CIC

1. SpectreRF Overview
SpectreRF is an optional feature added to Spectre ,and is represented by 6 analyses:
1. 2. 3. 4. PSS: Periodic Steady State Analysis PAC: Periodic AC Analysis PXF: Periodic Transfer Function Analysis PNOISE: Periodic Noise Analysis Tdnoise: Time Domain Noise QPNOISE: Quasi-Periodic Noise (not discuss here) 5. PDISTO: Periodic Distortion Analysis QPSS: Quasi-Periodic Steady State (not discuss here) 6. Envelope Analysis (not discuss here)

PAC, PXF, and PNOISE are similar in concept to AC, XF, and Noise. However, they are applied to periodically-driven circuits such as mixers and oscillators.
1-1

CIC

SpectreRF in a Design Flow

Design Schematic Models
The netlists include all components along with an analysis selection, simulation controls and statements to save, plot nodes or currents.

Analog Artist Environment Netlist

SPECTRE Engine

Spectre RF Control

Analog Artist Plot Results

Use Direct plot or the Calculator plot capabilities.

1-2

CIC

SpectreRF Tool Flow

SpectreRF
No

Stimuli is coperiodic
Yes

PDISTO Setup Spectre Engine PDISTO Analysis PDISTO Results

PSS setup Spectre Engine PSS Analysis PSS Results Spectre Engine -PAC Analysis -PXF Analysis -PNOISE Analysis Report Results
1-3

PSS is a large-signal analysis and determines the period of the small-signal analyses. PSS requires that multiple periodic stimuli be coperiodic.

PDISTO is also a largesignal analysis, and need not to be run after a PSS analysis. PDISTO does not require multiple periodic stimuli to be coperiodic.

CIC

SpectreRF Features
Compute a steady-state solution efficiently and directly Handles very large circuits (~ 10,000 transistors) Displays results in both time and frequency domains Use Discrete Fourier Transform (DFT) for better accuracy Displays standard RF measurements, such as s-parameter in Smith chart, NF, IP3, and 1dB compression point in the Analog Artist design environment. Performs oscillator analysis.

1-4

CIC

2. S-Parameter Analysis
Linear Simulation:
Entirely in the frequency domain A basic RF feature of the Spectre simulator

Ports:
Specify the port number on the psin ( or port); psin (or port) can act as a source port or a load. Required properties for linear analysis: Resistance & Port number

Noise Analysis:
Use Nfmin and NF for 2-port circuits ONLY.

2-1

CIC

Plotting S-Parameter Simulation Results

s-, z-, y-, and h-parameters group delay Voltage Standing Wave Ratio minimum noise figure reflection coefficient associated with Nfmin(also known as min, opt, or on) noise sensitivity parameter normalized equiv. Noise resistance noise figure stability terms transducer gain available gain, assuming conjugate matched output power gain, assuming conjugate matched input maximum available power gain maximum stable power gain maximum unilateral power gain impedance at port m noise circles available gain circles power gain circles load stability circles source stability circles
2-2

SP, ZP, YP, HP GD VSWR NFmin Gmin Rn rn NF Kf & B1f GT GA GP Gmax Gmsg Gumx ZM NC GAC GPC LSB SSB

CIC

Lab1 : S-parameter Analysis

Create a new library and a new schematic view. Use library analogLib & tsmc25rf to draw the scheme. After drawing, push Design Check and Save; then push Tools Analog Environment, and the window Affirma Analog Circuit Design Environment will appear.
create instance from library tsmc25rf

2-3

CIC

Setup Design Environment(1)

Push Setup Model Libraries then the window Model Library Setup appears. Setup the model library as shown right. Then click OK. Push Setup Simulator/Directory/Host to designate the project directory. The default project directory is ~/simulation .

Use Browse to access to the model files

2-4

CIC

Setup Design Environment(2)

You can use either an absolute model path or a relative model path IF you use the absolute approach, the setup is as shown right-upper. To use a relative path ,push Setup Simulation Files,than Setup Model Libraries .The setup is as shown right.

2-5

CIC

Setup Design Environment(3)

Push Analyses Choose then the window Choosing Analyses appears. Key in the values as right and push ok, then some information will appear in the Analyses domain of the window Affirma Analog Circuit Design Environment. Push Simulation Netlist and Run to run the simulation. The Netlist will be saved under a directory called ~/simulation.
Push Select button then to select the port on the schematic window

Netlist and Run

2-6

CIC

See the Results

Use the Direct Plot tool to look the results. In the S-parameter Results window choose some parameters to see their results.

2-7

CIC

Some Results

2-8

CIC

Save the results to *.s2p

Edit the S-Parameter Options, and enter the path to the output S-parameter file in the file field of the OUTPUT PARAMETERS section and OK the S-Parameter Options form. And Simulate again. Check if the file is created in the appointed directory.

2-9

CIC

S2P File

2-10

CIC

Simulation State
Push Session Save State to save simulation states under a directory called ~/.artist_states. Designate a new directory with the Session Options command in the simulation window. Push Session Save State to load saved states for a design.

2-11

CIC

Read the S2P file(1)

Create a new schematic view. Use library analogLib (n2port cell) to draw the scheme. Simulate if the results are the same as before.

2-12

CIC

3. Lab2: Swept DC Analysis

Create a new schematic view and use library analogLib & tsmc25rf to draw the scheme. . After Check and Save ; then call the window Affirma Analog Circuit Design Environment .

Setup up the Model Libraries. Push Variables Copy From Cellview, and the defined variables appear in the Design Variables section. Double click on the variable name or push Variables Edit, the window Editing Design Variables appears. Key in the appropriate value for the variables.
3-1

CIC

Set up the Design Environment(1)

Call the window Choosing Analyses and key in the values as right and push ok. To plot power or current at the end of the simulation, you must explicitly save the currents necessary for the calculation before the simulation. The voltages at each node are saved by default. Select Outputs To Be Saved Select On Schematic. In the schematic, select the NMOS. The terminals are circled in the schematic window after you select them. Press Esc to end the selections.

3-2

CIC

Set up the Design Environment(2)

In the window Design Environment select Tools Parametric Analysis ; the window Parametric Analysis appears, then key in the values as below . In the window Parametric Analysis select Analysis Start to start the simulation.

3-3

CIC

The Results

Select Results Direct Plot DC and select the terminal Drain of the nmos in the schematic window; then push ESC, and the results will be showed.

3-4

CIC

4. Periodic Steady State Analysis

Directly computes the periodic steady-state response of a circuit in the time domain. Iterative Shooting Newton method is employed. Calculate frequency translations using the saved matrices at every time point. The fundamental frequency of the circuit or system is determined, based on integer multiples of all source frequencies. The circuit is evaluated for one period of the common frequency, and the period is adjusted until all node voltages and all branch currents fall within a specified tolerance.

4-1

CIC

Shooting Newton Method

PSS operates by efficiently finding an initial condition that results in steady state.
v

vi vf

The signal starts at a The starting point is adjusted The first iteration is transient point vi doesn't by the shooting method to simulation from t=0 to t=1/PSSfund result in periodicity. result in periodic steady state.

by default. The tstab parameter can be adjusted to facilitate convergence. Transient Analysis The second iteration is PSS analysis between t=tstab to t=(tsatb+1/PSSfund) and compares all voltage and currents at the start and end of the shooting interval. Set the value of tstab to keep start-up behavior away.
PSS Analysis
tstab
Shooting Interval

All node voltages and Admittance Matrices are saved

t=0s t=1/PSSfundt=2/PSSfund

4-2

CIC

Shooting Newton Method(continued)

1 PSS fund

1 PSS fund

Transient Analysis

PSS Analysis
RF 2.4GHz IF 100MHz

Shooting method takes the last few point data at the end of the shooting interval to adjust the slopes of the waveform at the beginning of the next iteration. If 20 iterations do not yield a solution, this might indicate the circuit won converge to a PSS solution. t
4-3

LO 2.3GHz PSSfund=100MHz

CIC

PSS Analysis Assumptions

1st Assumption : Periodicity
All stimuli are periodic and coperiodic with the PSSfund ; All responses are periodic. PSSfund can be set to includes the subharmonics. If periodicity assumptions fail, PSS analysis will not converge.

2nd Assumption : Linearity

A near-linear relationship need to exist between initial and final points of the shooting interval.

4-4

CIC

The PSS Fundamental

150 MHz RF Input 900 MHz Mixer BPF Mixer 10 MHz BPF Output

IF1

IF2

1050 MHz Local Osc.

160 MHz Local Osc.

PSSfund = 10 MHz

4-5

CIC

PSS Operation
Start PSS Initial Transient (1 period or tstab) 1 Period of PSS Analysis

No

Periodicity Meet?
Yes

Final State = Initial State Exit

4-6

Refine Initial Guess

No

Yes

CIC

Simulator Accuracy Suggestions

Do not set conservative. This will dramatically extend the simulation time. V and I < reltol * lteratio * steadyratio The suggested settings are recommended for IP3 Analysis, Noise Analysis, or wherever high accuracy is needed. Choose the gear2only integration method. The default trap integration method yields Parameter Defaults Suggested Settings underdamping and reltol 1e-3 1e-5 gearOne yields too vabstol 1e-6 3e-8 much overdamping.
iabstol Method errpreset liberal moderate conservative reltol x10.0 x1.0 x0.1 relref allglobal sigglobal alllocal method gear2 traponly gear2only 4-7 1e-12 trap maxstep <0.4/maxacfreq <0.2/maxacfreq <0.1/maxacfreq 1e-13 gear2only steady-ratio 0.1 0.001 0.00001 lteratio 3.5 3.5 10.0

CIC

Normalized Convergence ratio

When the Conv norm is 1(unity) or less, the simulation meets the matching criterion. The PSS messages also display the number of PSS iterations, the number of accepted timesteps, and the total time required for PSS analysis.

Measured V between start and end of shooting interval Conv norm = reltol*lteratio*steadyratio

4-8

CIC

Lab3 : PSS and swept PSS Analysis

Create a new schematic view and use library analogLib & tsmc25rf to draw the scheme.
Port1: Frequency name: F1 Resistance: 50 Source type: sine Amplitude(dBm): -40 Frequency: frf Port2: Frequency name: F2 Resistance: 50 Source type: sine Amplitude(dBm): 8 Frequency: flo
4-9

Pif

CIC

Setup up the PSS Simulation(1)

Model library setup. Call the window Affirma Analog Circuit Design Environment; key in appropriate value for the variables in the Design Variables section.
Analyses Choose. In the window Choosing Analyses, select pss.

4-10

CIC

Setup up the PSS Simulation(2)

The Signal field is ONLY applicable to the pdisto analysis. Beat Frequency represents the PSS Fundamental (PSSfund) frequency. This fundamental is the highest frequency that evenly divides into all frequencies in the circuit. You may key in an appropriate value or push Auto Calculate button to get an autoresponded value. Set the value for number of harmonics. The number of harmonics won affect the simulation t accuracy or time. Make sure the Enabled field is on. Click the Options button and set the integration method to gear2only.

4-11

CIC

Setup up the PSS Simulation(3)

In the Analog Artist Simulation window, select Simulation Options Analog. Set the Tolerance Options as recommended. If it is hard to converge set the Tolerance Options looser. Finally, Select Simulation Netlst and Run to start the simulation. Note if the Conv norm is less than 1 or if the PSS simulation has a convergent result.

4-12

CIC

Display the Conversion Power Gainmethod 1

In the Analog Artist Simulation Window, select Results Direct Plot PSS. Note the prompts on the bottom of the schematic and PSS Results windows. The PSS Results window MUST be on the screen when probing the nodes in the schematic. Don push OK. t In the PSS Results form, use the cursor to select the Pif net and Prf nets on the schematic. Press Esc to end this command. Click the Switch Axis Mode icon on the Waveform Window or select Axes To Strip.

4-13

CIC

Display the Conversion Power Gainmethod 1(continued)

Click the Crosshair Marker A icon and place the marker on the 2.4GHz harmonic of Prf. Click the Crosshair Marker B icon and place the marker on the 100MHz harmonic of Pif. Prf: Magnitude: 4.0085m Power: -38 dBm Pif: Magnitude: 4.08038m Power: -37.8 dBm Conversion Power Gain 0.2dB + 3 dB = 3.2 dB

4-14

CIC

Display the Conversion Power Gain-method 2

Select Output Save All and the window Save Options appears. Set the buttons as below window in order to get the AC power! Select Outputs To Be Saved Select On Schematic. In the schematic, select the PORT1 and RL1. The terminals are circled in the schematic window after you select them. Press Esc to end the selections. Double click on the name in the Outputs section or select Outputs Setup. Set the outputs Will Be Plotted and Saved.

4-15

CIC

Display the Conversion Power Gainmethod 2(Continued)

Push Netlist and Run icon to run this simulation. Select Results Direct Plot PSS. Set the function and modifier as right; Select instance terminal(PORT1 & RL1) in the schematic window. Press Esc to end the selections. Compare the results to those of method 1.

4-16

CIC

1 dB Compression Point Simulation

Change the Amplitude(dBm) of PORT1 to a variable prf; Designate a value to prf in the Design Variables section. In the Choosing Analyses window, turn on the Sweep button as shown here. Type in prf for the Design Variable Name, or click the Select Design Variable button, and highlight prf from a list , then click OK. Remember to check in the INTEGRATION METHOD PARAMETERS the method is gear2only. Select Netlist and Run button.

4-17

CIC

P1 dB Simulation Results

Use Direct Plot function to see the results. Set up PSS Results form as shown here. Then select the Pif net in the schematic. With the cursor still in the schematic window, press ESC key to end the Direct Plot command.

4-18

CIC

Simulating IP3
PSS by itself is seldom used for IP3 simulation, because the separation between the 2-tone frequency is typically only a few Khz, and leads to a very long simulation time.

Edit PORT1 properties as right. So The Fundamental (Beat) Frequency is now 25MHz. Set up Choosing Analysis form appears as shown below and push OK Run the simulation

4-19

CIC

IP3 Results
3rd order intermodulation product will occur at (2 2.4GHz 2.425GHz) 2.3GHz = 75 MHz

Use Direct Plot function to see the results. Set up PSS Results form as shown here. Then select the Pif net in the schematic. Press ESC key to end the Direct Plot command.

4-20

CIC

5. PAC Analysis
PAC is a small-signal analysis like AC analysis, except the circuit is first linearized around a periodically varying operating point as opposed to a simple DC operating point. Linearizing around a periodically time-varying operating point allows analyzing transfer-functions that include frequency translation. When a small sinusoid is applied to a linear circuit that is periodically time-varying, the circuit responds with harmonics. PAC computes a series of transfer functions, one for each frequency. These transfer functions are unique because the input and output frequencies are offset by the harmonics of the LO.
5-1

CIC

PAC Analysis Overview

Input Output LO f out = f in + Ki PSS fund where fin represents the input frequency, and Ki are the PAC sidebands

PAC computes the transfer function from one input to many outputs. PAC is similar in concept to normal small-signal AC analysis, but it also calculates frequency conversion effects. By setting the maxsideband value to Kmax, PAC generates all 2Kmax +1 sidebands from Kmax to +Kmax. The small-signal frequency in a PAC analysis can be arbitrarily close or even equal to the LO frequency.

LO

State
PSSfund Harmonic no. 0 1 2 3 4

f f f
-2 -1 0 1 2 3

Input Output

PAC Sideband no. -3

5-2

CIC

Fundamental PAC Assumptions

The PAC small signal analysis assumes that the circuit responds in a small signal fashion to the sinusoidal stimulus. This is accomplished by keeping the magnitude of the PAC signal at least 10 dB below the 1 dB GCP. The harmonics of the small signal PAC tone are not computed, although small signals can be used to measure distortion caused by the large signals present in the PSS analysis. For the transfer function to be accurate, a large number of time steps, during the PSS analysis, are needed at the small signal frequency. If the analysis frequency of the small frequency analysis is too high, the accuracy degrades. The maxaxfreq parameter of the PSS analysis can be used to specify the highest frequency that SpecteRF uses in subsequent small signal analyses.
5-3

CIC

PAC Analysis Summary

The number of harmonics should be no less than the PAC harmonics. * Set type to dc and specify PAC magnitude Sweep, array or single point Plot results relative to output or absolute value of output frequency. Input is of little value and is not used.

Specify the following information when running a PAC analysis:

PSS fundamental Input port Input sweep frequency

Output frequencies of interest Sidebands or Array of Indices Results format

* When setting Output harmonics less than the PAC harmonics, be sure to set the maxacfreq parameter to assure that the simulator takes sufficient time points to accurately characterize the output waveform in the PSS analysis.
5-4

CIC

Lab4 : PAC Analysis

Use the same schematic as Lab3. Modify the parameter values of PORT1 as below table.
Note : When the source type is set to dc, this signal will not be checked for coperiodicity with the other signals; this source will be treated as a small signal. When the source is set to sine, it will be considered large signal.

Parameter Resistance Source type Frequency PAC magnitude Amplitude (dBm) Amplitude2 (dBm) Frequency2

Value 50 dc frf 1 prf (blank) (blank)

5-5

CIC

Setting Up the PAC Simulation

Call the window Choosing Analyses; In the pss form, fill in the form as left; then click Apply.
Note the number of harmonics is set to 0, because the PSS simulation is only run to calculate the large-signal, steady state solution.

Click on pac in the Choosing Analyses form, and setup the form as left; then click OK .
The Frequency Sweep Range sets the sweep range on the psin(port) component at the input port which has a PAC magnitude parameter value specified. The value for Maximum sideband is relative to the Fundamental frequency. Since the LO frequency and PSSfund are equal, you get the results of mixing the RF with the 0 through 3rd harmonic of the LO.

Select Netlist and Run.

5-6

CIC

Plotting the Conversion Gain

Note how much faster this simulation runs than the previous method used to calculate CG. Use Direct Plot function to see the results. In the schematic window, select the Pif node, and the result are plotted as next page:

5-7

CIC

Periodic Steady State Response

0

To measure the CG, move the marker to the 100MHz position in the waveform window and read the gain. -1
Note if the input and output port are both matched to 50ohm, we get conversion power gain; otherwise we get conversion voltage gain.

-2

1 -3 2 3

5-8

CIC

Running a Swept Small-Signal IP3 Simulation

Parameter Resistance Source type Frequency PAC magnitude PAC magnitude (dBm) Amplitude (dBm) Amplitude2 (dBm) Frequency2 Value 50 sine frf (blank) prf prf (blank) (blank)

Modify the parameter values of PORT1 as right table; then check and save! Select pss in the Choosing Analyses form, and setup the form as below : Note now the Fundamental Frequency is 100 MHz Set the Number of harmonics to 50 and you have the harmonics available to view; it won t affect the simulation time. Click Apply! The Choosing Analyses form is still active on the screen.

5-9

CIC

Setup up the PAC Simulation

In the Choosing Analyses form, select pac; then set up the form as right: This simulation applies a 2.425GHz tone in the PAC analysis to compare the results by the swept PSS. This PAC test tone is typically separated according to channel spacing. Click OK. Select Netlist and Run.
5-10

CIC

IP3 Results

LO: 2.3 G RF: 2.4 G & 2.425G 1st order harmonics: 100M & 125M 3 rd order harmonics: 75M & 150M

The only 1st and 3rd order pair available from this analysis (due to the 100MHz PSSfund) is 125M and 75M. Use Direct Plot function; select the Pif net in the schematic window. Compare the IP3 values using 2 different method!

5-11

CIC

6. PXF Analysis
The periodic transfer function (PXF) analysis directly computes such useful quantities as conversion efficiency(the transfer function from input to output at a preferred frequency), image and sideband rejection, and power supply rejection. The primary use of PXF analysis is to measure various conversion gains. This is very valuable when looking at different spurs on the input of a receiver. PXF can be a better choice for calculating CG than PAC, because PXF will provide information on all of the frequencies on the RF port that are converted to the IF band. When simulating oscillators, PXF can determine the tstab

value.
6-1

CIC

PXF Analysis Overview

Input Output
k LO f in = f out + k PSS fund where fout represents the output signal frequency; k is the PXF sidebands number

The PXF analysis computes the energy contributions from all source harmonic frequencies to a signal or swept output frequency. In this way, a single output response is the combination of all possible frequency components in the design. Set the maxsideband, or the sidebands parameters, to select the periodic small-signal input frequencies of interest, while sweeping the selected output frequency.
6-2

LO

State
PSSfund Harmonic no. 0 1 2 3 4

f f f
-1 0 1 2 3 4

Output Input

PXF Sideband no. -2

CIC

Fundamental PXF Assumptions

The PXF small signal analysis assumes that the circuit responds in a small signal fashion to sinusoidal stimulus. SpectreRF is not capable of computing the distortion caused by the small signals, although small signals can be used to measure distortion caused by the large signals present in the PSS analysis. To increase accuracy, choose a large number of time steps during PSS analysis. If the analysis frequency of the small signal analysis is too high, the accuracy of the results degrade. The maxacfreq parameter of the PSS analysis specifies the highest frequency uses in subsequent small signal analyses.
6-3

CIC

PXF Analysis Summary

The number of harmonics should be no less than the PXF harmonics. * Specify in form (To measure current, put a 0v battery in series with the branch. ) Sweep, array or single point Plot results relative to input or absolute input value of input frequency. Output is of little value and is usually not used.

Specify the information in this table when running a PXF analysis.

PSS fundamental Output net (v) or Voltage source (i) Output sweep frequency

Input frequencies of interest Sidebands Results format

* When setting Output harmonics to 0, be sure to set the maxacfreq parameter to assure that the simulator takes sufficient time points to accurately characterize the output waveform in the PSS analysis.
6-4

CIC

Lab5 : PXF Analysis

Because PXF is a small signal analysis, only one large signal tone, the LO, is required. Set the PORT1 as follows:
Parameter Resistance Source type Frequency PAC magnitude (dBm) Amplitude (dBm) Amplitude2 (dBm) Frequency2 Value 50 dc frf (blank) prf (blank) (blank)

Pif

Pif-

6-5

CIC

Setting Up the PXF Simulation(1)

In the Simulation window, select Analyses Choose; turn off the pac analysis. Then select the pss analysis, and set up the form as right: Note the number of harmonics is set to 0, because the PSS simulation is only run to calculate the large-signal, steady state solution. Therefore set a value for maxacfreq in the PSS Options form. Set maxacfreq to 4 GHz. Click Apply in the Choosing Analyses form.

6-6

CIC

Setting Up the PXF Simulation(2)

Click on pxf in the Choosing Analyses form, and setup the form as left; then click OK . The Frequency Sweep Range is specified from 1MHz to 300 MHz. The PXF analysis will calculate all inputs that produce this range of frequencies at the Pif port. To set the Positive Output Node, click the Select button, and select the Pif node in the schematic. Click the Netlist and Run.

6-7

CIC

Plotting the RF to IF Conversion Gain

Use Direct Plot function to see the results. In the PSS Results form, select pxf button. Follow the prompts at the bottom of the form, and select the port component (PORT1) in the schematic

0 -1 1

-2 2 -3 3

6-8

CIC

Power Supply Rejection

Double click on the pxf analysis in the window Design Environment, and the Choosing Analyses form appears. Change the Negative Output Node to Pif(/net016) in the pxf form, then click ok. Run the simulation.

6-9

CIC

Plotting the Power Supply Rejection

Use Direct Plot function to see the results. In the PSS Results form, select pxf button. Follow the prompts at the bottom of the form, and select the DC supply (vdc=2.5v) in the schematic

-1 1 -3 -2 2 0 3

6-10

CIC

7. PNOISE Analysis
PNOISE analysis, unlike conventional noise analysis, computes frequency convention effects, noise folding, aliasing. For noise sources that are bias dependent, such as shot noise sources, the time-varying operating point acts to modulate the noise sources. The transfer function from the noise source to the output is also periodically time-varying, and so acts to modulate the contribution of the noise source to the output. The effect of a periodically time-varying bias point on the noise generated by the various components in the circuit is also included. Include the effects of thermal noise, shot noise, and flicker noise.
7-1

CIC

PNOISE Analysis Overview(1)

The final result of the analysis is the sum of the noise contributions from both the upconverted and down-converted output frequency specified. By setting the maxsideband value to Kmax, all 2Kmax+1 sidebands from Kmax to +Kmax are generated. The number of requested sidebands has a small effect on the simulation time.
7-2

Input

Output

f noise_source = f out + K i PSS fund LO where fout represents the output signal frequency; Ki is the PNOISE sidebands no.

LO

State
PSSfund Harmonic no. 0 1 2 3 4

Output Input

PNOISE Sideband no. -2

-1

CIC

PNOISE Analysis Overview(2)

When the reference sideband has any value other 0, Single Sideband (SSB) NF is calculated. To determine the reference sideband, run a PXF analysis. The Noise Summary Table displays the following data:
Noise contribution (value and %) for each component in the circuit Total output noise Total input referred noise
7-3

CIC

Fundamental PNOISE Assumptions

The small signal analyses compute transfer function by using time-domain techniques. The time steps used in these timedomain computations are the same as those in PSS analysis. For accuracy, the PSS analysis needs to have many data points at the highest frequency that you want to analyze in the noise analysis. More sidebands yield greater accuracy, but they take longer to simulate and use more disk space. If the analysis frequency of the small signal analysis is too high, the Spectre simulator warns. Use the maxacfreq parameter of the PSS analysis to specify the highest frequency for SpectreRF to use in subsequent small signal analyses.
7-4

CIC

PNOISE Analysis Summary

Specify the information in this table when running a PNOISE analysis.
PSS fundamental The number of harmonics will likely be no less than the PNOISE harmonics. Sweep, array or single point Sidebands Port, voltage or current sources Noise figure and Input referred noise

Output net (v) or Voltage source (i) Specify in form Output sweep frequency Input frequency contributors Input Sources Reference Sideband

7-5

CIC

Lab6 : Noise Figure

Modify the parameter values of PORT1 as follows: In the Simulation window, select Analyses Choose; turn off the pxf analysis.
Parameter Resistance Source type Frequency PAC magnitude (dBm) Amplitude (dBm) Amplitude2 (dBm) Frequency2 Value 50 dc frf (blank) (blank) (blank) (blank)

7-6

CIC

Setting Up the PNOISE Simulation(1)

Then select the pss analysis, and set up the form as right: Set a value for maxacfreq in the PSS Options form. Set maxacfreq to 20GHz. Remember to set the integration method to gear2only. Click Apply in the Choosing Analyses form.

7-7

CIC

Setting Up the PNOISE Simulation(2)

Click on pnoise in the Choosing analyses form, and set up the form as right: A Maximum sideband of 8 implies PNOISE will calculate the noise out to 8 harmonics of the PSSfund, or 18.4 GHz. To set the Positive/Negative Output Node, click the Select button, and select the Pif/Pif- node in the schematic window. Click the Select button and select PORT1 component in the schematic to set the Input Port Source. To obtain the Reference Side-Band, run PXF analysis. Finally, push OK; then Netlist and Run.
7-8

CIC

Plotting the NF
Use Direct Plot function to see the results. In the PSS Results form, select pnoise button. Click Plot button, and the waveform window displays the results.

7-9

CIC

Printing the Noise summary Report

It is valuable to know the main contributions of noise in a system. This information is readily available from a PNOISE simulation. In the Analog Artist Simulation window, select Results Print PSS Noise Summary. The Noise Summary form appears. Fill the form as shown here.

7-10

CIC

The Noise Summary Table

Click OK in the Noise Summary form, and the Noise Summary Table displays. Note what are the main contributions of noise.

7-11

CIC

Time Domain Noise

The noise in RF circuits is generated by sources that can typically be modeled as periodically timevarying. Noise that is periodically time-varying is also called cyclostationary noise. Might or might mot be independent (correlated). Becomes intricate with nonlinear elements, with memory, or driven by time-varying signals.
7_time_domain-1

CIC

Time Domain Noise Overview

There have been 3 new noise type parameters added to PNOISE analysis:
1. sources:
Compute time-averaged total noise power at a signal output, in the frequency domain.

2. timedomain:
Calculates the time-varying instantaneous noise power in a circuit with periodically driven components Setting the NOISE Skip Count=N parameter will only compute the noise at every Nth timepoint in the PSS waveform.

3. correlations:
Calculate correlations in noise at different ports of a multiport circuit
7_time_domain-2

CIC

Lab7: Calculating Time-Varing Instantaneous Noise Power

Create a new schematic view. Use library analogLib & tsmc25rf to draw the scheme. After drawing, Check and Save!
7_time_domain-3

CIC

Setting Up the PNOISE Simulation(1)

Open the Design Environment window and set up a PSS analysis as shown right: Click the Options button and set the method to gear2only. Click Apply.

7_time_domain-4

CIC

Setting Up the PNOISE Simulation(2)

On the PNOISE analysis form, select timedomain in the NOISE Type field. Set up a PNOISE analysis as shown right:
Note: If the Noise Skip Count is set to an integer p, then noise will be calculated at every p+1 points. When the Noise Skip Count is 0 (default), it calculates the noise at every timepoint in the final PSS solution.
7_time_domain-5

CIC

Plotting Time Domain Results

Click the Netlist and Run icon to start the simulation. Use Direct Plot function to view the time domain plot of v(out)

7_time_domain-6

CIC

Plotting Time Domain Noise Results

In the PSS Results form, click tdnoise and set up the form as shown right: Click Plot button.
At about 4.1ns when waveform transition happens, the inverter is the most noisy.

7_time_domain-7

CIC

Plotting Time Domain Noise Results on Spectrum(1)

To display the spectrum of the noise results, set up the PSS Results form as show right: Click Plot. See the result in the next page.

7_time_domain-8

CIC

Plotting Time Domain Noise Results on Spectrum(2)

7_time_domain-9

CIC

8. Periodic Distortion Analysis

PDISTO is an analysis that invokes a series of PSS like analyses over all input frequencies, their harmonics, and the intermodulations of the frequencies and harmonics. Similar to PAC, the PDISTO analysis calculates the responses of circuits that exhibits frequency translations. However, instead of simulating small signal behavior, PDISTO models the response from moderately large input signals. Use PDISTO to calculate intermodulation distortion from two or more large input signals. PDISTO treats one particular input signal as the large signal, and the others as moderate signals. PDISTO allows arbitrary signal signal inputs, including sums of sinusoids that might not be periodic, it as a quasi-periodic extension of PSS. PDISTO can be thought of as an extension of PAC that allows signal signal inputs, capable of producing third-order products, to be used.
8-1

CIC

PNOISE Analysis Overview

Internal to the simulator, one input is treated as the large signal, which causes the most nonlinearity or the largest response in the circuit. Other signals are treated as moderate and do not need to be harmonically related to the large signal or integer multiples of each other. The moderate signals can be large enough to create distortion (near P-1dB point) The ability to sweep PDISTO provides a way to perform intermodulation distortion calculations with multiple input signals, considered as large signals.
8-2

CIC

PDISTO v.s. PAC

PDISTO analysis yields more information than PSS followed by a PAC analysis, when modeling intermodulation distortion.
PSS/PAC Results
895 1805 5 900 905 900 905 1800 PSS PAC

PDISTO Results
RF Amplifier
10 5 8-3 900 905 1800 1810 895 895 1805

Input Spectrum

CIC

Comparing PDISTO and PAC(1)

100 M

PSIN Source Type = sine RF1=900 MHz

PSS
0 hz

IF1=100 MHz IF2=99.8 MHz LO=1 GHz

PAC
RF2=900.2 MHz amplitude=-30dBm

199.8 M 200.2 M

99.8 M 100.2 M

200 M

899.8 M

100 M

PSIN Source Type = sine RF1=900 MHz moderate harms=2 or 3 RF1=900.2 MHz moderate harms=2 or 3 amplitude=-10dBm
8-4

100.4 M

99.6 M

199.6 M 200 M

200 K 600 K

99.4 M 99.8 M 100.2 M

899.6 M 900 M 900.4 M

199.4 M 199.8 M 200.2 M

LO=1 GHz large harms=5

900.6 M

0 hz 400 K

IF1=100 MHz IF2=99.8 MHz

harms=2

899.8 M 900.2 M

add for harms

900.2 M

200 K

900 M

PSS PAC

CIC

Comparing PDISTO and PAC(2)

The number of harmonics of the large signal does not affect the simulation time, where the number of harmonics on the moderate signals greatly affects simulation time. Always specify at least 1 harmonic on each signal in a PDISTO analysis. PDISTO analysis does not take as long as a PSS analysis with a small PSS Fundamental, but it is longer than a PSS/PAC analysis.

8-5

CIC

PDISTO Assumptions
Unlike PSS, PDISTO does not required multiple inputs be commensurate or coperiodic. However, they still must be periodic. For coperiodic, well separated signals, use PSS. For signals that are closely spaced or not coperiodic, use PDISTO. For circuits driven by 2 or more moderate signals or at unrelated frequencies, use PDISTO. If only one periodic signal is large enough to create distortion, choose PSS followed by PAC or PXF.

8-6

CIC

Lab8 : Simulation with PDISTO

Modify the parameter values of PORT1 as follows: In the Simulation window, select Analyses Choose; turn off the pss and pnoise analysis

Parameter Resistance Source type Frequency Amplitude (dBm) Amplitude2 (dBm) Frequency2

Value 50 sine frf prf prf frf +1M

8-7

CIC

Setting Up the PDISTO Simulation(1)

In the Choosing Analyses form, select pdisto for the analysis. Use the Clear/Add button to change the values in the Fundamental tones list box as shown right. Remember to select gear2only button in the Options form. Select Simulation-Options-Analog, and set the Tolerance Options as recommended. If the signals are truly large, relax reltol to 1e-4.
8-8

CIC

Setting Up the PDISTO Simulation(2)

Remember to select the output terminals to be saved and plotted before the simulation. Increase the power of the input RF signals from 40 dBm to 30 dBm.(P1dB for this circuit is 22 dBm) In the PSS/PAC analysis, you used a PAC tone that was at least 10 dB below the 1 dB compression point to prevent violating the small signal assumptions associated with the PAC analysis. This restriction does not apply to PDISTO. Select Netlist and Run button

8-9

CIC

Plotting Simulation Results

Use Direct Plot function to see the results. Follow the prompts at the bottom of the form, and select instance terminal (RL1) in the schematic

zoom in

8-10

CIC

Simulation IP3 with PDISTO(1)

The setup for this measurement is very similar to the one used for the swept PSS simulation, except you will be using PDISTO with two moderate tones and one large reference signal. Modify the parameter values of PORT1 as follows: Check and save.
Parameter Resistance Source type Frequency Amplitude (dBm) Amplitude2 (dBm) Frequency2 Value 50 sine frf prf prf frf +25M
8-11

CIC

Simulation IP3 with PDISTO(2)

In the PDISTO Analyses form, use the Clear/Add button to change the values in the Fundamental tones list box. Set up the Sweep Range as shown right. Remember to choose the gear2only method and set the Tolerance Options as recommended or relax reltol to appropriate value. Click OK. Run the simulation.
8-12

CIC

Displaying the IP3 Plot(1)

Use Direct Plot function to see the results. Set up the PDISTO Results form as shown right. Follow the prompts at the bottom of the form, and select instance terminal (RL1) in the schematic
LO: 2.3 G RF: 2.4 G & 2.425G

1st order harmonics: 100M & 125M 3rd order harmonics: 75M & 150M

8-13

CIC

Displaying the IP3 Plot(2)

-14.5 dBm
-16
8-14

CIC

9. Oscillator and Phase Noise Analysis

SpectreRF-PSS analysis can be performed on autonomous or nondriven circuits, such as oscillators. Oscillator analysis includes two phases:
The initial transient phase:
The PSS monitors the potential difference between the two nodes specified and the waveforms in the circuits, and this analysis develops a better estimate of the oscillation period of the circuit.

The shooting phase:

The circuit is simulated repeatedly while the length of the period and the initial conditions are adjusted to achieve a periodic steady state solution.
9-1

CIC

Troubleshooting Oscillators

Does not converge increase tstab Improve the estimate of the period. Be especially carefully that the period specified is not too short. Change the value of the method parameter from gear2only to trap or traponly. Does not converge increase maxperiods If the shooting iteration approaches convergence and fails, increase the value of the steadyratio parameter, but never set steadratio larger than 0.1. Change the value of the tolerance parameter.
9-2

CIC

Oscillator PSS Algorithm

Tonset 2PSSperoid tstab 4PSSperoid PSS

tstart

Ttran

Ttran_end

tstart - Start time for transient analysis.(default is 0) Tonset Time when the last stimulus waveform becomes periodic. PSSperiod the guess period entered by the user. tstab additional stabilization time entered by the user. maxstep = (Ttran / 50)(default). The algorithm then adds a further 4 periods of our guess frequency of transient analysis in order to measure the oscillator frequency.
9-3

CIC

Oscillator Algorithm and maxstep

Default maxstep > period if Ttran > 50 oscillator periods. The oscillator might not start correctly or a metastable state might be found by the simulator. Use tighter convergence criteria or set maxstep < 1/(200FreqOsc) In PSS shooting iterations stage, maxstep is the smallest of:
maxstep manual entry PSSperiod/(maxharm40) 1/(maxacfreq5) PSS Beat Frequency/200

Setting a high harmonic in the PSS analysis or setting maxacfreq will only effect the maxstep of the PSS shooting iterations but NOT the maxstep of the initial transient section.
9-4

CIC

Lab9 : Tunable Oscillator Transient Analysis

Create a new schematic view.

Use library analogLib & tsmc18rf to draw the scheme.

Vout1

Vout2

Use a vpulse source to kick-start the oscillator.

9-5

CIC

Set Up the Design Environment

In the Design Environment form select Setup Model Libraries to set up the model library as show below. Select Variables Copy From Cellview to set the variable vctrl to be some value.

9-6

CIC

Transient Simulation set up

Select Analyses Choose to set up the transient simulation as right window. Set up the form and option form as shown right: Push Netlist and Run button.

9-7

CIC

Display the Transient Results

In the Analog Artist Simulation window, select Results Direct Plot Transient Signal; then select Vout1 node in the schematic and press ESC key to end the selection. The Vout1 transient node voltage appears in the Waveform window.

9-8

CIC

Oscillator Notes
When applying initial conditions to start an oscillator, first run a transient analysis to get the voltages for a few nodes in the circuit. To set the initial conditions for the next run, select Simulation Convergence Aids Initial Condition. In the Transient Options form, set a value such as spectre.fc for the writefinal parameter in the STATE FILE PARAMETERS section. The spectre.fc file will have all of the final conditions on the nodes in the circuit. Before running another transient or PSS analysis, set readns to spectre.fc in the CONVERGENCE PARAMETERS section of the Options form.
9-9

CIC

spectre.fc file

9-10

CIC

Use the DFT Function

In the Waveform window, click the Add Subwindow icon, then a subwindow with a label of 2 in the upper right corner is added. Click the Calculator, then the calculator appears. Click the vt button in the Calaulator and follow the prompt at the bottom of the schematic window. Then select the Vout1 node in the schematic and press Esc; click and hold Special Functions and select dft form from the Special Function list.

Fill in the form as follows: And click OK.

9-11

CIC

The Frequency Domain Results

Finally, in the Calculator, click erplot Note the initial estimate of the oscillation frequency is developed.

2.33GHz

zoom in

9-12

CIC

PSS/PNOISE Analysis(1)
In the Choosing Analysis window, turn off the transient analysis; select the pss analysis and set up the form as right: An estimate of 2GHz was selected for Beat Frequency. It recommended to estimate a lower frequency than s expected to help in the convergence. The value of tstab is set to 100n to inform the simulator that the oscillation needs 100ns to stabilize to a steady-state waveform. Remember to choose the gear2only method in the options form. Click Apply.
9-13

CIC

PSS/PNOISE Analysis(2)
Next, click the pnoise button, and set up the PNOISE analysis as right: The phase noise from 1 Hz to 10 MHz, relative to the derived oscillation frequency, will be calculated. The Sidebands field is set to a Maximum sideband of 0. In this case, you are interested in the upconverted 1/f device noise to the oscillation frequency. To account for higher harmonics of the oscillator that also contribute noise, change this value. No Input Source is specified. Click OK.
9-14

CIC

Run PSS & PNOISE Simulation

Click the Run Simulation icon and use Direct function to see the results. Compare the oscillation frequency with the previous transient results. Click Plot icon, and the waveform window appears.

9-15