses, generate de-embedded S-parameters first using LIN analysis,
and then reuse them.

Small-Signal Parameter Data Frequency Table Model

The small-signal parameter data frequency table model (SP model) is a generic model that
frequency table modelsmall-signal parameter data-table modelSP modelsyntaxSP model

describes the frequency-varying behavior.

The following sections discuss these topics:
• SP Model Syntax
• Four Valid Forms of the SP Model
• Examples

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 195
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Small-Signal Parameter Data Frequency Table Model

SP Model Syntax
.MODEL name sp [N=val FSTART=val FSTOP=val NI=val
+ SPACING=val MATRIX=val VALTYPE=val INFINITY=matrixval
+ INTERPOLATION=val EXTRAPOLATION=val DC=val]
+ DATA=(npts ...)|DATAFILE=filename

Parameter Description

name Model name.

N Matrix dimension (number of signal terminals). Default is 1. If you use a value other than the
default, you must specify that value before you set INFINITY and DATA.

FSTART Starting frequency point for data. Default=0.

FSTOP Final frequency point for data. Use this parameter only for the LINEAR and LOG spacing
formats.

NI Number of frequency points per interval. Use this parameter only for the DEC and OCT
spacing formats. Default=10.

SPACING Data sample spacing format:

• LIN (LINEAR): uniform spacing with frequency step of (FSTOP-FSTART)/(npts-1). The
default.
• OCT: octave variation with FSTART as the starting frequency, and NI points per octave.
npts sets the final frequency.
• DEC: decade variation with FSTART as the starting frequency, and NI points per decade.
npts sets the final frequency.
• LOG: logarithmic spacing. FSTART and FSTOP are the starting and final frequencies.
• POI: nonuniform spacing. Pairs data
• (NONUNIFORM) points with frequency points.

MATRIX Matrix (data point) format:

• SYMMETRIC: symmetric matrix. Specifies only lower-half triangle of a matrix (default).
• HERMITIAN: similar to SYMMETRIC; off-diagonal terms are complex-conjugates of each
other.
• NONSYMMETRIC: non-symmetric (full) matrix.

VALTYPE Data type of matrix elements:

• REAL: real entry.
• CARTESIAN: complex number in real/imaginary format (default).
• POLAR: complex number in polar format. Specify angles in radians.

INFINITY Data point at infinity. Typically real-valued. This data format must be consistent with MATRIX
and VALTYPE specifications. npts does not count this point.

INTERPOLATION Interpolation scheme:

• STEP: piecewise step. This is the default.
• LINEAR: piecewise linear.
• SPLINE: b-spline curve fit.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 196
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Small-Signal Parameter Data Frequency Table Model

Parameter Description

EXTRAPOLATION Extrapolation scheme during simulation:

• NONE: no extrapolation is allowed. Simulation terminates if a required data point is
outside of the specified range.
• STEP: uses the last boundary point. The default.
• LINEAR: linear extrapolation by using the last two boundary points.
If you specify the data point at infinity, then simulation does not extrapolate and uses the
infinity value.

npts Number of data points.

DC Data port at DC. Normally real-valued. This data format must be consistent with MATRIX and
VALTYPE specifications. npts does not count this point. You must specify either the DC point
or the data point at frequency=0.

DATA Data points.

• Syntax for LIN spacing: .MODEL name sp SPACING=LIN [N=dim] FSTART=f0 +
FSTOP=f1 DATA=npts d1 d2 ...
• Syntax for OCT or DEC spacing: .MODEL name sp SPACING=DEC or OCT [N=dim] +
FSTART=f0 NI=n_per_intval DATA=npts d1 d2 ...
• Syntax for POI spacing: .MODEL name sp SPACING=NONUNIFORM [N=dim] +
DATA=npts f1 d1 f2 d2 ...

DATAFILE Data points in an external file. This file must contain only raw numbers without any suffixes,
comments, or continuation letters. The first number in the file must be an integer value to
indicate the number of sampling points in the file. Then, sampling data must follow. The
order of sampling data must be the same as in the DATA statement. This data file has no
limitation on line length so you can enter a large number of data points.

Note:
Interpolation and extrapolation occur after the simulator internally converts the
Z- and S-parameter data to Y-parameter data.

Four Valid Forms of the SP Model

The following four sample files are valid forms of the SP model:
SP Model 1: Symmetric complex matrices in linear frequency spacing:
.MODEL fmod SP N=2 FSTOP=30MegHz
+ DATA = 2
* matrix at f=0
+ 0.02 0.0
* Re(Y11) Im(Y11)
+ -0.02 0.0 0.02 0.0
* Im(Y21) Im(Y21) (= Y21) Re(Y22) Im(Y22)
* matrix at f=30MHz
+ 0.02 0.0
* Re(Y11) Im(Y11)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 197
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Small-Signal Parameter Data Frequency Table Model

+ -0.02 0.0 0.02 0.0

* Im(Y21) Im(Y21) (= Y21) Re(Y22) Im(Y22)

SP Model 2: Nonsymmetric complex matrices in linear frequency spacing:

.MODEL fmod SP N=2 FSTOP=30MegHz MATRIX=NONSYMMETRIC
+ DATA = 2
* matrix at f=0
+ 0.02 0.0 -0.02 0.0
* Re(Y11) Im(Y11) Re(Y12) Im(Y12)
+ -0.02 0.0 0.02 0.0
* Im(Y21) Im(Y21) Re(Y22) Im(Y22)
* matrix at f=30MHz
+ 0.02 0.0 -0.02 0.0
* Re(Y11) Im(Y11) Re(Y12) Im(Y12)
+ -0.02 0.0 0.02 0.0
* Im(Y21) Im(Y21) Re(Y22) Im(Y22)

SP Model 3: Symmetric complex matrices in nonuniform frequency spacing:

.MODEL fmod SP N=2 SPACING=POI
+ DATA = 1
+ 0.0 * first frequency point
* matrix at f=0
+ 0.02 0.0
* Re(Y11) Im(Y11)
+ -0.02 0.0 0.02 0.0
* Im(Y21) Im(Y21) (= Y21) Re(Y22) Im(Y22)
+ 30e+6 * second frequency point
* matrix at f=30MHz
+ 0.02 0.0
* Re(Y11) Im(Y11)
+ -0.02 0.0 0.02 0.0
* Im(Y21) Im(Y21) (= Y21) Re(Y22) Im(Y22)

SP Model 4: Nonsymmetric real matrices in linear frequency spacing:

.MODEL fmod SP N=2 FSTOP=30MegHz VALTYPE=REAL
+ MATRIX=NONSYMMETRIC
+ DATA = 2
* matrix at f=0
+ 0.02 -0.02
* Y11 Y12
+ -0.02 0.02
* Y21 Y22
* matrix at f=30MHz
+ 0.02 -0.02
* Y11 Y12
+ -0.02 0.02
* Y21 Y22

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 198
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Small-Signal Parameter Data Frequency Table Model

Examples
Example 1
**S-parameter example
.OPTION post=2
.probe v(n2)
V1 n1 0 ac=1v PULSE 0v 5v 5n 0.5n 0.5n 25n
.op
.ac lin 500 1Hz 30MegHz
.tran 0.1ns 10ns
*S1 n1 n2 0 mname=s_model
S1 n1 n2 0 mname=s_model
.model s_model S fqmodel=fmod Z0=50 50
*.model s_model S fqmodel=fmod2 Z0=50 100
* S parameter for Z0=(50 50)
.MODEL fmod SP N=2 FSTOP=30MegHz DATA = 1
+ 0.333333333 0.0 0.666666667 0.0 0.333333333 0.0
* S parameter for Z0=(50 100)
.MODEL fmod2 SP N=2 FSTOP=30MegHz MATRIX=NONSYMMETRIC
+ DATA = 1
+ 0.5 0.0 0.5 0.0
+ 1.0 0.0 0.0 0.0
Rt1 n2 0 50
.end

Example 2
Figure 57 illustrates a transmission line that uses a resistive termination, and Table 4
transmission linesresistive termination

shows a corresponding input file listing. In this example, the two outputs from the resistor
and S-parameter modeling must match exactly.
transmission linesS ElementS-elementtransmission line
Figure 57 Transmission Line with Resistive Termination

Four-conductor line
Ro, L, Go, C, Rs, Gd

+
v1
-

Reference conductor

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 199
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Small-Signal Parameter Data Frequency Table Model

Table 4 Input File Listing

Header, options, and *S-parameter x-line with a resistive positive termination.OPTION
sources POST V1 i1 0 ac=1v

Termination x1 o1 o2 o3 0 terminator

Transmission line W1 i1 i2 i3 0 o1 o2 o3 0 RLGCMODEL= N=3 + L=0.97 .MODEL wrlgc W

(W-element) MODELTYPE=RLGC N=3 + Lo = 2.78310e-07 + 8.75304e-08 3.29391e-07
+ 3.65709e-08 1.15459e-07 3.38629e-07 + Co = 1.41113e-10 +
-2.13558e-11 9.26469e-11 + -8.92852e-13 -1.77245e-11 8.72553e-11

Frequency model definition .MODEL fmod sp N=3 FSTOP=30MegHz DATA= 1 + -0.270166 0.0 + 0.322825
0.0 -0.41488 0.0 + 0.17811 0.0 0.322825 0.0 -0.270166 0.0

Resistor elements .SUBCKT terminator n1 n2 n3 ref

R1 n1 ref 75
R2 n2 ref 75
R3 n3 ref 75
R12 n1 n2 25
R23 n2 n3 25
.ends terminator

Analysis .AC lin 500 0Hz 30MegHz

.DC v1 0v 5v 1v

Equivalent S-parameter .ALTER S parameter case

element .SUBCKT terminator n1 n2 n3 ref S1 n1 n2 n3 ref
+ FQMODEL=fmod
.ENDS terminator
.END

Example 3
The transmission-line example shown here uses capacitive network termination. The two
outputs from the resistor and S-parameter modeling in Example 2 differ slightly due to the
linear frequency dependency relative to the capacitor. To remove this difference, use the
linear interpolation scheme in .MODEL.

Frequency model .MODEL fmod sp N=3 FSTOP=30MegHz

definition + DATA= 2
+ 1.0 0.0
+ 0.0 0.0 1.0 0.0
+ 0.0 0.0 0.0 0.0 1.0 0.0
+ 0.97409 -0.223096
+ 0.00895303 0.0360171 0.964485 -0.25887
+ -0.000651487 0.000242442 0.00895303
+ 0.0360171 0.97409 -0.223096

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 200
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
Small-Signal Parameter Data Frequency Table Model

Using capacitive .SUBCKT terminator n1 n2 n3 ref

elements C1 n1 ref 10pF
C2 n2 ref 10pF
C3 n3 ref 10pF
C12 n1 n2 2pF
C23 n2 n3 2pF
.ENDS terminator

Example 4
Figure 58 and Table 5 show an example of a transmission line that uses the S-parameter.
s-parametertransmission line
Figure 58 3-Conductor transmission line

3-conductor line

Ro, L, Go, C, Rs, Gd

+
v1
-
Reference conductor

Table 5 Input File Listing

Header, options, and *S parameter ex3: modeling x-line by using
sources + S parameter
.OPTION POST
vin in0 0 ac=1

Analysis .AC lin 100 0 1000meg

.DC vin 0 1v 0.2v

Transmission line W1 in1 in2 0 out1 out2 0 N=2 RLGCMODEL=m2

Termination R1 in0 in1 28

R2 in2 0 28
R3 out1 0 28
R4 out2 0 28

W-element RLGC model .MODEL m2 W ModelType=RLGC, N=2

W ElementsRLGC model definitionRLGC model definition, W_Element

definition + Lo= 0.178e-6 0.0946e-7 0.178e-6

+ Co= 0.23e-9 -0.277e-11 0.23e-9
+ Ro= 0.97 0 0.97
+ Go= 0 0 0
+ Rs= 0.138e-3 0 0.138e-3
+ Gd= 0.29e-10 0 0.29e-10

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 201
W-2024.09
 Feedback
Chapter 5: S-parameter Modeling Using the S-element
References

Table 5 Input File Listing (Continued)

Frequency model definition .MODEL SM2 sp N=4 FSTART=0 FSTOP=1e+09
+ SPACING=LINEAR DATA= 60 0.00386491 0
+ 0 0 0.00386491 0 0.996135 0 0 0 0.00386491 0
+ 0 0 0.996135 0 0 0 0.00386491 0
+ -0.0492864 -0.15301
+ 0.00188102 0.0063569 -0.0492864
+ -0.15301 0.926223 -0.307306 0.000630484
+ -0.00154619 0.0492864 -0.15301
+ 0.000630484 -0.00154619 0.926223
+ -0.307306 0.00188102 0.0063569
+ -0.0492864 -0.15301 -0.175236 -0.241602
+ 0.00597 0.0103297 -0.175236 -0.241602
+ 0.761485 -0.546979 0.00093508
+ -0.00508414 -0.175236 -0.241602
+ 0.00093508 -0.00508414 0.761485
+ -0.546979 0.00597 0.0103297 -0.175236
+ -0.241602
+ ...

Equivalent S-parameter .SUBCKT terminator n1 n2 n3 ref

element S1 n1 n2 n3 ref FQMODEL=SM2
.ENDS terminator
.END

References
[1] Dmitri Borisovich Kuznetsov and Jose E. Schutt-Aine, “Optimal Transient Simulation of
Transmission Lines”, IEE Transaction on Circuits and Systems-I: Fundamental Theory
and Applications. Vol. 43, No. 2, February 1996
[2] Bjorn Gustavsen and Adam Semlyen, “Rational Approximation of Frequency Domain
Responses by Vector Fitting,” IEEE Transaction on Power Delivery, Vol.14, No.3, pp.
1052-1061, July 1999

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 202
W-2024.09
 Feedback

6
W- and T-elements Transmission-Line Modeling

Describes how to use basic and advance transmission-line models and the various
methods available to input transmission-line descriptions.

PrimeSim HSPICE ships many examples for your use. For W-element-related demo files,
see Transmission (W-element) Line Examples in the PrimeSim HSPICE User Guide:
Demonstration Netlists.
The following sections discuss these topics:
• Overview of W-element Usage
• Using the W-element
• Controlling W-element Accuracy
• Understanding Frequency-dependent Matrices and Transmission Line Theory
• Overview of T-element Usage
• Using T-element for Predicting Simple Tx Line Effects
• References

Overview of W-element Usage

The W-element is a versatile transmission line model that you can apply to efficiently and
accurately simulate transmission lines, ranging from a simple lossless line to complex
frequency-dependent lossy-coupled lines. Unlike the U-element, the W-element can output
accurate simulation results without fine-tuning optional parameters. For more information
on U-elements, see Appendix B, Simulating Circuits with Xilinx FPGAs.
A transmission line is a passive element that connects any two conductors at any distance
apart. One conductor sends the input signal through the transmission line and the other
conductor receives the output signal from the transmission line. The signal that transmits
from one end of the pair to the other end is voltage between the conductors.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 203
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Overview of W-element Usage

Examples of transmission lines include:

• Traces on printed circuit boards and multichip modules (MCMs)
• Bonding wires in semiconductor IC packages
• On-chip interconnections
This chapter describes the basic transmission-line simulation equations. It explains how to
use these equations as an input to the transmission line model, the W-element. (For more
information about the W-element, see [1].)
For an optional method to compute the parameters of the transmission line equations
by using the field solver model, see Chapter 7, Using the Field Solver to Extract Tx Line
Parameters. This simple (but efficient and accurate) 2D electromagnetic field solver
calculates the electrical parameters of a transmission line system based on its cross-
section.
For less complex transmission line analysis, see Using T-element for Predicting Simple Tx
Line Effects at the end of this chapter.
The following section discusses these topics:
• W-element Capabilities

W-element Capabilities
The W-element is a multiconductor lossy frequency-dependent transmission line. It
provides advanced modeling capabilities for transmission lines. The W-element provides:
• Ability to extract analytical solutions for AC and DC.
• No limit on the number of coupled conductors.
• No restriction on the structure of RLGC matrices; all matrices can be full.
• No spurious ringing, such as is produced by the lumped model. (See Figure 59.)
• Accurate modeling of frequency-dependent loss in the transient analysis.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 204
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

• Built-in 2D field solver, which you can use to specify a physical line shape.

Figure 59 Spurious ringing in U-element

The W-element supports the following types of analysis:

• DC
• AC
• Transient
• RF analyses (HB, HBAC, HBNOISE, PHASENOISE, LIN)
• Parameter sweeps
• Optimization
• Monte Carlo
• StatEye

Using the W-element

For detailed information on the W-element syntax and examples, see W-element
(Distributed Transmission Lines) in PrimeSim Continuum User Guide: Elements.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 205
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

The following sections discuss these topics:

• Input Model 1: W-element, RLGC Model
• Input Model 2: U-element, RLGC Model
• Input Model 3: Built-in Field-Solver Model
• Input Model 4: Frequency-Dependent Tabular Model
• Input Model 5: S Model
• Input for W-element Passive Noise Model

Input Model 1: W-element, RLGC Model

Equations and Parameters describes the inputs of the W-element per unit length matrices:
Ro (DC resistance), L, G, C, Rs (skin effect), and Gd (dielectric loss).
The W-element does not limit any of the following parameters:
• Number of coupled conductors.
• Shape of the matrices.
• Line loss.
• Length or amount of frequency dependence.
The RLGC text file contains frequency-dependent RLGC matrices per unit length. The W-
element also handles frequency-independent RLGC, and lossless (LC) lines. It does not
support RC lines.
Because RLGC matrices are symmetrical, the RLGC model specifies only the lower
triangular parts of the matrices.
The following sections discuss these topics:
• Syntax
• Sample Netlist
• Simulation Results
• Specifying the RLGC Model in an External File

Syntax
The syntax of the RLGC model for the W-element is:

.MODEL name W MODELTYPE=RLGC N=val

+ Lo=matrix_entries

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 206
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

+ Co=matrix_entries [Ro=matrix_entries Go=matrix_entries]

+ Rs=matrix_entries wp=val Gd=matrix_entries Rognd=val
+ Rsgnd=val Lgnd=val

Parameter Description

N Number of conductors (same as in the element card).

Lo DC inductance matrix, per unit length .

Co DC capacitance matrix, per unit length .

Ro DC resistance matrix, per unit length .

Go DC shunt conductance matrix, per unit length .

Rs Skin effect resistance matrix, per unit length .

Gd Dielectric loss conductance matrix, per unit length .

wp Angular frequency of the polarization constant [radian/sec] (see Introducing the Complex
Dielectric Loss Model). When the wp value is specified, the unit of Gd becomes [S/m].
The wp parameter is an array property; you can specify as many polarization frequencies
as you want to achieve Equation 61.

Lgnd DC inductance value, per unit length for grounds (reference line).

Rognd DC resistance value, per unit length for ground .

Rsgnd Skin effect resistance value, per unit length for ground .

Sample Netlist
The following input netlist file shows RLGC input for the W-element:
* W-Element example, four-conductor line
W1 N=3 1 3 5 0 2 4 6 0 RLGCMODEL=example_rlc l=0.97
V1 1 0 AC=1v DC=0v pulse(4.82v 0v 5ns 0.1ns 0.1ns 25ns)
.AC lin 1000 0Hz 1GHz
.DC v1 0v 5v 0.1v
.tran 0.1ns 200ns

* RLGC matrices for a four-conductor lossy

.MODEL example_rlc W MODELTYPE=RLGC N=3
+ Lo=
+ 2.311e-6
+ 4.14e-7 2.988e-6

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 207
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

+ 8.42e-8 5.27e-7 2.813e-6

+ Co=
+ 2.392e-11
+ -5.41e-12 2.123e-11
+ -1.08e-12 -5.72e-12 2.447e-11
+ Ro=
+ 42.5
+ 0 41.0
+ 0 0 33.5
+ Go= + 0.000609
+ -0.0001419 0.000599
+ -0.00002323 -0.00009 0.000502
+ Rs=
+ 0.00135
+ 0 0.001303
+ 0 0 0.001064
+ Gd=
+ 5.242e-13
+ -1.221e-13 5.164e-13
+ -1.999e-14 -7.747e-14 4.321e-13
.end

Simulation Results
The following three figures show plots of the simulation results:
• Figure 60 shows DC sweep.
• Figure 61 shows AC response.
• Figure 62 shows transient waveforms.
These figures also demonstrate that the transmission line behavior of interconnects
has a significant and complicated effect on the integrity of a signal. This is why it is very
important to accurately model transmission lines when you verify high-speed designs.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 208
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

Figure 60 Simulation results: DC sweep

Figure 61 Simulation results: AC response

5

4
Frequency Responses (V)

V4

2
V1

V5
0
0 200 400 600 800 1000
Frequency (MHz)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 209
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

Figure 62 Simulation results: transient waveforms

6

V1
4
Transient Waveforms (V)

2
V4

0 V5

-2

-4
0 50 100 150 200
Time (ns)

Specifying the RLGC Model in an External File

You can also specify RLGC matrices in an RLGC file. Its file format is more restricted than
the RLGC model; for example:
• You cannot include any parameters.
• The file does not support ground inductance and resistance.
Note:
This format does not provide any advantage over the RLGC model so do not
use it unless you already have an RLGC file. It is supported for backward-
compatibility.
The RLGC file only specifies the lower-triangular parts of the matrices and is order-
dependent. Its parameters are in the following order:
Table 6 Parameters in RLGC File for W-element
Parameter Description

N Number of conductors (same as in the element card).

L DC inductance matrix, per unit length .

C DC capacitance matrix, per unit length .

Ro (Optional) DC resistance matrix, per unit length .

Go (Optional) DC shunt conductance matrix, per unit length .

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 210
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

Table 6 Parameters in RLGC File for W-element (Continued)

Parameter Description

Rs (Optional) Skin effect resistance matrix, per unit length .

Gd (Optional) Dielectric loss conductance matrix, per unit length .

Note:
You can skip the optional parameters, because they default to zero. But if you
specify an optional parameter, then you must specify all preceding parameters,
even if they are zero.
An asterisk (*) in an RLGC file comments out everything until the end of that line. You can
use any of the following characters to separate numbers:

space tab newline , ; ( ) [ ] { }

This RLGC file is for the same netlist example used for the RLGC model in the Input
Model 1: W-element, RLGC Model:
* W- Element example, four-conductor line

W1 N=3 1 3 5 0 2 4 6 0 RLGCfile=example.rlc l=0.97

V1 1 0 AC=1v DC=0v pulse(4.82v 0v 5ns 0.1ns 0.1ns 25ns)

.AC lin 1000 0Hz 1GHz

.DC v1 0v 5v 0.1v
.tran 0.1ns 200ns

.end

Calls this example.rlc file:

* RLGC parameters for a four-conductor lossy
* frequency-dependent line
* N (number of signal conductors)
3
* Lo
2.311e-6
4.14e-7 2.988e-6
8.42e-8 5.27e-7 2.813e-6
* Co
2.392e-11
-5.41e-12 2.123e-11
-1.08e-12 -5.72e-12 2.447e-11
* Ro
42.5

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 211
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

0 41.0
0 0 33.5
* Go
0.000609
-0.0001419 0.000599
-0.00002323 -0.00009 0.000502
* Rs
0.00135
0 0.001303
0 0 0.001064
* Gd
5.242e-13
-1.221e-13 5.164e-13
-1.999e-14 -7.747e-14 4.321e-13

The RLGC file format does not support scale suffixes, such as:
n (10^-9) or p (10^-12)

Input Model 2: U-element, RLGC Model

The W-element accepts the U Model as an input to provide backward compatibility with the
U-element. It also uses the geometric and measured-parameter interfaces of the U model.
To use the W-element with the U Model on the W-element card, specify:

Umodel=U-model_name

The W-element supports all U model modes, including:

• geometric, Elev=1
◦ planar geometry, Plev=1
◦ coax, Plev=2
◦ twin-lead, Plev=3
• RLGC, Elev=2
• measured parameters, Elev=3
• skin-effect, Nlay=2
The only exception is Llev=1, which adds the second ground plane to the U model. The
W-element does not support this. To model the extra ground plane, add an extra conductor
to the W-element in Elev=2, or use an external lumped capacitor in Elev=1 or Elev=3. For
information about the U model, see Appendix B, Simulating Circuits with Xilinx FPGAs

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 212
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

The following sections discuss these topics:

• Using RLGC Matrices
• Sample Netlist

Using RLGC Matrices

RLGC matrices in the RLGC model of the W-element are in the Maxwellian format. In
the U model, they are in self/mutual format. For conversion information, see Determining
Matrix Properties. When you use the U model, the W-element performs the conversion
internally. Table 7 shows how the RLGC matrices in the U Model are related to the RLGC
matrices in the W-element, and how the W-element uses these matrices.
Handling the Dielectric-loss Matrix
Because the U model does not input the Gd dielectric loss matrix, the W-element defaults
Gd to zero when it uses the U model input.
Handling the Skin-effect Matrix
The U- and W-elements use the Rs skin-effect resistance in different ways.
• In a W-element, the Rs matrix specifies the square-root dependence of the frequency-
dependent resistance (see the equations in Frequency-Dependent Matrices).
• In a U-element, R is the value of skin resistance at the frequency:

Equation 47

In the preceding equation, the core resistance (Rc) is equivalent to the DC resistance
(Ro) in the W-element. The frequency at which the U-element computes the R matrix
is:

Equation 48

Table 7 RLGC Matrices for U and W elements

For U models with W-element

RLGC input; Elev=2 Uses the Rs values that you specify in the U model.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 213
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

For U models with W-element

Geometric input; Elev=1 Divides the Rs (which the U model computes internally), by
to obtain the Rs value. For Elev=1, the Rs value in the U model
printout is not the same as the Rs value in the W-element.

Measured-parameter input; Does not support the skin effect.

Elev=3

If you do not specify the RISETIME option, the U-element uses Tstep from the .TRAN card.
Table 8 RLGC Matrices in the W-element and the U Model

W- U Model parameters
element
paramet
ers

L, C

Go, Gd

Ro Nlay=1 (no skin effect) Nlay=2 (skin effect present)

Rs Nlay=1 (no skin effect) Nlay=2 (skin effect present)

Sample Netlist
The following netlist is for a 4-conductor line as shown in Figure 63.
* W Element example, four-conductor line, U model
W1 1 3 5 0 2 4 6 0 Umodel=example N=3 l=0.97
.MODEL example U LEVEL=3 NL=3 Elev=2 Llev=0 Plev=1 Nlay=2
+ L11=2.311uH

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 214
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

+ L12=0.414uH L22=2.988uH
+ L13=84.2nH L23=0.527uH L33=2.813uH
+ Cr1=17.43pF
+ C12=5.41pF Cr2=10.1pF
+ C13=1.08pF C23=5.72pF Cr3=17.67pF
+ R1c=42.5 R2c=41.0 R3c=33.5
+ Gr1=0.44387mS
+ G12=0.1419mS Gr2=0.3671mS
+ G13=23.23uS G23=90uS Gr3=0.38877mS
+ R1s=0.00135 R2s=0.001303 R3s=0.001064
V1 1 0 AC=1v DC=0v pulse(4.82v 0v 5ns 0.1ns 0.1ns 25ns)
.AC lin 1000 0Hz 1GHz
.DC v1 0v 5v 0.1v
.TRAN 0.1ns 200ns
.END

Figure 63 4-Conductor line

Four-conductor line

v1 Ro, L, Go, C, Rs, Gd v2

+ v3 v4
-
v5 v6

+ Reference conductor +
_ _

Input Model 3: Built-in Field-Solver Model

Instead of RLGC matrices, you can directly use geometric data with the W-element
by using a built-in field solver. To use the W-element with a field solver, specify
FSmodel=model_name on the W-element card. For a description of the built-in field solver,
see Field Solver Model Syntax.

Input Model 4: Frequency-Dependent Tabular Model

You can use the tabular RLGC model as an extension of the analytical RLGC model to
model any arbitrary frequency-dependent behavior of transmission lines. (This model does
not support RC lines.)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 215
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

You can use this extension of the W-element syntax to specify a table model (use
a .MODEL statement of type w). To accomplish this, the .MODEL statement refers to .MODEL
statements where the “type” is SP (described in Small-Signal Parameter Data Frequency
Table Model), which contain the actual table data for the RLGC matrices.
Note:
To ensure accuracy, the W-element tabular model requires the following:
W elementtabular model accuracyaccuracy, W element tabular model
• R and G tables require zero frequency points.
• L and C tables require infinity frequency points as well as zero frequency points.
To specify a zero frequency point, you may use DC keyword or f=0 data entry in the DATA
field of the SP model. To specify an infinity frequency point, use the INFINITY keyword of
the SP model. See also, Small-Signal Parameter Data Frequency Table Model.
The following sections discuss these topics:
• Notation Used
• Table Model Card Syntax
• Examples: 4-Conductor Tx Line and RLGC Model List
• Introducing Causality Check for W-element RLGC Table Model

Notation Used
• Lower-case variable: Scalar quantity
• Upper-case variable: Matrix quantity
• All upper-case words: Keyword
• Parentheses and commas: Optional

Table Model Card Syntax

.MODEL name W MODELTYPE=TABLE [FITGC=0|1] N=val
+ LMODEL=l_freq_model CMODEL=c_freq_model
+ [RMODEL=r_freq_model GMODEL=g_freq_model]

Parameter Description

FITCG Keyword for W Model (w/ MODELTYPE=TABLE) 1=causality check on (eliminates ringings
for segmented lines), 0= causality check off (default)

N Number of signal conductors (excluding the reference conductor).

LMODEL SP model name for the inductance matrix array.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 216
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

Parameter Description

CMODEL SP model name for the capacitance matrix array.

RLMODEL SP model name for the resistance matrix array. By default, it is zero.

GMODEL SP model name for the conductance matrix array. By default, it is zero.

The following is an example netlist of a two-line system.

.MODEL ex1 W MODELTYPE=TABLE N=2 LMODEL=lmod1
+ CMODEL=cmod1 RMODEL=rmod1 GMODEL=gmod1
.MODEL cmod1 sp N=2 SPACING=NONUNIFORM VALTYPE=REAL
+ INFINITY=(5.602360e-11 -7.047240e-12 5.602360e-11)
+ DATA=( 1,
+ (0.000000e+00 5.602360e-11 -7.047240e-12 5.602360e-11)
+ )
.MODEL cmod1 sp N=2 SPACING=NONUNIFORM VALTYPE=REAL
+ INFINITY=(3.93346e-7 4.93701e-8 3.93346e-7)
+ DATA=( 34,
+ (0.000000e+00 3.933460e-07 4.937010e-08 3.933460e-07)
+ (3.746488e+06 4.152139e-07 4.937010e-08 4.151959e-07)
........
+ (4.000000e+09 3.940153e-07 4.937010e-08 3.940147e-07)
+ )
.MODEL lmod1 sp N=2 SPACING=NONUNIFORM VALTYPE=REAL
+ DATA=( 34,
+ (0.000000e+00 8.779530e-02 6.299210e-03 8.779530e-02)
+ (3.746488e+06 6.025640e-01 6.299210e-03 6.021382e-01)
........
+ (4.000000e+09 1.690795e+01 6.299210e-03 1.689404e+01)
+ )
.MODEL rmod1 sp N=2 SPACING=NONUNIFORM VALTYPE=REAL+ DATA=( 34,
+ (0.000000e+00 8.779530e-02 6.299210e-03 8.779530e-02)
+ (3.746488e+06 6.025640e-01 6.299210e-03 6.021382e-01)
........
+ (4.000000e+09 1.690795e+01 6.299210e-03 1.689404e+01)
+ )
.MODEL gmod1 sp N=2 SPACING=NONUNIFORM VALTYPE=REAL
+ DATA=( 34,
+ (0.000000e+00 5.967166e-11 0.000000e+00 5.967166e-11)
+ (3.746488e+06 1.451137e-05 -1.821096e-06 1.451043e-05)
........
+ (4.000000e+09 1.549324e-02 -1.944324e-03 1.549224e-02)
+ )

Examples: 4-Conductor Tx Line and RLGC Model List

Table 9 is an example of a four-conductor transmission line system, and Table 10 is a list
of a tabular RLGC model.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 217
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

Table 9 Input File Listing

Listing Type W-element Tabular Model Example

Header, options and .OPTION POST V1 7 0 ac=1v dc=0.5v pulse(0.5v 1.5v 0ns 0.1ns) V2 8 0
sources dc=1v

Analysis .DC v1 0.5v 5.5v 0.1v SWEEP length POI 2 1.2 2 .AC lin 200 0Hz 1GHz
SWEEP Ro POI 3 400 41.6667 400 .TRAN 0.1ns 50ns

Termination R1 7 1 50 R2 4 0 450 R3 5 0 450 R8 6 0 450 R5 4 5 10800 R6 5 6 10800 R7

4 6 1393.5

Analytical RLGC model .SUBCKT sub 1 2 3 4 5 6 7 8 W1 1 2 3 4 5 6 7 8 l=0.1 fgd=5e6

(W-element) RLGCMODEL=analymod
n=3 .MODEL analymod W MODELTYPE=RLGC
N=3 + Lo=2.41667e-6 + 0.694444e-6 2.36111e-6 + 0.638889e-6
0.694444e-6 2.41667e-6 + Co=20.9877e-12 + -12.3457e-12 29.3210e-12
+ -4.01235e-12 -12.3457e-12 20.9877e-12 +
Ro=41.6667 + 0 41.6667 + 0 0 41.6667 + Go=0.585937e-3 + 0 0.585937e-3
+ 0 0 0.585937e-3 +
Rs=0.785e-5 + 0 0.785e-5 + 0 0 0.785e-5 +
Gd=0.285e-6 + 0 0.285e-6 + 0 0 0.285e-6 .ENDS sub

Tabular RLGC model .ALTER Tabular Model .SUBCKT sub 1 2 3 4 5 6 7 8 W1 n=3 1 2 3 4 5 6 7 8

(W-element) l=0.1 fgd=5e6 tablem odel=trmod .INCLUDE table.txt .ENDS sub

Table 10 Tabular RLGC Model

Listing Type W-element Tabular Model Example

RLGC table model .MODEL trmod W MODELTYPE=TABLE N=3 + LMODEL=lmod CMODEL=cmod

definition RMODEL=rmod GMODEL=gmod

C model .MODEL cmod sp N=3 VALTYPE=REAL INTERPOLATION=LINEAR + DATA=( 1

2.09877e-11 -1.23457e-11 2.9321e-11 + -4.01235e-12 -1.23457e-11
2.09877e-11)

L model .MODEL lmod sp N=3 VALTYPE=REAL INTERPOLATION=LINEAR + INFINITY=

2.41667e-06 6.94444e-07 2.36111e-06 + 6.38889e-07 6.94444e-07
2.41667e-06 FSTOP=1e+07 + DATA=( 25 2.41667e-06 6.94444e-07 2.36111e-06
+ 6.38889e-07 6.94444e-07 2.41667e-06 2.41861e-06 + 6.94444e-07...
2.41707e-06 6.94444e-07 2.36151e-06 + 6.38889e-07 6.94444e-07
2.41707e-06 )

R model .MODEL rmod sp N=3 VALTYPE=REAL INTERPOLATION=LINEAR + FSTOP=1e+10

DATA=( 200 41.6667 0 41.6667 0 0 41.6667 + 41.7223 0 41.7223 0 0
41.7223 ... + 42.4497 0 42.4497 0 0 42.4497 42.4517 0 42.4517 0 0 +
42.4517)

G model .MODEL gmod sp N=3 VALTYPE=REAL INTERPOLATION=LINEAR + FSTOP=1e+08

+ DATA=( 100 0.000585937 0 0.000585937 + 0 0 0.000585937 0.282764 0
0.282764 0 0 0.282764 + ... 1.42377 0 1.42377 0 0 1.42377 1.42381 0
1.42381 + 0 0 1.42381)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 218
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

Introducing Causality Check for W-element RLGC Table Model

To improve the accuracy of W-element RLGC table model, you can introduce a complex
dielectric coefficient to assure causality for the W-element RLGC table model. You can use
the keyword FITGC=[1|0] when using the W-element RLGC table model to turn on or turn
off this method. By default, FITGC=0.
Although the dielectric properties have only a slight frequency-dependent character,
they have an impact on transmission line simulation accuracy in that not only does that
signal appear at the output port before the delay time is reached, but there is also a non-
consistency between the segmented lines and an integral line. Figure 64 and Figure 65
show that when a long transmission line is separated into 50 series connected segments,
the output signal at the far end of the lines has large discrepancies compared to a
single line, while they should be the same. When applying DELAYOPT=3, the discrepancy
becomes even bigger. This is due to the errors introduced by inter-/extrapolation at a high
frequency band which are more accurately fitted, and, therefore, more explicitly exposed in
the simulation results.

Figure 64 Non-consistency between segmented lines and one integral line; DELAYOPT=0

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 219
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

Figure 65 Non-consistency between segmented lines and one integral line; DELAYOPT=3

The solution to this issue lies in the dielectric properties of a transmission line system. In
an ideal capacitor, the current that flows through the capacitor is exactly 90 degrees out
of phase with the voltage sine wave. If the ideal capacitor were filled with an insulator with
a dielectric constant of , the capacitance would increase to . However, real
dielectric materials have some resistivity associated, which leads to leakage current. This
current is completely in phase with the voltage. It can be modeled as an ideal resistor.
By conventional transmission line theory, it is modeled by conductor G. Both of C and G
are frequency-dependent parameters in a real transmission line system. The frequency-
dependent character comes from the dipoles in dielectric material. Actually, both of these
two terms relate to the number of dipoles, how large they are and how they are able to
move. These characters are described by dielectric constant of the material.

Equation 49

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 220
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

The real part of corresponds to the motion of the dipoles that are out of phase with
the applied field and contributes to increasing capacitance, while the imaginary part
corresponds to the motion of the dipoles that are in phase with the applied voltage and
contribute to the losses.
Since the frequency-dependent character of both of these terms relates to the motion of
dipoles, which can be described by , the real and imaginary part of must satisfy the
Kramers-Kronig relationship. And therefore, we can fit with rational function.

Equation 50

Once you get , we can calculate accurate and interpolation and

extrapolation, using Equation 50.
Once PrimeSim HSPICE gets a successful fitting, you can get a and
matrix from the fitting result. Since only the RMS error of the real part is considered during
the fitting process, you get only the matrix from the fitting result. For ,
you still use the conventional linear interpolation and extrapolation because of its strong
dependency on .

Input Model 5: S Model

The W-element can accept the transmission line-based S-parameters as input. To use the
W-element with the S Model on the W-element card, specify the following line:

SMODEL=Smodel_name NODEMAP=XiYj...

where:
• Smodel_name is an S-model, which is normally used for an S-element. Use the
XLINELENGTH keyword in the S-model statement to indicate the line length of the
system where the S-parameters are extracted. This keyword is required only when
you use an S-model with a W-element. See S-element Syntax in PrimeSim Continuum
User Guide: Elements for more information.
• NODEMAP is a string that assign each index of the S-parameter matrix to one of the W-
element terminals. This string must be an array of pairs that consists of a letter and a
number, (for example, Xn), where
◦ X= I, i, N, or n to indicate near end (input side) terminal of the W-element
◦ X= O, i, F, or f to indicate far end (output side) terminal of the W-element.
For example, NODEMAP = I1I2O1O2 represents that the

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 221
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

◦ 1st port of the s-matrix corresponds to the 1st near end terminal of the W-element.
◦ 2nd port of the s-matrix corresponds to the 2nd near end terminal of the W-element.
◦ 3rd port of the s-matrix corresponds to the 1st far end terminal of the W-element.
◦ 4th port of the s-matrix corresponds to the 2nd far end terminal of the W-element.
NODEMAP = I1I2I3...InO1O2O3...On is the default setting.

The following sections discuss these topics:

• S-Model Conventions
• S-Model Example

S-Model Conventions
When specifying an S model, you must adhere to the following rules and conventions:
• The size of the NODEMAP array must be the same as twice the line number of the W-
elements and also must be the same as the port count of the S-parameter matrices.
• If the W-element input model is S-MODEL, an S-model definition must accompany that
input model.
• S-parameters must have even number of terminals.
• S-parameters must be symmetric.
• S-parameters must be passive.
• Transmission-line based S-parameters can be used with different lengths of a system
when the varying length keyword (L) in a W-element instance statement is present.
• The XLINELENGTH keyword must be set when used in S-models that use W-elements.

S-Model Example
The following input netlist file shows S-model input for the W-element:
**** W Element Example: S Model ***
rout out 0 50
vin in gnd LFSR (1 0 0 0.1n 0.1n 1g 1 [5,2] rout=50)
*+ pulse(0 1 0 0.1n 0.1n 0.9n 2n)
W1 in gnd out gnd SMODEL=smodel N=1 l=0.3
+ NODEMAP=I1O1
.MODEL smodel S TSTONEFILE=w.s2p
+ XLINELENGTH=0.3
.opt accurate post
.tran .01n 20n
.end

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 222
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

Input for W-element Passive Noise Model

The W-element is a passive transmission line model. When the transmission lines are
lossy, they generate thermal noise. The W-element passive noise model is used to
describe these noise effects. This model supports normal, 2-port and multiport (.NOISE
and .LIN noisecalc=1 [or 2 for N-port]). See Noise Parameters in 2-Port and N-Port
Networks.
The following sections discuss these topics:
• Input Interface
• Output Interface

Input Interface
To trigger a passive noise model, the NOISE and DTEMP keywords in an W-element
statement are used:

W i1 i2 ... iN iR o1 o2 ... oN oR N=val L=val

+ ...
+ [NOISE=[1|0]] [DTEMP=val]

Parameter Description

NOISE Activates thermal noise.

• 1 (default): element generates thermal noise
• 0: element is considered noiseless

DTEMP Temperature difference between the element and the circuit, expressed in °C. The default is 0.0.
Element temperature is calculated as:
T = Element temperature (°K)
= 273.15 (°K) + circuit temperature (°C)
+ DTEMP (°C)
Where circuit temperature is specified using either the .TEMP statement, or by sweeping the global
TEMP variable in .DC, .AC, or .TRAN statements. When a .TEMP statement or TEMP variable is
not used, the circuit temperature is set by .OPTION TNOM, which defaults to 25 °C unless you
use .OPTION SPICE, which raises the default to 27 °C.

When NOISE=1, PrimeSim HSPICE generates a 2N×2N noise-current correlation matrix

from the N-conductor W-element admittance matrix according to Twiss' Theorem. The

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 223
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using the W-element

result can be stamped into a PrimeSim HSPICE noise analysis as 2N-correlated noise
current sources: ji (i=1~2N), as shown in Equation 51:

Equation 51

where:
i=1~N corresponding to N input terminals
i=N+1~2N corresponding to N output terminals.
The noise-current correlation matrix represents the frequency-dependent statistical
relationship between 2N noise current sources, ji (i=1~2N), shown in the following figure.

Figure 66 Frequency-dependent relationship, 2N noise current sources

Lossy System Lossless System

in1 out1

j1 j
N+1
in1 out1 in2 out2
in2 out2
j2 j
N+2
...

...

...

...

...

inN outN inN ... outN

j
N
j2 N

Output Interface
PrimeSim HSPICE creates a .lis output list file that shows the results of a noise analysis
just as any other noisy elements. The format is as follows:
**** w element squared noise voltages (sq v/hz)
element 0:w1
N(i,j) data
r(N(i,j)) data
... i,j = 1~N ...
total data

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 224
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Controlling W-element Accuracy

where:
• N(i,j) = contribution of j1jj* to the output port
• r(N(i,j)) = transimpedance of ji to the output port
• total = contribution of total noise voltage of the W-element to the output port.

Controlling W-element Accuracy

The following sections discuss these topics:
• Setting Frequency Range of Interest for Greater Accuracy
• Using Time-Step Control

Setting Frequency Range of Interest for Greater Accuracy

This section describes the keywords you can use for achieving greater accuracy of the W-
element by controlling the frequency of interest.
The following sections discuss these topics:
• Setting .OPTION RISETIME
• Using DELAYOPT Keyword for Higher Frequency Ranges
• Using DCACC Keyword for Lower Frequency Ranges

Setting .OPTION RISETIME

The RISETIME option is used to improve accuracy of the W-element analysis by setting
the focal frequency. It is applicable to all W-element modeling methods. If not explicitly
set, PrimeSim HSPICE automatically determines RISETIME and the focal frequency from
independent sources by examining their edge rates and using the fastest of those as the
effective RISETIME setting.
If you know the focal frequency, you can explicitly set RISETIME. The focal frequency is
user-determined, but the knee frequency of the design is typically a valid approximation.
Use the reciprocal of the focal frequency to set RISETIME; e.g. 1/10GHz = 100p. The
minimum value for RISETIME is 0.1p and anything smaller defaults to that value. Do not
set RISETIME to a value of 0 (zero).
Be cautious when setting RISETIME as it can have a direct impact on transient simulation
results. If it is not properly set (too low or too high compared to the maximum frequency
contained in the actual signal), responses to the high frequency components might be
omitted or inaccurately modeled.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 225
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Controlling W-element Accuracy

It is generally a good idea to explicitly set RISETIME, especially if you are using behavioral
(controlled) sources or Verilog-A sources that do not have edge rates that can be
determined by statically parsing the netlist.
You may want to let PrimeSim HSPICE determine RISETIME under the following
conditions:
• The sources are independent and not subject to unexpected incidents that modify the
predicted edge rates.
• You don't want to risk forgetting to reset RISETIME when you modify your sources.
• You share the netlist others who may use it without knowing that the RISETIME is set
such as in the case of intellectual property.
• There is a potential for making a mistake in RISETIME estimation.

Using DELAYOPT Keyword for Higher Frequency Ranges

Long transmission lines fabricated in a high polymer insulator, such as PCB traces, show
high losses in high frequencies due to dielectric loss. In such cases, the propagation delay
of the system becomes a non-constant function of frequency. To take this phenomenon
accurately, a novel pre-process function was introduced for constructing W-element
transient (recursive convolution) model with a higher level of accuracy. To activate this
new function, you can add the DELAYOPT keyword to the W-element instance line. You can
use DELAYOPT=0|1|2 to deactivate, activate, and automatically determine, respectively.
The default value is 0 (deactivate). If this function is deactivated, the W-element behaves
identically to the previous versions.
You can use DELAYOPT=3 to achieve a level of accuracy up to a tens of GHz operation
and involve harmonics up to THz order. With this option, line length limits are removed,
which frees the simulation from segmenting, and allows independence in the behavior of
the RISETIME option setting. A setting of DELAYOPT=3 automatically detects whether or
not frequency-dependent phenomena need to be recorded, which makes it identical to the
DELAYOPT=0 setting if it produces a high enough accuracy.

Note:
The DELAYOPT=3 option activates additional evaluation functions in transient
analysis, which might take longer CPU time.
To set this parameter as a global option for all W-elements in a netlist, see .OPTION
WDELAYOPT in the PrimeSim HSPICE Reference Manual: Commands and Control
Options.

Using DCACC Keyword for Lower Frequency Ranges

The W-element can take an additional step in making a time domain model check the
accuracy of low frequency and DC coverage. It automatically adds rational function terms

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 226
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Controlling W-element Accuracy

if necessary. This process may cause slight additional computational cost and slight
difference in element behavior in DC offset. Should you choose to use this conventional
behavior, set DCACC=0 in the W-element instance or model line to deactivate this process.

Using Time-Step Control

This section describes using static and dynamic time-step controls in the time domain.
The following sections discuss these topics:
• Static Time-Step Control
• Dynamic Time-Step Control using WACC

Static Time-Step Control

The W-element provides accurate results with just one or two time steps per excitation
transient (0.1 ns in Figure 59). Like the T-element, the W-element supports the TLINLIMIT
parameter. The TLINLIMIT=0 default setting enables special breakpoint building, which
limits the maximum time step by the smallest transmission line delay in the circuit. This
timestepstatic control

improves transient accuracy for short lines, but reduces efficiency. Setting TLINLIMIT=1
disables this special breakpoint building.
Longer transmission lines might experience prolonged time intervals when nothing
happens at the terminals, while the wave propagates along the line. If you increase the
time step, the accuracy of the simulation decreases when the wave reaches the terminal.
To prevent this for longer lines excited with short pulses, set .OPTION DELMAX to limit the
time step to between 0.5 and 1 of the excitation transient.

Dynamic Time-Step Control using WACC

Static time step control achieves certain accuracy by setting static breakpoints. The
timestepdynamic control

TLINLIMIT=0 parameter limits the maximum time step by the minimum transmission line
delay, which results in poor performance for cases with ultra-short delay transmission lines
because too many redundant time points are calculated, especially when the transmission
line terminal signals do not vary rapidly. The same problem exists with the DELMAX option
where time steps are evenly set in spite of terminal signal variation. This is inefficient.
The WACC option was added to solve this problem by providing dynamic step control for W-
element transient analysis. Setting WACC to a positive value removes the static breakpoints
and the necessary time points are set dynamically according to the variations in terminal
currents and voltages.
The WACC option has the following syntax:

.OPTION WACC=value

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 227
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

where WACC is a non-negative real value between 0.0 and 10.0. PrimeSim HSPICE
assigns WACC -1 if you do not set a WACC option, or if you set .OPTION WACC. When a
value of 1 is specified, PrimeSim HSPICE assigns WACC with a positive value. If a non-
negative value is set in the .option line (.OPTION WACC=XXX), PrimeSim HSPICE uses
the specified WACC value for all the W-elements. When a positive WACC value is set, the
dynamic time step control algorithm is activated. When WACC is zero, the conventional
static time step control method is used. Larger WACC values result in less restriction in time
point intervals therefore faster simulation), while smaller values result in denser time points
with higher accuracy.
A positive WACC is selected by default to activate the dynamic time step control. PrimeSim
HSPICE automatically finds the optimum WACC value based on the netlist properties such
as transmission line system delay, risetime, and transient command configurations. Since
the W-elements in the netlist may have different properties, each has its own WACC values.
If a user-specified positive WACC value is found in the netlist, PrimeSim HSPICE uses the
user-defined WACC value for all the W-elements in the netlist. If the user-specified WACC is
larger than the automatic estimation, PrimeSim HSPICE outputs a warning message.
For cases containing IBIS, PKG, EBD, or ICM blocks, PrimeSim HSPICE turns WACC off
automatically. If you want to use the dynamic time step control algorithm for IBIS related
cases, you must set it explicitly in the netlist for example:
.option WACC $ Make PrimeSim HSPICE use automatically generated WACC
value for each W element

or:
.option WACC=value $ Use this value for all the W elements

Understanding Frequency-dependent Matrices and Transmission

Line Theory
The following sections discuss these topics:
• Equations and Parameters
• Frequency-Dependent Matrices
• Wave Propagation

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 228
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

Equations and Parameters

Maxwell’s equations for the transverse electromagnetic (TEM) waves on multiconductor
transmission lines reduce to the telegrapher’s equations. The general form of the
telegrapher’s equation in the frequency domain is:

Equation 52

Equation 53

The preceding equations use the following definitions:

• Lower-case symbols denote vectors.
• Upper-case symbols denote matrices.
• is the voltage vector across the lines.
• is the current vector along the lines.
For the TEM mode, the transverse distribution of electromagnetic fields at any instant of
time is identical to that for the static solution.
From a static analysis, you can derive the four parameter matrices for multiconductor TEM
transmission lines:
• resistance matrix,
• inductance matrix,
• conductance matrix,
• capacitance matrix,
The telegrapher’s equations and the four parameter matrices from a static analysis
completely and accurately describe TEM lines.
Not all transmission lines support pure TEM waves; some multiconductor systems
inherently produce longitudinal field components. In particular, waves propagating in
either the presence of conductor losses or the absence of dielectric homogeneity (but not
dielectric losses) must have longitudinal components.
However, if the transverse components of the fields are significantly larger than the
longitudinal components, the telegrapher’s equations (and the four parameter matrices
obtained from a static analysis) still provide a good approximation (a quasi-static
approximation).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 229
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

Multiconductor systems in which this approximation is valid take the name quasi-TEM
lines. For typical micro-strip systems the quasi-static approximation holds up to a few GHz.

Frequency-Dependent Matrices
The static (constant) L and C matrices are accurate for a wide range of frequencies. In
contrast, the static (DC) R matrix applies to only a limited frequency range, mainly due to
the skin effect. A good approximate expression of the R resistance matrix with the skin
effect is:

Equation 54

or:

Equation 55

where:
• is the DC resistance matrix.
• is the skin effect matrix.
The imaginary term depicts the correct frequency response at high frequency; however,
it might cause significant errors for low-frequency applications. In the W-element,
Equation 54 can be chosen by RSTYPE=1 (default) or Equation 55 by RSTYPE=2. Also, you
can exclude this imaginary term:
Wxxx i1 i2 ... iN iR o1 o2 ... oN oR N=val L=val INCLUDERSIMAG=NO

In contrast, the G (loss) conductance matrix often approximates as:

Equation 56

where:
• models the shunt current due to free electrons in imperfect dielectrics.
• models the power loss due to the rotation of dipoles under the alternating field (C.
A. Balanis, Advanced Engineering Electromagnetics, New York: Wiley, 1989).
• fgd is a cut-off frequency.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 230
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

If you do not set fgd, or if you set fgd to 0, then G(f) keeps linear dependency on the
frequency. In the W-element, the default fgd is zero (that is, G(f) does not use the fgd
value).
You can specify an alternate value in the W-element statement:

Wxxx i1 i2 ... iN iR o1 o2 ... oN oR N=val L=val fgd=val

If you prefer to use the previous linear dependency, set fgd to 0.

Note:
fgd estimates frequency-dependent shunt loss conductance described as
Equation 52 for the RLGC model without INCLUDEGDIMAG=yes only (see Fitting
Procedure Triggered by INCLUDEGDIMAG Keyword). When you specify
INCLUDEGDIMAG=yes, the RLGC model estimates frequency-dependent shunt
(C and G) parameters as described in Equation 56 and does not use the fgd
value. Both of these are ways to fit the RLGC model with actual measurements.
If you measure or computationally extract a tabular RLGC model, you are likely
to accurately extract parameters.
The following sections discuss these topics:
• Introducing the Complex Dielectric Loss Model
• Fitting Procedure Triggered by INCLUDEGDIMAG Keyword
• Example: Complex Dielectric Loss Model
• Determining Matrix Properties
• Using the PRINTZO Option
• Defining Frequency-dependent Impedance in Mixed Mode Printing
• File Description for *.wzo

Introducing the Complex Dielectric Loss Model

When the INCLUDEGDIMAG keyword = yes and there is no wp input, the W-element regards
the Gd matrix as the conventional model and then automatically extracts constants for
the complex dielectric model. In conventional use, with the PrimeSim HSPICE W-element
RLGC model, you approximate frequency-dependent conductance as Equation 56 where
the increase of shunt conductance is due to dielectric loss.
These pure real nonconstant functions of frequency violate causality [2]. As system
operating frequency becomes high, even for PCB systems that use high polymer
dielectric materials like FR4, the appearance of the dielectric loss becomes significant
and significant noncausality of Equation 56 appears. The frequency-dependent loss of the
shunt conductance in the dielectric is mainly due to dielectric polarization.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 231
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

This polarization loss leads to a complex permittivity, , for the dielectric material [3].

Equation 57

You can specify loss tangent of the dielectric material as the ratio of imaginary part of
to the real part:

Equation 58

You can write complex electric permittivity for a single dielectric dipolar moment:

Equation 59

where:
• and are low and high frequency limits of dielectric permittivity, which are real
numbers.
• is the angular frequency that corresponds to the polarization time constant of the
dielectric material.
In Equation 59, frequency-dependent complex shunt loss conductance is expressed as [4],

Equation 60

Where the imaginary part of the conductance contributes reactively. In cases of multiple
dielectric materials surrounding the system, the complex loss conductance extends as
linear combinations of multiple dipole moments as:

Equation 61

Since Equation 61 satisfies the Krong-Kramers condition, you can ensure the passivity/
causality of the system. Activating this new model changes the definition of Gd from
conventional [S/m*Hz] to [S/m].

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 232
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

Fitting Procedure Triggered by INCLUDEGDIMAG Keyword

A fitting procedure generates a complex dielectric model with as close behavior as
possible to the conventional pure real loss conductance model while preserving passivity.
The INCLUDEGDIMAG keyword is the trigger to activate the new complex dielectric loss
model. You activate the model with conventional Go/Gd input with INCLUDEGDIMAG=yes
without polarization constant (wp) input, and the W-element automatically generates the
new model by fitting.
In this fitting process, the W-element automatically computes and values for
Equation 61 where the real part of the function fits with conventional pure real dielectric
loss model, . Then the imaginary part of derived model contributes to
the frequency dependency of the capacitance.
Because the model ensures causality, frequency domain and time domain responses
maintain better consistency (see Figure 67). Also, you reduce the functional overhead
of the DELAYOPT=3 for passive transfer functions. Thus, you improve performance of the
DELAYOPT function.

Figure 67 Improved consistency using the INCLUDEGDIMAG keyword

To set the INCLUDEGDMAG keyword as a global option for all W-elements in a netlist,
see .OPTION WINCLUDEGDIMAG in the PrimeSim HSPICE Reference Manual:
Commands and Control Options.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 233
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

Example: Complex Dielectric Loss Model

This fully executable example shows use of two methods:
1. Method 1- Specifying Gd/wp pairs
2. Method 2 - Automatic parameter generation
* complex dielectric loss model example
* -- method 1: specify Gd/wp pairs -----
.model use_wp W modeltype=rlgc n=1
+ Lo= 3.842520e-07
+ Co= 1.212598e-10
+ Ro= 7.559055e+01
+ Go= 0.000000e+000
+ Rs= 1.510184e-03
+ Gd =
+ 0.0008840 $ for wp[1]
+ 0.00365 $ for wp[2]
+ 0.04848 $ for wp[3]
+ 0.39910 $ for wp[4]
+ wp=
+ 0.04396e9 $ wp[1]
+ 0.3768e9 $ wp[2]
+ 3.1e9 $ wp[3]
+ 30e9 $ wp[4]
W1 in1 0 out1 0 N=1 RLGCmodel=use_wp l=1
+ includegdimag=yes
P11 in1 0 port=1 ac=1 sin 0 1 1e8
P12 out1 0 port=2

* -- method 2: automatic parameter generation -----

* use INCLUDEGDIMAG=yes
* without polarization constant input
.model auto_fit W modeltype=rlgc n=1
+ Lo= 3.842520e-07
+ Co= 1.212598e-10
+ Ro= 7.559055e+01
+ Go= 0.000000e+000
+ Rs= 1.510184e-03
+ Gd= 5.983933e-11
W2 in2 0 out2 0 N=1 RLGCmodel=auto_fit l=1
+ INCLUDEGDIMAG=yes
P21 in2 0 port=3 ac=1 sin 0 1 1e8
P22 out2 0 port=4
.opt post
.ac dec 10 1 1e10
.tran 0.1n 20n
.end

Determining Matrix Properties

All matrices in Frequency-Dependent Matrices are symmetric.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 234
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

• The diagonal terms of and are positive, nonzero.

• The diagonal terms of , , , and are nonnegative (can be zero).

• Off-diagonal terms of the , impedance matrices are nonnegative.

can have negative off-diagonal terms, but a warning appears. Negative off-diagonal
terms normally appear when you characterize at a frequency higher than zero.
Theoretically, should not contain negative off-diagonal terms because these might
cause errors during analysis.
• By convention, off-diagonal terms of admittance matrices , , and are negative.

• Off-diagonal terms of all matrices can be zero.

The elements of admittance matrices relate to the self/mutual admittances (such as those
that the U-element generates):

Equation 62

Equation 63

In the preceding equations, Y stands for either , , or .

A diagonal term of an admittance matrix is the sum of all self and mutual admittance in this
row. This term is larger (in absolute value) than the sum of all off-diagonal terms in its row
or column. Admittance matrices are strictly diagonally dominant (except for a zero matrix).
For example, you can express diagonal terms for capacitance matrix c as shown in the
following figure:

C12 C13
2 1 3

Cr1

C11 = Cr1 + C12 + C13

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 235
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

You can obtain loop impedance matrix terms from the partial impedance matrix:

Equation 64

where partial impedance is a ratio of induced current versus applied voltage without
considering the series and coupling effect of the return path. Then in the preceding
equation, the g index denotes a reference node which does not appear in Maxwellian
format matrices.

Using the PRINTZO Option

The PRINTZO option outputs the W-element complex characteristic impedance matrix to
a .wzo file. For simplicity, since the W-element is a symmetric system, the Zo matrix is a
symmetric matrix. Therefore, PrimeSim HSPICE only outputs the lower half of the matrix.
For discussion of the *.wzo file, see File Description for *.wzo and the following sections.
For example, the following frequency-sweep example shows the use of the PRINTZO
option with the W-element to check for characteristic impedance.
Input:
W1 N=2 in1 in2 gnd out1 out2 gnd RLGCMODEL=line2 l=0.1
+ PRINTZO=POI 3 1e6 1e9 1e12

The file line2.wzo stores the output (see the following example external file).
* w-element model [2_line] Characteristic Impedance Matrix:
.MODEL ZO SP N=2 SPACING=POI MATRIX=SYMMETRIC
+ DATA=3
+ 1.0e6
+ 175.362 -156.577
+ 3.54758 -2.53246 175.362 -156.577
+ 1.0e9
+ 48.7663 -1.3087
+ 1.69417 -0.0073233 48.7663 -1.3087
+ 1e12
+ 48.9545 0.238574
+ 1.66444 0.0348332 48.9545 0.238574

The following example shows a PRINTZO statement with the MIXEDMODEZO option enabled.
The syntax is:
W1 N=2 in1 gnd out1 out2 gnd RLGMODEL=line2_mx 1=0.1
+ PRINTZO=POI 3 1e6 1e9 1e12
+ MIXEDMODEZO=1

Then the mixed mode characteristic impedance will be stored in: line2_mx.wzo. For more
information about *.wzo, see File Description for *.wzo.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 236
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

Defining Frequency-dependent Impedance in Mixed Mode Printing

This section discusses the PrimeSim HSPICE ability to print a complex characteristic
impedance matrix in differential and common modes at given frequency points. This
functionality supports high-speed network designs where differential data transfer systems
commonly achieve a higher data transfer rate with low loss. For such designs, impedance
information in mixed (differential and common) mode is more useful than single-ended
representation. To learn more about the SP model syntax which has a complex number
matrix by default, see Small-Signal Parameter Data Frequency Table Model.
Equation 65 provides a choice to output transmission-line characteristic impedance
in mixed mode. Equation 65 also gives the characteristic impedance as a frequency-
independent constant matrix for the ideal lossless transmission line system:

Equation 65

where and are inductance and capacitance matrices of the system, respectively,
and in this case, the characteristic impedance is a real matrix. When the system becomes
lossy, in other words, the system has nonzero resistance, , and/or nonzero shunt
loss conductance, . Equation 66 expresses characteristic impedance as a function of
frequency, :

Equation 66

In this case, becomes a complex matrix. Knowledge of the characteristic impedance

( ) matrix of the transmission line system at given frequency point enables circuit
designers to establish well matched signal transfer conditions. This ability preserves
integrity of the system, especially for high frequency operation. This feature allows you to
check the complex characteristic impedance matrix of the system.
Note:
The PRINTZO function does not evaluate Equation 66, but it obtains
directly from the W-element's AC model. It does this so you can get
from other types of W-element models as well as S-parameter models.
In cases of lossless transmission line structure, the PRINTZO result may differ
from Equation 66 at very low frequencies because the RLGC-based lossless
W-element adds a small amount of loss in the very low frequency range when
it initializes the AC model. The effect of this on your actual AC or transient
simulation is negligible, but is important to achieve stable simulation.
The PrimeSim HSPICE W-element creates its own frequency-dependent characteristics
when the construct uses RLGC parameters (RLGC or RLGC table model), structural (field

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 237
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

solver) mode, U-element model, or scattering (S-parameter) model. You can compute the
characteristic impedance, not only from the RLGC model, but also from any other of W-
element configurations when you use the keyword PRINTZO to specify the frequency point.

Figure 68 Definition of mixed-mode impedance and derivation from single-ended impedance

i4
L4
v4 i3
L3
v3 i2
L2
v2 i1

v1

L2n l2d,i2c
v2d,v2
L2p

L1n
l1d,i1c
L1p

Equation 68 defines differential and common voltage as:

where and are voltage and current transformation matrices. Both single-ended
and mixed mode representations satisfy relationships of voltage and current vector
through characteristic impedance matrices as:

Equation 67

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 238
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

Substitute Equation 67 for Equation 69, and the mixed-mode characteristic impedance can
relate to the single-ended one as shown in Equation 69:

Equation 68

Equation 69

thus:

Equation 70

For example, for a system with one differential pair of lines, the transformation matrix
would be:

Equation 71

Therefore, Equation 72 expresses mixed-mode characteristic impedance as:

Equation 72

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 239
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

where:
• is differential (mode) impedance

• is common mode impedance.

Differential impedance is useful for designers to check matching characteristics of
differential signal transfer systems.
Typically, for a symmetric two-line structure with weak coupling, single-ended
characteristic impedance matrix components become:
. Therefore, mixed-mode characteristic impedance is:

Equation 73

Equation 74

File Description for *.wzo

When you include the PRINTZO option in a W-element statement, you also generate the
*.wzo file. The prefix for this file is the W-element model name:
W1 N=2 i1 i2 gnd o1 o2 gnd RLGCMODEL=line l=10m PRINTZO=POI 2 10meg 10g
+ mixedmodezo=1 $ 1= mixed mode 0= common mode

The netlist fragment creates two impedance matrices, one at 10 MHz and the other at 10
GHz. The netlist fragment then writes them to the file line.wzo for this two-conductor W-
element.
The .wzo file contains complex impedances in the form R + jX, where the first term of each
pair is R and the second is X. The off-diagonal (negative) terms relate to the interactions
between conductors (common mode). For a single-ended analysis, the diagonal terms are
identical.
For the mixed-mode configuration, the [1,1] term is the differential mode impedance, and
the [2,2] term is the common mode impedance. It is typical for the matrices to show a 3 or
4 to 1 ratio of the [1,1] to the [2,2] terms. Loosely-coupled conductors means that the ratio
is close to 4:1. This is because and .

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 240
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

Wave Propagation
To illustrate the physical process of wave propagation and reflection in transmission lines,
Figure 69 shows lines where the voltage step excites simple termination.

Figure 69 Propagation of a voltage step in a transmission line

Z1 v2 Z2
v1

v
t=t1
x
v
t=t2
x
v
t=t3
x=0 x=l x
v1 v2

0 2t 4t 6t 8t t 0 t 3t 5t 7t t
t1, t2, t3 t1, t2, t3

• At time t=t1, a voltage step from the e1 source, attenuated by the Z1 impedance,
propagates along the transmission line.
• At t=t2, the voltage wave arrives at the far end of the transmission line, is reflected, and
propagates in the backward direction. The voltage at the load end is the sum of the
incident and reflected waves.
• At t=t3, the reflected wave arrives back at the near end, is reflected again, and again
propagates in the forward direction. The voltage at the source end is the sum of
attenuated voltage from the e1 source, the backward wave, and the reflected forward
wave.
The surface plot in Figure 70 shows voltage at each point in the transmission line. The
input incident propagates from the left (length = 0) to the right. You can observe both

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 241
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

reflection at the end of the line (length = 1) and a reflected wave that goes backward to the
near end.

Figure 70 Surface plot for the transmission line shown in

Figure 69

1
V1 [V]
1
0.05

0.5
0

0 0.1
0.2 0.05
0.4
0.6 Time [ns]
0.8 0
Length [cm]

You can find more information about transmission lines in this resource: H.B. Bakoglu,
Circuits, Interconnections and Packaging for VLSI. Reading, MA: Addison-Wesley, 1990.
The following sections discuss these topics:
• Propagating a Voltage Step
• Handling Line-to-Line Junctions

Propagating a Voltage Step

This section is a summary of the process in Figure 69 to propagate a voltage step in a
transmission line.
• Signals from the excitation source spread-out in the termination networks, and
propagate along the line.
• As the forward wave reaches the far-end termination, it does the following:
◦ Reflects.
◦ Propagates backward.
◦ Reflects from the near-end termination.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 242
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

◦ Propagates forward again.

◦ Continues in a loop.
• The voltage at any point along the line, including the terminals, is a superposition of the
forward and backward propagating waves.
Figure 71 shows the system diagram for this process, where:
• Wvr and Wvb are forward and backward matrix propagation functions for voltage waves.
• T1 and T2 stand for the near-end matrix transmission and reflection coefficients.
• , (Gamma_1,Gamma_2) stand for the far-end matrix transmission and reflection
coefficients.

Figure 71 System model for transmission lines

N+1 conductor line

[v1]1 R(f), L(f), G(f), C(f) [v2]1

+ + [e2]1
[e1]1 [v1]2 Signal Conductors [v2]2 -
-
. . .
Termination

Termination
n et wo r k2
+ . .
n et wo r k1

[e1]2 . + [e2]2
. - . . . -
. .
. [v1]N
.
+ [v2]N .
[e1]M + + +
- _ Reference conductor _ [e2]M
-

0 l x

vr1 vr2
Wvr  v2

v1 v2
+ +

e1 e2
Tv1 +  v1 Wvb
+ Tv2
vb1 vb2

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 243
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

This model reproduces the general relationship between the physical phenomena of wave
propagation, transmission, reflection, and coupling in a distributed system. It can represent
an arbitrarily-distributed system, such as:
• Transmission line
• Waveguide
• Plane-wave propagation
You can use this model for:
• System analysis of distributed systems, or
• Writing a macro solution for a distributed system without complicated mathematical
derivations.
Figure 60 shows how transmission lines and terminations form a feedback system.
Because the feedback loop contains a delay, both the phase shift, and the sign of the
feedback change periodically with the frequency. This causes oscillations in the frequency-
domain response of the transmission lines, such as those in Figure 60.

Handling Line-to-Line Junctions

A special case occurs when the line terminates in another line. Figure 61 shows the
system diagram for a line-to-line junction. Use this diagram to:
• Solve multi-layered plane-wave propagation problems.
• Analyze common waveguide structures.
• Derive generalized transmission and reflection coefficient formulas.
• Derive scattering parameter formulas.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 244
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Understanding Frequency-dependent Matrices and Transmission Line Theory

Figure 72 System model for a line-to-line junction

R1, L1, G1, C1 [v]1 R2, L2, G2, C2

[v]2

. . .
. . .
[v]N
+
-

Wvr1 T1 + Wvr2

v v
+ 1 2 +

Wvb1 + T2 Wvb2

The Wvr and Wvb propagation functions describe how propagation (from one termination
to another) affects a wave. These functions are equal for the forward (Wvr) and backward
(Wvb) directions. The off-diagonal terms of the propagation functions represent the
coupling between conductors of a multiconductor line.
As a wave propagates along the line, it experiences delay, attenuation, and distortion
(see Figure 73). Lines with frequency-dependent parameters (that is, all real lines) do not
contain the frequency-independent attenuation component.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 245
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Overview of T-element Usage

Figure 73 Propagation function transient characteristics (unit-step response)

Transient
characteristic
ww (t)

Frequency Larger
dependent losses
issues
Attenuation
Distortion

0 Delay Time, t

Overview of T-element Usage

Some applications do not require the sophisticated models used by the PrimeSim HSPICE
W-element or S-element to effectively predict transmission line effects. In such cases, the
less complex T-element and U-element transmission line models can provide adequate
solutions. For discussion of the rarely used U-element model, see PrimeSim Continuum
User Guide: Elements. For detailed information on the T-element syntax and examples,
see T-element (Ideal Transmission Lines) in PrimeSim Continuum User Guide: Elements.
For information regarding the physics of modeling transmission lines with ideal and
lumped equivalents, see Appendix A, Simulating Circuits with Signetics Drivers.
The following sections discuss these topics:
• Invoking Ideal T-element Transmission Lines
• Selecting Ideal or Lossy Transmission Line Elements

Using T-element for Predicting Simple Tx Line Effects

The following sections discuss these topics:
• T-element Capabilities
• Invoking Ideal T-element Transmission Lines

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 246
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using T-element for Predicting Simple Tx Line Effects

T-element Capabilities
The T-element is useful for modeling ideal transmission lines in terms of their impedance
and delay. The U-element is useful for modeling lossy, nonideal transmission lines in terms
of lumped-element equivalent circuits. The T- and U-elements can model effects such as:
• Time delay
• Phase shift
• Coupling and Crosstalk
• Distortion

Invoking Ideal T-element Transmission Lines

The following sections discuss these topics:
• Ideal Transmission Line Model
• Selecting Ideal or Lossy Transmission Line Elements
• Lossless Voltage and Current Propagation

Ideal Transmission Line Model

The ideal transmission model is based on a characteristic impedance (Z0), and
propagation delay (TD). The Z0 and TD values may be obtained from a U model.

Figure 74 Ideal element circuit

in out
Z0 Z0

vin vout
iin iout
refin refout

The input and output of an ideal transmission line have these relationships:

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 247
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using T-element for Predicting Simple Tx Line Effects

The signal (propagation) delays for ideal transmission lines are specified as one of the
following:
• TDeff=TD (if only TD is given)
• TDeff=TD*L if TD and L are given
• TDeff=NL/F if NL and F are given
• TDeff=TD if you use a U model
The ideal transmission line only delays the difference between the signal and the
reference. Note that although you can specify length as an input parameter, it is used
only to compute the effective propagation delay. Some applications, such as a differential
output driving twisted pair cable, require both differential and common mode propagation.
Use a W-element or U-element, if you need to model both propagation modes and
coupling. If not, you can use two T-elements (see Figure 75) to model uncoupled
differential signals. In this figure, the two lines are completely uncoupled but the delay and
impedance values are correctly modeled.

Figure 75 Use of Two T-elements for Differential Signals

out

in

outbar

You cannot implement coupled lines with the T-element so use W-elements or U-elements
for applications requiring two or three coupled conductors.
transmission linescoupledlines
Synopsys circuit simulation uses a transient timestep that does not exceed half the
minimum line delay. Very short transmission lines (relative to the analysis time step) cause
long simulation times. You can replace very short lines with a single R, L, or C-element
(see Figure 76).

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 248
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using T-element for Predicting Simple Tx Line Effects

Selecting Ideal or Lossy Transmission Line Elements

An electrical model that simulates the behavior of an interconnect or transmission line
must consider all of the following:
• Physical nature or electrical properties of the interconnect
• Bandwidth or risetime of signals of interest
• Interconnect’s actual time delay or equivalent electrical length
• Complexity and accuracy of the model
When accuracy is the utmost priority, and all high-frequency effects must be taken into
account, use the W-element or S-element with models based on measurements or EM
analyses. When basic models are adequate, you can choose from the following circuit
models for interconnects:
transmission linesmodelstransmission lineselements
• No model at all. Use a common node to connect two elements.
• Lumped models with R, L, and C-elements as described in the PrimeSim Continuum
User Guide: Elements. These include a series resistor (R), a shunt capacitor (C), a
series inductor and resistor (RL), or a series resistor and a shunt capacitor (RC).
• Transmission line models such as an ideal transmission line (T-element) or a lossy
transmission line (U-element)
T ElementsU Elements
The following tables and charts explain the factors involved when determining the best
choice between RLC-elements, the T-element, or U-element transmission line models.
The following sections discuss these topics:
• Source Properties
• Interconnect Properties
• Comparison of Ideal and Lossy Transmission Line Elements
• Transmission Lines: Example
Source Properties
trise = source risetime
Rsource = source output impedance
Interconnect Properties
Z0 = characteristic impedance
TD = time delay of the interconnection

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 249
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using T-element for Predicting Simple Tx Line Effects

or:
R = equivalent series resistance
C = equivalent shunt capacitor
L = equivalent series inductance
Figure 76 shows how to select a model based on source and interconnect properties.
wiremodel selectioncharttransmission linesmodelsselection
Figure 76 Wire model selection chart

Initial information required:

trise = source risetime
TD = time delay of the interconnect
Rsource = source resistance
Selection Criterion

R > 10% Rsource R

RC RL
(R + Rsource)*C > 10% trise C
trise >= 5 TD (low
frequency) L
Lumped model > 10% trise L
(R + Rsource)

Default U

Compatibility with
conventional SPICE T

Infinite bandwidth:
trise < 5 TD required for ideal T
(high frequency) sources
U model
Consider crosstalk U
effects

Consider line losses U

TDs are very long

or very short U

Default U

Use the U model with either the ideal T-element or the lossy U-element. You can also
use the T-element alone without the U model. PrimeSim HSPICE offers both a flexible
definition of the conventional SPICE T-element and an accurate U-element lossy
simulation.
transmission linesselectionelements
PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 250
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using T-element for Predicting Simple Tx Line Effects

Figure 77 U Model, T-element, and U-element relationship

Physical Geometry
U Model Calculated
Precalculated R, C, L Field Solution R, C, L

Impedance, Delay Inverse Solution Z0,TD

(Z0,TD)

Lossless (Ideal) Lossy

T-element U-element

The T- and U-elements do not support the [M=val] multiplier function. If a U- or T-element
multiplier function, U and T Elementsmultiplier parameter, U and T ElementsU Elementsmultiplier function MT Elementsmultiplier function M

is used in a subcircuit and an instance of the subcircuit has a multiplier applied, that
multiplier is not applied to the T- or U-element.
A warning message similar to the following is issued in both the status file (.st0) and the
output file (.lis) if the smallest transmission line delay is less than TSTOP/10e6:
**warning**: the smallest T-line delay (TD) = 0.245E-14 is too small
Please check TD, L and SCALE specification warnings, T-line delay too small
This feature is an aid to finding errors that cause excessively long simulations.
transmission linesdelay too small warning
Note:
All transmission lines have a ground reference for the signal conductors. In
this manual, the ground reference is called the reference plane so that it is not
confused with SPICE ground. The reference plane is the shield or the ground
plane of the transmission line element. The reference plane nodes may or may
not be connected to SPICE ground.
transmission linesHSPICEreference planereferenceplane, transmission linesgroundplane, transmission linestransmission linesground plane
Comparison of Ideal and Lossy Transmission Line Elements
The ideal and lossy transmission line models each have particular advantages, and they
may be used in a complementary manner. Both model types are fully functional in AC
analysis and transient analysis. Some of the comparative advantages and uses of each
type of model are listed in Table 11.
transmission linesidealpropertiestransmission lineslossy modelpropertieslossytransmission linesideal transmission linelosslesstransmission lines
Table 11 Ideal versus Lossy Transmission Line
Ideal Transmission Line Lossy Transmission Line

Lossless Includes loss effects

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 251
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using T-element for Predicting Simple Tx Line Effects

Table 11 Ideal versus Lossy Transmission Line (Continued)

Ideal Transmission Line Lossy Transmission Line

Used with voltage sources Used with buffer drivers

No limit on input risetime Prefiltering necessary for fast rise

Less CPU time for long delays Less CPU time for short delays

Differential mode only Supports common mode simulation

No ground bounce Includes reference plane reactance

Single conductor Up to five signal conductors allowed

AC and transient analysis AC and transient analysis

The ideal line is modeled as a voltage source and a resistor. The lossy line is modeled as
a multiple lumped filter section as shown in Figure 78.
transmission lineslumpedRC model
Figure 78 Ideal versus lossy transmission line model

in out in out

ref ref refin refout

Ideal Element Circuit Lossy Element Circuit

Because the ideal element represents the complex impedance as a resistor, the
transmission line impedance is constant, even at DC values. On the other hand, you must
prefilter the lossy element if ideal piecewise linear voltage sources are used to drive the
line.
Transmission Lines: Example
The following file fragment is an example of how both T-elements and U-elements can be
referred to a single U-model as indicated in Figure 77. The file specifies a 200 mm printed
circuit wire implemented as both a U-element and a T-element. The two implementations
share a U-model that is a geometric description (ELEV=1) of a planar structure (PLEV=1).
T1 in gnd t_out gnd micro1 L=200m
U1 in gnd u_out gnd micro1 L=200m
.model micro1 U LEVEL=3 PLEV=1 ELEV=1 wd=2m ht=2m th=0.25m
+ KD=5

The next section provides details of element and model syntax.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 252
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using T-element for Predicting Simple Tx Line Effects

Table 12 Transmission Line Parameters

Parameter Description

T1, U1 Element names

micro1 The model name

in, gnd, t_out, and u_out Nodes

L The length of the signal conductor

wd, ht, th Dimensions of the signal conductor and dielectric

KD The relative dielectric constant

Lossless Voltage and Current Propagation

For the ideal transmission line, voltage and current propagate without loss along the length
of the line (±x direction) with spatial and time-dependence given according to the following
equation:

The A represents the incident voltage, B represents the reflected voltage, Z0 is the
characteristic impedance, and β is the propagation constant. The latter is related to the
transmission line inductance (L) and capacitance (C) by the following equation:

The L and C terms are in per-unit-length units (Henries/meter, Farads/meter). The

following equation gives the phase velocity:

At the end of the transmission line ( ), the propagation term becomes the following
equation:

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 253
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
Using T-element for Predicting Simple Tx Line Effects

This is equivalent to an ideal delay with the following value:

where:
: absolute time delay (sec)
: physical length (L) (meters)
: phase velocity (meters/sec)
Using standard distance=velocity*time relationships, the PrimeSim HSPICE T-element
parameter values are related to these terms according to:

where:
: frequency
: wavelength
: relative time delay (TD) (sec/meter)

where:
: physical length (L) (meters)
: normalized length (NL)
: frequency at NL (F) (Hz)

The T-element is therefore an ideal behavioral model given in terms of Zo and delay, yet
you can relate it back to inductance, capacitance, velocity, and/or length quantities as
needed. PrimeSim HSPICE allows you to specify the T-element transmission line in four
different ways:
• Z0, TD
• Z0, TD, L

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 254
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
References

• Z0, NL, F

• L, with and values taken from a U model.

References

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 255
W-2024.09
 Feedback
Chapter 6: W- and T-elements Transmission-Line Modeling
References

[1] Dmitri Kuznetsov, “Optimal Transient Simulation of Transmission Lines,” IEEE Trans.,
Circuits Syst., vol.43, pp. 110-121, Feb., 1996.
[2] Luca Daniel and Joel Philips, “Model Order Reduction for Strictly Passive and Causal
Distributed Systems,” proceeding of DAC, June, 2002
[3] [2] Colin Gordon, Thomas Blazeck, and Raj Mittra, “Time-Domain Simulation of
Multiconductor Transmission Lines with Frequency-Dependent Losses,” IEEE
TRANSACTIONS ON COMPUTER-AIDED DESIGN, VOL. I I. NO. 11, pp. 1372-1387,
NOVEMBER 1992
[4] [3] Qingiian Yu and Omar Wing, “Computational Models of Transmission Lines
with Skin Effects and Dielectric Loss,” IEEE TRANSACTIONS ON CIRCUITS AND
SYSTEMS-I: FUNDAMENTAL THEORY AND APPLICATIONS. VOL. 41, NO. 2, pp.
107-119, FEBRUARY 1994
[5] Eric Bogatin, “Signal Integrity Simplified”, Prentice HallBjorn Gustavsen and Adam
Semlyen, “Rational Approximation of Frequency Domain Responses by Vector Fitting,”
IEEE Transaction on Power Delivery, Vol.14, No.3, pp. 1052-1061, July 1999
[6] Ravi Kollipara, et al “Practical Design Considerations for 10 to 25 Gbps Copper
Backplane Serial Links,” DesignCon 2006, Santa Clara, CA, U.S.A. E. Hammerstad
and O. Jensen, “Accurate models for microstrip computer aided design,” IEEE MTT-S
Int. Microwave Symp. Dig., 1980,. pp. 407-409.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 256
W-2024.09
 Feedback

7
Using the Field Solver to Extract Tx Line
Parameters

Describes how to use the built-in PrimeSim HSPICE Field Solver for transmission line
parameter extraction.

Transmission line simulation is challenging and time-consuming, because extracting

transmission line parameters from physical geometry requires a significant effort. To
minimize this effort, you can use a simple (but efficient and accurate) 2D electromagnetic
field solver, which calculates the electrical parameters of a transmission line system,
based on its cross-section.
The built-in 2-D electromagnetic field solver is highly-optimized for interconnects in
stratified media. This field solver uses the W-element, and it supports optimization and
statistical analysis within transient simulation.
The solver is based on:
• An improved version of the boundary-element method, and
• The filament method that is also implemented in the Synopsys product, Raphael.
See K. S. Oh, D. B. Kuznetsov, and J. E. Schutt-Aine, “Capacitance computations in a
multi-layered dielectric medium using closed-form spatial Green s functions,” IEEE Trans.
Microwave Theory and Tech., vol. 42, pp. 1443-1453, August 1994 for more information on
the boundary-element method.
To learn more about BEM and Green’s Function, see the Raphael Reference Manual.
The following sections discuss these topics:
• Using the Field Solver Model
• Field-Solver-Related Netlist Statements
• Accelerating the W-element Field Solver Using Multicore Technology
• Visualizing Cross-Sectional Geometric Information
• Field Solver Examples

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 257
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Using the Field Solver Model

Using the Field Solver Model

Use the field-solver model to specify a geometry model for the W-element transmission
newlink Extracting Transmission Line Parameters (Field Solver)

line. In the field-solver model:

• The list of conductors must appear last.
• Conductors cannot overlap each other.
• The Field Solver assumes that floating conductors are electrically disconnected, and
does not support non-zero fixed charges. Because the field solver is designed as 2-D,
it ignores displacement current in floating conductors.
• The Field Solver treats metal layers in the layer stack as the reference node.
• Conductors defined as REFERENCE are all electrically-connected, and correspond to the
reference node in the W-element.
• You must place signal conductors in the same order as the terminal list in the W-
element statement. For example, the ith signal conductor (not counting reference and
floating conductors) is associated with the ith input and output terminals specified in the
corresponding W-element.
• Floating and reference conductors can appear in any order.
Note:
It is strongly recommended that a separate model be created for complex
structures when an internal ground plane is introduced.
The following sections discuss these topics:
• Filament Method
• Modeling Geometries
• Solver Limitation

Filament Method
This section describes the filament method for the skin-effect resistance and inductance
solver. The 2-D filament method uses data about magnetic coupling when it extracts
frequency-dependent resistance and inductance. To use this solver, set COMPUTE_RS=yes
in a .FSOPTIONS statement.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 258
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Using the Field Solver Model

The following process explains the filament method:

1. The filament method divides the original conductor system into thin filaments.
2. From the coupling of these filaments, this method then derives the distributed magnetic
coupling of the inside and outside of the conductor.
3. After dividing the conductors into thin filaments, this method creates the impedance
matrix of the filament system:

4. This method uses the following equation to solve the current matrix ( ):

In the preceding equation, the vf vector excites the filament system.

5. The filament method uses the result of this equation to calculate the partial current
matrix of the conductor system ip as a sum of all filament currents:

Equation 75

6. The filament method uses the following equation to solve the partial impedance matrix
( ):
161
7. From the components of the partial impedance matrix, the filament method uses
the following relationship to calculate the components of the loop
impedance matrix:

Equation 76

In the preceding equation, n is the number of signal (non-reference) conductors in the

system.
Note:
W-element analysis uses these loop impedance components.
For full discussion of the .FSOPTIONS command, see Accelerating the W-element Field
Solver Using Multicore Technology or .FSOPTIONS in the PrimeSim HSPICE Reference
Manual: Commands and Control Options.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 259
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field-Solver-Related Netlist Statements

Modeling Geometries
In geometry modeling:
• The number of dielectric layers is arbitrary.
• You can arbitrarily shape the conductor cross-section, including an infinitely-thin strip.
• The number of conductors is unlimited.
• The current dielectric region must be planar.
• Conductors must not overlap each other.
• Magnetic materials are not supported.
Geometric modeling outputs the Maxwellian (short-circuit) transmission line matrices: C, L,
Ro, Rs, Go, and Gd. (See Equations and Parameters.)

Solver Limitation
When the field solver computes the conductance matrices (Go and Gd), if the media are
not homogeneous, then the solver uses the arithmetic average values of conductivities
and loss tangents.

Field-Solver-Related Netlist Statements

Table 13 describes the netlist statements that specifically relate to the field solver. For
the syntax and examples of these statements, see the links to the PrimeSim HSPICE
Reference Manual: Commands and Control Options.
Table 13 Field-Solver Statement Syntax
Statement Usage

.MATERIAL Use this statement to define the properties of a

material.

.LAYERSTACK Use this statement to define a stack of dielectric or

metal layers.

.SHAPE Use this statement to define a shape. The Field

Solver uses these shapes:
• Rectangle
• Circle
• Strip
• Polygon
• Trapezoid
to describe a cross-section of the conductor.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 260
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field-Solver-Related Netlist Statements

Table 13 Field-Solver Statement Syntax (Continued)

Statement Usage

.FSOPTIONS Use this statement to set various options for the field
solver.

.MODEL W MODELTYPE=FieldSolver Type of transmission-line model.

The following sections discuss these topics:

• Field Solver Model Syntax
• Using the Field Solver to Extract an RLGC Tabular Model
• Accounting for Surface Roughness Effect in W-element

Field Solver Model Syntax

Use a .MODEL statement to define a field solver.

.MODEL mname W MODELTYPE=FieldSolver

+ LAYERSTACK=name [FSOPTIONS=name]
+ [RLGCFILE=name] [COORD=0|DESCART|1|POLAR]
+ [OUTPUTFORMAT=RLGC|RLGCFILE]
+ CONDUCTOR=SHAPE=name [MATERIAL=name]
+ [ORIGIN=(val1,val2)] [TYPE=SIGNAL|REFERENCE|FLOATING] ...

Parameter Description

mname Model name.

LAYERSTACK Name of the associated layer stack.

FSOPTIONS Associated option name. If you do not specify this entry, the Field Solver uses
the default options.

RLGCFILE Use the output file for RLGC matrices, instead of the standard error output
device. An RLGC file name must start with an alphabetic letter (not a number).
If the specified file already exists, then the Field Solver appends the output. To
generate output, you must set PRINTDATA in .FSOPTIONS to YES; setting it to
APPEND, appends the extracted RLGC model to the specified file.

COORD The polar field solver is invoked only when COORD=1 or COORD=POLAR.

OUTPUTFORMAT Model syntax format for RLGC matrices in the W-element. Specified in the
RLGC file. Default format is an RLGC model.

SHAPE Shape name.

MATERIAL Conductor material name. If you do not specify this entry, the Field Solver
defaults to the predefined metal name PEC (perfect electrical conductor).
PEC
PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 261
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field-Solver-Related Netlist Statements

Parameter Description

ORIGIN=(val1, val2) val1, val2 can be (x,y) for regular coordinates and (radius, degree) for polar
coordinates.

TYPE One of the following conductor types:

• SIGNAL: a signal node in the W-element (the default).
• REFERENCE: the reference node in the W-element.
• FLOATING: floating conductor, no reference to W-element.

Using the Field Solver to Extract an RLGC Tabular Model

You can use the Field Solver to extract an RLGC tabular model which allows higher
flexibility of dependence on frequency. (When PRINTDATA=APPEND, RLGC model output is
appended to the specified output file.)
Note:
The default for ACCURACY is HIGH.
The following sections discuss these topics:
• Syntax for Extracting RLGC Tabular Model
• COMPUTE_TABLE/COMPUTETABLE Keyword
• Default
• RLGC Tabular Model Sample Output

Syntax for Extracting RLGC Tabular Model

.FSOPTIONS name [ACCURACY=HIGH|MEDIUM|LOW]+ [GRIDFACTOR=val] +
[COMPUTE_GO=YES|NO] [COMPUTE_GD=NO|YES]+ [COMPUTE_RO=YES|NO]
[COMPUTE_RS=NO|YES|DIRECT|ITER]
+ [COMPUTE_TABLE=frequency_sweep]
+ [PRINTDATA=NO|YES|APPEND]

Notes
• The forms of the following arguments are interchangeable:
COMPUTE_GO : COMPUTEGO
COMPUTE_GD : COMPUTEGD
COMPUTE_RO : COMPUTERO
COMPUTE_RS : COMPUTERS
COMPUTE_TABLE : COMPUTETABLE

• If you only set COMPUTE_GD=yes in the .FSOPTIONS statement, you do not generate
data in the Gd matrix of the RLGC file. Since Gd is the dielectric loss conductance

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 262
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field-Solver-Related Netlist Statements

matrix, you also need to define the loss tangent values for each dielectric material you
have in your layer stack. For example:
.MATERIAL die1 DIELECTRIC ER=4.1 LOSSTANGENT=.012

The Gd matrix is derived from Co matrix and loss tangent value as

when multiple loss tangent values are given to the .layerstack structure, the average
value (including background material property) is used.

COMPUTE_TABLE/COMPUTETABLE Keyword
Frequency sweep Definition

COMPUTE_TABLE or Specifies a type of frequency sweep. You can specify either LIN, DEC, OCT,
COMPUTETABLE POI. Specify the nsteps, start, and stop values using the following syntax for
each type of sweep:
• LIN nsteps start stop
• DEC nsteps start stop
• OCT nsteps start stop
• POI nsteps freq_values
Note: To reduce the risk of instability in subsequent simulations, the field
solver adds points to ensure that there is at least one frequency point per
decade up to 1 THz.

Once a frequency sweep is specified, the W-element computes transmission line

parameters at specified frequency points. In addition, since resistance and conductance
at DC (zero frequency) and capacitance and inductance at infinite frequency are essential
to ensure the accuracy, these four matrices are automatically computed regardless of the
sweep configuration.
In the table model extraction, series impedance,
is computed directly from the filament method solver. For shunt admittance,
, the static capacitance solver is still used. From the static
capacitance, , and corresponding dielectric loss term, , a complex
dielectric loss model is derived. For further detail about the complex dielectric loss model
generation, see Fitting Procedure Triggered by INCLUDEGDIMAG Keyword on page 233.

Default
If the COMPUTE_TABLE keyword is not specified, a conventional RLGC model is generated.
Note:
The direct solver is chosen (COMPUTERS=DIRECT) by default for table model
extraction as well as for the RLGC (non-table) model.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 263
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field-Solver-Related Netlist Statements

RLGC Tabular Model Sample Output

.MODEL rmod1 sp N=3 MATRIX=SYMMETRIC
+ SPACING=POI VALTYPE=REAL INTERPOLATION=LINEAR
+ DC = 3.850667e+04
+ 7.700000e+03 3.850667e+04
+ 7.700000e+03 7.700000e+03 3.850667e+04
+ INFINITY = 1.444000e+05
+ 3.082000e+04 1.402667e+05
+ 2.389333e+04 3.082000e+04 1.444000e+05
+ DATA = ( 20
+ 0.000000e+00
+ 3.850667e+04
+ 7.700000e+03 3.850667e+04
+ 7.700000e+03 7.700000e+03 3.850667e+04
+ .....
+
.MODEL lmod1 sp N=3 MATRIX=SYMMETRIC
+ SPACING=POI VALTYPE=REAL INTERPOLATION=LINEAR
+ INFINITY = 6.624667e-07
+ 2.603333e-07 7.253333e-07
+ 1.406667e-07 2.603333e-07 6.624667e-07
+ DATA = ( 20
+ 0.000000e+00
+ 9.733333e-07
+ 4.876000e-07 9.806667e-07
+ 3.454000e-07 4.876000e-07 9.733333e-07
+ .....
+
.MODEL gmod1 sp N=3 MATRIX=SYMMETRIC
+ SPACING=POI VALTYPE=REAL
+ DC = 0.000000e+00
+ 0.000000e+00 0.000000e+00
+ 0.000000e+00 0.000000e+00 0.000000e+00
+ DATA=( 1
+ 0.000000e+00
+ 0.000000e+00
+ 0.000000e+00 0.000000e+00
+ 0.000000e+00 0.000000e+00 0.000000e+00
+ ......
+
.MODEL cmod1 sp N=3 MATRIX=SYMMETRIC
+ SPACING=POI VALTYPE=REAL+ INFINITY = 3.930057e-17
+ -1.300346e-17 3.969922e-17 + -3.026179e-18 -1.300788e-17
3.929558e-17
+ DATA=( 20+ 0.000000e+00
+ 3.930057e-17 + -1.300346e-17 3.969922e-17
+ -3.026179e-18 -1.300788e-17 3.929558e-17 + .....

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 264
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field-Solver-Related Netlist Statements

Accounting for Surface Roughness Effect in W-element

In real devices operating at high frequency range, skin effect causes a secondary effect
since the surfaces of conductors are not flat but have some roughness. When a majority of
current propagates across the rough surfaces, there is a non-negligible increase in series
impedance. Since the influence of this phenomenon depends on the dominance of current
around the conductor surface, this effect is also frequency-dependent.
PrimeSim HSPICE provides two ways to take this frequency-dependent increase of series
impedance into account:
• Scaling RS matrix
• Calculating root mean square (RMS) surface roughness height
To scale the RS matrix use the SCALE_RS keyword with scaling factor and apply it to the
skin-effect (Rs) matrix. Thus, the series impedance is expressed as:

Equation 77

To calculate the RMS surface roughness height, the ratio of resistivity increment, , may
be empirically estimated as:

Equation 78

Where and are the RMS surface roughness height and skin depth of the conductor
material.
When ROUGHNESS_METHOD=1 (default) is chosen, the W-element field solver will re-scale
the metal resistivity of each conductor filament at each frequency point using the surface
roughness factor as:

Equation 79

When ROUGHNESS_METHOD=2 is chosen, the W-element field solver first solves frequency
dependent impedance without surface roughness effect. Then roughness factor (SR in
Equation 62) will be applied to scale the impedance as:

Equation 80

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 265
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field-Solver-Related Netlist Statements

Where Zrough (f), Zo and Z(f) are impedance with surface roughness effect, DC impedance
and impedance without surface roughness effect, respectively.
The following sections discuss these topics:
• Syntax for Scaling RS Matrix
• Syntax for Taking RMS Surface Roughness of Conductor Materials
• Syntax for Choosing Surface Roughness Method
• Example: Transmission Line Surface Roughness

Syntax for Scaling RS Matrix

Wxxx ni1 ni2… ref_in no1 no2… ref_out
+ [SCALE_RS=value]

Keyword
SCALE_RS: Scaling factor to the RS matrix

Syntax for Taking RMS Surface Roughness of Conductor

Materials
.material copper metal conductivity=value [roughness=value]

Keyword
ROUGHNESS: RMS surface roughness height.

Syntax for Choosing Surface Roughness Method

.fsoptions [ROUGHNESS_METHOD=1|2]

Keyword
ROUGHNESS_METHOD: Surface roughness computation method.

Example: Transmission Line Surface Roughness

*** no surface roughness ***
P1 in1 0 port=1 ac=1
P2 out1 0 port=2
W1 in1 gnd out1 gnd FSmodel=line1 N=1 l=0.1

*** use copper_roughs w/ roughness=2um ***

P3 in2 0 port=3 ac=1
P4 out2 0 port=4
W2 in2 gnd out2 gnd FSmodel=line1_rough N=1 l=0.1

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 266
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Accelerating the W-element Field Solver Using Multicore Technology

*** use SCALE_RS=1.1 ***

P5 in3 0 port=5 ac=1
P6 out3 0 port=6
W3 in3 gnd out3 gnd FSmodel=line1 N=1 l=0.1 SCALE_RS=1.1

.material diel dielectric er=4.3

.material copper metal conductivity=57.6meg
.material copper_rough metal conductivity=57.6meg
+ROUGHNESS=2e-6

.shape rect rectangle width=400e-6 height=40e-6

.layerstack stack1 background=air
+layer=(copper,10e-6)
+layer=(diel,200e-6)
.fsoptions opt1 printdata=yes computegd=no computers=yes
.model line1 W Modeltype=fieldsolver,
+layerstack=stack1,
+fsoptions=opt1,
+Rlgcfile=line1.rlgc,
+conductor=(shape=rect,origin=(0,110e-6),material=copper)
.model line1_rough W Modeltype=fieldsolver,
+layerstack=stack1,
+fsoptions=opt1,
+Rlgcfile=line1_rough.rlgc,
+conductor=(shape=rect,origin=(0,110e-6),material=copper_rough)
.opt post
.ac dec 100 1e6 1e10
.end

Accelerating the W-element Field Solver Using Multicore

Technology
You can increase the speed of the W-element magnetic coupling field solver with the
multithread option when a multicore system is available. Since the skin effect solver
has to solve multiple frequency points to capture the frequency-dependent effect, it
consumes a majority of field solver run-time. When the multicore option (-mt #) is used
in the command-line invocation, PrimeSim HSPICE parallelizes frequency sampling in the
tabular RLGC model extraction.
Syntax

FSOPTIONS name [ACCURACY=HIGH|MEDIUM|LOW+ [GRIDFACTOR=val]

[PRINTDATA=NO|YES|APPEND]+ [COMPUTE_GO=NO|YES] [COMPUTE_GD=YES|NO]+
[COMPUTE_RO=YES|NO] [COMPUTE_RS=NO|YES|DIRECT|ITER]
+ [COMPUTE_TABLE=frequency_sweep]

Keyword
COMPUTE_RS | COMPUTERS

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 267
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Visualizing Cross-Sectional Geometric Information

COMPUTE_RS Options Definition

YES Activate filament solver with direct matrix solver

NO Do not to perform filament solver

DIRECT Activate filament solver with direct matrix solver (same as YES)

ITER Activate filament solver with iterative matrix solver

For full discussion of the .FSOPTIONS command, see .FSOPTIONS in the PrimeSim
HSPICE Reference Manual: Commands and Control Options.

Visualizing Cross-Sectional Geometric Information

When PrimeSim HSPICE runs with the W-element field solver model, it generates the
model_name.str file, which contains Tcl/Tk scripts to display cross-sectional geometrical
information of the target structure. Once PrimeSim HSPICE is executed, locate the
model_name.str file. To invoke Tcl/Tk graphics, at the command prompt, enter either:
% wish model_name.str

or
% chmod +x model_name.str
% model_name.str

Example output is seen in Figure 79.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 268
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

Figure 79 Structural output example

The following dynamic functions are available on the structural display window:
• Move the mouse cursor over an object to display corresponding material properties.
• Drag the mouse cursor to display geometrical locations.
Note:
To use this function, the Tcl/Tk environment must be installed on users'
systems. On a Linux system, Tcl/Tk is included in the default installation
package. Manual installations are needed for the other platforms. The Tcl/Tk
environment is available for all the platforms supported by PrimeSim HSPICE.
Limitation: the POLAR coordinate is not currently supported.
For further information go to: http://www.tcl.tk/software/tcltk/8.0.html

Field Solver Examples

The following examples show you how to use the Field Solver. All of the examples shown
in this section run with the HIGH accuracy mode and with GRIDFACTOR = 1.
• Example 1: Cylindrical Conductor Above a Ground Plane
• Example 2: Stratified Dielectric Media
• Example 3: Two Traces Between Two Ground Planes

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 269
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

• Example 4: Using Field Solver with Monte Carlo Analysis

• Example 5: Modeling Coaxial and Shielded Twin-Lead Lines Using the Polar Field
Solver

Example 1: Cylindrical Conductor Above a Ground Plane

This is an example of a copper cylindrical conductor above an ideal (lossless) ground
plane.
With these formulas, you can derive the exact analytical formulas for all transmission line
parameters:

Equation 81

Equation 82

Equation 83

Figure 80 shows the geometry of a copper cylindrical conductor above an ideal ground
plane.

Equation 84

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 270
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

Figure 80 Cylindrical conductor above a perfect electrical conductor ground plane

c = 5.76e7

r = 4.0 d = 1 mm H = 3 mm

tan = 1.2e-3

Ground Plane

Table 14 lists the corresponding netlist.

Table 14 Input File Listing
Listing Type Field Solver Cylindrical Example

Header, options and * Example: cylindrical conductor .OPTION PROBE POST VIMPULSE in1
sources gnd PULSE 4.82v 0v 5n 0.5n 0.5n 25n

W-element W1 in1 gnd out1 gnd FSmodel=cir_trans N=1 l=0.5

Materials .MATERIAL diel_1 DIELECTRIC ER=4, LOSSTANGENT=1.2e-3 .MATERIAL

copper METAL CONDUCTIVITY=57.6meg

Shapes .SHAPE circle_1 CIRCLE RADIUS=0.5mm

Defines a half-space .LAYERSTACK halfSpace + BACKGROUND=diel_1,LAYER=(copper,1mm)

Option settings .FSOPTIONS opt1 PRINTDATA=YES, + COMPUTE_RS=yes, COMPUTE_GD=yes

Model definition .MODEL cir_trans W MODELTYPE=FieldSolver + LAYERSTACK=halfSpace,

FSOPTIONS=opt1, RLGCFILE=ex1.rlgc + CONDUCTOR=(SHAPE=circle_1,
ORIGIN=(0,4mm), + MATERIAL=copper)

Analysis, outputs and end .TRAN 0.5n 100n .PROBE v(out1) .END

Compare the computed results with the analytical solutions in Table 15. The Field Solver
computes the resistance and conductance at the frequency of 200 MHz, but does not
include the DC resistance (Ro) and conductance (Go) in the computed values.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 271
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

Table 15 Comparison Result

Value Analytical Computed

C (pF/m) 89.81 89.66

L (nH/m) 494.9 496.4

G (mS/m) 0.1354 0.1352

R ( W /m) 1.194 1.086

Note:
The analytical resistance value does not consider ground plane resistivity.

Example 2: Stratified Dielectric Media

This is an example of three traces immersed in a stratified dielectric media (see
Figure 81).

Figure 81 Three traces immersed in stratified dielectric media

AIR
150  150 
2 3

r = 3.2 100 
1 70 

350 
r = 4.3 200 

Table 16 shows the input file.

Table 16 Input File for Three Traces Immersed in Stratified Dielectric Media
Listing Type Field Solver Stratified Dielectric Example

Header, options * Example: three traces in dielectric .OPTION PROBE POST VIMPULSE in1 gnd PULSE 4.82v
and sources 0v 5n 0.5n 0.5n 25n

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 272
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

Table 16 Input File for Three Traces Immersed in Stratified Dielectric Media (Continued)
Listing Type Field Solver Stratified Dielectric Example

W-element W1 in1 in2 in3 gnd out1 out2 out3 gnd + FSmodel=cond3_sys N=3 l=0.5

Materials .MATERIAL diel_1 DIELECTRIC ER=4.3

.MATERIAL diel_2 DIELECTRIC ER=3.2
.MATERIAL copper METAL CONDUCTIVITY=57.6meg

Shapes .SHAPE rect_1 RECTANGLE WIDTH=0.35mm, HEIGHT=0.07mm

Uses the default .LAYERSTACK stack_1

AIR background + LAYER=(copper,1um),LAYER=(diel_1,0.2mm),
+ LAYER=(diel_2,0.1mm)

Option settings .FSOPTIONS opt1 PRINTDATA=YES

Three .MODEL cond3_sys W MODELTYPE=FieldSolver,

conductors + LAYERSTACK=stack_1, FSOPTIONS=opt1, RLGCFILE=ex2.rlgc
share the same + CONDUCTOR=(SHAPE=rect_1,MATERIAL=copper, ORIGIN=(0,0.201mm)),
shape + CONDUCTOR=(SHAPE=rect_1,MATERIAL=copper,ORIGIN=(0.5mm,0.301mm)),
+ CONDUCTOR=(SHAPE=rect_1,MATERIAL=copper, ORIGIN=(1mm,0.301mm))

Analysis, .TRAN 0.5n 100n

outputs and end .PROBE v(out1)
.END

Note:
W. Delbare and D. D. Zutter, “Space-domain Green’s function approach to the
capacitance calculation of multi-conductor lines in multi-layered dielectrics with
improved surface charge modeling,” IEEE Trans. Microwave Theory and Tech.,
vol. 37, pp. 1562-1568, October 1989.
Figure 82 shows the results of convergence analysis, based on the total capacitance of
the first conductor with respect to the GRIDFACTOR parameter.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 273
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

Figure 82 Convergence of accuracy modes

Accuracy Comparison

1.5
Error [%]

0.5 1
2
Grid Factor
0 3
LOW MEDIUM HIGH
Accuracy Mode

Example 3: Two Traces Between Two Ground Planes

This is an example of two traces between two ground planes (in other words, a coupled
strip line) (see Figure 83).

Figure 83 Example of a coupled strip line

r = 2.5 0.2 mm 1 mm
3 mm

0.2 mm

r = 10 2 mm

Table 17 lists the complete input netlist.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 274
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

Table 17 Input Netlist for Two Traces Between Two Ground Planes
Listing Type Field Solver Ground Planes Example

Header, options and * Example: two traces between gnd planes .OPTION PROBE POST VIMPULSE
sources in1 gnd PULSE 4.82v 0v 5n 0.5n 0.5n 25n

W-element W1 in1 in2 gnd out1 out2 gnd FSmodel=cond2_sys +N=2 l=0.5

Materials .MATERIAL diel_1 DIELECTRIC ER=10.0 .MATERIAL diel_2 DIELECTRIC

ER=2.5 .MATERIAL copper METAL CONDUCTIVITY=57.6meg

Shapes .SHAPE rect RECTANGLE WIDTH=1mm, + HEIGHT=0.2mm,

Top and bottom ground .LAYERSTACK stack_1, + LAYER=(copper,1mm), LAYER=(diel_1,2mm), +

planes LAYER=(diel_2,3mm), LAYER=(copper,1mm)

Option settings .FSOPTIONS opt1 PRINTDATA=YES

Two conductors share the .MODEL cond2_sys W MODELTYPE=FieldSolver, +

same shape LAYERSTACK=stack_1, FSOPTIONS=opt1 RLGCFILE=ex3.rlgc +
CONDUCTOR=(SHAPE=rect,MATERIAL=copper, ORIGIN=(0,3mm)), +
CONDUCTOR=(SHAPE=rect,MATERIAL=copper, ORIGIN=(1.2mm,3mm))

Analysis, outputs and end .TRAN 0.5n 100n .PROBE v(out1) .END

Table 18 compares the computed result with the Finite Element (FEM) solver result.
Table 18 Comparison Between Computed and FEM Solver
Results
Computed
(pF/m)

FEM Solver
(pF/m)

Example 4: Using Field Solver with Monte Carlo Analysis

The following example shows how to use Monte Carlo transient analysis to model
variations in the manufacturing of a microstrip.
The transient output waveforms are shown in Figure 84.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 275
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

Figure 84 Monte carlo analysis with field solver and W-element

Table 18 shows the input listing with the W-element.

Table 19 Input File Listing with the W-element
Listing Type Field Solver Monte Carlo Example

Header, options and PETL *Example: with Monte Carlo

sources .OPTION PROBE POST
+ VIMPULSE in1 gnd AC=1v PULSE 4.82v 0v 5ns
+ 0.5ns 0.5ns 25ns

Parameter definitions .PARAM x1=Gauss(0,0.02,1) x2=Gauss(0.5mm,0.02,1)

+ x3=Gauss(1mm,0.02,1)
.PARAM dRef=1u dY1=Gauss(2mm,0.02,1)
+ dY2=Gauss(1mm,0.02,1)

W-element W1 in1 in2 in3 0 out1 out2 out3 0

+ FSMODEL=cond3_sys N=3 l=0.5

Materials .MATERIAL diel_1 DIELECTRIC ER=4.3

.MATERIAL diel_2 DIELECTRIC ER=3.2
.MATERIAL copper METAL CONDUCTIVITY=57.6meg

Shapes .SHAPE r1 RECTANGLE WIDTH=0.35mm, HEIGHT=0.070mm

Uses the default AIR .LAYERSTACK stack_1 LAYER= (copper,dRef),

background + LAYER=(diel_1,dY1), LAYER= (diel_2,dY2)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 276
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

Table 19 Input File Listing with the W-element (Continued)

Listing Type Field Solver Monte Carlo Example

Three conductors share the .MODEL cond3_sys W MODELTYPE=FieldSolver,

same shape + LAYERSTACK=stack1,
+ CONDUCTOR=(SHAPE=r1,MATERIAL=copper,ORIGIN=(x1,'dRef+dY1')),
+ CONDUCTOR=(SHAPE=r1,MATERIAL=copper,ORIGIN=(x2,'dRef+dY1+dY2')),
+ CONDUCTOR=(SHAPE=r1,MATERIAL=copper,ORIGIN=(x3,'dRef+dY1+dY2'))

Analysis, outputs and end .PROBE TRAN v(in1) v(out1) v(in3)

.PROBE AC v(out1) v(out3)
.PROBE DC v(in1) v(out1) v(out3)
.AC LIN 200 0Hz 0.3GHz
.DC VIMPULSE 0v 5v 0.01v
.TRAN 0.5ns 100ns SWEEP MONTE=3 .END

Example 5: Modeling Coaxial and Shielded Twin-Lead Lines Using

the Polar Field Solver
The following examples show how to model a coaxial line and a twin-lead line. The
keyword COORD=polar (or COORD=1) invokes the polar field solver. When the polar field
solver is used, the conductor position is defined in polar coordinates (radius, angle in
degrees). Only one dielectric is permitted and the dielectric layer is surrounded by ground.
• Coax Line
• Shielded Twin-Lead Line

Coax Line
Figure 85 Polar field solver for modeling coaxial lines

*PETL Example: Coaxial Line

.OPTION PROBE POST
VIMPULSE in1 gnd AC=1v PULSE 4.82v 0v 5ns
+0.5ns 0.5ns 25ns

*W element

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 277
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

W1 in1 gnd out1 gnd FSMODEL=coax N=1, L=1

R1 out1 gnd 50

* [[ Material List ]]
.MATERIAL diel_1 DIELECTRIC ER=4
.MATERIAL copper METAL CONDUCTIVITY=57.6meg

* [[ Shape List ]]
.SHAPE circle_1 CIRCLE RADIUS=0.5m

* [[ Layer Stack ]]
.LAYERSTACK coaxial LAYER=(diel_1 11m) $ only one

* [[ Field solver option ]]

.FSOPTIONS myOpt printdata=yes compute_rs=yes compute_gd=yes
+ compute_go=yes

* [[ Field solver model ]]

.MODEL coax W MODELTYPE=FIELDSOLVER FSOPTIONS=myOpt COORD=polar
+ LAYERSTACK=coaxial, RLGCFILE=coax.rlgc
+ CONDUCTOR = ( SHAPE=circle_1, MATERIAL=copper, ORIGIN=(0, 0) )

.TRAN 0.5n 100n

.PROBE v(in1) v(out1)
.END

Shielded Twin-Lead Line

Figure 86 Polar field solver for modeling twin-lead line

*PETL Example: Shield twin-lead lines

.OPTION PROBE POST
VIMPULSE in1 gnd AC=1v PULSE 4.82V 0v 5ns
+0.5ns 0.5ns 25ns
*W element
W1 in1 in2 0 out1 out2 0 FSMODEL=twin, N=2, L=1
R1 out1 gnd 50
R2 out2 gnd 50
R3 in2 gnd 50

* [[ Material List ]]
.MATERIAL diel_1 DIELECTRIC ER=4

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 278
W-2024.09
 Feedback
Chapter 7: Using the Field Solver to Extract Tx Line Parameters
Field Solver Examples

.MATERIAL copper METAL CONDUCTIVITY=57.6meg

* [[ Shape List ]]
.SHAPE circle_1 CIRCLE RADIUS=0.5m

* [[ Layer Stack ]]
.LAYERSTACK coaxial LAYER=(diel_1 11m)) $ only one

* [[ Field solver option ]]

.FSOPTIONS myOpt printdata=yes compute_rs=yes compute_gd=yes
compute_go=yes
* [[ Field solver model ]]
.MODEL twin W MODELTYPE=FIELDSOLVER FSOPTIONS=myOpt COORD=polar
+ LAYERSTACK=coaxial, RLGCFILE=twin.rlgc
+ CONDUCTOR = ( SHAPE=circle_1, MATERIAL=copper,
ORIGIN=(4.5m, 0) )
+ CONDUCTOR = ( SHAPE=circle_1, MATERIAL=copper,
ORIGIN=(4.5m, 180) )

.TRAN 0.5n 100n

.PROBE v(in1) v(out1) v(out2)
.END

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 279
W-2024.09
 Feedback

8
Using IBIS Files

Describes how IBIS model components are included in a PrimeSim HSPICE simulation,
with usage models for IBIS buffer, component package, board-level components (EBD),
and Interconnect Specification (ICM).

PrimeSim HSPICE ships numerous of examples for your use. Find paths to IBIS-related
demo files under IBIS Examples in the PrimeSim HSPICE User Guide: Demonstration
Netlists.
For information about IBIS-Algorithmic Modeling Interface (AMI) usage with PrimeSim
HSPICE, see Chapter 9, Using IBIS-AMI Equalizer Models with StatEye.
The Input/Output Buffer Information Specification (IBIS) specifies a standard ASCII format
bufferIBIS

for presenting information describing the behavior of various I/O buffers that send electrical
signals outside the silicon chip or receive such signals.
IBIS promotes tool independent I/O models for system level Signal Integrity work and has
become:
• The ANSI/SAE-ITC and IEC 62014-1 standard for describing the analog behavior of
the buffers of digital devices using plain ASCII text-formatted data.
• An alternative to models - IBIS files are not models, they contain the data to be used by
the behavioral models and algorithms inside the simulator.
Except for SPICE or Verilog-A formatted buffers, IBIS specifies formats for the following
types of information:
• Output I-V curves for output buffers in LOW and HIGH states.
• V(t) curves describing the exact form of transitions from LOW to HIGH states and from
HIGH to LOW states for a given load.
• Values for die capacitance.
The IBIS standard was developed by the IBIS Open Forum, affiliated with the Society of
IBISmodels

Automotive Engineers - Industry Technologies Consortia (SAE-ITC). IBIS specifies only

the “form” for the information; it does not specify how the information is processed or used
by the simulator.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 280
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Terminology

For introductory information on using IBIS with PrimeSim HSPICE, see Simulating Circuits
with IBIS Models in PrimeSim HSPICE in this user guide.
The following sections discuss these topics:
• Terminology
• Verifying IBIS Files with the Golden Parser
• Specifying Models for IBIS Buffers
• Troubleshooting Signal Propagation Issues/Overclocking
• Multiple Rising and Falling Waveforms Support
• Differential Pins
• Buffers in Subcircuits
• Netlist Example with Output Buffer, Transmission Line, and Input Buffer
• Using the IBIS Component Command
• Using IBIS Package Modeling
• Using IBIS Board-Level Components
• Using IBIS Interconnect Modeling (ICM)
•

Terminology
The following terms are frequently used in this chapter:
Table 20 Terminologies Used in This Chapter
Term Definition

B-element The B-elementis a single instance of an IBIS buffer, whereas the .IBIS
component defines a component that has multiple buffers and can also include
package and pin information. Model selection is not available when using the
B-element.

.IBIS component The .IBIS component enables model selection using mod_sel. The mod_sel
keyword in the .IBIS statement supports multiple model assignments. If you
want to model package effects, the .IBIS component allows the use of package
models that are defined in the IBIS file or in a separate package file.

card, buffer card Denote lines from the netlist that specifies the buffer name (should begin with
the letter b), a list of external nodes, required keyword, and optional keywords.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 281
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Verifying IBIS Files with the Golden Parser

Table 20 Terminologies Used in This Chapter (Continued)

Term Definition

buffer, I/O buffer, One of 14 IBIS models as specified in the standard, version 5.0, and
input/output buffer implemented in the Synopsys IBIS device models.

RWF, FWF Rising waveform, falling waveform

I/O Input/Output

I/V curve Current-voltage curve

PU, PD Pullup, pulldown

PC, GC Power clamp, ground clamp

Verifying IBIS Files with the Golden Parser

The IBIS standard contains a section devoted to recommendations on how to derive
information from either the simulation or silicon measurement. The IBIS Open Forum has
sponsored development of a parser for IBIS files—called the golden parser. The golden
golden parser (IBIS)

parser is freely available as an executable and should be used for verification of IBIS files.
The golden parser is incorporated into Synopsys circuit simulators. When processing
an IBIS file, the golden parser produces warnings or error messages that appear in the
PrimeSim HSPICE output by default.

Specifying Models for IBIS Buffers

This section describes IBIS buffer usage in PrimeSim HSPICE. The I/O buffer element
type is called buffer and in PrimeSim HSPICE is commonly referred to as the B-element
or b-element. For information on IBIS syntax conventions for I/O buffers and buffer types,
see Modeling Input/Output Buffers Using IBIS Files in PrimeSim Continuum User Guide:
Elements.
Using buffers is similar to using other simulation elements, such as transistors: specify a
name for the buffer, specify a list of nodes that connect the buffer to the rest of the circuit,
and specify parameters. Only parameters that specify a model for the buffer (file name and
model name) are required. The IBIS model name is limited to 60 characters. Messages
similar to the following are issued if the name exceeds 60 characters.
**error** Model
'EDD2516AETA_NORMAL_iobuf_xxxxxx_xxxxx_xxxxx_xxxxx_xxxxx_xxxxx_xxxxx' is
not found
**error** .\md8.sp:174) difficulty in reading input

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 282
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Troubleshooting Signal Propagation Issues/Overclocking

Except for SPICE or Verilog-A formatted buffers, there are two significant differences from
the use of other elements:
• You can specify a total of 3 to 8 external nodes depending on the buffer type.
• If nodes are supposed to connect to power or ground rails, do not connect them in the
netlist because the simulation connects the nodes by default.
This chapter is not intended to introduce the IBIS standard, because it is a large
document; familiarity with the standard is assumed. A significant amount of information is
available on the Internet.
The official IBIS Open Forum web site is located at:
http://www.ibis.org
This site contains articles introducing IBIS, text of the IBIS standard, examples of IBIS
files, and tools such as the golden parser. The site also links to other web sites devoted to
IBIS.
Three types of analysis are supported for input/output buffers:
• DC analysis
• Transient analysis
• AC analysis

Troubleshooting Signal Propagation Issues/Overclocking

If you connect a source to your buffer using the correct nodes and do not see any output
or have a distorted waveform, your input voltage swings may be out of range, or your input
stimulus (digital signal) is switching too fast, such that overclocking occurs.
In the case of overclocking, you may see a “jump” when a transition is triggered before the
previous one is completed. This shows up as an artificial skew in simulation.
PrimeSim HSPICE can handle overclocking issues without making modifications to an
IBIS model by using IBIS keywords that remove initial delays and flat tails from the rising
and falling waveforms.
The keywords are as follows:
• rm_dly_rwf= default | rdly_time_value
• rm_dly_fwf= default | fdly_time_value
• rm_tail_rwf= default | rtail_time_value
• rm_tail_fwf= default | ftail_time_value

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 283
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Multiple Rising and Falling Waveforms Support

rm_dly_rwf and rm_dly_fwf remove initial delays from [Rising Waveform] and [Falling
Waveform], respectively.
rm_tail_rwf and rm_tail_fwf remove flat tails from [Rising Waveform] and [Falling
Waveform], respectively.
For each keyword, if the value is set as a positive number, then that amount of time is
removed from the corresponding V-T waveforms.
For each keyword, if the value is set as default, then for each V-T waveform, PrimeSim
HSPICE computes the amount of no-switching time based on the criteria that the voltage
change should be less than 0.2% of full voltage swing, then selects the minimum one as
the final value used.
The removal of initial no-switching time by 0.2% criteria is the default for both B-
element and .IBIS cards. So there is no need to set rm_dly_rwf=default or
rm_dly_fwf=default. These two keywords are just for backward compatibility.

For detailed information on the keywords, see Specifying Required and Optional Common
Keywords in PrimeSim Continuum User Guide: Elements.

Multiple Rising and Falling Waveforms Support

An IBIS model specification can contain more than one rising or falling VT waveform
table. If more than two rising or falling VT waveforms exist in the model, the PrimeSim
HSPICE simulator selects two suitable waveforms for accurate modeling. The selection is
based on the initial voltage of the VT waveforms at time zero. Before the buffer transition,
PrimeSim HSPICE sorts all rising or falling waveforms by initial voltage and then selects
two waveforms so that the current die pad voltage lies between the initial voltages of these
two waveforms. PrimeSim HSPICE uses the PU and PD scaling coefficients derived from
the two VT waveforms for the transient simulation.

Differential Pins
Differential pins refer to the relationship between buffers. This facilitates creating pseudo-
bufferdifferential pins

differential buffer by two B-elements. Specify these pins in the “Component Description”
section of the IBIS standard. Figure 87 and Figure 88, and the examples that follow these
figures, show how you can simulate differential pins using the Synopsys implementation of
IBIS.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 284
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Differential Pins

Figure 87 Output buffers

(Logical) Direction of signal

Inverter
Output
Buffer

input for
output buffer Differential pins

Output
Buffer
Transmission line

You must use two separate buffers, each specified in a separate card in the netlist. They
are related only through their input, which is differential. To implement the inverter in this
situation, you must specify two independent voltage sources that have opposite polarity.
You must specify the out_of_in_1, and out_of_in_2 nodes, even if they are not used.
Differential input buffers probe the voltage between nd_in_1 and nd_in_2. A voltage-
dependent voltage source processes the voltage.
V_diff is a differential voltage parameter from the IBIS file (default is 200 mV). Add a
definition of the V_diff parameter, the E_diff_out_of_in voltage controlled voltage source,
and a R_diff_out_of_in resistor.

Figure 88 Input buffers

(Logical) Direction of signal

nd_in_1 Transmission line

out_ot_in_1 Input
Buffer

Differential pins

out_ot_in_2 Input
Buffer
nd_in_2
Transmission line

For example,
.PARAM V_diff = 0.2
E_diff_out_of_in diff_out_of_in 0 PWL(1) nd_in_1 nd_in_2

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 285
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Differential Pins

+ '- V_diff' 0 '+ V_diff' 1

R_diff_out_of_in diff_out_of_in 0 1x

Use the voltage across R_diff_out_of_in as the output of the differential input buffer as
shown below:
If V(nd_in_1) - V(nd_in_2) < V_diff, V(diff_out_of_in) = 0
if V(nd_in_1) - V(nd_in_2) > V_diff, V(diff_out_of_in) = 1

In IBIS 4.1, a true differential buffer with SPICE-formatted external model is introduced.
You can use it with the rules in Multilingual Model Support and SPICE or Verilog-A
Formatted B-element Naming Rules sections.

Differential and Pseudo-Differential Buffers

While you use the traditional IBIS model for a single-ended IO buffer, you can use two B-
elements for low voltage differential signal (LVDS). For this case:
• Use inverse sitimulus, if the same IBIS model is used.
• Use the same stimulus, if two IBIS models are used (for the same model data except
for a polarity flag)
• For the receiver, add an E-element for A2D translation (from diffV to 0/1 signal).

Figure 89 Pseudo-differential buffer

• For greater accuracy, use a series buffer for differential mode characteristics.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 286
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Buffers in Subcircuits

Figure 90 Series buffer for best accuracy

• [External Model] is a good choice for differential buffer modeling.

Buffers in Subcircuits
The following examples demonstrate usage of buffers in subcircuit definition. Also,
modification in the circuit description, by changing the power keyword value, is explained.
**********************************************
* example 1 * buffers in subcircuit, power=on
**********************************************
v_in1 nd_in1 0 pulse
+ ( 0V 1.0V CLK_Q_PRD DLT_TIME DLT_TIME CLK_H_PRD CLK_PRD )
v_en1 nd_en1 0 1V
v_in2 nd_in2 0 pulse
+ ( 1.1V 0V CLK_Q_PRD DLT_TIME DLT_TIME CLK_H_PRD CLK_PRD )
v_en2 nd_en2 0 1V
x1 nd_out1 nd_in1 nd_en1 nd_outofin1 buffer11
x2 nd_out2 nd_in2 nd_en2 nd_outofin2 buffer11

R_load nd_out1 nd_out2 50

.subckt buffer11 nd_out0 nd_in0 nd_en0 nd_outofin0

b_io_0 nd_pu0 nd_pd0 nd_out nd_in0 nd_en0 nd_outofin0 nd_pc0
+ nd_gc0
+ file = '92lv090b.ibs'
+ model = 'DS92LV090A_DOUT'
+ typ=typ power=on
+ buffer=3
+ interpol=1
xpin nd_out nd_out0 pin22
.ends

.subckt pin22 nd_out nd_out0

R_pin nd_out_c nd_out0 50m
C_pin nd_out_c 0 0.3p

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 287
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Buffers in Subcircuits

L_pin nd_out nd_out_c 2n

.ends

In this example, buffers are connected to power sources implicitly, inside the subcircuit.
Subcircuit external terminals does not need to include nd_pu, nd_pd, nd_pc, and nd_gc.
**********************************************
* example 2* buffers in subcircuit, power=off
**********************************************
v_in1 nd_in1 0 pulse
+ ( 0V 1.0V CLK_Q_PRD DLT_TIME DLT_TIME CLK_H_PRD CLK_PRD )
v_en1 nd_en1 0 1V
v_in2 nd_in2 0 pulse
+ ( 1.1V 0V CLK_Q_PRD DLT_TIME DLT_TIME CLK_H_PRD CLK_PRD )
v_en2 nd_en2 0 1V

x1 nd_power 0 nd_out1 nd_in1 nd_en1 nd_outofin1 nd_power 0

+ buffer11
x2 nd_power 0 nd_out2 nd_in2 nd_en2 nd_outofin2 nd_power 0
+ buffer11

R_load nd_out1 nd_out2 50

.subckt buffer11 nd_pu0 nd_pd0 nd_out0 nd_in0 nd_en0

+ nd_outofin0 nd_pc0 nd_gc0
r_0 nd_pu0 nd_pd0 1.23456789x
b_io_0 nd_pu0 nd_pd0 nd_out nd_in0 nd_en0 nd_outofin0 nd_pc0
+ nd_gc0
+ file = '92lv090b.ibs'
+ model = 'DS92LV090A_DOUT'
+ typ=typ power=off
+ buffer=3
+ interpol=1
xpin nd_out nd_out0 pin22
.ends

.subckt pin22 nd_out nd_out0

R_pin nd_out_c nd_out0 50m
C_pin nd_out_c 0 0.3p
L_pin nd_out nd_out_c 2n
.ends

V_power nd_power 0 3.3V

In the example above, only one voltage source (V_power) is used to power all buffers.
Specify the power nodes, nd_pu, nd_pd, nd_pc, and nd_gc. In other words, if subcircuit
external terminals include nd_pu, nd_pd, nd_pc, or nd_gc, take care that the two X
elements calling the same subcircuit are not be connected to the same node of pu, pd, pc,
or gc. Otherwise, this results in a voltage loop error.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 288
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Netlist Example with Output Buffer, Transmission Line, and Input Buffer

Netlist Example with Output Buffer, Transmission Line, and Input

Buffer
An example of a netlist that contains an output buffer, transmission line, and input
transmission linesexample input bufferexample

buffer is shown below. A digital signal is supplied to the nd_in node. The output buffer
bufferinputexamplebufferoutputexample outputbufferexample

transmits to a network, goes through a transmission line, is received by an input buffer. It is

transformed into digital form and available on the out_of_in node.
* IBIS Buffer Test
.option post
.tran 0.05n 70n
*
* input source
v1 in 0 pulse ( 0V 1V 1n 1n 1n 9n 20n )
r1 in in1 50
* tristate enable
v5 enable 0 1V

* transmission line
wline1 n1 0 n2 0 RLGCmodel=pcb N=1 L=0.3

* IBIS buffers
b1 nd_pc nd_gc n2 out_of_in
+ file = 'at16245.ibs'
+ model = 'AT16245_IN'

b4 nd_pu nd_pd n1 in1 enable

+ file = 'at16245.ibs'
+ model = 'AT16245_OUT'
+ ramp_fwf=0 ramp_rwf=0

* load
rload n2 0 50

* RLCG parameters for W-element

.model pcb w modeltype=rlgc n=1
+L0=3.94266e-7
+C0=1.12727e-10
+R0=5.7739
+G0=0
+Rs=0.00141445
+Gd=0

.end

Using the IBIS Component Command

The .IBIS command creates all buffers, package and interconnection for an entire
component in an IBIS file.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 289
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

For the syntax of the .IBIS command, see .IBIS in the PrimeSim HSPICE Reference
Manual: Commands and Control Options.
The following sections discuss these topics:
• How .IBIS Creates Buffers
• Required Keywords
• Optional Keywords
• Component Calls for SPICE or Verilog-A Formatted Pins
• Component Calls for SPICE or Verilog-A Formatted [External Circuit]
• Buffer Power

How .IBIS Creates Buffers

The .IBIS command adds a buffer to the netlist for every pin, according to the
signal_name and model_name defined in the [Pin] keyword in the *.ibs file.

Note:
The .IBIS command does not create a buffer if the pin name is a reserved
model name, such as POWER, GND, or NC.

buffer_name = 'cname'_'pin_name'

• cname is defined in the .IBIS card in the *.sp netlist.

• pin_name is defined in the [Pin] keyword in the *.ibs file

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 290
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

Figure 91 Package and component combined

Component Package

Buffer1
node1 node2 node3

PrimeSim HSPICE connects the new buffers that the .IBIS command creates to the
following nodes:
• name of buffer 1: ‘cname'_'pin_name’
• name of node 1: 'cname'_'pin_name'_i
• name of node 2: 'cname'_'pin_name'_o
• name of node 3: 'cname'_'pin_name'
Note that:
• 'cname'_'pin_name'_en is the enable node if the buffer has enable node.
• 'cname'_'pin_name'_i is the outofin node if the buffer is an input buffer; for an input/
output buffer, the outofin node is 'cname'_'pin_name'_outofin.

Required Keywords
file=’file_name’
This keyword identifies the IBIS file. The file_name parameter must be lower case and
must specify either the absolute path for the file or the path relative to the directory from
which you run the simulation.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 291
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

Example
file = '.ibis/at16245.ibs'
file = '/home/oneuser/ibis/models/abc.ibs'

component=’component_name’
This keyword identifies the component for an .IBIS command from the IBIS file, specified
using the file='...' keyword. The component_name keyword is case-sensitive, and it
must match one of the components from the IBIS file.
Example
component = 'procfast'
component = 'Virtex_SSTL_3-I_BG432'

Optional Keywords
The following sections discuss these optional keywords:
• package
• pkgfile=’pkg_file_name’
• [Model Selector] Support
• Other Optional Keywords

package
This keyword specifies the type of package to add.

package = [0|1|2|3|4]

When package is set to

• ◦ 0, then the RLC package is not added into the component.
• 1, then [Package] (in the ibs file) is added.
• 2, then [Pin] (in the ibs file) is added. If the package information is not set in [Pin], set
the package with [Package] as a default.
• 3 (default), and if [Package Model] is defined, set package with a package model. If the
[Package Model] is not defined, set the package with [Pin]. If the package information
is not set in [Pin], set the package with [Package] as a default. You can define the
[Package Model] in IBIS or PKG files.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 292
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

• 4, then the package model of [Package Model] is added.

You can define the [Package Model] in an IBIS file specified by the file keyword or a
PKG file specified by the pkgfile keyword. The pkgfile keyword is useful only when
package =3 or 4.
Example
.ibis p_test
+ file = 'comp.ibs'
+ component = 'cpu_133mhz_ff'
+ hsp_ver = 2002.4 nowarn
+ package = 3
+ pkgfile = 'test.pkg'

This card combines the .pkg and .ibs files. If you use [Package Model] in the
cpu_133mhz_ff component, HSPICE searches for the pkg model in the comp.ibs and
test.pkg files.

pkgfile=’pkg_file_name’
You can define the package model file using the pkgfile keyword.

pkgfile = 'pkg_file_name'

If you cannot find a package model defined within a component in both the PKG and IBIS
files, HSPICE issues an error message.

[Model Selector] Support

You can select a model using the mod_sel keyword.

mod_sel = 'selName_1=modName_1, selName_2=modName

_2,...,selName_n=modName_n'

If selName and modName cannot be found in a related IBIS file or they cannot be matched,
then an error occurs. In addition, the input string should be less than 1024 characters. If
a model selector is used for a pin of a component, but mod_sel is not set, then the first
model under the corresponding [Model Selector] is selected as default.

Other Optional Keywords

The following keywords are all optional keywords of the .IBIS command. They have
the same function as for the B-element (I/O buffer). For detailed function descriptions,
see Specifying Required and Optional Common Keywords in PrimeSim Continuum User
Guide: Elements.
• typ
• interpo

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 293
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

• ramp_rwf
• ramp_fwf
• ss_state
• time_control
• dt_scale
• pd_scal
• pu_scal
• pc_scal
• gc_scal
• rwf_scal
• fwf_scal
• nowarn
• hsp_ver
• c_com_pd
• c_com_pu
• c_com_pc
• c_com_gc
• detect_oti_mid
If any keyword above is used, then it will be applied to all buffers created by the .IBIS
command.

Component Calls for SPICE or Verilog-A Formatted Pins

The syntax to call a [Component] having SPICE or Verilog-A formatted [Pin] is:

.IBIS ibis_command_name
+ file='ibis_filename' component='component_name'
+ [package=3|0|1|2] [pkgfile='pkg_file_name']
+ [nowarn]
+ [mod_sel = 'selName_1=modName_1,
+ selName_2=modName_2,...,selName_n=modName_n']

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 294
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

In the preceding syntax,

• The component_name specification must involve one or more SPICE formatted [Pin].
• Usage of the other parameters follow .ibis conventions.
Other parameters not shown above have no affect on a SPICE formatted buffer. If the
buffer is a series buffer or a SPICE or Verilog-A formatted true differential buffer in which
two pins are involved, then:

buffer_name='cname'_'pin1_name1'_'pin2_name'

For related node names, refer to SPICE or Verilog-A Formatted B-element Naming Rules.

Component Calls for SPICE or Verilog-A Formatted [External

Circuit]
[External Circuit] in a component can be divided into two types. One is used to describe
the buffer and is similar to [External Model]; the other is used to describe interconnection
inside a component. For the former, [Pin Mapping] can not be used in order to avoid
ambiguous connections. For the latter, a die node declared in Port_map of [Circuit Call]
can be probed in a netlist by the naming rule: 'cname'_'die_node_name'. Here, cname is
defined in the .ibis card in the .sp netlist. In addition, die pad must not occur in Port_map
since related connection syntax is still not available in the latest IBIS spec. If a buffer is
described by [External Circuit], then it is set as power off, so voltage sources are not be
created automatically to connect to external terminals of Pullup, Pulldown, Powerclamp
and Groundclamp.

Buffer Power
The buffer component creates buffers that are always connected to the power sources
that are specified in the IBIS file. You can specify power sources in two different ways, as
described in the following sections:
• Buffer Power ON
• Buffer Power OFF

Buffer Power ON
If [Pin Mapping] is not defined in the ibs file or hsp_ver <= 2002.2, PrimeSim HSPICE
automatically sets the power of buffers to ON. Use the [Voltage Range], [Pullup
Reference], [Pulldown Reference], [POWER Clamp Reference], and [GND Clamp
Reference] keywords in the ibs file to set the voltage source inside the buffer.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 295
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

The buffer nodes expect input and output node names in this format:

'buffer_name'_node_name
buffer_name = 'cname'_'pin_name'

• cname is defined in the .ibis card in the .sp netlist.

• pin_name is defined in the [Pin] keyword in the ibs file
• node_name is different for different types of buffers as shown in the following list:

INPUT pc , gc

OUTPUT pu , pd , pc , gc

INPUT_OUTPUT pu , pd , en , outofin , pc , gc

THREE_STATE pu , pd , en , pc , gc

OPEN_DRAIN pu , pd , pc , gc

IO_OPEN_DRAIN pu , pd , en , outofin , pc , gc

OPEN_SINK pu , pd , pc , gc

IO_OPEN_SINK pu , pd , en , outofin , pc , gc

OPEN_SOURCE pu , pd , pc , gc

IO_OPEN_SOURCE pu , pd , en , outofin , pc , gc

INPUT_ECL pc , gc

OUTPUT_ECL pu , pc , gc

IO_ECL pu , en , outofin , pc , gc

THREE_STATE_ECL pu , en , pc , gc

Note:
For more information about nodes for different buffers, see Buffer Types in
PrimeSim Continuum User Guide: Elements.
Table 21 shows the names of the input and output nodes for the buffers:
Table 21 Input and Output Node Names for Buffers
Buffer Type Node Names

INPUT and INPUT_ECL buffers 'cname'_'pin_name' (for in

node)'cname'_'pin_name'_i (for outofin node)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 296
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

Table 21 Input and Output Node Names for Buffers (Continued)

Buffer Type Node Names

Other types of buffers ‘cname’_’pin_name’_i (for in

node)‘cname’_’pin_name’ (for out node)

If the buffer has an enable terminal, you must create a node named buffer_name_en to
enable the buffer.
You can test the following sample cases using the same method explained.
In this example, you can test the buffers with the B-element.
*********
.tran 50p 30n
.option post probe
.subckt twobus_b out1 out2
boutput1 nd_pu1 nd_pd1 out1 nd_in1 nd_pc1 nd_gc1
+ file = 'pinmap.ibs'
+ model = 'out50v'
+ buffer=2 power_on
+ nowarn
boutput2 nd_pu2 nd_pd2 out2 nd_in2 nd_pc2 nd_gc2
+ file = 'pinmap.ibs'
+ model = 'out50v'
+ buffer=2 power=on
+ nowarn
Vin1_b nd_in1 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Vin2_b nd_in2 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
.ends

Xtwobus_b out1_b out2_b twobus_b

Rout1_b out1_b 0 50
Rout2_b out2_b 0 50

.probe tran
+ out1_of_output_b = v(out1_b)
+ out2_of_output_b = v(out2_b)
+ in_of_output_b = v(Xtwobus_b.nd_in1)
.end

In this example, you can test the buffers that the .IBIS command creates.
*********
.tran 50p 30n
.option post probe
* add component
.ibis pcomp
+ file = 'pinmap.ibs'
+ component = 'NO_PINMAPPING'

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 297
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

+ nowarn
+ package = 0
Vin1 pcomp_1_i 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Vin2 pcomp_2_i 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Rout1_1 pcomp_1 0 50
Rout2_1 pcomp_2 0 50

.probe tran
+ out1_of_nopm = v(pcomp_1)
+ out2_of_nopm = v(pcomp_2)
+ in_of_output = v(pcomp_1_i)
.end

The following ibs file is:

|*********************************************************
| IBIS file pinmap.ibs
|*********************************************************|
[IBIS ver] 3.1
[File name] pinmap.ibs
[File Rev] 1.0
[Date] 5/24/2002
[Source] spice models
[Notes] This ibis file tests the pinmap
results of HSPICE .ibis command. Pinmapping
is considered only if hsp_ver > 2002.2.

|*********************************************************
| Component NO PINMAPPING
|*********************************************************
|
[Component] NO_PINMAPPING
|
[Manufacturer] TEST
|

[Package] typ min max

R_pkg 0 NA NA

L_pkg 0 NA NA

C_pkg 0 NA NA

[Pin] signal_name model_name R_pin L_pin C_pin

OUT1 out50v 50.0m 2.01554n 0.33960p

OUT2 out50v 50.0m 1.83759n 0.31097p

GND1 GND 50.0m 1.89274n 0.32012p

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 298
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

VCC1 POWER 50.0m 1.74394n 0.31941p

GND2 GND 50.0m 1.89274n 0.32012p

VCC2 POWER 50.0m 1.74394n 0.31941p

.....................
|model information
....................
[END]

Buffer Power OFF

If [Pin Mapping] is defined both in the ibs file and hsp_ver > 2002.2, PrimeSim HSPICE
automatically uses the [Pin Mapping] keyword in the ibs file to load voltage sources to the
buffers. PrimeSim HSPICE turns OFF the power of buffers that the component created.
In this case, the nd_pc, nd_pu, nd_gc, and nd_pd nodes connect to the power and ground
bus according to the information in [Pin Mapping]. All other nodes are the same as in
Buffer Power ON.
The following two .sp files are equivalent. The first file uses the B-element to describe
buffers:
* test for IO,OUTPUT,3-STATE,INPUT BUFFER in component w/pin
mapping
.option post probe
.tran 50p 30n
.param vhi = 2.5

.subckt io_o_3st_i_b in st gnd1 vcc1 io output

bin vcc1 gnd1 in nd_out1
+ file = 'pinmap.ibs'
+ model = 'DS92LV090A_IN1'
+ buffer=1 power=off
+ nowarn
b3st vcc1 gnd1 st nd_in1 nd_en1 vcc1 gnd1
+ file = 'pinmap.ibs'
+ model = 'DS92LV090A_RO'
+ buffer=4 power=off
+ nowarn
bio vcc1 gnd1 io nd_in2 nd_en2 nd_ofi vcc1 gnd1
+ file = 'pinmap.ibs'
+ model = 'DS92LV090A_DOUT'
+ buffer=3 power=off
+ nowarn
boutput vcc1 gnd1 output nd_in3 vcc1 gnd1
+ file = 'pinmap.ibs'
+ model = 'out50v'
+ buffer=2 power=off
+ nowarn

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 299
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

Vin1_b nd_in1 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )

Ven1_b nd_en1 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Vin2_b nd_in2 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Ven2_b nd_en2 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Vin3_b nd_in3 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Rout_b nd_out1 0 50
.ends

Xio_o_3st_i_b in_b st_b gnd1_b vcc1_b io_b output_b io_o_3st_i_b

Vvcc1_b vcc1_b 0 3.3v
Vgnd1_b gnd1_b 0 0v
Vin_b in_b 0 0V PULSE (0 vhi 1ns 0.5ns 0.5ns 5ns 10ns)
Rout1_b st_b 0 50
Rout2_b io_b 0 50
Rout3_b output_b 0 50
.probe tran
+ out_of_in_b = v(Xio_o_3st_i_b.nd_out1)
+ out_of_3st_b = v(st_b)
+ out_of_io_b = v(io_b)
+ out_of_output_b = v(output_b)
+ in_of_output_b = v(Xio_o_3st_i_b.nd_in3)
.end

The second file uses a the .IBIS command to create buffers:

* test with component
.option post post_version=9601 probe
.tran 50p 30n
.param vhi = 2.5
.ibis pcomp
+ file = 'pinmap.ibs'
+ component = 'IO_OUTPUT_3ST_INPUT'
+ nowarn
+ package = 0

Vin1 pcomp_3_i 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )

Ven1 pcomp_3_en 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Vin2 pcomp_6_i 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Ven2 pcomp_6_en 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Vin3 pcomp_7_i 0 0V pulse ( 0V 1V 1n 0.1n 0.1n 7.5n 15n )
Rout pcomp_2_i 0 50

Vvcc1 pcomp_5 0 3.3v

Vgnd1 pcomp_4 0 0v
Vin pcomp_2 0 0V PULSE (0 vhi 1ns 0.5ns 0.5ns 5ns 10ns)
Rout1 pcomp_3 0 50
Rout2 pcomp_6 0 50
Rout3 pcomp_7 0 50

.probe tran
+ out_of_in = v(pcomp_in_2)
+ out_of_3st = v(pcomp_3)
+ out_of_io = v(pcomp_6)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 300
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using the IBIS Component Command

+ out_of_output = v(pcomp_7)
+ in_of_output = v(pcomp_7_i)
.end

The .ibs file is:

|*********************************************************
| IBIS file pinmap.ibs
|*********************************************************|
[IBIS ver] 3.1
[File name] pinmap.ibs
[File Rev] 1.0
[Date] 5/24/2002
[Source] spice models
[Notes] This ibis file tests the pinmap
results of HSPICE .ibis command. Pinmapping
is considered only if hsp_ver > 2002.2.
|*********************************************************
| Component IO,OUTPUT,3-STATE,INPUT BUFFER
|*********************************************************

[Component] IO_OUTPUT_3ST_INPUT[Manufacturer]

TEST

[Package] typ min max

R_pkg 0 NA NA

L_pkg 0 NA NA

C_pkg 0 NA NA

|
|*********************************************************
|

[Pin] signal_name model_name R_pin L_pin C_pin

NC NC 50.0m 2.11433n 0.27505p

IN DS92LV090A_IN1 50.0m 2.01554n 0.33960p

ST DS92LV090A_RO 50.0m 1.94357n 0.32802p

GND1 GND 50.0m 1.89274n 0.32012p

VCC1 POWER 50.0m 1.74394n 0.31941p

IO DS92LV090A_DOUT 50.0m 1.83759n 0.31097p

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 301
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Package Modeling

OUTPUT out50v 50.0m 1.83759n 0.31097p

|*********************************************************|

[Pin Mapping] pulldown_ref pullup_ref gnd_clamp_ref power_clamp_ref

NC NC NC NC

NC NC GND1 VCC1

GND1 VCC1 GND1 NC

NC VCC1 NC NC

GND1 VCC1 GND1 VCC1

GND1 VCC1 GND1 VCC1

.....................
|model information
....................
[END]

Using IBIS Package Modeling

In addition to .IBIS, PrimeSim HSPICE provides the .PKG command to parse package
model definition in either .ibs or .pkg files. The .PKG command automatically creates a
series of W-elements or discrete R, L and C components. For the complete description
of this command, see the PrimeSim HSPICE Reference Manual: Commands and Control
Options.
The following sections discuss these topics:
• Syntax
• Examples
• Accessing Nets inside a Package Model

Syntax
.PKG pkgname
+ pkgfile='pkgfilename'
+ model='pkgmodelname'

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 302
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Package Modeling

Examples
Example 1
.pkg p_test
+ pkgfile='processor_clk_ff.ibs'
+ model='FCPGA_FF_PKG'

Example 2—This example shows how in the second line, pin1 is referenced as the
element name changes:
p_test_pin1_dia and p_test_pin1
w_p_test_pin1_? ? or r_p_test_pin1_? ? ...

Accessing Nets inside a Package Model

You can access the nets (internal pins) of the package model defined in the IBIS file. The
node name is compName_pinName. Where compName is the component name defined in the
.IBIS command and pinName is the Pin name given by the [Pin] definition in the package
model. For example
.ibis buff
+ file = 'buff1.ibs'
+ component = 'Comp12'

The IBIS file contains the following package information for the pins that you want to
access.
[Package Model] pkg_model
[Pin] signal_name model_name R_pin L_pin
C_pin
vdd vdd POWER
gnd gnd GND
out out driver
out2 out2 driver

Using .PROBE statements, you can probe the nets as follows:

.probe v(buff_vdd) $ pin vdd defined in the package model
+ v(buff_gnd) $ pin gnd defined in the package model
+ v(buff_out) $ pin out defined in the package model
+ v(buff_out2) $ pin out2 defined in the package model

Note:
If [pin mapping] is not defined in the IBIS file, then there is no connection
to the pins of vdd and gnd and the nodes of buff_vdd and buff_gnd are not
created. PrimeSim HSPICE creates voltage sources automatically for pins of
out and out2.If [pin mapping] is defined in the IBIS file, then PrimeSim HSPICE
connects PU, PD, PC, and GC of out and out2 to pins of vdd and gnd (in fact,

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 303
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

to nodes of buff_vdd_o and buff_gnd_o, due to package) according to [pin

mapping].

Using IBIS Board-Level Components

A board-level component is used to describe a printed circuit board (PCB) or substrate
that can contain components or even other boards and can connect to another board
through a set of user visible pins. The electrical connectivity of such a board level
component is referred to as an “Electrical Board Description” (EBD). The .EBD command
provides connectivity descriptions of chip to chip, or multiple-chips on a single package.
The .EBD command is associated with .IBIS command, because the EBD file contains
the reference designator of the component. PrimeSim HSPICE search paths between
the component pins and the EBD pins are displayed in the [Path Description] of the EBD
file. The component pins are designated with keyword:component in the .EBD command,
which automatically create Lumped RLCs and distributed RLCs (W-elements) to express
those paths. PrimeSim HSPICE then simulates the newly-created circuit.
In the following netlist, you can see the important elements and internal nodes that are
created in the naming rules of the .EBD command.
**********************************************************
* Analysis And Options
**********************************************************

.op
.tran 10p 30n

.option probe
**********************************************************
* Stimulus
**********************************************************

Vin cmp1_5_i 0 0V pulse ( 0V 3V 2n 0.1n 0.1n 7.5n 15n )

Vpu cmp1_1 0 2.5v

vpc cmp1_2 0 2.5v
vpd cmp1_3 0 0v
vgc cmp1_4 0 0

**********************************************************
* Rload
**********************************************************

Rd1 ebd1_a3 0 50
Rd2 ebd1_a3 0 50

**********************************************************
* Define Component

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 304
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

**********************************************************
.ibis cmp1
+ file = 'pinmap.ibs'
+ component = 'Component1'
+ package = 0

*********************************************************
* Define EBD
*********************************************************

.ebd ebd1
+ file = 'pinmap.ebd'
+ model = 'Board1'
+ component = 'cmp1:u21'

********************************************************
* Output
********************************************************

.probe tran
+ cmp1_5_out = v(cmp1_5) $ buf_5(Output): cmp1_5
+ cmp1_5_in = v(cmp1_5_i) $ no package , so the out node is cmp1_5,
$ not cmp1_5_o

+ ebd1_a3 = v(ebd1_a3)
+ ebd1_a4 = v(ebd1_a4)

.end

The following sections discuss these topics:

• .EDB and .IBIS Command Syntax
• Circuit Topology Created by the .EBD and .IBIS Commands
• B-element Naming Rules
• Circuit Topology Created with SPICE or Verilog-A Formatted Pins
• IBIS Board-Level Component Examples

.EDB and .IBIS Command Syntax

Use the following syntax when using IBIS board-level components:

.EBD 'ebd_name'
+ file = 'ebd_file_name' $ case-sensitive
+ model = 'ebd_model_name' $ case-sensitive
+ component = 'ibis_or_ebd_name:ref_des' $ case-insensitive
+ [component = 'ibis_or_ebd_name:ref_des'[...]]

.IBIS 'ibis_name'

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 305
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

+ file = 'ibis_file_name' $ case-sensitive

+ component = 'component_name' $ case-sensitive
+ [package = 0|1|2|3]
+ [other_keyword = value [...]]

Note:
Parameters surrounded by arrow brackets (< >) are not required.
Table 22 .EBD/.IBIS Required Argument Descriptions
Parameter Description

ebd_name Specifies the name of the .EBD command. This argument is case-insensitive.

ibis_or_ebd_name Specifies the name of associated .IBIS or .EBD command. This argument is
case-insensitive.

ebd_file_name Identifies the EBD file. You must specify either the absolute path for the file or the
path relative to the directory from where you run the simulation. This argument is
case-sensitive. For example:
file = './ebd/test.ebd' file = '/home/usr/ebd/test.ebd'

ebd_model_name Identifies the Board-Level Component. Visible in the [Begin Board Description] keyword
in the EBD file. This argument is case-sensitive.

ibis_or_ebd_name: Maps an IBIS buffer component or EBD board to a reference designator.

ref_des • ibis_or_ebd name: Name of the associated .IBIS command or associated .EBD
command to identify IBIS buffer component or EBD board.
• ref_des: Reference designator. Visible in the Node sub-parameter of the [Path
Description] keyword in the EBD file.
For example: component = 'cmp2:u22'

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 306
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

Circuit Topology Created by the .EBD and .IBIS Commands

The circuit topology in Figure 92 is created by the .EBD and IBIS. commands:

Figure 92 Circuit topology created by the .EBD and .IBIS commands

D_1
Component EBD

A_1 B_1 C_1 D_i

Buf_1 Pkg

S_Buf_i
Die

A_n C_n D_m

Buf_n Pkg
B_n

B-element Naming Rules

The following rules in Table 23 apply when naming B-elements in IBIS board-level
components:
Table 23 B -element Naming Rules
Buffer Name Note

Buf_# ‘ibis_name’_’ibis_pin_name’ If model_name shows NC/POWER/GND in the column

of the [Pin] keyword in the IBIS file, no Buf_# buffer
exists.

S_Buf_# ‘ibis_name’_’pin1_name’_’pin2_name’ If the [Series Pin Mapping] keyword is not in the IBIS
file, no S_Buf_# series/series switch buffer exists.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 307
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

Table 24 Node Naming Rules

Node Name Corresponding Buffer Node

A_# ‘ibis_name’_’ibis_pin_name’_i Terminator buffer: No A_#

nodes
Input/Input_ecl buffer:
nd_out_of_in
Other buffers: nd_inIf no
Buf_#, no A_# nodes

B_# ‘ibis_name’_’ibis_pin_name’_o Input/input_ecl buffer: nd_in

Other buffers: nd_out
Note: If no package (i.e. the
package keyword in the .IBIS
command is set to 0 or no
relative sub- parameter is
given in [Pin]/[Package] in
IBIS file), then no B_# nodes.
The corresponding nodes are
named as C_# nodes.

C_# ‘ibis_name’_’ibis_pin_name’

D_# ‘ebd_name’_’ebd_pin_name’

Enable ‘ibis_name’_’ibis_pin_name’_en Buffer that has the nd_en

Enable node.
Other buffers: No

Out_of_in ‘ibis_name’_’ibis_pin_name’_outofin Buffer that has out_of_in node

except the Input/Input_ECL
buffer

Pulldown of Buf_# ‘ibis_name’_’ibis_pin_name’_pd Buffer that has the Pulldown

or nd_pd node.
‘ibis_name’_’ibis_pin_pd_name_o’(if pin mapping Other buffers: No
is defined in the IBIS file)

Pullup of Buf_# ‘ibis_name’_’ibis_pin_name’_pu Buffer that has the Pullup

or nd_pu node.
‘ibis_name’_’ibis_pin_pu_name_o’ (if pin mapping Other buffers: No
is defined in the IBIS file)

GND Clamp of Buf_# ‘ibis_name’_’ibis_pin_name’_pd Buffer that has the GND

or Clamp nd_gc node.
‘ibis_name’_’ibis_pin_gc_name_o’ (if pin mapping Other buffers: No
is defined in the IBIS file)

POWER Clamp of Buf_# ‘ibis_name’_’ibis_pin_name’_pc Buffer that has POWER

or Clamp nd_pc node.
‘ibis_name’_’ibis_pin_pc_name_o’ (if pin mapping Other buffers: No
is defined in the IBIS file)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 308
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

In the following node names, the first matched pin name in [Pin] consists of ground/power
connections given in [Pin Mapping] in the *.ibs file.
• 'ibis_name': Name of .IBIS command.
• 'ebd_name': Name of .EBD command.
• 'ibis_pin_name': Pin Name, the first column in Keyword: [Pin] in the IBIS file.
• 'ebd_pin_name': Pin Name, the pin name given in [Path Description] in the EBD file.
• 'pin1_name': One of two pins joined by a series model, the first column in Keyword:
[Series Pin Mapping] in the IBIS file.
• 'Pin2_name': The other of two pins joined by a series model, the second column in
Keyword: [Series Pin Mapping] in the IBIS file.
• 'ibis_pin_pd_name'
• 'ibis_pin_pu_name'
• 'ibis_pin_gc_name'
• 'ibis_pin_pc_name': The first matched pin name in [Pin] which is the ground/power
connections given in [Pin Mapping] in the IBIS file.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 309
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

Circuit Topology Created with SPICE or Verilog-A Formatted Pins

The circuit topology in Figure 93 is created by the .EBD and .IBIS commands where the
IBIS component model contains SPICE or Verilog-A formatted pins:

Figure 93 Circuit topology with SPICE or Verilog-A formatted pins

Component

A_1 B_1 C_1 D_1

Buf_1 Pkg

A_1_2 S_Buff_1
Die (SPICE-Formatted EBD
Series Switch)

A_2 C_2 D_i

Buf_2 Pkg
B_2
B_3 C_3
A_3_4 Pkg D_m
TDBuf_3
Pkg
(SPICE or Verilog-A-Formatted B_4 C_4
True Differential)

SPICE or Verilog-A Formatted B-element Naming Rules

The following rules (Table 25) apply when naming SPICE or Verilog-A formatted B-
elements in IBIS board level components:
Table 25 SPICE or Verilog-A Formatted B-element Naming Rules
Buffer Name Comment

Buf_# ‘ibis_name’_’ibis_pin_name’ If model_name shows NC/POWER/GND in the column

of the [Pin] keyword in the IBIS file, no Buf_# buffer
exists.

TDBuf_# 'ibis_name'_'pin1_name'_'pin2_name’ If [Diff Pin] keyword is not in the IBIS file, or it is not a
true differential buffer, no TDBuf_# buffer exists.

S_Buf_# 'ibis_name'_'pin1_name'_'pin2_name' If the [Series Pin Mapping] keyword is not in the IBIS
file, no S_Buf_# series/series switch buffer exists.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 310
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

The following rules (Table 26) apply when naming SPICE or Verilog-A formatted nodes in
IBIS board level components:
Table 26 Node naming rules—SPICE/Verilog-A formatted buffers in component
Node Name Corresponding buffer
node/port

A_# ‘ibis_name’_’ibis_pin_name’_i Terminator buffer: No A_#

nodes Input/Input_ecl
buffer: nd_out_of_inOther
buffers: nd_in

B_# ibis_name'_'ibis_pin_name'_o Input/input_ecl buffer: nd_in

Other buffers: nd_out
Note: If no package (i.e.
the package keyword
in the .IBIS command
is set to 0 or no relative
sub-parameter is given in
[Pin] / [Package] in IBIS
file), then no B_# nodes.
The corresponding nodes
are named as C_# nodes.

C_# 'ibis_name'_'ibis_pin_name'

D_# 'ebd_name'_'ebd_pin_name'

A_#_# 'ibis_name'_'pin1_name'_'pin2_name'_i Input_diff buffer:

nd_out_of_in
Other buffers: nd_in

enable 'ibis_name'_'ibis_pin_name'_en or Buffer that has an enable

'ibis_name'_'pin1_name'_'pin2_name'_en (if buffer is node (nd_en)
true differential buffer)

Out_of_in 'ibis_name'_'ibis_pin_name'_outofin Buffer that has out_of_in

or'ibis_name'_'pin1_name'_'pin2_name'_outofin(if node except Input,
buffer is true differential buffer) Input_ECL and, Input_diff
buffers

Pulldown of Buf_# 'ibis_name'_'ibis_pin_name'_pd A_pdref

or'Ibis_name'_'ibis_pin_pd_name'_o' (if pin mapping
is defined in the IBIS file)

Pullup of Buf_# 'ibis_name'_'ibis_pin_name'_pu or A_puref

'ibis_name'_'ibis_pin_pu_name'_o (if pin mapping is
defined in the IBIS file)

GND Clamp of Buf_# 'ibis_name'_'ibis_pin_name'_gc or A_gcref

'ibis_name'_'ibis_pin_gc_name'_o (if pin mapping is
defined in the IBIS file)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 311
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

Table 26 Node naming rules—SPICE/Verilog-A formatted buffers in component (Continued)

Node Name Corresponding buffer
node/port

POWER Clamp of Buf_# clamp'ibis_name'_'ibis_pin_name'_pc or A_pcref

'ibis_name'_'ibis_pin_pc_name'_o (if pin mapping is
defined in the IBIS file)

External Reference of Buf_# reference 'ibis_name'_'ibis_pin_name'_ext or A_extref

'ibis_name'_'ibis_pin_ext_name'_o (if pin mapping is
defined in the IBIS file)

Global reference 0 A_gnd

Non-inverting node of S_Buf_# 'ibis_name'_'pin1_name'_o A_pos

Inverting node of S_Buf_# 'ibis_name'_'pin2_name'_o A_neg

Non-inverting node of TDBuf_# 'ibis_name'_'pin1_name'_o A_signal_pos

Inverting node of TDBuf_# 'ibis_name'_'pin2_name'_o A_signal_neg

other nodes of Buf_# 'ibis_name'_'ibis_pin_name'_'port_name' other ports defined by user

other nodes of TDBuf_# 'ibis_name'_'pin1_name'_'pin2_name'_ 'port_name' other ports defined by user

The following names are the same as in Table 24 except that for TDBuf_#, 'pin1_name',
and 'pin2_name' are two pins joined by a true differential buffer model. They are the first
and second columns in the keyword: [Diff Pin] in the IBIS file:
• 'ibis_name'
• 'ibis_pin_name'
• 'pin1_name'
• 'pin2_name'
• ’ibis_pin_pd_name'
• 'ibis_pin_pu_name'
• 'ibis_pin_gc_name'
• 'ibis_pin_pc_name’

IBIS Board-Level Component Examples

The following examples show how you can use the .IBIS and .EBD commands for board-
level components.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 312
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

Example 1: .IBIS Command File

The IBIS model for this example is the file 4mx32.ibs. To use the .IBIS command, it
is necessary to find the [Component] keyword in the .ibs file as in the following part of
the .ibs file:
|
[Component] K4D263238A |
[Manufacturer] TEST
[Package]
| variable typ min max
R_pkg 0.17 77.00m 0.77
L_pkg 2.84nH 0.56nH 6.04nH
C_pkg 3.20pF 0.55pF 5.20pF
|
[Pin] signal_name model_name R_pin L_pin C_pin
B3 DM0 CTRL_PIN 0.71 5.75nH 5.20pF
B5 DQ3 DQ_PIN 0.12 0.86nH 0.57pF
B6 DQ2 DQ_PIN 83.00m 0.64nH 0.92pF
B7 DQ0 DQ_PIN 88.00m 0.56nH 0.53pF
C6 DQ1 DQ_PIN 0.11 0.78nH 0.56pF
M11 CK CLK_PIN 0.36 1.64nH 2.52pF

In the example above, the component name is K4D263238A. In addition to the

[Component] keyword, this portion of the .ibs file also shows the pin list. This information
becomes important later.
Note:
There are three models relating to this component: CTRL_PIN, DQ_PIN, and
CLK_PIN. If ramp_rwf and ramp_fwf keywords are set, then all models take the
keyword values. Of course, such keywords may be nonsense for some model
types.
This component only uses [Ramp] to define the V-t characteristics of the buffers so; the
ramp_rwf and ramp_fwf keywords can be set to 0 for this example. There is package
information for this component that is contained in the package file 4mx32.pkg. The
package information can be specified in the .IBIS command. The .IBIS command
requires an instance name of the component to be specified. For this case, the name bc1
is used. The ibis component definition can then be written as:
* ibis component
.ibis bc1 component='K4D263238A'
+ file='4mx32.ibs'
+ ramp_rwf=0 ramp_fwf=0
+ package=3
+ pkgfile='4mx32.pkg'

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 313
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

Note:
The component keyword is case sensitive and the specified component name
must match the case of [Component] in the .ibs file or PrimeSim HSPICE
prints an error message that the component is not defined.
Example 2: .IBIS Command File
** HSPICE Netlist **
.ibis cmp1
+ file = 'test1.ibs'
+ component = 'Component1'
+ package = 1

.ebd ebd1
+ file = 'test.ebd'
+ model = 'Board1'
+ component = 'cmp1:u21'

** IBIS file: test1.ibs **

[Component] Component1
| ... ...
[Pin] signal_name model_name R_pin L_pin C_pin
|
1 RAS0 Buffer1 200.0m 5.0nH 2.0pF
2 gnd GND 209.0m NA 2.5pF
3 pwr POWER NA 6.3nH NA
| ... ...
[Pin Mapping] pulldown_ref pullup_ref gnd_clamp_ref power_clamp_ref
ext_ref
|
1 GNDBUS1 PWRBUS1 |
2 GNDBUS1 NC |
3 NC PWRBUS1
| ... ...
[Series Pin Mapping] pin_2 model_name function_table_group
2 3 CBTSeries 1 | Four independent groups

The nodes listed in Table 27 are used with B-elements in the example above.
Table 27 B -element Nodes
Buffer/Node Name

A_1 cmp1_1_i

B_1 cmp1_1_o

B_2 cmp1_2_o

B_3 cmp1_3_o

C_1 cmp1_1

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 314
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

Table 27 B -element Nodes (Continued)

Buffer/Node Name

C_2 cmp1_2

C_3 cmp1_3

D_1 ebd1_a1

D_2 ebd1_a2

D_3 ebd1_a3

Pulldown of Buf_1 cmp1_2_o

Pullup of Buf_1 cmp1_3_o

GND Clamp of Buf_1 cmp1_1_gc

POWER Clamp of Buf_1 cmp1_1_pc

Buf_1 cmp1_1

S_Buf_1 cmp1_2_3

Example 3: .IBIS Command File

The following is an example where the component keyword of use the map of another ebd
to reference designator:
** HSPICE Netlist **
.ibis cmp1
+ file = 'testibis.ibs'
+ component = 'testibis'
+ nowarn
.ebd ebd1
+ file = 'testebd1.ebd'
+ model = 'testebd1'
+ component = 'cmp1:u2'
+ component = 'ebd2:u20'
.ebd ebd2
+ file = 'testebd2.ebd'
+ model = 'testebd2'
+ component = 'cmp1:u2'
** IBIS file: testibis.ibs **
[Component] testibis
| ...
[Pin] signal_name model_name R_pin L_pin C_pin
|C8 DQ1 IO 0.0980 1.1970nH 0.4340pF

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 315
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

Example 1: .EBD Command File

To use the .EBD command, it is necessary to find the name of the board as given by the
[Begin Board Description] keyword in the .ebd file.
The complete netlist, IBIS model, package and .ebd files for this example can be found
in $installdir/demo/hspice/ibis/ebd. The name of the board is used for the model
name in the .EBD command. In this example, the name of the board description is Board1.
The reference designators of the components on the board are given by the [Reference
Designator Map] keyword and are used by the component keyword of the .EBD command.
In this example, there is only one reference designator, U1 which is part of the .ebd file:
[Begin Board Description] Board1
[Manufacturer] Test
|
[Number Of Pins] 2
|
[Pin List] signal_name
B7 DQ0
C6 DQ1
[Path Description] Trace1
Pin B7
Len = 0.5 L=8.35n C=3.34p R=0.01 /
Len = 0.1 L=2.1n C=1.15p R=0.01 /
Node u1.B7
[Path Description] Trace2
Pin C6
Len = 0.5 L=8.35n C=3.34p R=0.01 /
Len = 0.1 L=2.1n C=1.15p R=0.01 /
Node u1.C6
[Reference Designator Map]
| Ref Des File name Component name
u1 4mx32.ibs K4D263238A
[End Board Description]

The .EBD command requires an instance name of the board to be specified. For this case,
the name sb1 is used. The board definition can be written as:
* board description
.ebd sb1
+ file='4mx32.ebd'
+ model='board1'
+ component=bc1:u1

Now that both the ibis component and board description are defined, the input, enables,
and terminations for the buffers can be defined. The node naming convention follows the
node naming rules defined in Table 24. The complete netlist can be written as:
* EBD Usage Example
* ibis component
.ibis bc1 component='K4D263238A'
+ file='4mx32.ibs'

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 316
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Board-Level Components

+ ramp_rwf=0 ramp_fwf=0
+ pkgfile='4mx32.pkg'
+ package=3

* board description
.ebd sb1
+ file='4mx32.ebd'
+ model='board1'
+ component=bc1:u1

* enables
v3 bc1_b7_en 0 1
v4 bc1_c6_en 0 1
* data
vdq0 bc1_b7_i 0 pulse(0 1 1n 0.1n 0.1n 5n 20n)
vdq1 bc1_c6_i 0 pulse(0 1 1n 0.1n 0.1n 5n 15n)

* terminations
r1 sb1_b7 0 50
r2 sb1_c6 0 50

Example 2: .EBD Command File

Here is an example of an .ebd file:
[Begin Board Description] Board1
|... ...
[Pin List] signal_name
A1 GND
A2 data1
A3 data2
|... ...
[Path Description] CAS_2
Pin A1
Len = 0.5 L=8.35n C=3.34p R=0.01 /
Node u21.1
Len = 0.5 L=8.35n C=3.34p R=0.01 /
Node u21.2
Len = 0.5 L=8.35n C=3.34p R=0.01 /| this section is
| discarded because
| the Node u23.3
| reference
| designator:u23 is not
| designated to IBIS
| component in the .ebd
| ebd1 command.

Example 3: .EBD Command Files

The following are examples of two .ebd files:
** EBD file: testebd1.ebd **
[Begin Board Description] testebd1

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 317
W-2024.09
 Feedback
Chapter 8: Using IBIS Files
Using IBIS Interconnect Modeling (ICM)

[Manufacturer] Test
|...
[Pin List] signal_name
12 RDQ8
[Path Description] 12
Pin 12
Len=0.10604 L=8.39999e-009 C=2.74272e-012 R=0.25139 /
Len=0 R=15.00000 /
Fork
Len=0.07935 L=8.39999e-009 C=2.74272e-012 R=0.25139 /
Node U2.C8
Endfork
Len=0.07063 L=8.39999e-009 C=2.74272e-012 R=0.25139 /
Node U20.A13

** EBD file: testebd2.ebd **

[Begin Board Description] testebd2
[Manufacturer] Test
|...
[Pin List] signal_name
A13 DQ1
[Path Description] A13
Pin A13
Len=0.18690 L=8.38871e-009 C=2.32868e-012 R=0.12121 /
Node U2.C8

Using IBIS Interconnect Modeling (ICM)

ICM is a standard for behavioral descriptions of interconnect electrical characteristics. The
text describes a consistent format that can be parsed by software, allowing the simulation
tool and interconnect vendors to create models compatible with their own products.
The PrimeSim HSPICE .ICM command automatically creates port names that reference
the pin name of an ICM model and generates a series of element nodes on the pin.
Syntax

.ICM icmname
+ file='icmfilename'
+ model='icmmodelname'

Example
.ICM icm1
+ file='test1.icm'
+ model='FourLineModel1'

For the complete description of the command, see the PrimeSim HSPICE Reference
Manual: Commands and Control Options, .ICM.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 318
W-2024.09
 Feedback

9
Using IBIS-AMI Equalizer Models with StatEye

Describes use of the Algorithmic Modeling Interface (AMI) to enable StatEye analysis of
systems with Serializer-Deserializer (SERDES) devices for equalizers in both transmitter
and receiver sides.

PrimeSim HSPICE supports an interface that uses the IBIS Version 5.0 interface to the
C/C++ language-based compiled shared object called Algorithmic Modeling Interface
(AMI). The AMI models complex linear and non-linear equalizers and enables the StatEye
AMI utilityAlgorithmic Modeling InterfaceIBISAMI utilityStatEyeAMI usage

analysis of systems with Serializer-Deserializer (SERDES) devices for equalizers in both

transmitter (TX) and receiver (RX) sides.

Figure 94 High-speed serial links can be composed of three parts: Transmitter algorithmic
part (Tx), the analog channel, and Receiver algorithmic part (Rx)

See Chapter 3, Statistical Eye Analysis, for a full discussion of the .STATEYE command
and application examples.
These topics are discussed in the following sections:
• IBIS-AMI Model Structure
• Using the Standalone AMI Object Checking Program ’amicheck’
• How AMI Works in PrimeSim HSPICE
• StatEye Analysis with AMI Using Demo Example
• Parametric Control over AMI File for Quoted Expressions
• Using an IBIS-AMI Mid-Channel Repeater Model
• References

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 319
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
IBIS-AMI Model Structure

IBIS-AMI Model Structure

An IBIS-AMI model is an executable “black box” that is:
• Written in C/C++
• Object files are platform dependent
• Limited to one IBIS-AMI model per device
• Accompanied by parameter definition file
• Supports statistical simulation
In the IBIS file (*.ibs), the model is defined by the keywords:
[Algorithmic Model]
[End Algorithmic Model]

Example
[Algorithmic Model]
Executable Linux_gcc_32 libtx_ami.so
tx_ami_params.ami
[End Algorithmic Model]

IBIS-AMI Data Flow

Each IBIS-AMI object file contains the following sections:
• AMI_Init
◦ Initializes impulse response filter
◦ Sets up data structures
• AMI_GetWave
◦ Enables waveform processing
◦ Sets up clock and data recovery
• AMI_Close
◦ Frees memory
◦ Closes files, etc.
• Parameter file (*.ami) for model and simulation control

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 320
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
Using the Standalone AMI Object Checking Program ’amicheck’

Requirements for Using IBIS-AMI Models with PrimeSim HSPICE

The following are the requirements for PrimeSim HSPICE-AMI usage:
• IBIS-AMI models are to be used only with PrimeSim HSPICE .STATEYE analysis.
• The IBIS-AMI parameter file (*.ami) can be encrypted using 3DES encryption.
• Required files:
◦ AMI parameter file – *.ami
◦ PrimeSim HSPICE netlist
◦ Either C/C++ files that define the model, or Makefile – required to compile the
model
◦ Or platform-specific binary AMI object files, *.so or *.dll

Using the Standalone AMI Object Checking Program ’amicheck’

PrimeSim HSPICE includes a standalone utility program which provides a simple way to
examine PrimeSim HSPICE's AMI filtering flow without composing a full simulation netlist.
By combining this utility program with a memory profiling tool such as valgrind on the
Linux platform, AMI model creators may detect possible memory access violations and
problems in filtering operations in advance of the model release to users.
For more information on using valgrind, see http://valgrind.org/
Caution:
Consider the following before using the amicheck utility:
• Ensure that the amicheck executable bitness matches with that of the object
file.
• Ensure to check all AMI files received from a model supplier using the
amicheck utility.

To run the checker program, on the command line type:

% amicheck object_filename parameter_filename

For the 64 bit version of amicheck, the syntax is:

% amicheck -64 object_filename parameter_filename

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 321
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
Using the Standalone AMI Object Checking Program ’amicheck’

Note:
PrimeSim HSPICE and amicheck uses standard shared object loader
function which attempts to find the target library in the path specified in the
LD_LIBRARY_PATH environment variable.

PrimeSim HSPICE and amicheck fails to locate the target library if you do not
have your current directory (“.”) included in the LD_LIBRARY_PATH environment
variable.
The following sections discuss these topics:
• amicheck Flow
• Windows OS ’amicheck’ Setup
• amicheck Output

amicheck Flow
On launch, the amicheck flow follows this process:
1. Applies linear time invariant (LTI) filters to the entire system using AMI_init
2. When supplied in AMI object, applies non-LTI filters to the bit stream response
waveforms using the AMI_GetWave Closes the AMI interface with AMI_Close
3. Calls AMI_Init, AMI_GetWave (if available) and AMI_Close with falling step input
4. Calls AMI_Init, AMI_GetWave (if available) and AMI_Close with pulse input
5. Unloads the AMI object and exits

Windows OS ’amicheck’ Setup

On the Windows OS, use the following steps to set up amicheck.
1. Go to My Computer or the Computer folder on your desktop and click the right mouse
button.
2. Select Properties and then Advanced system settings.
3. On the Advanced tab click the Environment Variables button.
4. In the System Variables field, select the Path variable and edit the value to add your
PrimeSim HSPICE installation path. For example, if the path to the PrimeSim HSPICE

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 322
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
How AMI Works in PrimeSim HSPICE

executable is C:\synopsys\Hspice_G-2012.09\BIN then the amicheck.exe utility is

also in this path.
5. Execute amicheck by using the DOS command line. For example:
C:\users\test> amicheck test_ami.dll test_param.ami

amicheck Output
The amicheck program generates these output files:
• parameter_filename.lis: program status output.
• parameter_filename.tree: parsed parameter tree
• object_filename.rise: rise step response
• object_filename.fall: fall step response
• object_filename.pulse: pulse response
WaveView can be used to display the waveform response files, object_filename.{rise,
fall, pulse}. These files contain the following columns:

• Input signal to AMI Object

• Output of AMI_Init command
• Output of AMI_GetWave command

How AMI Works in PrimeSim HSPICE

The compact standalone utility program (amicheck) detects potential problems in AMI
objects and parameter files before running PrimeSim HSPICE simulations. The amicheck
program takes the AMI object file name and parameter file name as command line
arguments to load and filter processes as used in actual PrimeSim HSPICE simulations.
AMI functionality is called at appropriate step(s) from incident and/or probe port(s) in
StatEye analysis when these port elements are configured with an AMI object (shared
library) file name and an AMI parameter file (*.ami) which contains specific keywords to
control the behavior of the AMI objects.
When a port (P-) element includes the keywords AMI_OBJ=filename and
AMI_PARAM=filename, StatEye opens interface(s) to the AMI object(s) and applies
waveform filtering processes using AMI_Init, AMI_GetWave, and AMI_Close functions at
appropriate analysis steps.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 323
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
How AMI Works in PrimeSim HSPICE

These topics are presented in the following sections:

• Input Syntax
• Saving AMI Pulse/Step Responses and Filters
• Visualizing Eye Diagram Output of an IBIS-AMI RX Model
• Usage Example

Input Syntax
Pi n1 n2 port=1 …
+ [AMI_OBJ=filename]
+ [AMI_PARAM=filename]
+ [AMI_CDR=0|1]
+ [AMI_CORNER=idx]

Keyword Description

AMI_OBJ Specifies an AMI shared object (typically, lib *.so for UNIX, *.dll for Windows

AMI_PARAM Specifies an AMI parameter file (*.ami)

AMI_CDR When AMI_CDR=1, StatEye analysis with specific pattern (LFSR/PAT) input takes
into account the clock data recovery (CDR) output information from the receiver side
AMI_GetWave function. Default = 0.

AMI_CORNER Specifies an index number. The IBIS keyword "...(Corner val1 val2 ...)" defines the
corner values in the IBIS-AMI file specified by AMI_PARAM. When you use AMI_CORNER,
idx is used to chose the idxth value defined by the "...(Corner ...)" keyword.
For example, if AMI_CORNER =2 is specified in a port element and the associated AMI
parameter file has the following:
(My_AMI_param1 ... (Corner 1 2 3))
(My_AMI_param2 ... (Corner 4 5 6))
(My_AMI_param3 ... (Corner 7 8 9))
Then, 2, 5 and 8 are chosen for My_AMI_param{1,2 and 3} respectively.

See also Statistical Eye Analysis Setup.

Note:
PrimeSim HSPICE's AMI parameter parsing rule—When GUI form contents
are given in the AMI parameter file, PrimeSim HSPICE uses your specified

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 324
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
How AMI Works in PrimeSim HSPICE

Default value. If you do not specify a Default value in the GUI content, then
PrimeSim HSPICE:
• Tries to set a “typical” value based on the Value/Range/List/Step/
Increment keyword. Then, when a value is determined, PrimeSim HSPICE
describes it in the *.lis file.
• Issues a warning.
For example, for this syntax:
(my_param
(Range 30 0 100)(Usage Info)(Type Integer)(Default 10)
)

PrimeSim HSPICE sets(my_param 10), while for this syntax:

(my_param
(Range 30 0 100)(Usage Info)(Type Integer)
)

PrimeSim HSPICE issues a warning and sets (my_param 30).

Saving AMI Pulse/Step Responses and Filters

When SAVE_TR=ascii is specified with AMI configurations, any system step/pulse
responses after AMI filtering are saved in netlist.saveAMI#/ directory where #
represents the .alter index number. If AMI filters are active, when LOAD_TR=ascii is
specified, StatEye loads pre-AMI filtered pulse/step responses from the netlist.save#/
directory, then applies the AMI filters again. This way, users can efficiently examine
multiple AMI filters with the same netlist.
Note:
When an explicit directory path is not given for the AMI object file name (for
example, only the file name is given), PrimeSim HSPICE and the amicheck
utility search the object file under directories given in the LD_LIBRARY_PATH
environment variable. Typically, when the current working directory (“.”) is not
a part of the LD_LIBRARY_PATH environment variable, specifying only the file
name with the target object located under the current directory causes a load
failure. For such cases, add the current directory to your LD_LIBRARY_PATH
variable or use the explicit path.

Visualizing Eye Diagram Output of an IBIS-AMI RX Model

To see the statistical eye at output of the IBIS-AMI RX model when you connect an
IBIS-AMI Rx model in your circuit, you need to properly assign port elements to the
inputs and outputs of the system and identify them in the .STATEYE command using the

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 325
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
StatEye Analysis with AMI Using Demo Example

INCIDENT_PORT and PROBE_PORT lists. You also need to add .PROBE statements to output
the desired eye diagram.
For example:
.STATEYE T=200p trf=25p incident_port=1 probe_port=2
.PROBE stateye eye(2) $ eye diagram at port 2

Usage Example
This example demonstrates PrimeSim HSPICE's StatEye analysis setup with a receiver
side equalizer described by AMI.
*
* StatEye with an AMI model
*
.param _BW=0.05n
.param trf = '_BW*0.1'
P1 in 0 port=1
P2 out 0 port=2
+ AMI_OBJ='./my_AMI_lib.so' AMI_PARAM='./
my_AMI_params.ami'
Xdut in out DUT
stateye T='_BW' trf='trf' incident_port=1
probe_port=2_port=2
.subckt DUT _in _out
*** your DUT
.ends
.end

StatEye Analysis with AMI Using Demo Example

The PrimeSim HSPICE installation includes an example to perform .StatEye analysis
with an AMI object to model a simple receiver side Finite Impulse Response (FIR) filter.
Locate the demo example in the $installdir/demo/hspice/si/stateyeAMI directory. The
example contains the PrimeSim HSPICE netlist (stateye_ami.sp), AMI parameter file
(SNPS_Rx.ami), C/C++ source code, and make systems (Makefiles) for each platform with
a pre-compiled AMI object. In the stateye_ami.sp file, the second port element (declared
as the probe port for .StatEye analysis) contains AMI object assignment statements
which are originally commented out as:
P2 out 0 port=2
+ AMI_OBJ='libSNPS_Rx.so' AMI_PARAM=SNPS_Rx.ami
**+ AMI_OBJ='./Linux64/libSNPS_Rx.so' AMI_PARAM=SNPS_Rx.ami

By activating one AMI assignment statement for your platform, you can examine the
effect of the AMI DFE module in the .StatEye analysis. Also with AMI parameter file,

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 326
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
StatEye Analysis with AMI Using Demo Example

SNPS_Rx.ami, you can modify the DFE behavior by changing the number of taps, tap
locations, and tap weights described in following part of the file.
(Model_Specific
(TAP
(Description "TAP (location weight) ")
(-1 -0.200)
(0 0.900)
(1 -0.030)
(2 0.035)
(3 -0.010)
)
)

Taking these tap coefficients, the example DFE model outputs:

Equation 85

where:
• : DFE output (equalized waveform)
• : DFE input (pre-equalized waveform)
• : i-th tap location
• : i-th tap weight
• : Unit interval

Figure 95 Output statistical eye diagram from the demo example: (left) without equalization
(right) with equalization

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 327
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
Parametric Control over AMI File for Quoted Expressions

Note:
Descriptions about definitions of controllable model specific parameters must
be provided by each AMI model provider. Follow vendor guidelines before
changing the contents of vendor supplied *.ami parameter files.

Parametric Control over AMI File for Quoted Expressions

StatEye analysis can evaluate quoted expressions in the AMI parameter file (*.ami file) by
using parameter values defined in a PrimeSim HSPICE netlist. This capability increases
usability for optimizing transmitter and receiver equalizations in high speed serial links.
In a PrimeSim HSPICE netlist, you can define an arbitrary number of parameters by use of
the .PARAM command. For example,
.PARAM param_name=value

Then, in the AMI parameter file (.ami file), you can insert expressions as (single) quoted
strings. When loading the AMI parameter file, PrimeSim HSPICE evaluates these quoted
strings as expressions, then builds a string to be passed to the AMI_Init function. Once
PrimeSim HSPICE parses the AMI parameter tree evaluates expressions the tool prints
the resulting parameter tree and the parameter string for the AMI_Init function to the
listing file.
Example
The following PrimeSim HSPICE netlist performs two .StatEye runs with different
parameter sets for an AMI parameter file.
*
* StatEye with SNPS Rx Equalizer Model in AMI
*
.param _BW=0.05n
.param trf = '_BW*0.01'
.param opfreq = '1.0/_BW'
P1 in 0 port=1
*+ LFSR (1 -1 0 'trf' 'trf''opfreq' 1 [14,5,2] )
P2 out 0 port=2
+ AMI_OBJ='./libSNPS_Rx.so'
AMI_PARAM=SNPS_Rx_param.ami
X1 in out CONN
.stateye T='_BW' trf='0.1*_BW' incident_port=1 probe_port=2
.opt post accurate delmax=1e-12
.subckt CONN _in _out
Rtest _in _out 50 RS=1e-3 $ skin effect resistor
Cin _in 0 0.3p
Cout _out 0 0.3p
.ends

.probe stateye eye(2)

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 328
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
Parametric Control over AMI File for Quoted Expressions

.param ami_filt_mag = 1.0

.param ami_tap_n1 = 0.0 ami_tap_0 = 1.0
.param ami_tap_1 = 0.0 ami_tap_2 = 0.0
.param ami_tap_3 = 0.0
.alter 2nd param set
.param ami_filt_mag = 0.9
.param ami_tap_n1 = -0.18 ami_tap_0 = 1.0
.param ami_tap_1 = -0.03 ami_tap_2 = 0.04
.param ami_tap_3 = -0.01
.end

The resulting AMI parameter file, SNPS_Rx_param.ami, contains expressions using

parameters which are defined in the PrimeSim HSPICE netlist.
(SNPS_Rx
(Description "SNPS Rx Equalizer Model")
(Reserved_Parameters
(Ignore_Bits 10)
(Max_Init_Aggressors 0)
(Init_Returns_Impulse False)
(GetWave_Exists True)
(Use_Init_Output False)
)
(Model_Specific
(TAP
(Description "TAP (loc weight) ")
(-1 ' ami_filt_mag * ami_tap_n1 ' )
(0 ' ami_filt_mag * ami_tap_0 ' )
(1 ' ami_filt_mag * ami_tap_1 ' )
(2 ' ami_filt_mag * ami_tap_2 ' )
(3 ' ami_filt_mag * ami_tap_3 ' )
)
)
)

This is the resulting parameter tree with expression evaluation at first alter (to be printed
out in the .lis file):
--------------------------------
, SNPS_Rx
, , Description "SNPS Rx Equalizer Model"
, , Reserved_Parameters
, , , Ignore_Bits 10
, , , Max_Init_Aggressors 0
, , , Init_Returns_Impulse False
, , , GetWave_Exists True
, , , Use_Init_Output False
, , Model_Specific
, , , TAP
, , , , Description "TAP (loc weight) "
, , , , -1 0
, , , , 0 1
, , , , 1 0

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 329
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
Parametric Control over AMI File for Quoted Expressions

, , , , 2 0
, , , , 3 0
--------------------------------

The resulting parameter tree with expression evaluation at the second alter is also printed
to the .lis file):
--------------------------------
, SNPS_Rx
, , Description "SNPS Rx Equalizer Model"
, , Reserved_Parameters
, , , Ignore_Bits 10
, , , Max_Init_Aggressors 0
, , , Init_Returns_Impulse False
, , , GetWave_Exists True
, , , Use_Init_Output False
, , Model_Specific
, , , TAP
, , , , Description "TAP (loc weight) "
, , , , -1 -0.162
, , , , 0 0.9
, , , , 1 -0.027
, , , , 2 0.036
, , , , 3 -0.009
--------------------------------

Here, the resulting eye diagrams show the effect of equalizer tap weight.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 330
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
Backchannel Communication Interface for IBIS-AMI’s Link Training Protocol

Figure 96 Effect of equalizer tap weight

Backchannel Communication Interface for IBIS-AMI’s Link

Training Protocol
The target feature adds keywords for the Backchannel Communication Interface (BCI) to
the IBIS-AMI simulation flow to allow information exchange between IBIS-AMI equalizer
models at Tx and Rx.
PrimeSim HSPICE interprets the new AMI reserved parameters such as BCI_Protocol,
BCI_State, BCI_ID, BCI_Message_Interval_UI, and BCI_Training_UI from *.ami files
and implements channel training flow.
The BCI equalizer training flow is implemented in the edge mode only. The repeater
support implementation is planned after upcoming release depending on the customer
requirements.

Input Syntax
No additional netlist keyword is necessary. PrimeSim HSPICE reads the reserved
keywords in the *.ami file and automatically activates channel training flow.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 331
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
Using an IBIS-AMI Mid-Channel Repeater Model

Output
There is no change in the output. The output from AM_Init and AMI_GetWave may include
BCI_State for either one of the following Off, Training, Converged, Failed, and Error
outputs.

BIC_ID Naming Convention

PrimeSim HSPICE decides and substitutes the string value for the BCI_ID keyword.
The ID is used as a prefix of the file names and used by Tx and Rx equalizers for
communication. StatEye uses the following naming convention.
Syntax
(BCI_ID (netlist_name.bci#/BCI_protocol_name_))

where
netlsti_name: main (PrimeSim HSPICE) netlist body name (without suffix, for
example, .sp)
#:alter index

For example, my_stateye.bic0/channel_ for the netlist, my_stateye.sp.

Using an IBIS-AMI Mid-Channel Repeater Model

The IBIS version 6.0 and later specification allows placing an IBIS-AMI equalizer model to
realize a mid-channel repeater. PrimeSim HSPICE's StatEye allows this analysis flow by
specifying the redriver=(rx_index, tx_index) or the retime=(rx_index, tx_index)
keywords to connect two channels in series.
The redriver keyword specifies a probe port index and an incident port index pair. The
mid-channel repeater configured as a re-driver will then be placed in between the specified
probe port and the incident port. The output of the upstream channel as the waveform at
the probe port will be fed into the incident port as an input to the downstream channel.
The retimer keyword specifies a probe port index and incident port index pair. The mid-
channel repeater configured as a re-timer will be placed between the specified probe port
and incident port. The output of the upstream channel as the waveform at the probe port
will be used to generate a digital stream where the CDR clock information will be used
when an AMI model is attached to the probe port. The generated digital stream will be fed
into the incident port as an input to the downstream channel.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 332
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
Using an IBIS-AMI Mid-Channel Repeater Model

Note:
The mid-channel repeater analysis, using either of the two preceding keywords,
requires StatEye's convolution mode (mode=conv) to be set.

Figure 97 Mid-Channel repeater

Mid-Channel Repeater Example

***
.param _UI = xxx

.param trf = '_UI*0.05'

.param opfreq = '1.0/_UI'

P1 in1 0 port=1
+ LFSR (-1 1 0 'trf' 'trf' 'opfreq' 2 [14 7 2] )
X1 in1 out1 CONN
P2 out1 0 port=2
+ AMI_OBJ='my_rx1.so' AMI_PARAM=my_rx1.ami

P3 in2 0 port=3
+ AMI_OBJ='my_tx2.so' AMI_PARAM=my_tx2.ami
X2 in2 out2 CONN
P4 out2 0 port=4
+ AMI_OBJ='my_rx2.so' AMI_PARAM=my_rx2.ami

.stateye T='_UI' trf='trf'

+ incident_port=1 3 probe_port=1 2 3 4
+ retimer = (2,3)
+ mode=conv
+ VD_RANGE=2.0 0.8 2.0 0.8

.probe stateye eye(1) eye(2) eye(3) eye(4)

.option post runlvl=6

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 333
W-2024.09
 Feedback
Chapter 9: Using IBIS-AMI Equalizer Models with StatEye
References

.subckt CONN _in _out

:
.ends
.end

Results

Upstream Channel Tx (port 1) Output Upstream Channel Rx (port 2) Output

Downstream Channel Tx (port 3) Output Downstream Channel Rx (port 4) Output

References
[1] IBIS Version 5.0 Specification:https://ibis.org/ver5.0/ver5_0.txt
[2] IBIS (I/O Buffer Information Specification) home page: http://www.eigroup.org/ibis/
default.htm
[3] IBIS Version 7.0 Specification:(Training/Analysis flow for channels with no repeater):
https://ibis.org/ver7.0/ver7_0.pdf
[4] IBIS Specification v7.0 :http://ibis.org/specs/

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 334
W-2024.09
 Feedback

Part 3: Feed-Forward Equalization with the

PrimeSim HSPICE Tool

Describes how the Feed-Forward Equalization can be modeled in a variety of methods in

PrimeSim HSPICE .

The following sections discuss this topic:

• Introduction to Feed-Forward Equalization

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 335
W-2024.09
 Feedback

10
Introduction to Feed-Forward Equalization
Equalization is important in high-speed data channels where frequency dependent losses
and dispersion result in inter symbol interference (ISI). Equalizers are used at both
transmit and receive positions to compensate for channel distortion.
Feed-forward equalization (FFE) is useful at the transmitter (TX) end, where it can be used
to increase the transmitted ratio of high-frequency to low-frequency symbol power to help
offset channel losses at higher frequencies. The basic FFE equalizer concept is shown in
Figure 98. Delayed and scaled versions of the transmitted signal are combined to reshape
its power spectrum.

Figure 98 A 4-tap FFE

This demo example shows three different approaches

• FFE Modeling With Verilog-A
• Built-in FFE Model Using Port Emphasis Parameters
• E-element Delay Based Modeling

FFE Modeling With Verilog-A

The time-sampled (Z-domain) transfer function of a4-tapFFE can be written as

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 336
W-2024.09
 Feedback
Chapter 10: Introduction to Feed-Forward Equalization
Built-in FFE Model Using Port Emphasis Parameters

Using Verilog-A, the transfer function is easily implemented in PrimeSim HSPICE using
a linear discrete-time Z-transform filter operator. The core of an example Verilog-A model
implementation is below:
// Z-transformFFE modelin Verilog-A
...
analog begin
V(out) <+ zi_nd( V(in),{a0,a1,a2,a3},{1},sampleT );
end
...

The zi_nd() function implements the numerator-denominator form of a Z-transform filter.

The real parameter values a0 to a3 represent the FFE tap coefficients, and sampleT is
the sampling interval. An advantage of this linear Verilog-A implementation is that it can
be analyzed for both linear and nonlinear simulations. In fact, the Z-domain model has a
continuous linear time response in PrimeSim HSPICE. This means its evaluation for both
TranEye and StatEye simulations are the same, and it can be used in StatEye with no
topology restrictions.

Built-in FFE Model Using Port Emphasis Parameters

PrimeSim HSPICE also includes a simple built-in 2-tap analog continuous-time feed-
forward equalizer for sources specified with the Port element. This built-in FFE is useful
for adding emphasis or de-emphasis to clock and data streams. The 2-tap configuration is
shown in

Figure 99 2-tap FFE

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 337
W-2024.09
 Feedback
Chapter 10: Introduction to Feed-Forward Equalization
E-element Delay Based Modeling

This FFE has a continuous input/output relationship given by

where the Emphasis_Time parameter sets the delay value (typically one-half the bit-
width or UI/2) and the Emphasis_Level or parameter controls both tap coefficients
according to

Note that negative values of results in de-emphasis, while positive values give
emphasis. This 2-tap combination restricts the emphasis/de-emphasis waveform effect to
the Emphasis_Time interval on rising and falling edges.

E-element Delay Based Modeling

Another option for constructing FFEs is to use the PrimeSim HSPICE E-element as a
delay and scaling element:
Exxx n+ n- [VCVS] DELAY in+ in- TD=val SCALE=val

where the propagation delay is specified with the input parameter TD and the scaling
is given by the SCALE parameter. By using a combination of E-elements as delays and
VCVS components(for scaling and summing), a two-tap FFE can be realized as shown in
Figure 100.

Figure 100 2-tap FFE using stacked E-elements

In this example, the main tap has an initial propagation delay of 0.5Unit Intervals(UI). The
subsequent tap has an additional delayof UI/2. Note that this example corresponds to the
sampled Z-domain transfer function.

But with an additional propagation delay and with a sampling interval of UI/2. E-element
FFE implementations are also linear, and can be used with all PrimeSim HSPICE analyses
including StatEye.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 338
W-2024.09
 Feedback

Appendices

Describes how to simulate circuits with Signetics drivers and Xilinx FPGAs.

The following topics are covered in this appendix:

• Simulating Circuits with Signetics Drivers
• Simulating Circuits with Xilinx FPGAs

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 339
W-2024.09
 Feedback

A
Simulating Circuits with Signetics Drivers

This appendix describes how to simulate circuits with Signetics drivers.

The following topics are covered in this appendix:

• Simulating Circuits with Signetics Drivers—Examples

Simulating Circuits with Signetics Drivers—Examples

PrimeSim HSPICE includes a Signetics I/O buffer library in the $installdir/parts/
signet directory. You can use these high-performance parts in the backplane design. An
example describes of how to combine these parts with transmission line models.

Figure 101 I/O drivers/receivers with package lead inductance, parallel 4" lossy microstrip
connectors
5.5 v

driver receiver

+
_ z0 = 75 z0 = 75 75
_
+
vin

The schematic in Figure 101 shows a pair of drivers driving 4 inches of transmission
line to a pair of receivers that drive 4 inches of transmission line. A cross-section of the
transmission line is shown in Figure 102.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 340
W-2024.09
 Feedback
Appendix A: Simulating Circuits with Signetics Drivers
Simulating Circuits with Signetics Drivers—Examples

Figure 102 Planar transmission line DLEV=2: microstrip sea of dielectric

Upper Ground Plane

Insulator

SP12
WD1=8 mil (5 mil) WD1=8 mil
TS=32 mil
TH1=1.3 mil line 1 line 1 TH1=1.3 mil

W1eff
(6 mil) HT1=10 mil

Lower Ground Plane

The following is an example of a chip model with package inductance:

.SUBCKT IO_CHIP IN OUT VCC XGND PIN_VCC=7n PIN_GND=1.8n
X1 IN1 INVOUT VCC1 XGND1 ACTINPUT
X2 INVOUT OUT1 VCC1 XGND1 AC109EQ
* Package inductance
LIN_PIN IN IN1 PIN_IN
LOUT_PIN OUT1 OUT PIN_OUT
LVCC VCC VCC1 PIN_VCC
LGND XGND1 XGND PIN_GND
.ENDS
$ TLINE MODEL - 2 SIGNAL CONDUCTORS WITH GND
$ PLANE
.MODEL USTRIP U LEVEL=3 ELEV=1 PLEV=1
+ TH1=1.3mil HT1=10mil TS=32mil KD1=4.5 DLEV=0 WD1=8mil
+ XW=-2mil KD2=4.5 NL=2 SP12=5mil
$ ANALYSIS / PRINTS
.TRAN .1NS 100NS
.PROBE V(STIM1) V(STIM2) $ Inputs
.PROBE V(TLOUT1) V(TLOUT2) V(TLOUT3) V(TLOUT4) $ Outputs
.END

Simulation results for this model are shown in Figure 103.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 341
W-2024.09
 Feedback
Appendix A: Simulating Circuits with Signetics Drivers
Simulating Circuits with Signetics Drivers—Examples

Figure 103 Simulation results of I/O Chips connected with Tlines

Here’s a full netlist example of how I/O chips connect with transmission lines:
.OPTION SEARCH='$installdir/parts/signet'
.OPTION POST=2 TNOM=27 NOMOD LIST METHOD=GEAR
.TEMP 27
$ DEFINE PARAMETER VALUES
.PARAM LV=0 HV=3 TD1=10n TR1=3n TF1=3n TPW=20n
+ TPER=100n TD2=20n TR2=2n TF2=2n LNGTH=101.6m
$ POWER SUPPLY
VCC VCC 0 DC 5.5
$ INPUT SOURCES
VIN1 STIM1 0 PULSE LV HV TD1 TR1 TF1 TPW TPER
VIN2 STIM2 0 PULSE LV HV TD2 TR2 TF2 TPW TPER
$ FIRST STAGE: DRIVER WITH TLINE
X1ST_TOP STIM1 OUTPIN1 VCC GND IO_CHIP PIN_IN=2.6n
+ PIN_OUT=4.6n
X1ST_DN STIM2 OUTPIN2 VCC GND IO_CHIP PIN_IN=2.9n
+ PIN_OUT=5.6n
U_1ST OUTPIN1 OUTPIN2 GND TLOUT1 TLOUT2 GND USTRIP L=LNGTH
$ SECOND STAGE: RECEIVER WITH TLINE
X2ST_TOP TLOUT1 OUTPIN3 VCC GND IO_CHIP PIN_IN=4.0n
+ PIN_OUT=2.5n
X2ST_DN TLOUT2 OUTPIN4 VCC GND IO_CHIP PIN_IN=3.6n
+ PIN_OUT=5.1n
U_2ST OUTPIN3 OUTPIN4 GND TLOUT3 TLOUT4 GND USTRIP L=LNGTH
$ TERMINATING RESISTORS

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 342
W-2024.09
 Feedback
Appendix A: Simulating Circuits with Signetics Drivers
Simulating Circuits with Signetics Drivers—Examples

R1 TLOUT3 GND 75
R2 TLOUT4 GND 75
$ IO CHIP MODEL - SIGNETICS
.SUBCKT IO_CHIP IN OUT VCC XGND PIN_VCC=7n PIN_GND=1.8n
X1 IN1 INVOUT VCC1 XGND1 ACTINPUT
X2 INVOUT OUT1 VCC1 XGND1 AC109EQ
*Package Inductance
LIN_PIN IN IN1 PIN_IN
LOUT_PIN OUT1 OUT PIN_OUT
LVCC VCC VCC1 PIN_VCC
LGND XGND1 XGND PIN_GND
.ENDS
$ TLINE MODEL - 2 SIGNAL CONDUCTORS WITH GND
$ PLANE
.MODEL USTRIP U LEVEL=3 ELEV=1 PLEV=1
+ TH1=1.3mil HT1=10mil TS=32mil KD1=4.5 DLEV=0 WD1=8mil
+ XW=-2mil KD2=4.5 NL=2 SP12=5mil
$ ANALYSIS / PRINTS
.TRAN .1NS 100NS
.PROBE V(STIM1) V(STIM2)
.PROBE V(TLOUT1) V(TLOUT2) V(TLOUT3) V(TLOUT4)
.END

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 343
W-2024.09
 Feedback

B
Simulating Circuits with Xilinx FPGAs

This appendix describes how to simulate circuits with Xilinx FPGAs.

The following topics are covered in this appendix:

• Introduction to Simulating Circuits with Xilinx FPGAs
• Syntax for IOB (xil_iob) and IOB4 (xil_iob4)
• Ground-Bounce Simulation
• Coupled Line Noise

Introduction to Simulating Circuits with Xilinx FPGAs

Synopsys and Xilinx maintain a library of PrimeSim HSPICE device models and transistor-
level subcircuits for the Xilinx 3000 and 4000 series Field Programmable Gate Arrays
(FPGAs). These subcircuits model the input and output buffer. See Signal Integrity
field programmable gate arraysFPGA’s

Examples in the PrimeSim HSPICE User Guide: Demonstration Netlists.

The following simulations described in this appendix use the Xilinx input/output buffer
(xil_iob.inc) to simulate:
• Ground bounce, as a function of package, temperature, part speed, and technology
• Coupled noise, both on-chip and chip-to-chip
• Full transmission line effects at the package level and the printed circuit board level
• Peak current and instantaneous power consumption for power supply bus
considerations and chip capacitor placement

Syntax for IOB (xil_iob) and IOB4 (xil_iob4)

Example of call for 1.2u PART:
X1 I O PAD TS FAST PPUB TTL VDD GND XIL_IOB
+ XIL_SIG=0 XIL_DTEMP=0 XIL_SHRINK=0

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 344
W-2024.09
 Feedback
Appendix B: Simulating Circuits with Xilinx FPGAs
Ground-Bounce Simulation

Example of call for 1.08u part:

X1 I O PAD TS FAST PPUB TTL VDD GND XIL_IOB
+ XIL_SIG=0 XIL_DTEMP=0 XIL_SHRINK=1

Nodes Description

I (IOB only) output of the TTL/CMOS receiver

O (IOB only) input pad driver stage

I1 (IOB4 only) input data 1

I2 (IOB4 only) input data 2

DRIV_IN (IOB4 only)

PAD bonding pad connection

TS three-state control input (5 V disables)

FAST slew rate control (5 V fast)

PPUB (IOB only) pad pull-up enable (0 V enables)

PUP (IOB4 only) pad pull-up enable (0 V enables)

PDOWN (IOB4 only) pad pull-up enable (5 V enables)

TTL (IOB only) CMOS/TTL input threshold (5 V selects TTL)

VDD 5-volt supply

GND ground

XIL_SIG model distribution: (default 0)

-3==> slow 0==> typical +3==> fast

XIL_DTEMP Buffer temperature difference from ambient. The default = 0 degrees

if ambient is 25 degrees, and if the buffer is 10 degrees hotter than
XIL_DTEMP=10.

XIL_SHRINK Old or new part; (default is new):

0==>old1==>new

All grounds and supplies are common to the external nodes for the ground and VDD. You
can redefine grounds to add package models.

Ground-Bounce Simulation
Ground-bounce simulation duplicates the Xilinx internal measurements methods. It
simultaneously toggles 8 to 32 outputs. The simulation loads each output with a 56 pF

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 345
W-2024.09
 Feedback
Appendix B: Simulating Circuits with Xilinx FPGAs
Ground-Bounce Simulation

capacitance. Simulation also uses an 84-pin package mode and an output buffer held at
chip ground to measure the internal ground bounce.
groundbounce
Figure 104 Ground bounce simulation

< <

84plcc
pkg

PrimeSim HSPICE adjusts the simulation model for the oscilloscope recordings so you
can use it for the two-bond wire ground. For example, the following netlist simulates
ground bounce:
qabounce.sp test of xilinx i/o buffers
.OPTION SEARCH='$installdir/parts/xilinx'
.op
.option post list
.tran 1ns 50ns sweep gates 8 32 4
.measure bounce max v(out1x)
*.tran .1ns 7ns
.param gates=8
.print v(out1x) v(out8x) i(vdd) power
$.param xil_dtemp=-65 $ -40 degrees c
$ (65 degrees from +25 degrees)
vdd vdd gnd 5.25
vgnd return gnd 0
upower1 vdd return iob1vdd iob1gnd pcb_power
+ L=600mil
* local power supply capacitors
xc1a iob1vdd iob1gnd cap_mod cval=.1u
xc1b iob1vdd iob1gnd cap_mod cval=.1u
xc1c iob1vdd iob1gnd cap_mod cval=1u
xgnd_b iob1vdd iob1gnd out8x out1x xil_gnd_test
xcout8x out8x iob1gnd cap_mod m=gates
xcout1x out1x iob1gnd cap_mod m=1
.model pcb_power u LEVEL=3 elev=1 plev=1 nl=1 llev=1
+ th=1.3mil ht=10mil kd=4.5 dlev=1 wd=500mil xw=-2mil
.macro cap_mod node1 node2 cval=56p
Lr1 node1 node1x L=2nh R=0.05
cap node1x node2x c=cval
Lr2 node2x node2 L=2nh R=0.05

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 346
W-2024.09
 Feedback
Appendix B: Simulating Circuits with Xilinx FPGAs
Ground-Bounce Simulation

.eom
.macro xil_gnd_test vdd gnd outx outref
+ gates=8
* example of 8 iobuffers simultaneously switching
* through approx. 4nh lead inductance
* 1 iob is active low for ground bounce measurements
vout drive chipgnd pwl 0ns 5v, 10ns 5v, 10.5ns 0v,
$+ 20ns 0v, 20.5ns 5v, 40ns 5v R
x8 I8 drive PAD8x TS FAST PPUB TTL chipvdd chipgnd
+ xil_iob xil_sig=0 xil_dtemp=0 xil_shrink=1 M=gates
x1 I1 gnd PAD1x TS FAST PPUB TTL chipvdd chipgnd
+ xil_iob xil_sig=0 xil_dtemp=0 xil_shrink=1 m=1
*Control Settings
rts ts chipgnd 1
rfast fast chipvdd 1
rppub ppub chipgnd 1
rttl ttl chipvdd 1
* pad model plcc84 rough estimates
lvdd vdd chipvdd L=3.0nh r=.02
lgnd gnd chipgnd L=3.0nh r=.02
lout8x outx pad8x L='5n/gates' r='0.05/gates'
lout1x outref pad1x L=5nh r=0.05
c_vdd_gnd chipvdd chipgnd 100n
.eom
.end

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 347
W-2024.09
 Feedback
Appendix B: Simulating Circuits with Xilinx FPGAs
Coupled Line Noise

Figure 105 Results of ground bounce simulation

Coupled Line Noise

This example uses coupled noise to separate IOB parts. The output of one part drives the
input of the other part through 0.6 inches of PCB. This example also monitors an adjacent
quiet line.

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 348
W-2024.09
 Feedback
Appendix B: Simulating Circuits with Xilinx FPGAs
Coupled Line Noise

Figure 106 Coupled noise simulation

m
V

V
V

Here is an example netlist for coupled noise simulation:

Input File, for qa8.sp test of xilinx 0.8u i/o buffers
.OPTION SEARCH='$installdir/parts/xilinx'
.op
.option nomod post=2
*.tran .1ns 5ns sweep xil_sig -3 3 3
.tran .1ns 15ns
.print v(out1x) v(out3x) i(vdd) v(irec)
vdd vdd gnd 5
vgnd return gnd 0
upower1 vdd return iob1vdd iob1gnd pcb_power L=600mil
upower2 vdd return iob2vdd iob2gnd pcb_power L=600mil
x4io iob1vdd iob1gnd out3x out1x outrec irec xil_iob4
cout3x out3x iob1gnd 9pf
u1x out1x outrec iob1gnd i_o_in i_o_out iob2gnd pcb_top
+ L=2000mil
xrec iob2vdd iob2gnd i_o_in i_o_out xil_rec
.ic i_o_out 0v
.model pcb_top u LEVEL=3 elev=1 plev=1 nl=2 llev=1
+ th=1.3mil ht=10mil sp=5mil kd=4.5 dlev=1 wd=8mil xw=-2mil
.model pcb_power u LEVEL=3 elev=1 plev=1 nl=1 llev=1
+ th=1.3mil ht=10mil kd=4.5 dlev=1 wd=500mil xw=-2mil
.macro xil_rec vdd gnd tri1 tri2
* example of 2 iobuffers in tristate
xtri1 Irec O pad_tri1 TSrec FAST PPUB TTL
+ chipvdd chipgnd xil_iob xil_sig=0 xil_dtemp=0 xil_shrink=1
+ m=1

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 349
W-2024.09
 Feedback
Appendix B: Simulating Circuits with Xilinx FPGAs
Coupled Line Noise

xtri2 Irec O pad_tri2 TSrec FAST PPUB TTL

+ chipvdd chipgnd xil_iob xil_sig=0 xil_dtemp=0
+ xil_shrink=1 m=1
*Control Setting
rin_output O chipgnd 1
rtsrec tsrec chipvdd 1
rfast fast chipvdd 1
rppub ppub chipgnd 1
rttl ttl chipvdd 1
* pad model plcc84 rough estimates
lvdd vdd chipvdd L=1nh r=.01
lgnd gnd chipgnd L=1nh r=.01
ltri1 tri1 pad_tri1 L=3nh r=0.01
ltri2 tri2 pad_tri2 L=3nh r=.01
c_vdd_gnd chipvdd chipgnd 100n
.eom
.macro xil_iob4 vdd gnd out3x out1x outrec Irec
* example of 4 iobuffers simultaneously switching
* through approx. 3nh lead inductance
* 1 iob is a receiver (tristate)
vout O chipgnd pwl 0ns 0v, 1ns 0v, 1.25ns 4v, 7ns 4v,
+ 7.25ns 0v, 12ns 0v R
x3 I3 O PAD3x TS FAST PPUB TTL chipvdd chipgnd xil_iob
+ xil_sig=0 xil_dtemp=0 xil_shrink=1 m=3
x1 I1 O PAD1x TS FAST PPUB TTL chipvdd chipgnd xil_iob
+ xil_sig=0 xil_dtemp=0 xil_shrink=1 m=1
xrec Irec O PADrec TSrec FAST PPUB TTL chipvdd chipgnd xil_iob
+ xil_sig=0 xil_dtemp=0 xil_shrink=1 m=1
* control settings
rts ts chipgnd 1
rtsrec tsrec chipvdd 1
rfast fast chipvdd 1
rppub ppub chipgnd 1
rttl ttl chipvdd 1
* pad model plcc84 rough estimates
lvdd vdd chipvdd L=1nh r=.01
lgnd gnd chipgnd L=1nh r=.01
lout3x out3x pad3x L=1nh r=.0033
lout1x out1x pad1x L=4nh r=0.01
loutrec outrec padrec L=4nh r=.01
c_vdd_gnd chipvdd chipgnd 100n
.eom
.end

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 350
W-2024.09
 Feedback
Appendix B: Simulating Circuits with Xilinx FPGAs
Coupled Line Noise

Figure 107 Results of coupled noise simulation

PrimeSim™ HSPICE® User Guide: Signal Integrity Modeling and Analysis 351
W-2024.09