Electrodynamics and Insulating Materials B

Dr.-Ing. Wilfred Njoroge Mwema

18th August 2018 (Rev)

ii
Contents

1 Introduction 1

2 Wave classification 5
2.1 Reduced Maxwell’s equations - A recap . . . . . . . . . . . 5
2.2 Application . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 TEM or Plane waves . . . . . . . . . . . . . . . . . . 8
2.2.2 TE (H) waves . . . . . . . . . . . . . . . . . . . . . . 11
2.2.3 TM (E) waves . . . . . . . . . . . . . . . . . . . . . 13

3 Theory and applications of multi-conductor transmission

lines 17
3.1 TEM wave propagation along a two-conductor transmission
line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2 The TEM wave along a parallel-plate transmission line . . . 21
3.2.1 A lossy parallel-plate line . . . . . . . . . . . . . . . 24
3.2.2 Imperfect dielectric . . . . . . . . . . . . . . . . . . . 25
3.2.3 Solution using the perturbation theory . . . . . . . . 25
3.2.4 Finite conductivity of the conductors . . . . . . . . . 26
3.2.5 Lossy parallel-plate transmission line . . . . . . . . . 28
3.3 Important practical planar transmission lines . . . . . . . . 30
3.3.1 The microstrip . . . . . . . . . . . . . . . . . . . . . 30
3.3.2 Coupled microstriplines . . . . . . . . . . . . . . . . 33
3.3.3 Coplanar lines . . . . . . . . . . . . . . . . . . . . . 34
3.3.4 Coplanar striplines . . . . . . . . . . . . . . . . . . . 34
iv Contents

3.3.5 Suspended and inverted suspended microstrip lines . 34

3.3.6 Open and shielded slot line . . . . . . . . . . . . . . 35
3.3.7 Stripline . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.8 Substrate properties . . . . . . . . . . . . . . . . . . 37
3.4 A distributed circuit analysis . . . . . . . . . . . . . . . . . 38
3.5 Propagation characteristics on an infinitely long transmis-
sion line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.5.1 Limiting cases . . . . . . . . . . . . . . . . . . . . . 43
3.5.2 Attenuation estimation from power relations . . . . 45
3.6 Propagation characteristics on finite-length transmission lines 46
3.6.1 Finite-length transmission line as a circuit element . 50
3.6.2 Lines terminated in resistive loads . . . . . . . . . . 54
3.6.3 Lines terminated in arbitrary loads . . . . . . . . . . 57
3.6.4 Transmission line circuits . . . . . . . . . . . . . . . 60
3.7 Transients on transmission lines . . . . . . . . . . . . . . . . 62
3.7.1 Reflection or bounce diagrams . . . . . . . . . . . . 64
3.7.2 Transients on an initially charged line . . . . . . . . 66
3.7.3 Line terminated in a reactive load . . . . . . . . . . 68
3.8 The Smith-chart . . . . . . . . . . . . . . . . . . . . . . . . 72
3.8.1 Transmission line matching . . . . . . . . . . . . . . 78
3.9 ♣Transmission line parameters . . . . . . . . . . . . . . . . 82
3.9.1 Capacitance per unit length of line on a general two-
conductor line . . . . . . . . . . . . . . . . . . . . . 82
3.9.2 Inductance per unit length on a general two-conductor
line . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.9.3 Conductance per unit length on a general two-conductor
line . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.9.4 Resistance per unit length on a general two-conductor
line . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.9.5 Coaxial line . . . . . . . . . . . . . . . . . . . . . . . 85
3.9.6 Two-wire transmission line . . . . . . . . . . . . . . 86

4 Waveguides 89
4.1 General wave characteristics along uniform guiding structures 89
4.1.1 TEM modes . . . . . . . . . . . . . . . . . . . . . . . 91
4.1.2 TM modes or E modes . . . . . . . . . . . . . . . . . 92
4.1.3 TE modes or H modes . . . . . . . . . . . . . . . . . 95
4.2 Wave velocities and dispersion in guiding structures . . . . 97
4.3 Group velocity . . . . . . . . . . . . . . . . . . . . . . . . . 99

5 Waveguide modes in the parallel-plate transmission line 103

5.1 Waveguide modes in a parallel-plate transmission line . . . 103
5.1.1 TM modes . . . . . . . . . . . . . . . . . . . . . . . 103
5.1.2 TE modes . . . . . . . . . . . . . . . . . . . . . . . . 106
5.2 Energy transport velocity . . . . . . . . . . . . . . . . . . . 108
 Contents v

5.3 Attenuation in the parallel-plate the guide . . . . . . . . . . 110

5.3.1 TEM modes . . . . . . . . . . . . . . . . . . . . . . . 110
5.3.2 TM modes . . . . . . . . . . . . . . . . . . . . . . . 110
5.3.3 TE modes . . . . . . . . . . . . . . . . . . . . . . . . 112

6 Rectangular section waveguides 113

6.1 Rectangular section waveguides . . . . . . . . . . . . . . . . 113
6.1.1 TM or E Modes . . . . . . . . . . . . . . . . . . . . 113
6.1.2 TE or H modes . . . . . . . . . . . . . . . . . . . . . 117
6.1.3 Power transport in rectangular waveguides . . . . . 120
6.1.4 Attenuation in rectangular waveguides . . . . . . . . 121
6.1.5 Discontinuities in rectangular waveguides . . . . . . 125
6.2 Power capacity of guiding structures . . . . . . . . . . . . . 127
6.3 Some useful ideas for waveguide analysis . . . . . . . . . . . 128
6.3.1 Transmission line analogy of waveguides . . . . . . . 128
6.3.2 TEm0 modes of a partially loaded rectangular waveguide132
6.3.3 Transverse-resonance techniques . . . . . . . . . . . 133
6.3.4 Mode properties in rectangular waveguides . . . . . 136
6.4 Rectangular waveguide cavity . . . . . . . . . . . . . . . . . 136
6.4.1 TMmnp modes . . . . . . . . . . . . . . . . . . . . . 137
6.4.2 TEmnp modes . . . . . . . . . . . . . . . . . . . . . . 138
6.4.3 Quality factor of the rectangular cavity . . . . . . . 139

7 Circular section waveguides 143

7.1 Circular section waveguides . . . . . . . . . . . . . . . . . . 143
7.1.1 TM modes . . . . . . . . . . . . . . . . . . . . . . . 144
7.1.2 TE modes . . . . . . . . . . . . . . . . . . . . . . . . 151
7.1.3 Mode Properties in Circular Waveguides . . . . . . . 155
7.2 Waveguide Modes in a Coaxial Transmission Line . . . . . . 156
7.2.1 Coaxial connectors for microwave systems . . . . . . 157
7.3 Circular waveguide cavity . . . . . . . . . . . . . . . . . . . 158

8 Dielectric waveguides 161

8.1 Planar dielectric slab in free space . . . . . . . . . . . . . . 161
8.1.1 TM modes . . . . . . . . . . . . . . . . . . . . . . . 161
8.1.2 TE modes . . . . . . . . . . . . . . . . . . . . . . . . 164
8.1.3 A grounded dielectric slab . . . . . . . . . . . . . . . 168
8.1.4 Finding a solution for the TM surface wave modes
by reverse resonance technique . . . . . . . . . . . . 169
8.2 ♣The cylindrical optical fibre . . . . . . . . . . . . . . . . . 171
8.2.1 Step-index cylindrical waveguide . . . . . . . . . . . 172
8.3 ♣Dielectric-coated conducting wire . . . . . . . . . . . . . . 175
8.4 ♣Inhomogeneously-filled parallel-plate transmission line . . 178
8.4.1 Low frequency solution . . . . . . . . . . . . . . . . 181
8.4.2 High-frequency solution . . . . . . . . . . . . . . . . 183
vi Contents

9 Antenna theory 185

9.1 Radiation from elemental dipoles . . . . . . . . . . . . . . . 186
9.1.1 The Hertzian dipole . . . . . . . . . . . . . . . . . . 186
9.1.2 Elemental magnetic dipole . . . . . . . . . . . . . . . 189
9.2 Thin linear antennas . . . . . . . . . . . . . . . . . . . . . . 192
9.2.1 The half-wave dipole . . . . . . . . . . . . . . . . . . 194
9.3 Antenna patterns and parameters . . . . . . . . . . . . . . . 195
9.3.1 Antenna effective length . . . . . . . . . . . . . . . . 201
9.4 Antenna array theory . . . . . . . . . . . . . . . . . . . . . 203
9.4.1 General uniform linear arrays . . . . . . . . . . . . . 206
9.4.2 The broadside array, (ϕ = 0, φ0 = π/2) . . . . . . . 208
9.4.3 Endfire array (ϕ = −βd, φ0 = 0) . . . . . . . . . . . 209
9.4.4 Phase-scanning of Arrays . . . . . . . . . . . . . . . 210
9.4.5 Arrays with unequally spaced array elements . . . . 211
9.5 Antennas for reception . . . . . . . . . . . . . . . . . . . . . 213
9.5.1 Effective aperture or receiving cross-section . . . . . 215
9.5.2 Back scatter cross-section . . . . . . . . . . . . . . . 217
9.5.3 Transmit-receive systems . . . . . . . . . . . . . . . 219
9.6 Wave propagation near the Earth’s surface . . . . . . . . . . 222
9.6.1 A thin-wire dipole above a perfectly conducting pla-
nar Earth . . . . . . . . . . . . . . . . . . . . . . . . 222
9.6.2 A thin-wire antenna over a finite conductivity flat
Earth . . . . . . . . . . . . . . . . . . . . . . . . . . 224
9.7 Other types of practical antenna systems . . . . . . . . . . . 225
9.7.1 Traveling-wave antennas . . . . . . . . . . . . . . . . 225
9.7.2 Helical antennas . . . . . . . . . . . . . . . . . . . . 227
9.7.3 The Yagi-Uda antenna . . . . . . . . . . . . . . . . . 228
9.7.4 The log-periodic array . . . . . . . . . . . . . . . . . 232
9.8 Radiating apertures . . . . . . . . . . . . . . . . . . . . . . 237
9.8.1 Plane wave sources . . . . . . . . . . . . . . . . . . . 239
9.8.2 Apertures with plane wave excitation . . . . . . . . 241
9.8.3 Radiation from a rectangular aperture . . . . . . . . 242
9.8.4 Radiation from a long slit . . . . . . . . . . . . . . . 247
9.8.5 Circular aperture with uniform illumination . . . . . 249
9.8.6 Parabolic reflector . . . . . . . . . . . . . . . . . . . 252
9.9 Feed techniques for wire antennas . . . . . . . . . . . . . . . 255
9.9.1 Monopole (whip antenna) perpendicular to a con-
ducting plane . . . . . . . . . . . . . . . . . . . . . . 255
9.9.2 The folded dipole . . . . . . . . . . . . . . . . . . . . 255
9.9.3 BALUN . . . . . . . . . . . . . . . . . . . . . . . . . 256
9.10 Circuits of a size comparable to signal wavelength . . . . . 258
9.10.1 Fields in a radiating structure . . . . . . . . . . . . . 261
9.11 Overview of practical radiating systems . . . . . . . . . . . 262
9.11.1 Dipole antennas . . . . . . . . . . . . . . . . . . . . 262
9.11.2 Loop antennas . . . . . . . . . . . . . . . . . . . . . 262
 Contents vii

9.11.3 Traveling-wave antenna . . . . . . . . . . . . . . . . 262

9.11.4 Slot and aperture antennas . . . . . . . . . . . . . . 263
9.11.5 Reflectors and lenses . . . . . . . . . . . . . . . . . . 264
9.11.6 Integrated-circuit (patch) antennas . . . . . . . . . . 264
9.12 Antenna analysis by field computation . . . . . . . . . . . . 264
9.13 Waves guided by biconical guide . . . . . . . . . . . . . . . 269

10 Fundamentals of electromagnetic compatibility 271

10.1 Basic information on EMC . . . . . . . . . . . . . . . . . . 272
10.1.1 EMI sources and victims . . . . . . . . . . . . . . . . 273
10.1.2 Propagation of interference . . . . . . . . . . . . . . 273
10.1.3 Characteristics of interference . . . . . . . . . . . . . 275
10.1.4 Interference measurement . . . . . . . . . . . . . . . 275
10.1.5 Propagation of conducted interference . . . . . . . . 277
10.1.6 Cross-talk . . . . . . . . . . . . . . . . . . . . . . . . 279
10.1.7 Grounding and referencing . . . . . . . . . . . . . . 282
10.1.8 Antenna effect . . . . . . . . . . . . . . . . . . . . . 283
10.2 The EMC guidelines . . . . . . . . . . . . . . . . . . . . . . 284
10.2.1 Government requirements . . . . . . . . . . . . . . . 284
10.2.2 Radiated emissions measurement . . . . . . . . . . . 289
10.2.3 Measurement requirements for conducted emissions . 293
10.2.4 Additional product requirements . . . . . . . . . . . 294

11 Standards and standardisation 297

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
11.1.1 Types of standards . . . . . . . . . . . . . . . . . . . 299
11.2 Status of standards . . . . . . . . . . . . . . . . . . . . . . . 299
11.3 How standards are developed . . . . . . . . . . . . . . . . . 300
11.3.1 Pre-normative and co-normative research . . . . . . 300
11.3.2 Standardisation . . . . . . . . . . . . . . . . . . . . . 300
11.3.3 Formal standards . . . . . . . . . . . . . . . . . . . . 301
11.3.4 Costs of standardisation . . . . . . . . . . . . . . . . 301
11.3.5 Decision tree for standards development . . . . . . . 303
11.3.6 Practical details of standards development . . . . . . 303
11.3.7 Consensus . . . . . . . . . . . . . . . . . . . . . . . . 308

12 Liquid- and solid-state insulators 311

12.1 Liquid insulation . . . . . . . . . . . . . . . . . . . . . . . . 311
12.1.1 Breakdown in liquid insulators . . . . . . . . . . . . 313
12.1.2 Henry’s law . . . . . . . . . . . . . . . . . . . . . . . 314
12.2 Solid-state insulating materials . . . . . . . . . . . . . . . . 316
12.2.1 Organic solid-state insulators . . . . . . . . . . . . . 317
12.2.2 Inorganic solid-state insulators . . . . . . . . . . . . 319
12.2.3 Composite insulating materials . . . . . . . . . . . . 319
12.2.4 Breakdown in solid-state insulators . . . . . . . . . . 319
viii Contents
Preface

This forms the second part of the course "Electrodynamics and Insulating
Materials" in which we will treat topics in

• Transmission lines supporting TEM wave propagation. We will dedi-

cate some time to look into important transmission lines supporting
the so-called quasi-TEM modes such as the microstrip and the copla-
nar waveguide.
• Waveguiding structures supporting TE and TM modes such as the
rectangular and circular section hollow conducting pipes as well as
the dielectric rod (optical fibre).
• Dielectric waveguiding structures supporting the so-called surface
waves such as the dielectric slab.
• Important properties of waveguiding structures including cut-off fre-
quency, dispersion and wave velocities, power propagation and loss.
• Radiating structures or antennas which will include the dipole wire
antennas, waveguide antennas such as the rectangular and circular
horn antennas, radiating apertures and antenna arrays
• Electromagnetic interference (EMI) and electromagnetic compatibil-
ity (EMC) issues
• Liquid- and solid-phase dielectric materials.
x Contents
1
Introduction

Transmission lines and waveguides guide electromagnetic energy from one

part of a system to a another ideally without radiation into the surrounding
medium. The main important desirable characteristics of transmission lines
include

• Single mode propagation over a wide frequency range

• Low attenuation of the signals over a wide range of frequency

Although there is a wide variety of waveguiding structures, these can be

classified broadly under

• Structures supporting TEM mode of propagation e.g. two- or multi-

conductor transmission lines
• Closed cylindrical conducting tubes e.g. the rectangular and circular
section hollow waveguides supporting TM and TE modes of propa-
gation
• Open-boundary structures supporting surface wave propagation such
as the dielectric slab on a conducting plane

Transmission lines consist of two or more parallel conductors and exam-

ples of such structures include

• Two- and multi-conductor line

• Shielded two-conductor line
2 1. Introduction

FIGURE 1.1. Part of the electromagnetic spectrum and allocated applications

TABLE 1.1. Some applications in the low-frequency end of the electromagnetic

spectrum
Application Frequency range
AM broadcast 535-1605 kHz
Shortwave broadcast 3-30 M Hz
FM broadcast 88-108 M Hz
VHF TV (2-4) 54-72 M Hz
VHF TV (5-6) 76-88 M Hz
UHF TV (7-13) 174-216 M Hz
UHF TV (14-83) 470-890 M Hz
Microwave ovens 2, 45 GHz

• Coaxial line
• Microstrip
• Coplanar line

Waveguides include

• The rectangular and circular waveguides which may or may be com-

pletely hollow, partially filled or completely filled with a dielectric
material.
• The dielectric waveguide e.g., the optical fibre
• Surface guides such as a dielectric slab backed by a grounded con-
ducting plane

Fig. 1.1 shows a section of the electromagnetic spectrum and the desig-
nated application. Table 1.1 presents some traditional applications of the
low-frequency end of the electromagnetic spectrum.
Note that Channels 54 − 69 (734M Hz − 862 M Hz) have now been
switched off for analogue TV (European Union). The frequency band from
734 MHz−790 MHz is now allocated to digital terrestrial TV (DTT)/terrestrial
 1. Introduction 3

UHF Ananalogue TV Plan (before switch off)

8MHz
... CH54 CH55 CH56 CH57 CH58 CH59 CH60 CH61 CH62 CH63 CH64 CH65 CH66 CH67 CH68 CH69 ...
734MHz 742MHz750MHz758MHz 766MHz 774MHz782MHz 790MHz798MHz 806MHz814MHz 822MHz830MHz 838MHz 846MHz 854MHz 862MHz

IMT 800MHz Band

UHF DTT (DVB-T) after switch off Technology neutral (LTE FDD)
8MHz 10MHz 11MHz
... FDD1 FDD2 FDD3 Duplex FDD1 FDD2 FDD3 ...
CH54 CH55 CH56 CH57 CH58 CH59 CH60
DL DL DL gap UL UL UL
734MHz 742MHz750MHz758MHz766MHz 774MHz 782MHz 801MHz 811MHz 821MHz 832MHz 842MHz 852MHz 862MHz

790MHz 791MHz

1MHz guard band

FIGURE 1.2. Old and new frequency plan for the 800 M Hz band (European
Union)

TABLE 1.2. Traditional designation of some microwave bands

Designation Frequency range
L-band 1-2 GHz
S-band 2-4 GHz
C-band 4-8 GHz
X-band 8-12 GHz
Ku-band 12-18 GHz
K-band 18-26 GHz
Ka-band 26-40 GHz
V-band 40-60 GHz

digital video broadcasting (DVB-T) applications. The band from 791 MHz−
862 M Hz has been allocated for international mobile telecommunications.
An important point to note is that the 4G long term evolution (LTE) ap-
plication is now collocated with DTT/DVB-T applications and presents
potential interference problems. This new frequency plan is shown in Fig.
1.2.
An alternative designation of some of the microwave bands are presented
in Table 1.2.
4 1. Introduction
2
Wave classification

2.1 Reduced Maxwell’s equations - A recap

We consider the homogeneous vector Helmholtz equations
−
→ −
→ −
→ −
→
∇2 E + k2 E = 0, ∇2 H + k2 H = 0

and seek a solution in which the wave is constrained to propagate in the

+z-axis. We shall assume harmonic time-dependent solutions and a an ex-
ponential spatial dependence on the axial coordinate of the form exp (−γz).
−
→
Recall that ∇ can be decomposed into an axial and transverse component
as
−
→ − → −
→ −
→ ∂ −
→
∇ = ∇ t + ∇ z = ∇ t + z
= ∇ t − z
γ
∂z
where in the Cartesian coordinate system,
−
→ ∂ ∂
∇t = x

+ y

∂x ∂y
We also decompose the electric and magnetic field intensities of a wave
propagating in the positive z-direction in an unbounded region of space
into axial and transverse components as,
→ −
− → 
E = E t + z
Ez exp (−γz)

→ −
− → 
H = H t + z
Hz exp (−γz)
6 2. Wave classification

where we drop the factor exp (jωt) and use the phasor formulation for
convenience.
Now consider Faraday’s induction law in a source-free region of space.
We can write
−
→ − → −→  − →  −→ 
∇ × E = ∇ t − z
γ × E t + z
Ez = −jωµ H t + z
Hz

or
−
→ −
→ −
→ −
→ −
→
∇ t × E t + ∇ t × z
Ez − γ

z × E t − γ

z × z
Ez = −jωµ H t − jωµ

z Hz

We see that this separates the fields into transverse and axial components
as,
−
→ −
→
∇ t × E t = −jωµ
z Hz , axial field components
and
−
→ →
− −
→
z
× ∇ t Ez + γ E t = jωµ H t , transverse field components

Similarly for Ampere’s circuital law, we may write,

−
→ − → − →  − →  −→ 
∇ × H = ∇ t − z
γ × H t + z
Hz = jωǫ E t + z
Ez

or
−
→ −
→ −
→ −
→ −
→
∇ t × H t + ∇ t × z
Hz − γ

z × H t − γ

z × z
Hz = jωǫ E t + jωǫ

z Ez

Thus,
−
→ −
→
∇ t × H t = jωǫ

z Ez , axial field components
and −
→ →
− −
→
z
× ∇ t Hz + γ H t = −jωǫ E t , transverse components

In the divergence equations, we write,

−
→ − → −
→  −→ 
∇ · B = µ ∇ t − z
γ · H t + z
Hz = 0

or
−
→ − → −
→ −
→
∇ t · H t + ∇ t · z
Hz − γ

z · H t − γ

z · z
Hz = 0
Since
−
→ −
→
∇ t · z
Hz = z
· H t = 0
then
−
→ − → −
→ − →
∇ t · H t − γ

z · z
Hz = 0 ⇒ ∇ t · H t = γHz
In a region of space containing free charges, we can write,
−
→ − → −
→  −→ 
∇ · D = ǫ ∇ t − γ
z · E t + z
Ez = ρ
 2.1 Reduced Maxwell’s equations - A recap 7

or
−
→ − → −
→ −
→ ρ
∇ t · E t + ∇ t · z
Ez − γ

z · E t − γ

z · z
Ez =
ǫ
or
→ −
− → ρ −
→ − → ρ
z · z
Ez = ⇒ ∇ t · E t = γEz +
∇ t · E t − γ

ǫ ǫ
When the region contains no free charges, we have
−
→ − →
∇ t · E t = γEz

Thus −
→ −
→
∇t × −E t = −jωµ
z Hz
→ →
− −
→
z
× ∇ t Ez + γ E t = jωµ H t
−
→ −
→
∇t × −H t = jωǫ

z Ez
→ −
→ −
→
z
× ∇ t Hz + γ H t = −jωǫ E t
−
→ − →
∇ t · H t = γHz
−
→ − →
∇ t · E t = γEz
We can express the transverse fields as functions of the axial field com-
ponents alone. In the transverse curl equations, we write
 −→ →  −
− → →
− −
→
z
× z
× ∇ t Ez + γ E t = ∇ t Ez + γ E t = −jωµ
z × Ht

and  −
→ → 
− −
→
−jωµ z
× ∇ t Hz + γ H t = (−jωµ) (−jωǫ) E t
or  →
− ω2 µǫ −
→ jωµ −
→
−jωµ z
× H t = − Et + z
× ∇ t Hz
γ γ
Thus, subtracting, we obtain
−
→ →  ω 2 µǫ −
− → jωµ −
→
∇ t Ez + γ E t + Et − z
× ∇ t Hz = 0
γ γ
or 
ω2 µǫ −
→ −
→ jωµ −
→
γ+ E t = − ∇ t Ez + z
× ∇ t Hz
γ γ
Hence,
−
→ γ− → jωµ −
→
E t = − 2 ∇ t Ez + 2 z
× ∇ t Hz
kc kc
where
kc2 = γ 2 + ω 2 µǫ = γ 2 + k2
Similarly,
 −
→ →  −
− → −
→ −
→
z
× z
× ∇ t Hz + γ H t = ∇ t Hz + γ H t = jωǫ

z × Et
8 2. Wave classification

and  −
→ → 
− −
→
jωǫ z
× ∇ t Ez + γ E t = (jωǫ) (jωµ) H t
or
−
→ ω 2 µǫ −
→ jωǫ −
→
jωǫ

z × Et = − z
× H t − z
× ∇ t Ez
γ γ
Thus, subtracting,


−
→ ω2 µǫ −→ jωǫ −
→
∇ t Hz + γ + Ht + z
× ∇ t Ez = 0
γ γ
or
−
→ γ− → jωǫ −
→
H t = − 2 ∇ t Hz − 2 z
× ∇ t Ez
kc kc

2.2 Application
2.2.1 TEM or Plane waves
−
→ −
→
For TEM waves, E and H lie in a plane transverse to the axis along which
the wave propagates. Thus, Ez = Hz = 0. Hence
−
→ −
→
∇t × E t = 0
−
→ −
→
z
× E t = j ωµ
γ Ht
−
→ −
→
∇t × H t = 0
−
→ −
→
z
× H t = −j ωǫγ Et
−
→ − →
∇t · H t = 0
−
→ − →
∇t · E t = 0
−
→ −
→ −
→ −
→
The fact that ∇ t × E t = 0, ∇ t × H t = 0 implies that the line integral
−
→ −
→
of either E t or H t along a closed path in the xy-plane is zero equivalent
to the case of static electromagnetic fields. This has a physical basis since
there is neither an axial magnetic nor electric field threading the contour.
We can thus define the transverse field components as gradients of scalar
potentials i.e.,
−
→ −
→ −
→ −
→
E t = − ∇ t Φe , H t = − ∇ t Φm
where Φe and Φm are analytical scalar potential functions.
Consider the transverse divergence of the transverse field components.
We obtain
−
→ − → −
→ − →
∇ t · E t = − ∇ t · ∇ t Φe = ∇2t Φe = 0
and
−
→ − → −
→ − →
∇ t · H t = − ∇ t · ∇ t Φm = ∇2t Φm = 0
 2.2 Application 9

which are the two-dimensional Laplace’s equations for the transverse elec-
tric and magnetic fields in a sourceless region of space.
These fields must satisfy the Helmholtz equation, i.e.,
−
→ −
→
∇2 E t + k2 E t = 0

and
−
→ −
→
∇2 H t + k2 H t = 0
where k2 = ω2 µǫ. Thus, since
−→  −
→ 
∇2 = ∇ t − γ T EM z
· ∇ t − γ T EM z
= ∇2t + γ 2T EM

then for the electric field,

−
→ −
→ −
→  −
→
∇2 E t + k2 E t = ∇2t E t + k2 + γ 2 E t = 0

or
→  2
− 
∇ t ∇t Φe + k2 + γ 2T EM Φe = 0
Thus, for TEM waves,
√
γ T EM = jβ = ±jk = ±jω µǫ
−
→
We can then obtain H t from
−
→ jωµ − →
z
× E t = Ht
γ T EM
or
−
→ −
→
−
→ γ T EMT EM −
→ z
× E t z
× ∇ t Φe
Ht = z
× E t = =−
jωµ ηT EM ηT EM
where 
jωµ µ
η T EM = = =η
γ ǫ
η is the intrinsic wave impedance of the unbounded medium. For a TEM
wave propagating in the +z-direction, we hence have,
−
→ −
→
E t = − ∇ t Φe
−
→ −
→
−
→ z
× E t z
× ∇ t Φe
Ht = =−
η η
ηT EM is called the wave impedance of the medium and is used in situations
where the wave is confined. For the TEM waves, we see that the wave
impedance is equal to the intrinsic impedance of the unbounded medium.
10 2. Wave classification

We can derive equations that resemble the telegrapher’s voltage and

current wave equations for the TEM waves by defining,

Φe = V (z) exp (−γz) , Φm = I (z) exp (−γz)

In the transverse curl equations, we can therefore write,

−
→ −
→ −
→ jωǫ −→
γ T EM E t = −jωµ

z × H t , z
× H t = − Et
γ T EM
or

−
→ jωǫ −
→
γ T EM E t = (−jωµ) − Et
γ T EM
or
−
→ −
→
γ 2T EM E t = (−jωµ) (−jωǫ) E t
Since
∂2
γ 2T EM =
∂z 2
then we can write,
∂2 − → −
→
E t = (−jωµ) (−jωǫ) E t
∂z 2
or
∂2  − →   − → 
− ∇ t Φ e = (−jωµ) (−jωǫ) − ∇ t Φe
∂z 2
Since Φe is analytic, we can interchange the order of differentiation and
write,
2 
−
→ ∂ −
→
∇t 2
Φe = (jωµ) (jωǫ) ∇ t Φe
∂z
or
∂2 ∂2
2
Φe = −ω 2 µǫΦe ⇒ 2 V (z) = −ω2 µǫV (z) = −ω 2 L′ C ′ V (z)
∂z ∂z
where we deduce that, L′ = µ, C ′ = ǫ.
Similarly for the magnetic field,
−
→ −
→ −
→ jωµ − →
γ T EM H t = jωǫ

z × E t , z
× E t = Ht
γ T EM
or


−
→ jωµ −
→ −
→ −
→
γ T EM H t = (jωǫ) E t ⇒ γ 2T EM H t = (jωµ) (jωǫ) H t
γ T EM
then we can write,
∂2 − → −
→
2
H t = (−jωµ) (−jωǫ) H t
∂z
 2.2 Application 11

or
∂ 2 −
→  −
→ 
− ∇ t Φm = − (jωµ) (jωǫ) ∇ t Φm
∂z 2
or
∂2 ∂2
2
Φm = −ω 2 µǫΦm ⇒ 2 I (z) = −ω 2 µǫI (z) = −ω2 L′ C ′ I (z)
∂z ∂z

2.2.2 TE (H) waves

For TE waves, Ez = 0, Hz = 0 and
−
→ γ − →
H t = − T2E ∇ t Hz
kc
and
−
→ jωµ −
→ jωµ −
→
Et = − z
× H t = 2 z
× ∇ t Hz
γT E kc
Hz thus plays the role of a potential function from which all other field
components can be derived.
√
We note that with k = ω µǫ, we can write,

jωµ k µ k
− = −j = −j η
γT E γT E ǫ γT E
Thus,
−
→ k −
→ −
→
E t = −j η

z × H t = −jηT E z
× H t
γT E
where
k
ηT E = η
γT E
is the wave impedance of the TE or H modes.
−
→
Since H t is curl-free, then we can write


−
→ −
→ γ − → −
→ ∂ 1
H t = − ∇ t Φm = − T2E ∇ t Hz = ∇ t Hz
kc ∂z kc2
Multiplying through by jωµ, we obtain


−
→ −
→ ∂ jωµ
−jωµ ∇ t Φm = ∇ t Hz
∂z kc2
or

∂ jωµ
Hz = −jωµΦm
∂z kc2
and from
−
→ −
→ −
→ −
→ ω 2 µǫ −
→
γ H t = − ∇ t Hz + jωǫ

z × E t = − ∇ t Hz + 2 ∇ t Hz
kc
12 2. Wave classification

we can write


2 

−
→ ∂Φm ω µǫ −
→ jωµ kc2 −→
∇t = 2
− 1 ∇ t Hz = − 2 jωǫ + ∇ t Hz
∂z kc kc jωµ
or


∂Φm k2 jωµ
= − jωǫ + c Hz
∂z jωµ kc2

We note that the quantity jωµHz /kc2 has dimensions of voltage which
we designate V (z) while Φm has those of current which we designate I (z)
and therefore express the equations as

∂
V (z) = −jωµI (z) = −Z ′ I (z)
∂z
and 
∂Iz kc2
=− + jωǫ V (z) = −Y ′ V (z)
∂z jωµ
where
jωµ
V (z) = Hz
kc2

I (z) = Φm

Z ′ = jωµ

and
kc2
Y′ = + jωǫ
jωµ
are the per unit length series impedance and shunt admittance of a trans-
mission line respectively. The characteristic impedance of the line is hence,
 
Z′ µ 1 η
Z0,T E = =  =
Y′ ǫ
1 − (kc /k)2 1 − (kc /k)2

and the propagation constant,


2
√  kc
γT E = Z ′ Y ′ = (kc2 − ω 2 µǫ) = k −1
k

and for these waves, γ T E = 0 at k = kc in which case the wave does not
propagate. kc is hence called the cut-off wavenumber of a TE mode.
The equivalent circuit for TE wave propagation is shown in Fig. 6.7.
 2.2 Application 13

jωµ

k2c
jωε
jωµ

FIGURE 2.1. Transmission equivalent circuit for TE modes in a rectangular

waveguide

2.2.3 TM (E) waves

Here we have Ez = 0, Hz = 0. Thus,
−
→ γ −
→
E t = − T2M ∇ t Ez
kc

and
−
→ jωǫ −
→ jωǫ −
→
Ht = z
× E t = − 2 z
× ∇ t Ez
γT M kc
We note that 
jωǫ k ǫ k
=j =j
γT M γT M µ γT M η
so that
−
→ k −
→ 1 −
→
Ht = j z
× E t = j z
× E t
γT M η ηT M
where
γT M
ηT M = η
k
is the wave impedance of the E-waves.
−
→
Because E t is curl-free, we can express it as the gradient of an analytical
scalar function Φe as


−
→ −
→ γ −
→ ∂ 1−→
E t = − ∇ t Φe = − T2M ∇ t Ez = ∇ E
t z
kc ∂z kc2

Thus, multiplying through by jωǫ, we obtain,



−
→ −
→ ∂ jωǫ
−jωǫ ∇ t Φe = ∇ t Ez
∂z kc2
or

∂ jωǫ
Ez = −jωǫΦe
∂z kc2
14 2. Wave classification

and from


−
→ −
→ −
→ jωǫ −
→ ω2 µǫ −
→
∇ t Ez +γ T M E t = −jωµ
z × − 2 z
× ∇ t Ez = 2 ∇ t Ez
z × H t = −jωµ

kc kc
we can write
−
→ ∂Φe ω 2 µǫ −
→ −
→
−γ T M ∇ t Φe = = 2 ∇ t Ez − ∇ t Ez
∂z kc
or
2 

∂Φe ω µǫ 1 jωǫ
= 2
− 1 Ez = − jωµ + Ez
∂z kc jωǫ kc2
We note that the quantity jωǫEz represents an axial displacement current
density while 1/kc2 has dimensions of area. Thus jωǫEz /kc2 is a current in
the axial direction which we denote by an equivalent current I (z). Hence


∂ jωǫ ∂
Ez = I (z) = −jωǫV (z)
∂z kc2 ∂z
and 
∂ kc2
V (z) = − jωµ + I (z)
∂z jωǫ
These have the same form as the differential equations for a lossless trans-
mission line with series impedance per unit length given by
kc2
Z ′ = jωµ +
jωǫ
and shunt admittance per unit length
Y ′ = jωǫ
Thus, we can represent the wave propagation characteristics of the TM
waves as shown in Fig. 6.6.We note that TE and TM waves are duals of
each other since
η T E ηT M = η2
as long as γ is the same for both waves.
For energy flow along the z-axis, power is defined in terms of the trans-
verse field components alone. Thus, for TE waves

1 −
→ −
→
Pave = Re E t × H ∗t · z
dx dz
2
S
 
1 → −
− →
= Re ηT E z
× H t × H ∗t · z
dx dz
2
S
 
ηT E −
→ − → 1 −
→ − →
= H t · H ∗t dx dz = E t · E ∗t dx dz
2 2ηT E
S S
 2.2 Application 15

k2
c
jωµ jωε

jωε

FIGURE 2.2. Transmission line equivalent circuit for TM modes in a rectangular

waveguide

Thus, ηT E allows power to be expressed in terms of one of the transverse

field components alone.
16 2. Wave classification
3
Theory and applications of
multi-conductor transmission lines

The most common transverse electromagnetic (TEM) guiding structures

include

• The parallel plate transmission line

• The two-conductor and the shielded two-conductor transmission lines

−
→ −
→
• The coaxial transmission line in which the E and H fields are com-
pletely confined in the dielectric region

Fig. 3.1 shows cross-section examples of common transmission line struc-

tures.

FIGURE 3.1. Example of common two-conductor transmission lines

18 3. Theory and applications of multi-conductor transmission lines

FIGURE 3.2. A two-conductor line showing the electric and magnetic fields

3.1 TEM wave propagation along a two-conductor

transmission line
Consider a TEM wave propagating in the +z-direction along a two-conductor
transmission line of arbitrary cross-section with perfect conductors and air
dielectric as shown in Fig. 3.2. We assume that the wave has a harmonic
time-dependence and exponential dependence on the axial coordinate of
−
→
the form exp (−γz). We consider a point in the dielectric at which J = 0
and ρ = 0. The reduced Maxwell’s equations are then,
−
→ −
→ −
→ −
→
∇ t × E t = 0, ∇ t × H t = 0
−→ →
− −
→
z
× ∇ t Ez + γ E t = jωµH t
−→ →
− −
→
z
× ∇ t Hz + γ H t = −jωǫ E t
→ −
− → −
→ − →
∇ t · E t = 0, ∇ t · H t = 0

We assume the conductor surfaces S1 and S2 are at potentials Φ = −V0 /2

−
→ −
→
and Φ = V0 /2 respectively. Since E t and H t are curl-free in the TEM
case, we see that this is the exact condition for a static electromagnetic
field. Hence, either the electric or magnetic fields can be expressed as the
gradient of a scalar potential function Φ. Hence, Φ (x, y) must be a solution
of the transverse Laplace’s equation
∇2t Φ = 0
and satisfy the boundary conditions
Φ = V0 /2 on S2
= −V0 /2 on S1
 3.1 TEM wave propagation along a two-conductor transmission line 19

Φ is unique only to within an additive constant which will be resolved by

application of the appropriate boundary conditions. Consider the electric
field intensity. For propagation in the +z-direction, we have
−
→ − → −
→
E = E t exp (−jkz) = − ∇ t Φ exp (−jkz)
−
→
The line integral of E t along a path connecting the two conducting surfaces
is then
S2 S2 S2
−
→ −
→ −
→ −
→ dΦ
E t · d l = − ∇ tΦ · d l = − dl = {Φ (S2 ) − Φ (S1 )} = −V0
dl
S1 S1 S1

Associated with the electric field is a unique voltage wave

V = V0 exp (−jkz)
−
→
and since the line integral of E t is independent of the path chosen between
−
→
the conductors, then E t is the gradient of the scalar potential.
−
→
The line integral of H t around one conductor, say S2 is given by
 
−
→ −
→
Ht · d l = Js dl = I0
C C

−
→ −
→
where n
× H t = J s due to the assumption of a perfect conductor with
n

the outward unit normal vector on the conductor surface. Since n
and
−
→ −
→
H t are coplanar in the transverse plane, J s is axial. Far away from the

−
→ − → → −
− →
conductor surfaces, ∇ × H = 0 but the line integral H t · d l = 0 because
C
−
→
of J s . Associated with the magnetic field is a unique current wave

I = I0 exp (−jkz)
−
→ −
→
Since Φ is independent of frequency, both E t and H t are as well.
We now consider the coaxial line of the transverse cross-section shown in
Fig. 3.3. In cylindrical coordinates, the transverse Laplace’s equation takes
the form

1 ∂ ∂Φ 1 ∂Φ
r + 2 =0
r ∂r ∂r r ∂φ
and if Φ has angular symmetry, ∂Φ/∂φ = 0 and



1 ∂ ∂Φ d dΦ
r = r =0
r ∂r ∂r dr dr
20 3. Theory and applications of multi-conductor transmission lines

FIGURE 3.3. Coaxial line carrying an electromagnetic wave

Hence,
dΦ C1
=
dr r
or
Φ = C1 ln r + C2
Since Φ = V0 @ r = a and Φ = 0 @ r = b, then C2 = −C1 ln b and
C1 = V0 / ln (a/b) so that
ln (r/b)
Φ= V0
ln (a/b)
and
−
→ dΦ −
r V0
Et = −

r exp (−jkz) = exp (−jkz)
dr r ln (a/b)
r
V0
= exp (−jkz)
r ln (b/a)
and
−
→ 1 −
→

φ V0
H t = r
× E t = exp (−jkz)
η r η ln (b/a)
Since the potential difference between the outer and inner conductors is
V0 , then the unique voltage wave associated with the electric field is
V = V0 exp (−jkz)
and the current density on the inner conductor is
−
→ −
→ −
→ z
V0
Js=n

× H t = r
× H t = exp (−jkz)
a η ln (b/a)
The magnitude of the current on the conductor is thus,
2π
1 V0 V0
|I0 | = a dφ = 2π
a η ln (b/a) η ln (b/a)
0
 3.2 The TEM wave along a parallel-plate transmission line 21

where dS = a dφ. The unique current wave associated with the magnetic
field is thus
I = I0 exp (−jkz)
The rate of energy flow or power along the line is from Poynting theorem

b 2π
1 −
→ −
→
P = Re E t × H ∗t · r
r dr dφ
2
a 0

2 b 2π
1 V0 dφ dr V02
= =π
2η ln (b/a) r η ln (b/a)
a 0

and the characteristic impedance given by

V0 η
Z0 = = ln (b/a)
I0 2π

which is a function of the cross-sectional shape of the transmission line. We

see that
1 1 V02
P = Z0 I02 =
2 2 Z0
Note that since the conductors are perfect, the electric field inside the
conductors is of course zero. This means that the power flow is zero in the
conductors and energy transport thus occurs in the dielectric region!

3.2 The TEM wave along a parallel-plate

transmission line
Now consider the parallel-plate structure shown in Fig. 3.4 assumed to be of
infinite extent in the +z-direction. The plates are assumed to be sufficiently
wide so that fringing effects at the edges of the plates can be ignored. In
a source-free region in the dielectric, the fields of a wave with a y-directed
time-harmonic electric field and propagating in the +z-direction can be
expressed as
−
→
E = y
Ey = y
E0 exp (−γz)
−
→
−
→ z
× E E0
H= = −

x exp (−γz)
η η
If the plates are perfectly conducting and the dielectric lossless, we have,
√
γ = α + jβ = jβ = jω µǫ
22 3. Theory and applications of multi-conductor transmission lines

h y z
x

FIGURE 3.4. A parallel-plate transmission line

and − 
→ 
E  µ
η = − −
→ = ǫ
H 

At both y = 0 and y = h, the continuity of the tangential components of

−
→ −
→
E requires that these vanish as must the normal components of H . Thus,
−
→
E t = 0, Hn = 0 ⇒ Ex = Ez = 0, Hy = 0

At y = 0, the outward unit normal on the conductor surface is n

= y
. Thus,
at y = 0,
−
→
y
· D = ρs,0 ⇒ ρs,0 = ǫEy = ǫE0 exp (−jβz)
−
→ − → −
→ E0
y
× H = J s,0 ⇒ J s,0 = −

z Hx = z
exp (−jβz)
η
At y = h, n

= −
y, hence
−
→
−

y · D = ρs,h ⇒ ρs,h = −ǫEy = −ǫE0 exp (−jβz)

−
→ − → −
→ E0
−
y × H = J s,h ⇒ J s,h = z
Hx = −
z exp (−jβz)
η
Thus, the surface current and free surface charge densities on the plates
vary sinusoidally with z.
The electric and magnetic field intensities in the dielectric satisfy Maxwell’s
equations, i.e.
−
→ − → −
→ dEy −
→ − → −
→ dHx
∇ × E = −jωµ H ⇒ = jωµHx , ∇ × H = jωǫ E ⇒ = jωǫEy
dz dz
−
→ − →
Thus, integrating ∇ × E over y from y = 0 to y = h, we obtain
h h
d
Ey dy = jωµ Hx dy
dz
0 0
 3.2 The TEM wave along a parallel-plate transmission line 23

or

d h
− V (z) = jωµJs,h h = jω µ (Js,h W ) = jωL′ I (z)
dz W
where
h
V (z) = − Ey dy = −hEy (z)
0

is the potential difference between the plates and

I (z) = W Js,h

is the total current flowing in the plate at y = b in the +z-direction.

h
L′ = µ
W
is the external inductance per unit length of the parallel-plate transmission
line structure.
−
→ − →
We similarly integrate ∇ × H over x from x = 0 to x = W to obtain

W x
d
Hx dx = jωǫ Ey dx
dz
0 0

or


d W
− I (z) = −jωǫEy (z) W = jω ǫ (−Ey (z) h) = jωC ′ V (z)
dz h

where
W
C′ = ǫ
h
is the capacitance per unit length of the parallel-plate transmission line
structure.
These equations can be combined into a single second-order differential
equation in either V (z) or I (z) as

d2
V (z) = −ω2 L′ C ′ V (z)
dz 2
or
d2
I (z) = −ω2 L′ C ′ I (z)
dz 2
which have general solutions of the form

V (z) = V + exp (−jβz) + V − exp (jβz)

I (z) = I + exp (−jβz) + I − exp (jβz)

24 3. Theory and applications of multi-conductor transmission lines

where √ √
β = ω L′ C ′ = ω µǫ
Since the line is infinitely long, then
V (z) = V + exp (−jβz)
I (z) = I + exp (−jβz)
We define a line impedance at a point z along the line as

∆ V (z) V+ L′
Z (z) = = + =
I (z) I C′
If only the −z-propagating wave exists on the line, we have
V (z) = V − exp (jβz)
I (z) = I − exp (jβz)
Thus,
d  −
− V exp (jβz) = −jβV − exp (jβz) = jωL′ I − exp (jβz)
dz
or 
V (z) V− ωL′ L′
Z (z) = = − =− =−
I (z) I β C′
We note that the line impedance is independent of the position along the
line for the case of voltage and current waves propagating in only one
direction. The line impedance for this special case is called the characteristic
impedance Z0 of the line and depends only on the line’s parameters and
geometry. Using the values for L′ and C ′ of a parallel-plate transmission
line, we obtain, 
h µ h
Z0 = =η , Ω
W ǫ W
This is the impedance that will be seen at any location on the line looking
towards an infinitely long (reflection-free) line and is known as the charac-
teristic impedance of the line. The wave propagates along the line with a
phase velocity
ω 1 1
vp = = √ = √ , ms−1
β LC′ ′ µǫ

3.2.1 A lossy parallel-plate line

Loss in the line can arise in any of two ways
• Through a lossy (imperfect) dielectric
• Through finite conductivity of the conducting plates
 3.2 The TEM wave along a parallel-plate transmission line 25

3.2.2 Imperfect dielectric

We account for the dielectric losses by requiring that the dielectric have a
permittivity ǫ = ǫ′ − jǫ′′ and assume that the conductors are perfect. Thus,
the propagation constant is given by,
 1/2  1/2
 ǫ′′ ǫ′′
γ = jω µǫ′ 1 − j ′ = jk 1 − j ′
ǫ ǫ

and   −1/2
µ/ǫ′ ǫ′′
Z0 = = η 1 − j
1 − jǫ′′ /ǫ′ ǫ′

When the losses are small such that ǫ′′ /ǫ′ << 1, the propagation constant
can be expanded in a binomial series as

ǫ′′ ǫ′′
γ = α + jβ = jk 1 − j ′ ∼ jk + k ′
ǫ 2ǫ
Hence
ǫ′′
α=k
2ǫ′
and
β=k
and the characteristic impedance is given by


ǫ′′
Z0 ∼ η 1 + j ′
2ǫ

Thus, as the wave propagates in the +z-direction, its amplitude decays as

exp (−αz).

3.2.3 Solution using the perturbation theory

We arrive at the same result using perturbation theory although the present
problem does not call for such a derivation since it is simple enough. It is
however instructive to give such a treatment to demonstrate the technique
for the cases when the situation is much more complex.
The perturbation theory technique assumes that the losses are sufficiently
low such that the field is not significantly perturbed from its lossless value
or distribution. Thus, we use the lossless values to calculate the line losses
from which the attenuation coefficient can be computed.
For the lossless case, we have seen
−
→ −
→ −
→ −
→
E t = − ∇ t Φ exp (−jkz) H t = Y z
× E t
26 3. Theory and applications of multi-conductor transmission lines

where
β = k Z0 = η, ǫ′′ ∼ 0
We saw previously that if ǫ′′ is small but different from zero, it can be
expressed as an equivalent conductivity σd = ωǫ′′ which results in a shunt
−
→ −
→
current J = σd E t between the conductors. The power loss per unit length
of the line is hence,
 
1 −
→ − → ωǫ′′ → −
− →
Pdiel = J · J ∗ dS = E t · E ∗t dS
2σd 2
S S

with the integration understood to be over the cross-section of the line.

Because of the non-zero power loss Pdiel , power propagation along the
line must decrease proportionally to the square of the electric field, i.e.,
according to exp (−2αz) along the line. From conservation of energy, the
rate of decrease of power propagating down the line must be equal to the
power loss. Letting Pin (0) be the power incident on the line at z = 0, then
the power at z down the line is given by

P (z) = Pin (0) exp (−2αz)

and
∂P
− = Pdiel = 2αPin (0) exp (−2αz) = 2αP (z)
∂z
which states that the power loss at any plane z is directly proportional
to the total power present at this plane. From Poynting’s theorem, the
time-average power propagated along the line is given by,

1 −
→ −
→
P (z) = Re E t × H ∗t · z
dS
2
S
  
1 −
→ → 
− 1 −
→ − →
= Re E t × z
× E ∗t · z
dS = Re E t · E ∗t dS
2η 2η
S S

Thus,
Pdiel σdη ωǫ′′ η kǫ′′
αdiel = = = =
2P (z) 2 2 2
the same as before.

3.2.4 Finite conductivity of the conductors

When σc of the conductor is finite, the conducting medium exhibits a sur-
face impedance,
1+j
Zm =
σc δ s
 3.2 The TEM wave along a parallel-plate transmission line 27

where 
2
δs =
ωµc σ c
is the skin depth.
The tangential electric field at the conductor surface must therefore be
−
→ −
→ −
→
Zm J s where J s = n
× H is the surface current density. Thus, in the case
of a finite σ, the axial electric field is no longer zero and the wave then no
longer truly TEM. This axial field gives rise to a component of Poynting’s
vector directed into the conductor which accounts for the conductor power
−
→
loss. If |Zm | is much smaller than Z0 , Zm J s is also small relative to the
transverse components so that the field solution is very nearly TEM. This
mode of wave propagation is referred to as quasi-TEM.
In applying the perturbation theory, we let
−
→ −
→
Js=n

× Ht
−
→
where n
is the unit outward normal to the conductor surface and H t the
lossless transverse field. Power loss per unit length of line due to the surface
impedance is then,
  
1 −
→ − → Rm → 
− → 
−
Pc = Re Zm J s · J ∗s dl = n
× H ∗t dl

× Ht · n
2 2
C1 +C2 C1 +C2

where C1 and C2 describe the peripheries of conductors S1 and S2 resp.

Thus,

Rm −
→ − →
Pc = H t · H ∗t dl
2
C1 +C2
→ −
− → − → −
→ − → −→
since with A · B × C = A × B · C , then
 → 
− → 
−  →
−  −→
n
× H ∗t

× Ht · n = n

× Ht × n
· H ∗t
 −
→  − →  − →
= (
n·n

) H t − n
n

· Ht · H ∗t
−
→ − → −
→ − →
= (

n·n
) H t · H ∗t = H t · H ∗t
−
→
and n

· H t = 0 in the case of σ → ∞. Thus

−
→ − →
Rm H t · H t dl
Pc C1 +C2
αc = = 
2P (z) −
→ − →
2Z0 H t · H ∗t dS
S
28 3. Theory and applications of multi-conductor transmission lines

where 
µ
Z0 = η =
ǫ
in the lossless case.
When both the conductors and dielectric are imperfect, the total atten-
uation coefficient is the sum of the individual attenuations provided these
are separately small. Thus, in the coaxial line, with permittivity ǫ = ǫ′ −jǫ′′
and conductors of conductivity σc , the pertinent quantities in the loss-free
case are replaced by
Z0 ∼ η, β ∼ k

Thus

−
→ r
V0 −
→

φ V0
Et = exp (−jkz) H t = exp (−jkz)
r ln (b/a) r Z0 ln (b/a)

and
b 2π
1 −
→ −
→ V02
P (z) = Re E t × H ∗t · r
r dr dφ = π
2 Z0 ln (b/a)
a 0

Power loss from lossy dielectric

b 2π
ωǫ′′ −
→ −
→ ωǫ′′ V02
Pdiel = E t × H ∗t · r
r dr dφ = π
2 ln (b/a)
a 0

and power loss due to the conductors of finite conductivity

2 2π

2

Rm Y V0 1 1 V0 1 1
Pc = + dφ = πRm +
2 ln (b/a) a b Z0 ln (b/a) a b
0

Hence, the attenuation for the lossy coax line is given by


Pdiel + Pc ωǫ′′ Z0 Rm 1 1
α = = + +
2P (z) 2 2Z0 ln (b/a) a b


kǫ′′ Rm 1 1
= + +
2ǫ′ 2Z0 ln (b/a) a b

3.2.5 Lossy parallel-plate transmission line

From the definition, the capacitance per unit length between the plates of
a parallel-plate line is given by the charge per unit length Q′ and voltage
 3.2 The TEM wave along a parallel-plate transmission line 29

between the conductor V as


→ −
− →
ǫE · d l
Q′ C2
C′ = =
V S2
→ −
− →
− E ·d l
S1

where C1 and C2 are the peripheries of conductors S1 and S2 . The resistance

per unit length to the current through the dielectric of conductivity σd by,
S2
→ −
− →
− E ·d l
V S1
R′ = = 
I → −
− →
σd E ·d l
C2

Thus,
ǫ
R′ C ′ =
σd
or the conductance per unit length of line,
1 σd ′
G′ = = C
R′ ǫ
For the parallel-plate line, we thus have
W
G′ = σd
h
If the plates have a large but finite conductivity σ c , the ohmic losses in
the conductor require a non-zero tangential electric field at the conductor
surface such that
−
→ 1
P ave = Re [
z Ez × x

Hx ]
2
is non-zero and is equal to the average power per unit area dissipated in
each conducting plate.
Consider the plate at y = h having a surface current density Js,h = Hx .
We can thus define a surface impedance on the conductor as
∆ Etan
Zm = , Ω
Js
so that, for the plate at y = h,
Ez Ez
Zm = = = ηc
Js,h Hx
30 3. Theory and applications of multi-conductor transmission lines

where ηc is the intrinsic impedance of the boundless plate conductor. We

assume that the current is confined in a vanishingly thin surface layer and
can therefore be represented by a surface current density Js,h . Thus, we
can write 
πfµc
Zm = Rm + jXm = (1 + j)
σc
where µc is the permeability of the conductor. Hence
−   
→  1 → 2
− 1 −
→ 2
 P ave  = Re  J s,h  Zm =  J s,h  Rm
2 2

The ohmic power dissipated in a unit length of plate of width W is hence

− 
→  1 Rm
Pc = W  P ave  = I 2 , W m−1
2 W
where
I = W Js,h
The effective series resistance per unit length for both plates is hence

 
′ Rm 2 πfµc
R =2 =
W W σc

The ratio of the axial to transverse components of the electric field is given
by     
 Ez  |ηc Hx | ǫ ωµc ǫ ωǫ
 = = |η | = =
 Ey  |ηHx | c
µ µσ c σc
so that for Cu in air with σCu ∼ 5, 8 × 107 Sm−1 , ǫ ∼ ǫ0 at f = 3 GHz,

|Ez | ∼ 5, 3 × 10−5 |Ey | << |Ey |

for all practical purposes.

3.3 Important practical planar transmission lines

3.3.1 The microstrip
Fig. 3.5 shows the geometry of the microstrip line found in many microwave
integrated circuits (MICs). By image theory, the line is equivalent to a two-
conductor line with parallel conductor strips 2h apart. Since the dielectric
does not extend everywhere in the vicinity of the conducting strip, this line
does not support pure TEM propagation and the solution of the fields in
the vicinity of the line is a hybrid of both TM and TE fields. When h << λ,
the axial fields are so small that they can be neglected and the line thus
 3.3 Important practical planar transmission lines 31

FIGURE 3.5. The microstrip line

considered a quasi-TEM line. Analysis of this line is complicated and we

give only some of the important design formulae.
Fig. 3.6 shows several versions of the stripline transmission line structure.
For W >> h, the results obtained for the parallel-plate transmission line
will approximate the field configuration in a stripline.
The inhomogeneous dielectric can be replaced by an effective homoge-
neous one with an effective dielectric coefficient

ǫr + 1 ǫr − 1 1
ǫef f = +
2 2 1 + 12h/W

For the quasi-TEM case, the characteristic impedance of the line is


60 8h W W
Z0 = √ ln + , ≤1
ǫef f W h h
120π W
= √ W W , ≥1
ǫef f h + 1, 393 + 0, 667 ln h + 1, 444 h

For a given ǫr and Z0

W 8 exp (A) W
= , <2
h exp (2A) − 2 h

 
2 ǫr − 1 0, 61 W
= B − 1 − ln (2B − 1) + ln (B − 1) + 0, 39 − , >2
π 2ǫr ǫr h
where 

Z0 ǫr + 1 ǫr − 1 0, 11
A= + 0, 23 +
60 2 ǫr + 1 ǫr
and
377π 60π2
B= √ ∼ √
2Z0 ǫr Z0 ǫr
32 3. Theory and applications of multi-conductor transmission lines

W
t grounded
h
ε h t W
ε
h
grounded grounded
microstrip line triplate line
W W

ε h ε h

slotline
coplanar strip line
a
W

s
ε h

coplanar waveguide (CPW)

FIGURE 3.6. Versions of the strip line

 3.3 Important practical planar transmission lines 33

Since we have a quasi-TEM propagation, the attenuation due to a lossy

dielectric is given by
kǫr (ǫef f − 1) tan δ
αdiel ∼ √ N p/m
2 ǫeff (ǫr − 1)
where
ǫr (ǫef f − 1) tan δ
F =
ǫef f (ǫr − 1)
is the fill-factor that accounts for field lines lying partly in the dielectric
and partly in air. The attenuation associated with finite-conductivity con-
ductors is
Rm
αc ∼ N p/m
Z0 W
Example: Neglecting losses and fringing effects and assuming the sub-
strate of a stripline to be of thickness 0, 4 mm and ǫr = 2, 25,
(a) Determine the required width W of the metal strip in order for the
stripline to have a characteristic impedance of 50 Ω
(b) Determine L′ and C ′ of the line
(c) Determine vp along the line
(a) 
h µ 0, 4 × 10−3 m η
W = = √ = 2, 0 mm
Z0 ǫ 50 Ω ǫr
(b)
h 0, 4
L′ = µ = 4π × 10−7 Hm−1 × = 0, 251 µHm−1
W 2
W
C′ = ǫ = 99, 9 pF m−1
h
(c)
1
vp = √ = 2 × 108 ms−1
L′ C ′
Note that W ∝ 1/Z0 .

3.3.2 Coupled microstriplines

These are useful in the design of directional couplers. Because of the two
conductors, they support two modes of propagation
• Even modes when the E-fields on the two strips are in opposite di-
rections
• Odd modes when the E-fields on the two strips are in the same di-
rection.
Fig. 3.7 shows the cross-section of such a line.
34 3. Theory and applications of multi-conductor transmission lines

FIGURE 3.7. Th coupled microstrip line

FIGURE 3.8. The coplanar line

3.3.3 Coplanar lines

These are especially suited for connection of shunt circuit elements since
the ground conductors are on the same side as the signal conductor. They
also offer the ease of series circuit element connection of the microstrip
and have found extensive use in modern microwave monolithic integrated
circuits (MMICs). The cross-section of this line is shown in Fig. 3.8.

3.3.4 Coplanar striplines

This line with the cross-section shown in Fig. 3.9 is similar to the coplanar
line but is less desirable since it cannot be balanced with respect to the
ground plane and propagation is hence more sensitive to nearby conductors
e.g. shielding enclosures.

3.3.5 Suspended and inverted suspended microstrip lines

The cross-section of these lines are shown in Fig. 3.10. These lines are
similar to the conventional microstrip line but require less dielectric.
 3.3 Important practical planar transmission lines 35

FIGURE 3.9. Coplanar stripline

FIGURE 3.10. Suspended and shielded suspended microstrip line

3.3.6 Open and shielded slot line

The open slot line is not as frequently used as the coplanar or microstrip
lines. The shielded slotline, also referred to as the fin line, finds uses in
many circuits. Only shunt connected components can be used however.
The cross-sections of these lines are illustrated in Fig. 3.11.

3.3.7 Stripline
Finds common use in microwave filters but has been largely replaced by the
coplanar and microstrip lines. It is advantageous for fabricating couplers
but much more difficult to fabricate. The two-conductor structure and ho-
mogeneous dielectric allows TEM propagation which is the desired mode.

FIGURE 3.11. The slot- and finlines

36 3. Theory and applications of multi-conductor transmission lines

Higher-order mode propagation is suppressed by short circuiting the two

ground planes using shorting screws and having a ground plane separation
less than λ/4. The characteristic impedance is approximately,

30π b
Z0 = √
ǫr Wef f + 0, 441b

where Wef f is an effective width of the central conductor given by

Wef f W W
= , > 0, 35
b b b

2
W W
= 0, 35 − , < 0, 35
b b

which assumes strips of zero thickness. Thus, the design formulas can be
expressed as

W √
= x, Z0 ǫr < 120
b  √
= 0, 85 − 0, 6 − x, Z0 ǫr > 120

where
30π
x= √ − 0, 441
Z0 ǫr
The attenuation due to dielectric loss is the same as for other TEM lines.
The attenuation due to conductors may be obtained by perturbation theory
or Wheeler’s incremental inductance rule
2, 7 × 10−3 √
αc = Rm ǫr Z0 A, Z0 ǫr < 120
30π (b − t)
0, 16 √
= Rm B, Z0 ǫr > 120
Z0 b
where

2W 1 b+t 2b − t
A=1+ + ln
b−t π b−t t
and

b 0, 414t 1 4πW
B = 1+ 0, 5 + + ln
0, 5W + 0, 7t W 2π t
where t is the strip thickness. The cross-section of the stripline is illustrated
in Fig. 3.12.
Example: Find the width for a 50 Ω copper stripline conductor with b =
0, 32 and ǫr = 2, 20. If the dielectric has a loss tangent of 0, 001 and the
operating frequency is 10 GHz, calculate the attenuation in dB/λ. Assume
conductor thickness of t = 0, 01 mm.
 3.3 Important practical planar transmission lines 37

FIGURE 3.12. The stripline

√ √
Z0 ǫr = 74, 2 < 120 ⇒ x = 30π/ (Z0 ǫr ) − 0, 441 = 0, 83 cm

W = bx = 0, 266 cm
2πf √
k= ǫr = 310, 6 m−1
c
k tan δ
αdiel = = 0, 155 Np/m
2
2, 7 × 10−3 Rm ǫr Zc A
Rm = 0, 026 Ω ⇒ αcond = = 0, 122 N p/m, A = 4, 74
30π (b − t)
α = αdiel + αcond = 0, 277 N p/m ⇒ αdB = −20 log10 eα = 2, 41 dB/m
c
λ = √ = 2, 02 cm ⇒ αdB /λ = (2, 41 dB/m) (0, 0202 m) = 0, 049 dB/λ
f ǫr

3.3.8 Substrate properties

Fabrication techniques for planar transmission lines are compatible with
those for integrated circuits. The dielectric used must have,

• Low loss

• High dielectric coefficient for more compact circuits

• Good mechanical strength

• Good machinability

• Good thermal conduction since heavy metal heat sinks are undesir-
able at microwave frequencies as they interact undesirably with the
electromagnetic fields.
38 3. Theory and applications of multi-conductor transmission lines

• High uniformity of dielectric coefficient and thickness must be high

due to the dependence of the propagation coefficient on these pa-
rameters. High thickness and dielectric constant uniformity is more
important in filter and impedance matching network designs.

Common dielectrics include

• Teflon (Polytetrafluoroethylene-PTFE) with ǫr ∼ 2, 1 and tan δ ∼

0, 0002 @ 1 MHz and ∼ 0, 0005 @ microwave frequencies. Teflon has
excellent resistance to the chemicals used in the etching process.
The mechanical strength is improved by integrating a fibre glass mat
or glass microparticles which also improve ǫr to 2, 2 − 3. However
this introduces anisotropy in ǫr . During manufacture, the glass fibres
are aligned parallel to the substrate. The dielectric coefficient along
the substrate is then generally 5 − 10% higher than in the direction
normal to the dielectric.
Use of ceramic powders, e.g. titanium oxide allows much larger di-
electric constants to be achieved, typically 5 − 15.

• Aluminium oxide (alumina), boron nitride sapphire

Very difficult to machine. The excellent thermal conductivity of alu-
mina is responsible for its frequent use in hybrid and MIC’s.

• High resistivity silicon, germanium and semi-insulating gallium ar-

senide find frequent use in microwave IC’s. These have high ǫr and
may exhibit some conductivity depending on doping.

Substrate materials are usually plated with copper at 0, 5 - 1- or 2 oz

weights (amounts of Cu/ft2 ) where 1-oz is equivalent to 0, 0014′′ thick
copper lamina. Gold plating is also used over the copper to prevent oxi-
dation. The most common substrate is 1 12 -oz copper-clad substrate which
corresponds to ∼ 18 µm metal thickness.

3.4 A distributed circuit analysis

TEM propagation is characterised by unique voltage and current waves as
well as a field distribution similar to the static case. The transmission line
can hence be regarded as a distributed electrical network.
Energy storage in the electric and magnetic fields is accounted for by
a shunt capacitance C ′ and a series inductance L′ per unit length of line
respectively. Power loss due to a lossy dielectric is accounted for by a shunt
conductance G′ per unit length of line while the losses due to finite con-
ductivity of the conductors is accounted for by a series resistance R′ per
 3.4 A distributed circuit analysis 39

FIGURE 3.13. A short transmission line model

unit length of line. For the moment, we satisfy ourselves with the following
expressions for the transmission line parameters

µ −
→ − →
L′ = H · H ∗ dS
II ∗
S

derived by equating
1 ∗ ′
II L = Um
4

ǫ′ −
→ − →
C′ = E · E ∗ dS
VV∗
S

derived by equating
1
V V ∗ C ′ = Ue
4

′ Rm −
→ − →
R = ∗ H · H ∗ dl
II
S1 +S2

and

ωǫ′′ −
→ − →
G′ = E · E ∗ dS
VV∗
S

For the analysis, we consider a short section of line dz as shown in Fig.

3.13. In terms of the line parameters, the equivalent circuit is shown in Fig.
3.14.
Let the current and voltage at the sending end be i (z, t) and v (z, t)
respectively and at the receiving end be i (z + ∆z, t) and v (z + ∆z, t) re-
spectively. Then from Kirchhof’s laws, we can write

∂
v (z, t) − v (z + ∆z, t) = i (z, t) R′ ∆z + L′ ∆z i (z, t)
∂t
40 3. Theory and applications of multi-conductor transmission lines

L’∆z R’∆z
i(z,t) i(z+∆z,t)

v(z,t) C’∆z G’∆z v(z+∆z,t)

FIGURE 3.14. Model of a short section of a typical transmission line

We use Taylor series expansion for small ∆z to write

∂
v (z + ∆z, t) ∼ v (z, t) + ∆z v (z, t)
∂z
Hence
∂ ∂
−∆z v (z, t) = i (z, t) R′ ∆z + L′ ∆z i (z, t)
∂z ∂t
or
∂ ∂
v (z, t) = −i (z, t) R′ − L′ i (z, t)
∂z ∂t
Similarly
∂
i (z, t) − i (z + ∆z, t) = v (z + ∆z, t) G′ ∆z + C ′ ∆z v (z + ∆z, t)
∂t
or



∂ ∂ ∂ ∂
−∆z i (z, t) = v (z, t) + ∆z v (z, t) G′ ∆z + C ′ ∆z v (z, t) + ∆z v (z, t)
∂ ∂z ∂t ∂z
∂
= v (z, t) G′ ∆z + C ′ ∆z v (z, t)
∂t
since ∆z is small. Thus
∂ ∂
i (z, t) = −v (z, t) G′ − C ′ v (z, t)
∂z ∂t
We differentiate the first equation w.r.t. z and the second one w.r.t. t to
obtain
∂2 ′ ∂ ′ ∂
2
v (z, t) = −R i (z, t) − L i (z, t)
∂z 2 ∂z ∂z ∂t
∂2 ∂ ∂2
i (z, t) = −G′ v (z, t) − C ′ 2 v (z, t)
∂t ∂z ∂t ∂t
 3.4 A distributed circuit analysis 41

or



∂2 ′ ′ ′ ∂ ′ ′ ∂ ∂2
v (z, t) = −R −v (z, t) G − C v (z, t) −L −G v (z, t) − C v (z, t)
∂z 2 ∂t ∂t ∂t2
or
∂2 ∂ ∂2
2
v (z, t) = R′ G′ v (z, t) + (R′ C ′ + L′ G′ ) v (z, t) + L′ C ′ 2 v (z, t)
∂z ∂t ∂t
or
∂2 ′ ′ ′ ′ ′ ′ ∂ ′ ′ ∂
2
v (z, t) − R G v (z, t) − (R C + L G ) v (z, t) − L C v (z, t) = 0
∂z 2 ∂t ∂t2
For harmonic time-dependent voltage and current, we can write
 2 
γ − R′ G′ − jω (R′ C ′ + L′ G′ ) + ω2 L′ C ′ V (z) = 0

This is in the form of a wave equation in terms of the voltage at z down

the line and is referred to as the telegrapher’s equation for a voltage wave.
The solution of this equation is a propagating voltage wave of the form

v (z, t) = Re [V (z) exp (jωt)]

where
V (z) = V + exp (−γz) + V − exp (γz)
is the line voltage in phasor form at z down the line and

γ = jω (R′ C ′ + L′ G′ ) − ω 2 L′ C ′ + R′ G′



′
√ 1 R′ G′ 1 R′ G
′
= jω L C 1 + ′ + ′ − 2
jω L ′ C ω L′ C′

If R′ and G′ are small as would often be the case,



√ 1 R′ G′
′ ′
γ ∼ jω L C 1 + + ′
jω L′ C
or 
′ 
√ 1 R G′
γ ∼ jω L′ C ′ 1 + +
j2ω L′ C′
by binomial expansion. Hence


L′ C ′ R ′ G′ √
γ = α + jβ = + ′ + jω L′ C ′
2 L′ C
The line current phasor at z is given by
γ  + 
I (z) = I + exp (−γz)+I − exp (γz) = V exp (−γz) − V − exp (γz)
R′ + jωL′
42 3. Theory and applications of multi-conductor transmission lines

In the forward going and reflected waves, I + , V + and I − , V − are the inci-
dent and reflected quantities respectively.
The characteristic impedance of the line is given by
  
V+ V− R′ + jωL′ R′ + jωL′ L′ 1 + (R′ /L′ ) /jω
Zc = + = − − = = =
I I γ G′ + jωC ′ C ′ 1 + (G′ /C ′ ) /jω
or  
′ 
L′ 1 R G′
Zc ∼ 1+ − ′
C′ j2ω L′ C
In the loss-free case √
γ = jω L′ C ′
and 
L′
Zc = Z0 =
C′

3.5 Propagation characteristics on an infinitely

long transmission line
On an infinitely long line, the −z-directed waves must vanish. Hence

V (z) = V + exp (−γz)

I (z) = I + exp (−γz)

In this case, the voltage-current ratio is independent of position on the line
and is called the characteristic impedance of the line,

R′ + jωL′ γ R′ + jωL′
Zc = = ′ =
γ G + jωC ′ G′ + jωC ′

Example: Demonstrate the analogy between wave characteristics on an

infinitely long transmission line and uniform plane waves in a lossy and
unbounded medium.
In a lossy medium, ǫ = ǫ′ −jǫ′′ and µ = µ′ −jµ′′ . Maxwell’s curl equations
are then modified as
−
→ − → −
→
∇ × E = −jω (µ′ − jµ′′ ) H
−
→ − → −
→
∇ × H = jω (ǫ′ − jǫ′′ ) E
−
→
We assume E = x

Ex (z), then
d
− Ex (z) = (ωµ′′ + jωµ′ ) Hy (z)
dz
 3.5 Propagation characteristics on an infinitely long transmission line 43

d
− Hy (z) = (ωǫ′′ + jωǫ′ ) Ex (z)
dz
Comparing with
d
− V (z) = (R′ + jωL′ ) I (z)
dz
and
d
− I (z) = (G′ + jωC ′ ) V (z)
dz
Then the equations governing Ex (z), Hy (z) and V (z), I (z), are similar
in form. We combine to obtain
d2
Ex (z) = γ 2 Ex (z)
dz 2
d2
Hy (z) = γ 2 Hy (z)
dz 2
with 
γ = α + jβ = (ωµ′′ + jωµ′ ) (ωǫ′′ + jωǫ′ )
and the intrinsic impedance of the lossy medium for a wave propagating in
the +z-direction is 
µ′′ + jµ′
η=
ǫ′′ + jǫ′

3.5.1 Limiting cases

(a) Lossless line (R′ = 0, G′ = 0). Hence
√
γ = α + jβ = jω L′ C ′
From which we deduce that
√
α = 0, β = ω L′ C ′
ω 1
vp = =√
β L′ C ′

L′
Z0 =
C′
(b) Low-loss line (R << ωL′ , G′ << ωC ′ )



√ R′ G′
γ = α + jβ = jω L C ′ ′ 1+ 1+
jωL′ jωC ′



√ R′ G′
∼ jω L′ C ′ 1 + ′
1+
j2ωL j2ωC ′

′ 
√ 1 R G′
∼ jω L′ C ′ 1 + +
j2ω L′ C′
44 3. Theory and applications of multi-conductor transmission lines

Hence
   
 
1 ′ C′ L′ 1 R′ L′ √
α∼ R + G′ = ′
+ G Z0 , Z0 = , β = ω L′ C ′
2 L′ C′ 2 Z0 C′

ω 1
=√ vp ∼
β L′ C ′
 

−1/2  ′ 
′ 
L′ R′ G′ L 1 R G′
Zc = 1+ 1+ ∼ 1+ − ′
C′ jωL′ jωC ′ C′ j2ω L′ C
or 
L′
Zc ∼ = Z0
C′
since the frequency dependent imaginary part is negligible when R′ <<
ωL′ and G′ << ωC ′ .
(c) Distortionless line (R′ /L′ = G′ /C ′ )

′ ′   ′
RC C
′
γ = α + jβ = (R + jωL ) ′ + jωC = ′ (R′ + jωL′ )
L′ L′

′ C′ √
α=R ′
, β = ω L′ C ′
L
ω 1
vp ∼ = √
β L′ C ′
 
R′ + jωL′ L′
Zc = ′ ′ ′ ′
= = Z0
R C /L + jωC C′

In general, a lossy line is dispersive.

Example: It is found that the attenuation on a 50 Ω distortionless line is
0, 01 dBm−1 . The line has a capacitance of 0, 1 nF m−1

(a) Determine the resistance, inductance and conductance per unit length
of the line
(b) Find the velocity of wave propagation and the velocity factor
(c) Determine the percentage to which the amplitude of a travelling voltage
wave decreases in 1 km and 5 km
(a) For a distortionless line,
R′ G′
′
= ′
L C
Thus 
L′
Z0 = = 50 Ω
C′
 3.5 Propagation characteristics on an infinitely long transmission line 45

Note: X dB = 20xNp log10 e = 8, 686xNp .


′ C′ 0, 01
α=R = 0, 01 dBm−1 = N pm−1 = 1, 15 × 10−3 N pm−1
L′ 8, 69

Thus
R′ = αZ0 = 0, 057 Ωm−1
L′ = C ′ Z02 = 0, 25 µHm−1
R′ C ′
G′ = = 22, 8 µSm−1
L′
(b)

1
vp = √ = 2 × 108 ms−1
L′ C ′
and
∆ v 2 × 108 ms−1
velocity factor = = = 0, 7
c 3 × 108 ms−1

(c)

V2
= exp (−αz) = 31, 7% @ 1 km
V1
= 0, 32% @ 5 km

3.5.2 Attenuation estimation from power relations

We note that for a travelling wave on a transmission line, the attenuation
constant is
α = Re [γ] = Re [(R′ + jωL′ ) (G′ + jωC ′ )]
This can also be obtained from a power relation where the voltage and cur-
rent on an infinitely long transmission line have the phasor representation,

V (z) = V + exp {− (α + jβ) z} = V + exp (−αz) exp (−jβz)

V+ V+
I (z) = exp {− (α + jβ) z} = exp (−αz) exp (−jβz)
Z0 Z0
The time-average power propagated down the line across the plane at z is
given by

1 1 V +2
P (z) = Re [V (z) I ∗ (z)] = Re [Z0 ] exp (−2αz)
2 2 |Z0 |2
46 3. Theory and applications of multi-conductor transmission lines

From the principle of energy conservation, the rate of decrease of P (z)

with distance down the line must be equal to the time-average power loss
Pl per unit length at z. Hence
∂
− P (z) = Pl (z) = 2αP (z)
∂z
Example:

(a) Find the attenuation constant of a lossy transmission line with R′ , L′ ,

G′ , C ′ using the power relation
(b) Specialise the results in (a) to obtain the attenuation constant for a
low-loss and a distortionless line

(a) For a lossy transmission line, the time-average power loss per unit
length is

1 2  V02  ′ 
Pl (z) = I (z) R′ + |V (z)|2 G′ = 2 R + G′ |Z0 |2 exp (−2αz)
2 2 |Z0 |

Thus
1  ′  
α= R + G′ |Z0 |2 , N pm−1
2Z0

(b) For a low-loss line, Z0 ∼ L′ /C ′

    
1 R′ ′ 1 ′ C′ ′ L′
α∼ + G Z0 = R ′
+G
2 Z0 2 L C′

and for a distortionless line, Z0 = L′ /C ′ , R′ /L′ = G′ /C ′ so that

′ C′
α=R
L′

3.6 Propagation characteristics on finite-length

transmission lines
The general solutions to the Helmholtz equations governing the voltage and
current waves on a transmission line are of the form

V (z) = V + exp (−γz) + V − exp (γz)

I (z) = I + exp (−γz) + I − exp (γz)

where γ is in general complex and both terms of the solutions are present.
 3.6 Propagation characteristics on finite-length transmission lines 47

Zg
Ii IL

Vg Vi Zi ZL
VL
γ,Z0

-l -z 0

FIGURE 3.15. A terminated finite-length transmission line

The reflected waves V − (z) exp (γz) and I − (z) exp (γz) vanish when the
line is terminated in its characteristic impedance. Consider a load ZL con-
nected to a sinusoidal generator Vg of source impedance Zg through a
finite-length line of length l as depicted in Fig. 3.15. For convenience, we
assume the load to be at z = 0 so that the source is at z = −l. Thus, at
the load, we can write
V+ V−
V (0) = VL = V + + V − , I (0) = IL = I + + I − = −
Z0 Z0
where the characteristic impedance of the line Z0 is defined as
V+ V−
Z0 = +
=− −
I I
By Ohm’s law,
VL = IL ZL
Thus, we can eliminate the wave amplitudes by writing
VL = IL ZL = V + + V −
IL Z0 = V + − V −
from which we see that
IL IL
V+ = (ZL + Z0 ) , V − = (ZL − Z0 )
2 2

At any distance z from the load, we can then write

IL IL
V (z) = (ZL + Z0 ) exp (−γz) + (ZL − Z0 ) exp (γz)
2 2
= IL ZL cosh γz − IL Z0 sinh γz

IL IL
I (z) = (ZL + Z0 ) exp (−γz) − (ZL − Z0 ) exp (γz)
2Z0 2Z0
IL IL
= Z0 cosh γz − ZL sinh γz
Z0 Z0
48 3. Theory and applications of multi-conductor transmission lines

Zg Ii

Vg Vi Zi

FIGURE 3.16. Equivalent circuit for finite-length line at the source-end

and the ratio

V (z) ZL cosh γz − Z0 sinh γz ZL − Z0 tanh γz
Z (z) = = Z0 = Z0
I (z) Z0 cosh γz − ZL sinh γz Z0 − ZL tanh γz
is the impedance seen looking towards the load at a distance z from the
load. Hence at the generator end (source-end) of the line, i.e. at z = −l,
V (−l) Vi ZL + Z0 tanh γl
Zi = Z (−l) = = = Z0
I (−l) Ii Z0 + ZL tanh γl
This is the impedance the generator would see looking towards the load.
Thus, as far as the generator is concerned, the terminated finite-length line
can be replaced by this impedance as shown in Fig. 3.16. The input voltage
and current can then be computed as
Zi
Vi = Vg
Zi + Zg
Vg
Ii =
Zi + Zg

The time-average power delivered by the generator at the input terminals

of the line is given by

∗ 
1 ∗ 1 Zi Vg Vg
Pin,ave = Re [Vi Ii ] = Re
2 2 Zi + Zg Zi + Zg
2
Ri Vg 1 2
= 2 = 2 |Ii | Ri
2 |Zi + Zg |
and that delivered to the load is
 2
1 1  VL  1
PL,ave = Re [VL IL∗ ] =   RL = |IL |2 RL
2 2 ZL 2
If the line is lossless, then PL,ave = Pin,ave . If ZL = Z0 , the impedance
seen by the generator at any distance z from the load is Z0 . In this case,
the current and voltage at any point is given by
V (z) = IL Z0 exp (−γz) = VL exp (−γz)
 3.6 Propagation characteristics on finite-length transmission lines 49

I (z) = IL exp (−γz)

and no reflections then occur on the line. Thus, when a finite length loss-
less transmission line is terminated in its own characteristic impedance,
the voltage and current distribution on the line is as though the line were
infinitely long. In this case, the line is said to be matched.
Example: A signal generator  with a source
resistance of 1 Ω and an open-
circuit voltage vg = 0, 3 cos 2π × 108 t V is connected to a 50 Ω lossless
transmission line. If the line is 4 m long and the measured velocity of the
wave on the line is 2, 5 × 108 ms−1 . then for a matched load, determine
(a) The instantaneous voltage and current at an arbitrary location on the
line
(b) The expression of the instantaneous voltage and current at the load
(c) The average power transmitted to the load
(a) We have
VG = 0, 3∠0◦ , Zg = Rg = 1 Ω, Z0 = 50 Ω, ω = 2π×108 rads−1 , vp = 2, 5×108 ms−1 , l = 4 m
Since the line is matched, Zi = Z0 = 50 Ω. The voltage and current
at the input are thus
Zi
Vi = Vg = 0, 294∠0◦ V
Zi + Zg
VG
Ii = = 5, 9∠0◦ mA
Zi + Zg
Since only the forward wave exists on the lossless line, we have
ω
γ = jβ = j = j0, 8π radm−1
vp
and
V (z) = 0, 294 exp (−j0, 8πz) V
I (z) = 5, 9 exp (−j0, 8πz) mA
so that

v (z, t) = Re [0, 294 exp {j (ωt − 0, 8πz)}] = 0, 294 cos π 2 × 108 t − 0, 8z V

i (z, t) = Re [5, 9 exp {j (ωt − 0, 8πz)}] = 5, 9 cos π 2 × 108 t − 0, 8z mV
(b) At the load, z = l = 4 m. Hence

v (z, t) = 0, 294 cos π 2 × 108 t − 3, 2 V

i (z, t) = 5, 9 cos π 2 × 108 t − 3, 2 mV
(c)
1
Pin,ave = Po,ave = Re [V (z) I ∗ (z)] = 0, 87 mW
2
50 3. Theory and applications of multi-conductor transmission lines

Xi,oc

inductive

λ/2 3λ/4 λ
l
0 λ/4 5λ/4

capacitive

FIGURE 3.17. Input reactance of an open-circuited short transmission line

3.6.1 Finite-length transmission line as a circuit element

At ultrahigh frequencies (300 M Hz-3 GHz), lumped-element components
are difficult to fabricate as effects of stray fields become important. Sec-
tions of transmission lines can be used to simulate capacitance and in-
ductance and may be used to match arbitrary terminating loads to the
source impedance of the generator for maximum power transfer operation.
In most practical cases, short sections of line can be considered lossless with
γl ∼ jβl, Z0 is real and tanh (γl) ∼ j tan βl. Hence the impedance seen at
the input of the line section looking toward the load becomes

ZL + jZ0 tan βl
Zi = Z0
Z0 + jZL tan βl

An open-circuited short line

With ZL = ∞, we have

Z0
Zi,oc = = −jZ0 cot βl
j tan βl

Thus, depending on the sign of cot βl, Zi,oc can have either inductive or
capacitive characteristics. If βl is small, tan βl ∼ βl and

Z0 1 L′ 1
Zi,oc ∼ = √ ′
=
jβl jω L′ C ′ l C jωC ′ l

i.e. a capacitive impedance. In practice, ZL can only be made large but

remains finite especially at high frequencies due to coupling to nearby ob-
jects and radiation at the open end. Fig. 3.17 illustrates the variation of
the input impedance of an open-circuited short line section.
 3.6 Propagation characteristics on finite-length transmission lines 51

Xi,sc

inductive

λ/4 λ 5λ/4
l
0 λ/2 3λ/4

capacitive

FIGURE 3.18. Input impedance of a short-circuited short transmission line

A short-circuited short line

With ZL = 0, we have
Zi,sc = jZ0 tan βl

which for very short lines can be approximated as


√ L′
Zi,sc ∼ jZ0 βl = jω L′ C ′ l = jωL′ l
C′

which is an inductive impedance. The input reactances of open- or short-

circuited transmission lines are the same if their lengths differ by (2n + 1) λ/4, n =
0, 1, 2, . . .. Fig. 3.18 illustrates the variation of the input impedance of a
short-circuited short line section.

A quarter-wavelength line
If l is such that
λ
l = (2n + 1)
4
then
2π λ π
βl = (2n + 1) = (2n + 1)
λ 4 2
and
π
tan βl = tan (2n + 1) = ±∞
2
In this case,
R20
Zi =
ZL
52 3. Theory and applications of multi-conductor transmission lines

A half-wavelength line
For l = nλ/2, n = 1, 2, . . .,
2π nλ
βl = = nπ
λ 2
and
tan βl = 0
so that
Zi = ZL
We can determine Z0 and γ of a given line by measuring the input
impedance of a section of the line with open- and short-circuit terminations.
For ZL → ∞, Zi,oc = Z0 coth γl and for ZL = 0, Zi,sc = Z0 tanh γl. Thus

Z0 = Zi,oc Zi,sc
and 
1 Zi,sc
γ = tanh−1
l Zi,oc
These expressions hold regardless of whether the line is lossy or not.
Example: The open- and short-circuit impedances measured at the input
terminals of a lossless transmission line of length 1, 5 m less than λ/4 are
−j54, 6 Ω and j103 Ω resp.
(a) Determine Z0 and γ of the line
(b) Without changing the operating frequency, determine the input im-
pedance of a short-circuited section of this line that is twice as long
as the one given.
(c) How long should the short-circuited line be in order for it to appear as
an open circuit at the input terminals at this frequency?
(a)  
Z0 = Zi,oc Zi,sc = −j54, 6 × j103 = 75 Ω
and
 
1 Zi,sc 1 j103
γ = tanh−1 = tanh−1 = j0, 628 m−1
l Zi,oc 1, 5 −j54, 6

(b)
γl = j0, 628 × 2 × 1, 5 m = j1, 884 rad
Zi,sc = 75 tanh (j1, 884) Ω = j75 tan 1, 884 = −j231 Ω
Thus
λ
1, 5 m < < 3, 0 m
4
 3.6 Propagation characteristics on finite-length transmission lines 53

(c) For a short-circuited line to appear open at the input, it should be an

odd multiple of λ/4 long. We have
2π 2π
λ= = = 10 m
β 0, 628 m−1
Hence
λ λ
l= + (n − 1) = 2, 5 + 5 (n − 1) , n = 1, 2, . . .
4 2

When a lossy line is terminated in a short-circuit, the input impedance

seen at the input will not vanish when l = nλ/2. In this case, we have
sinh (α + jβ) l sinh αl cos βl + j sin βl cosh αl
Zi,sc = Z0 tanh γl = Z0 = Z0
cosh (α + jβ) l cosh αl cos βl + j sinh αl sin βl
For l = nλ/2, βl = nπ, sin βl = 0 and
sinh αl cos βl
Zi,sc = Z0 = Z0 tanh αl ∼ αlZ0
cosh αl cos βl
for αl small. We therefore have a series resonant circuit when l = nλ/2.
Similarly, when l = (2n + 1) λ/4, Zi,sc will not be infinite. With l = nλ/4,
n odd, cos βl = 0 and
sin βl cosh αl Z0 Z0
Zi,sc = Z0 = ∼
sinh αl sin βl tanh αl αl
for small αl which is a parallel resonant circuit.
The quality factor of the circuit is determined by first determining its
half-power bandwidth. We let f = f0 + ∆f /2 where ∆f << f0 is the shift
from the resonance frequency in one direction. Thus


2πf 2π (f0 + ∆f/2) nπ nπ ∆f
βl = l= l= + , n odd
vp vp 2 2 2f0



nπ ∆f nπ ∆f
cos βl = − sin ∼−
2 2f0 2 2f0


nπ ∆f
sin βl = cos ∼1
2 2f0
Thus

−1
2 
−2
nπ ∆f 2 |Z0 |  nπ ∆f 
Zi,sc = Z0 αl + j ⇒ |Zi,sc | =  1 + j
2 2f0 αl 2αl 2f0 
or
2
|Z0 |
|Zi,sc (f0 )|2 = |Zi,sc,max |2 =
αl
54 3. Theory and applications of multi-conductor transmission lines

and  −2
|Zi,sc |2  nπ 2
∆f 2
= 1+
|Zi,sc,max |2 2αl 2f0
Therefore, the half-power frequencies f1 and f2 , where f2 > f1 are where
this ratio is 1/2. i.e.,
nπ ∆f β ∆f
= = 1, n odd
2αl 2f0 2α f0
The Q of the parallel-resonant circuit is hence,
f0 β
Q= =
∆f 2α
and assuming low loss,
 ′   
ωL′ R G′ 1 ′ L′ ′
Q= = + = R + G
R′ + G′ L′ /C ′ ωL′ ωC ′ ωL′ C′
For a well insulated line, G′ L′ /C ′ << R′ . Hence
ωL′
Q∼
R′
Example: The measured attenuation of an air-dielectric coaxial line at
400 M Hz is 0, 01 dBm−1 . Determine the Q and half-power bandwidth of
a quarter-wavelength section of this line terminated in a short-circuit.
At f = 4 × 108 Hz, λ = 0, 75 m, β = 8, 38 radm−1
0, 01
α =, 01 dBm−1 = N pm−1
8, 69
Thus
β 8, 38 × 8, 69
Q= = = 3641
2α 2 × 0, 01
f0
∆f = = 0, 11 × 106 Hz
Q

3.6.2 Lines terminated in resistive loads

Recall that the voltage at a point z from the load is given by
IL IL
V (z) = (ZL + Z0 ) exp (−γz) + (ZL − Z0 ) exp (γz)
2  2
IL ZL − Z0
= (ZL + Z0 ) exp (−γz) 1 + exp (2γz)
2 ZL + Z0
IL
= (ZL + Z0 ) exp (−γz) {1 + ΓL exp (2γz)}
2
 3.6 Propagation characteristics on finite-length transmission lines 55

where
ZL − Z0
ΓL = = |ΓL | exp (jθ Γ )
ZL + Z0
is the voltage reflection coefficient of the load. Similarly, for the current
wave,
IL IL
I (z) = (ZL + Z0 ) exp (−γz) − (ZL − Z0 ) exp (γz)
2Z0 2Z0

IL ZL − Z0
= (ZL + Z0 ) exp (−γz) 1 − exp (2γz)
2Z0 ZL + Z0
IL
= (ZL + Z0 ) exp (−γz) {1 − ΓL exp (2γz)}
2Z0
Note that the current reflection coefficient defined as
∆ I− V−
ΓI = = −
I+ V+
is equal to the negative of the voltage reflection coefficient. Thus
IL
V (−z) = (ZL + Z0 ) exp (γz) {1 + ΓL exp (−2γz)}
2
and
IL
I (−z) = (ZL + Z0 ) exp (γz) {1 − ΓL exp (−2γz)}
2Z0
For a lossless transmission line, γ = jβ. Hence
IL
V (−z) = (ZL + Z0 ) exp (jβz) [1 + |ΓL | exp {−j (2βz − θ Γ )}]
2
IL
I (−z) = (ZL + Z0 ) exp (jβz) [1 − |ΓL | exp {−j (2βz − θΓ )}]
2Z0
The line current and voltage phasors are obtained as

V (−z) = VL cos βz − jIL Z0 sin βz

VL
I (−z) = IL cos βz − j sin βz
Z0
where VL = IL ZL .
If ZL = RL ,


2
Z0
|V (−z)| = VL cos2 βz + sin2 βz
RL


2
RL
|I (−z)| = IL cos2 βz + sin2 βz
Z0
56 3. Theory and applications of multi-conductor transmission lines

which are standing wave patterns with minima and maxima located at fixed
points along the line. The voltage standing wave ratio (VSWR) is defined
as,
∆ |Vmax | 1 + |ΓL |
S= =
|Vmin | 1 − |ΓL |
|Vmax | and |Imin | occur together at z = −zM such that

θ Γ − 2βzM = 2nπ, n = 0, 1, 2, . . .

while |Vmin | and |Imax | occur together at z = −zm such that

θΓ − 2βzm = (2n + 1) π, n = 0, 1, 2, . . .

Example: The standing wave ratio on a transmission line is an easily

measured quantity.

(a) Show how the value of the terminating resistance on a lossless line of
known characteristic impedance Z0 can be determined by measuring
V SW R.

(b) What is the impedance of the line looking towards the load at a distance
λ/4 where λ is the operating wavelength?

(a) If there are voltage maxima at z = 0, −λ/2, −λ, . . . then RL < Z0 .

If RL > Z0 , θΓ = 0 and |Vmax |, |Imin | occur at βz = 0 and |Vmin |,
|Imax | at βz = −π/2. Thus

Z0
|Vmax | = VL , |Vmin | = VL
RL

RL
|Imin | = IL , |Imax | = IL
Z0
Thus
|Vmax | |Imax | RL
= =S=
|Vmin | |Imin | Z0
or
RL = SZ0
If RL < Z0 , θ Γ = π and |Vmin |, |Imax | occur at βz = 0 and |Vmax |,
|Imin | at βz = −π/2.

Z0
|Vmin | = VL , |Vmax | = VL
RL

RL
|Imax | = IL , |Imin | = IL
Z0
 3.6 Propagation characteristics on finite-length transmission lines 57

Thus
|Vmax | |Imax | Z0
= =S=
|Vmin | |Imin | RL
or
Z0
RL =
S
(b) λ is determined from twice the distance between to neighbouring voltage
(or current) maxima or minima. At z = −λ/4, βz = −π/2, cos βz =
0, sin βz = −1. Thus



λ λ VL
V − = jIL Z0 , I − =j
4 4 Z0


λ V (−λ/4) Z2
Z − = Zi = = 0
4 I (−λ/4) RL

3.6.3 Lines terminated in arbitrary loads

A voltage minimum or maximum cannot occur at the termination when ZL
is complex because θΓ = 0 or π. Let ZL = RL +jXL . If the standing wave is
extended by an extra distance lm beyond the load, the wave will eventually
reach a minimum. The voltage minimum is hence where it should be had
the termination been replaced by a section of length lm terminated in a
resistance Rm < Z0 as shown in Fig. 3.19. This does not change the voltage
distribution to the left of the actual termination where z < 0.
Using Rm in place of ZL , and lm in place of l, we have
Rm + jZ0 tan βlm
Ri + jXi = Z0
Z0 + jRm tan βlm
and the real and imaginary parts yield values of Rm and lm . Experimentally,
ZL is obtained by measuring the standing wave ratio and the distance
z = −zm of the closest extremum of the voltage wave to the load, i.e.
λ
−zm + lm =
2
Thus,

(i) Determine |ΓL | from the V SW R as

S−1
|ΓL | =
S+1

(ii) Determine θ Γ from zm as

θ Γ = π − 2βzm , n = 0
58 3. Theory and applications of multi-conductor transmission lines

Z0 ZL

z’m z’=0
lm

Z0 Rm

z’m z’=0
λ/2

FIGURE 3.19. Finite-length transmission line with an arbitrary load termination

 3.6 Propagation characteristics on finite-length transmission lines 59

(iii) Determine ZL as
1 + |ΓL | exp (jθΓ )
ZL = RL + jXL = Z0
1 − |ΓL | exp (jθΓ )

(iv) The distance of a voltage maximum to the termination, z = −zM

could also be used for which

θΓ − 2βzM = −2nπ, n = 0

Example: The standing wave ratio of a lossless 50 Ω line terminated

in an unknown load impedance is found to be 3, 0. The distance between
successive voltage minima is 20 cm and the first minimum occurs at 5 cm
from the load. Determine

(a) The reflection coefficient ΓL

(b) The load impedance ZL
(c) The equivalent length and terminating resistance of line such that the
input impedance is ZL
(a) The distance between successive voltage minima is λ/2. Thus

λ = 2 × 0, 2 m = 0, 4 m
2π
β= = 5π rads−1
λ
S−1
|ΓL | = = 0, 5
S+1
3
θ Γ = π − 2βzm = π rad
2
ΓL = |ΓL | exp (jθΓ ) = −j0, 5

(b)
1 − j0, 5
ZL = 50 = 30 − j40 Ω
1 + j0, 5
(c)
Rm + j50 tan βlm
30 − j40 = 50
50 + jRm tan βlm
Equate real and imaginary parts or recall
λ Z0
|zm | + lm = , Rm =
2 S
Hence
λ
lm = − |zm | = 0, 15 m
2
60 3. Theory and applications of multi-conductor transmission lines

50
Rm = = 16, 7 Ω
3
Note that
′ λ
lm = lm − = 0, 05 m
4
and
′
Rm = Z0 S = 150 Ω
are also valid when we use the locations of the maxima instead.

3.6.4 Transmission line circuits

We seek to express the voltages on a line terminated in ZL in terms of the
source characteristics Vg , Zg and line characteristics γ, Z0 , l and the load
impedance ZL . At z = −l, we have
IL
Vi = Vg − Ii Zg = (ZL + Z0 ) exp (γl) {1 + ΓL exp (−2γl)}
2
and
IL
Ii = (ZL + Z0 ) exp (γl) {1 − ΓL exp (−2γl)}
2Z0
Thus
IL Z0 Vg 1
(ZL + Z0 ) exp (γl) =
2 Z0 + Zg 1 − Γg ΓL exp (−2γl)
where
Zg − Z0
Γg =
Zg + Z0
is the voltage reflection coefficient at the generator end of the line. Thus,
at any point z from the load, we have

Z0 Vg 1 + ΓL exp (2γz)
V (z) = exp {−γ (z + l)}
Z0 + Zg 1 − Γg ΓL exp (−2γl)

and
Vg 1 − ΓL exp (2γz)
I (z) = exp {−γ (z + l)}
Z0 + Zg 1 − Γg ΓL exp (−2γl)
Example: A 100 M Hz sinusoidal generator with Vg = 10∠0◦ and source
resistance 50 Ω is connected to a lossless 50 Ω air line 3, 6 m long and
terminated in a 25 + j25 Ω load. Determine

(a) V (z) at the location z from the generator

(b) Vi at the input terminals of the line and VL at the load

(c) The VSWR on the line

 3.6 Propagation characteristics on finite-length transmission lines 61

(d) The average power delivered to the load

(a) We have

Vg = 10∠0◦ , Zg = 50 Ω, f = 108 Hz, Z0 = 50 Ω, ZL = 25+j25 Ω = 35, 36∠45◦ , l = 3, 6 m

Thus
2π 2πf 2π
β= = = radm−1 ⇒ βl = 2, 4π rad
λ c 3
ZL − Z0
ΓL = = 0, 447∠116, 6◦ , Γg = 0
ZL + Z0
At −z from the load, we have
Z0 Vg
V (z) = (1 + ΓL exp (−j2βz)) exp {−jβ (l − z)}
Z0 + Zg
Thus, measured from the generator, this will be the point z − l and
we therefore have
Z0 Vg
V (z − l) = [1 + ΓL exp {j2β (z − l)}] exp {jβ (z − l)}
Z0 + Zg
where z ≥ 0 and measured from the generator.
(b)
Z0 Vg
Vi = V (0) = {1 + ΓL }
Z0 + Zg
Z0 Vg
= {1 + 0, 447 exp (−j4, 8π + j0, 6478π)} exp (−j2, 4π)
Z0 + Zg
Z0 Vg
= {1 + 0, 447∠ − 27, 4◦ } exp (−j2, 4π)
Z0 + Zg
= 5 (1 + 0, 3969 − j0, 2057) exp (−j2, 4π) V
= 7, 06∠ − 80, 4◦ V

and
Z0 Vg
VL = V (l) = [1 + ΓL ]
Z0 + Zg
= 5 {1 + 0, 447∠116, 6◦ } V
= 5 {1 − 0, 2001 + j0, 3997}
= 4, 47∠26, 6◦ V

(c)
1 + |Γ|
S= = 2, 62
1 − |Γ|
62 3. Theory and applications of multi-conductor transmission lines

(d)
 2
1  VL 
Pave =  RL = 0, 2 W
2 ZL 
If ZL = 50 Ω, Γ = 0 and |VL | = |Vi | = Vg /2. Maximum power would
thus be delivered to the load and is called the available power. This
has the value
V2
Pave,max = L = 0, 25 W
2RL
The reflected power is hence
2
|Γ| Pave,max = 0, 05 W

3.7 Transients on transmission lines

The concept of reactance, wavelength, wavenumber or phase constant lose
meaning under transient conditions or non-sinusoidal line operation. Impor-
tant practical situations where signals and sources are not time-harmonic
and the conditions not steady-state include,
• Line excitation with digital signals in computer networks
• Sudden surges in power on telephone or power lines
We consider lossless lines and start with the simplest situation where a
DC source is connected to a line terminated in its characteristic impedance
at t = 0 as shown in Fig. 3.20. The impedance looking into the line from
the source-end is hence Z0 and a voltage wave of magnitude
Z0
V1+ = VDC
Z0 + Rg
propagates
√ down the line in the +z-direction with phase velocity v =
1/ L′ C ′ and a corresponding current
V1 VDC
I1+ = =
Z0 Z0 + Rg
at z1 . The plot of V (z1 ) as a function of time is a delayed step function
at t = z1 /v and the current has a similar shape. When the voltage and
current waves arrive at the load end (z = 0), no reflections occur since the
line is matched. Steady-state is established with the entire line charged to
a voltage V1+ .
If both Rg and RL are different from Z0 , then when the switch is closed
at t = 0, the source-end will send a voltage wave of magnitude
Z0
V1+ = VDC
Z0 + Rg
 3.7 Transients on transmission lines 63

Zg t=0

VDC Z0 Z0

z1
z=-l z=0

FIGURE 3.20. DC excited terminated line

√
in the +z-direction which propagates at v = 1/ L′ C ′ proceeding as if the
line were infinitely long. At t = T = l/v, this wave arrives at the load end
and a reflected wave travelling in the −z-direction is established. This has
a magnitude
RL − Z0
V1− = ΓL V1+ , ΓL =
RL + Z0
The reflected wave arrives at the source end at t = 2T where it suffers
reflection again and a new wave propagating in the +z-direction established
with a magnitude
Rg − Z0
V2+ = Γg V1− = Γg ΓL V1+ , Γg =
Rg + Z0
This goes on indefinitely with reflection at both ends of the line at t =
nT, n = 1, 2, . . .. Some of the reflected waves may have negative amplitudes
since Γg or ΓL could be negative. Unless the termination is a short- or open-
circuit, Γg and ΓL are less than unity so that the reflections grow weaker
with time.
At any location z1 on the line from the source, the voltage and current
on the line at a given time interval are

V (z1 ) = V1+ + V1− + V2+ + V2− + . . .

and
I (z1 ) = I1+ + I1− + I2+ + I2− + . . .
The ultimate voltage across the load as time progresses will be
 
VL = V1+ 1 + ΓL + Γg ΓL + Γg Γ2L + . . .
  
= V1+ 1 + Γg ΓL + Γ2g Γ2L + . . . + ΓL 1 + Γg ΓL + Γ2g Γ2L + . . .

or
1 + ΓL
VL = V1+
1 − Γg ΓL
64 3. Theory and applications of multi-conductor transmission lines

For RL = 3Z0 , Rg = 2Z0 say, Γg = 1/3, ΓL = 1/2 and V1+ = VDC /3. Thus
VL → 95 V1+ = 35 VDC as t → ∞. Similarly

1 − ΓL V1+ 3 V1+ 1 VDC

IL = = =
1 − Γg ΓL Z0 5 Z0 5 Z0

3.7.1 Reflection or bounce diagrams

The reflection diagram is a useful tool in visualising the voltage and current
at a given time or the variation with time at an arbitrary point on the line.
Fig. 3.21 illustrates one such diagram. To obtain the voltage distribution at
t = t4 say, we note that 3T < t4 < 4T . We thus, mark t4 on the time scale
and extend a horizontal line to intersect the directed line below t = 4T at
P4 which has a value Γg Γ2L V1+ . A vertical line drawn from P4 to the z-axis
intersects it at z1 . Note that in the range 0 < z < z1 from the source,
the voltage has value V1+ + V1− + V2+ = V1+ (1 + ΓL + Γg Γ L ) and in the
range z1 < z < l, the value is V1+ 1 + ΓL + Γg ΓL + Γg Γ2L . The voltage
discontinuity is hence Γg Γ2L V1+ at z = z1 from the source. Hence, at t = t4 ,
V (z, t4 ) has the form shown in Fig. 3.22.
Now consider the line excited by a rectangular pulse of the form

vg (t) = V0 {u (t) − u (t − T0 )}

The transient response of the line is the superposition of the two step
functions.
Example: A rectangular pulse of an amplitude 15 V and duration 1 µs is
applied through a series resistance of 25 Ω to the inputs of a 50 Ω lossless
coaxial line. If the line is 400 m long and is terminated in a short-circuit
at the far end, determine the voltage response midway down the line as a
function of time up to 8 µs. The dielectric constant is 2, 25.
ΓL = −1, Γg = −1/3. We have
  
vg (t) = 15 u (t) − u t − 10−6

c
v = √ = 2, 0 × 108 ms−1
ǫr

l
T = = 2 µs
v
Z0
V1+ = 15 V = 10 V
Z0 + RG
Fig. 3.23 illustrates
 the solution
process. The values indicated in brackets
apply to the −15u t − 10−6 function applied 1 µs later.
 3.7 Transients on transmission lines 65

t5 P5
Γ2g Γ2L V+1

4T
P4
t4 ΓgΓ2L V+1

3T

t3 P3
ΓgΓLV+1

2T
P2
t2 ΓLV+1

T
V+1
P1
t1 slope=1/v

z1 z
0 l

FIGURE 3.21. Reflection diagram

66 3. Theory and applications of multi-conductor transmission lines

ΓgΓ2L V+1

t4 l

FIGURE 3.22. Transient voltage on a lossless transmission line

3.7.2 Transients on an initially charged line

Any disturbance or change in a transmission line circuit will start transients
along the line even without external excitation as long as initial voltages
or currents exist on the line.
Example: A lossless air-dielectric, open-circuited line of resistive charac-
teristic impedance Z0 and length l is initially charged to a voltage V0 . At
t = 0 the line is connected to a resistance R. Determine the voltage across
and the current through R as a function of time. Assume R = Z0 .
Fig. 3.24(a) illustrates the situation. Since the voltage across R is ini-
tially zero (no current flows through R), we can model the situation as
shown in Fig. 3.24(b) which can be redrawn as in Fig. 3.24(c) and (d).
The line is assumed to be uncharged with the initial charge represented by
a voltage source.
When the switch is closed, a wave of amplitude V1+ is sent down the line
in the +z-direction with an amplitude

Z0 V0
V1+ = − V0 = −
Z0 + R 2

At t = l/v, this wave arrives at the open-circuit having lowered the voltage
everywhere on the line to V0 /2. A reflected wave V1− is sent back to the
sending end where
V0
V1− = ΓV1+ = −
2
and arrives here at t = 2l/v and lowers the line voltage to zero.
The current through the resistor

IR = −I1
 3.7 Transients on transmission lines 67

9 +1/9

(+1/3)

8
-1/3

+1/3 6

5 (-1/3)

(+1)
4
-1

+1 2

1 (-1)

z
0 200 400

FIGURE 3.23. Reflection diagram solution

68 3. Theory and applications of multi-conductor transmission lines

Z0 t=0 t=0
IR I+1

R VR Z0 LINE

z=-l z=0 z=0

(a) OPEN-CIRCUITED LINE (b) EQUIVALENT CIRCUIT

t=0 t=0
I1

IR

R LINE R LINE

z=0 z=0
(c) VOLTAGES OPPOSE HENCE NO TRANSIENTS (d) INTERESTING PART

FIGURE 3.24. Charged open-circuited line and its equivalent circuit

where
V1+ V0
I1 = I1+ = =−
Z0 2Z0

in the time interval 0 ≤ t ≤ l/v. At t = l/v, I1+ reaches the open circuit
and the reflected current must change sign to make the total current zero
as required by the open circuit. Thus

V0
I1− = −I1+ =
2Z0

This arrives at the sending end at t = 2l/v reducing both I1 and IR to

zero. Since R = Z0 , no further reflection takes place and the transient
ends. This circuit is useful as a pulse generator with the pulse duration
adjustable through l.

3.7.3 Line terminated in a reactive load

Consider a line terminated in an inductance L as shown in Fig. 3.25. When
the switch is closed at t = 0, a voltage wave of an amplitude

V0
V1+ =
2
 3.7 Transients on transmission lines 69

Z0 t=0 Z0 t=0

iL(t) iL(t)

2V+1 Z0 2V+1 vL(t) L vL(t)

vL(t) L

z=-l z=0 z=0

INDUCTIVELY TERMINATED LINE EQUIVALENT CIRCUIT

FIGURE 3.25. Inductively terminated transmission line

travels towards the load arriving there at l = l/v = T . A reflected wave

V1− (t) is generated due to the mismatch. At z = l, we can write

v (t) = V1+ + V1− (t)

for all t ≥ T and the current through the inductor is given by

1  +
iL (t) = V − V1− (t)
Z0 1
and from circuit theory, the load voltage can be expressed as,
d
vL (t) = L iL (t)
dt
Hence
V1− (t) = vL (t) − V1+
and
1  + 
iL (t) = V1 − vL (t) + V1+
Z0
Thus
vL (t) = 2V1+ − Z0 iL (t)
At the load end, the equivalent circuit for t ≥ T is then of the form shown
in Fig. 3.25.
We thus have
d
L iL (t) + Z0 iL (t) = 2V1+ , t ≥ T
dt
or   
2V1+ Z0
iL (t) = 1 − exp − (t − T ) , t≥T
Z0 L
2V1+
iL (T ) = 0, iL (∞) =
Z0
70 3. Theory and applications of multi-conductor transmission lines

iL(t) vL(l,t)

2V+1 /Z0 2V+1

t t
T T

V1- (l,t)
v(z,t)
V+1
2V+1
t
T
V+1
-V+1
z
z1 l

FIGURE 3.26. Transient on an inductively terminated line

and the voltage across the load is hence

  
d + Z0
vL (t) = L iL (t) = 2V1 1 − exp − (t − T ) , t≥T
dt L
and
  
Z0 1
V1− (t) = vL (t) − V1+ = 2V1+ exp − (t − T ) − , t>T
L 2
At any z = z1 along the line, the voltage is V1+ before the reflection from
the termination reaches this point at
l − z1
t−T <
v
It then equals
V1+ + V1− (t − T )
thereafter. Fig. 3.26 illustrates the form of the voltage waveform.Note that
l − z1 z1
t1 − T = ⇒ 2T −
v v
For a capacitive termination as shown in Fig. 3.27,
d
iC (t) = C
vC (t)
dt
at z = l and for t ≥ T , we need to solve
d 1 2V +
C vC (t) + vC (t) = 1
dt Z0 Z0
 3.7 Transients on transmission lines 71

Z0 t=0

iC(t)

V0 Z0 C vC(t)

z=-l z=0

FIGURE 3.27. Transient on a capacitively terminated line

vC(l,t) iC(t)

2V+1 2V+1 /Z0

t t
T T

V1- (l,t)
v(z,t)
V+1
2V+1
t
T

-V+1 z
z1 l

FIGURE 3.28. Transient on a capacitively terminated line

where V1+ = V0 /2. Thus



t−T
vC (t) = 2V1+ 1 − exp −
Z0 C
and

2V1+ t−T
iC (t) = exp −
Z0 Z0 C
Thus, 


1 t−T
V1− = 2V1+
− exp − , t≥T
2 Z0 C
Fig. 3.28 illustrates the form of the voltage on the line.Note
l − z1 z1
t1 − T = ⇒ t1 = 2T −
v v
72 3. Theory and applications of multi-conductor transmission lines

3.8 The Smith-chart

The reflection coefficient at the load ZL terminating a lossless transmission
line of characteristic impedance Z0 is given by
ZL − Z0
ΓL = = |ΓL | exp (jθ Γ )
ZL + Z0
where 
L′
Z0 =
C′
We can normalise the load to the characteristic impedance of the line and
define a dimensionless quantity zL ≤ 1 as
ZL
zL = = rL + jxL
Z0
Thus, if Γr = Re [ΓL ] and Γi = Im [ΓL ], then

(rL − 1) + jxL
Γr + jΓi =
(1 + rL ) + jxL
or equivalently
(1 + Γr ) + jΓi
rL + jxL =
(1 − Γr ) − jΓi
From which we see that
1 − Γ2r − Γ2i 2Γi
rL = 2 , xL =
(1 − Γr ) + Γi 2 (1 − Γr )2 + Γ2i
For a given value of rL , the locus of the reflection coefficient describes a
circle specified through the equation

2
2
rL 1
Γr − + Γ2i =
1 + rL 1 + rL
with the centre located in the complex Γ-plane at


rL
,0
1 + rL
and a radius
1
1 + rL
xL forms another family of circles

2
2
1 1
(Γr − 1)2 + Γi − =
xL xL
 3.8 The Smith-chart 73

x=1.0

phase angle of reflection coeff

x=2.0
x=0.5

r=0
r=0.5 r=1.0 r=2.0

x=-0.5

x=-1.0 x=-2.0

unit circle
FIGURE 3.29. Simplified Smith chart

of radius
1
xL
and centred at

1
1,
xL
The Smith-chart is thus a graph of r- and x-circles in the complex Γ-plane
for|Γ| ≤ 1. Fig. 3.29 presents an example of the impedance Smith-chart.
The fractional distance from the center of the chart to a given point
relative to the radius |Γ| = 1 of the circle is a measure of the value of |Γ|
at that point. θΓ is the angle the line from the centre passing through the
given point makes with the positive real axis. Thus,

• A |Γ| circle intersects the real axis at two points: at PM towards the
open-circuit point at (r = 1, x = 0), and Pm towards the short-circuit
point at (r = 0, x = 0)
• PM and Pm represent a case of pure resistive loading on the line with
PM representing RL > Z0 and Pm representing resistive loading with
RL < Z0 . Since
RL
S= = rL
Z0
74 3. Theory and applications of multi-conductor transmission lines

then for RL > Z0 , the r-circle passing through PM is numerically

equal to the standing wave ratio. The value of the r-circle passing
through Pm is equal to 1/S.

For a lossless line therefore, the impedance seen looking toward the load
at a point z < 0 from the load is given by

V (z) 1 + Γ exp (−j2βz)

Zi (z) = = Z0
I (z) 1 − Γ exp (−j2βz)

where like before, we locate the load at the origin of the z-axis. Using
normalised impedance, we can write

1 + Γ exp (−j2βz) 1 + |Γ| exp (jϕ)

zi (z) = =
1 − Γ exp (−j2βz) 1 − |Γ| exp (jϕ)

where
ϕ = θΓ − 2βz
Thus keeping |Γ| constant and subtracting (rotating clockwise) from θΓ
through 2βz = 4πz/λ locates the point |Γ| exp (jϕ), the reflection coeffi-
cient of the input impedance a distance z from the load zL .
Example: Use the Smith-chart to determine the input impedance of a
section of a 50 Ω lossless line that is 0, 1λ long and terminated in a short
circuit.
We have zL = 0, Z0 = 50 Ω, −z = 0, 1λ. On the Smith-chart, mark Psc
at the intersection of r = 0 and x = 0. Move along the circumference of
the chart ( |Γ| = 1) a distance 0, 1λ towards the generator in a clockwise
direction to point P1 as shown on Fig. 3.30. At P1 read off r = 0 and
x ∼ 0, 725 or zi = j0, 725 ⇒ Zi = zi Z0 = j36, 3 Ω. Confirm this result
using
Zi = jZ0 tan βl, (Ω)

Example: A lossless transmission line of length 0, 434λ and characteristic

impedance 100 Ω is terminated in an impedance 260 + j180 Ω. Determine

(a) The voltage reflection coefficient

(b) The standing-wave ratio
(c) The input impedance
(d) The location of a voltage maximum on the line

Given
−l = 0, 434λ, Z0 = 100 Ω, ZL = 260 + j180 Ω
Thus
 3.8 The Smith-chart 75

P1

Psc

FIGURE 3.30. Smith chart solution

(a) We have
ZL
zl = = 2, 60 + j1, 8
Z0
 
 zl − 1, 0 
|Γ| =   = 0, 60
zl + 1, 0 
We plot on the Smith chart, the point P2 = (2.6, 1.8) and draw a
circle of radius OP2 = |Γ| centred at the origin of the chart ( (0, 0)).
We then draw the line OP2 and extend it to the point P2′ on the unit-
circle and read off the length of line toward the generator as 0, 220.
Phase angle of reflection coefficient is hence

4π (0, 250 − 0, 220) = 0, 12π = 0, 38 rad = 22◦

Multiplication by 4π is because angles on the Smith chart are in

4π
2βl = l
λ
Thus
Γ = 0, 60∠22◦

(b) The |Γ| = 0, 60 circle intersects the positive real axis OPoc at r = S =
4 so that the voltage standing-wave ratio is 4
76 3. Theory and applications of multi-conductor transmission lines

(c) For the impedance, advance P2′ at 0, 220 through l = 0, 434λ towards
the generator to point P3′ , first to 0, 50 equivalent to 0, 0 and a further
0, 154
0, 50 − 0, 220 = 0, 280
and then
0, 280 + 0, 154 = 0, 434
since a complete revolution around the Smith chart corresponds to a
length of λ/2.
Join the origin with P3′ which then intersects the |Γ| = 0, 6 circle at
point P3 and read off the value of r and x at P3

zi = 0, 69 + j1, 2

Input impedance is hence

Zi = Z0 zi = 69 + j120 Ω

(d) Going from P2 to P3 , the |Γ| = 0, 6 circle intersects the positive real
axis OPoc at PM where the voltage is maximum. Hence, maximum
voltage occurs at
(0, 25 − 0, 22) λ = 0, 03λ
from the load.

Fig. 3.31 illustrates the Smith chart solution process.

Example: The input impedance of a short-circuited lossy transmission
line of length 2 m and Z0 = 75 Ω approximately real is 45 + j225 Ω.

(a) Determine α and γ of the line

(b) Determine the input impedance if the short-circuit is replaced by a load

ZL = 67, 5 − j45 Ω.

(a) For a lossy line of a sufficient length l such that 2αl cannot be ignored,
the expression for input impedance must be modified to

1 + Γ exp (−2αz) exp (−j2βz) 1 + |Γ| exp (−2αz) exp (jϕ)

zi = =
1 − Γ exp (−2αz) exp (−j2βz) 1 − |Γ| exp (−2αz) exp (jϕ)

and we must therefore perform extra calculations to account for the

exp (−2αz) factor.
A short circuit on the Smith-chart is represented by the point Psc at
(0, 0). Thus, plot
1
zi = {45 + j225} = 0, 6 + j3, 0
75
 3.8 The Smith-chart 77

P’3

P3

P2 P’2
toward generator

Psc Poc
O PM
toward load

FIGURE 3.31. Smith chart solution

on the chart as point P1 as shown in Fig. 3.32 and draw a straight line
from the centre of the circle through P1 to P1′ on the circumference
of the chart. Now measure OP1 /OP1′ = exp (−2αl) = 0, 89. Hence
determine α as
1 1 1
α= ln = ln 1, 124 = 0, 029 N pm−1
2l 0, 89 4

Record the length of the arc Psc P1′ in wavelengths toward the generator
(i.e. clockwise) which gives the number of wavelengths l/λ = 0, 20 and
l
2βl = 4π = 4π × 0, 20 = 0, 8π ⇒ β = 0, 20π
λ

(b) Plot zL = ZL /Z0 = 0, 9 − j0, 6 on the chart as P2 and draw a straight

line from centre to P2′ on the circumference of the chart and read
off the number of wavelengths towards the generator as 0, 364. Now
draw the |Γ|-circle centred at the centre of the chart having radius
OP2 . Move P2′ along the circumference by l = 0, 20 in wavelengths
toward the generator to point P3′

0, 364 + 0, 20 − 0, 5 = 0, 064

and join P3′ to the centre of the chart by a straight line. This line
intersects the |Γ|-circle at P3 . Mark on OP3 a point Pi such that
78 3. Theory and applications of multi-conductor transmission lines

P’3

P1 P’1

P3
Pi

P2

P’2

FIGURE 3.32. Smith chart solution

OPi /OP3 = exp (−2αl) = 0, 89. At Pi read off zi = 0, 64 + j0, 27.

Hence
Zi = Z0 zi = 48, 0 + j20, 3 Ω

3.8.1 Transmission line matching

Matching using the λ/4 impedance transformer
We insert a λ/4 impedance transformer to match RL to a line of char-
acteristic impedance Z0 (real). The characteristic impedance of the λ/4
transformer is then given by

Z0′ = Z0 RL
Example: A signal generator is to feed equal power through a lossless
air line of Z0 = 50 Ω to two separate resistive loads of 64 Ω and 25 Ω.
Quarter-wave transformers are used to match the loads to the 50 Ω line as
shown in Fig. 3.33.
(a) Determine the required characteristic impedance of the impedance trans-
former sections
(b) Determine S on the matching line sections
(a) The input impedance at the junction with the main line looking into
each load must be 2Z0 since they are in parallel. Thus, we must have
 3.8 The Smith-chart 79

λ/4

64Ω
B
Z’01

Z0

Z’02
B’ 25Ω

λ/4

FIGURE 3.33. A matching scheme using a quarter-wavlength impedance trans-

former

Ri2 = Ri1 = 100 Ω and

 
′ ′
Z01 = Z0 RL1 = 80 Ω, Z02 = Z0 RL2 = 50 Ω

(b) Under matched conditions, there are no standing waves on the main
line. On the matching lines however, we have
′
RL1 − Z01 1 + |Γ1 |
Γ1 = ′ = −0, 11 ⇒ S1 = 1 − |Γ | = 1, 25
RL1 + Z01 1

′
RL2 − Z02 1 + |Γ2 |
Γ2 = ′ = −0, 33 ⇒ S2 = 1 − |Γ | = 1, 99
RL2 + Z02 2

Single-stub matching of arbitrary loads

Consider Fig. 3.34. We seek to determine l and d such that the impedance
seen looking into the junction is Z0 . The parallel combination of a line ter-
minated in ZL and a stub at the points B − B ′ suggests use of admittances
for convenience. We have
1 1
Yi = = YB + Ys = Y0 =
Zi Z0
In terms of normalised admittances, then

yB + ys = 1
80 3. Theory and applications of multi-conductor transmission lines

d
B

Z0 yB

yi yL
ys
Z0
B’

FIGURE 3.34. A single-stub matching scheme

where
yB = Z0 YB , ys = Z0 Ys
with YB and Ys the admittances of the load section and short-circuited
stub respectively.
The short-circuited stub is purely susceptive so that the normalised ad-
mittance is satisfied only if

yB = 1 + jbB , ys = −jbB

Thus, we seek d such that yB looking towards the load at the junction has
unity real part and find l for the stub such that its susceptance cancels that
of the load. Analytically, we can write

(rL + jxL ) + jt
zB =
1 + j (rL + jxL ) t

where t = tan βd. The normalised input admittance to the right of B − B ′

is
1
yB = = gB + jbB
zB
where
rL (1 − xL t) + rL t (xL + t)
gB = 2 + (x + t)2
rL L
 3.8 The Smith-chart 81

and
2
rL t − (1 − xL t) (xL + t)
bB = 2 + (x + t)2
rL L

Perfect match requires simultaneously satisfying

yB = 1 + jbB , ys = −jbB

Thus
2
2xL t rL − rL − x2L
gB = 1 ⇒ t2 − + =0
rL − 1 rL − 1
or
   
1 
2 2
t = xL ± rL (1 − rL ) + xL , rL = 1
rL − 1
xL
= − , rL = 1
2
and
d 1
= tan−1 t, t ≥ 0
λ 2π
1 
= π + tan−1 t , t < 0
2π

l 1 1
= tan−1 , bB ≥ 0
λ 2π bB


1 −1 1
= π + tan , bB < 0
2π bB

Double-stub matching
Consider Fig. 3.35. d0 is chosen arbitrarily e.g. λ/16, λ/8, 3λ/16, 3λ/8,
etc. lA and lB are adjusted to match ZL to the main line. Total admittance
looking into the terminals B − B ′ looking towards the load must be equal
to the conductance of the main line, i.e.
1
Yi = YB + YSB = Y0 =
Z0
or
1 = yB + ySB
where ySB is purely imaginary. This hence requires that

yB = 1 + jbB , ySB = −jbB

82 3. Theory and applications of multi-conductor transmission lines

d0
B A

yA
Z0
yB ZL
yi
ySA
ySB
Z0
B’ Z0
A’

lB lA

FIGURE 3.35. A double-stub matching scheme

On the Smith-chart, yB must lie on the g = 1 circle and is translated

by d0 /λ wavelengths towards the load. This means that yA must be on the
g = 1 circle also but rotated through an angle 4πd0 /λ counterclockwise.
Since ySA is also purely imaginary, the real part of yA must come from
Re [zL ] = gL . Solutions are the intersections of the gL -circle and the rotated
g = 1 circle.

3.9 ♣Transmission line parameters

3.9.1 Capacitance per unit length of line on a general
two-conductor line
Consider the two-conductor structure in Fig. 3.36. In the loss-free case, once
the potential has been determined, the charge on the conductor surfaces
can be determined from the normal component of the transverse electric
field as
−
→ −
→ ∂Φe
ρs = ǫ

n · E t = −ǫ

n · ∇ t Φe = −ǫ
∂n
and the total charge per unit length of line on one conductor say S2 is

−
→
Q′ = ǫ

n · E t dl
C2

and that on S1 is then −Q′ , where C1 and C2 are the peripheries of con-
ductors S1 and S2 . Since the potential difference between the conductors
 3.9 ♣Transmission line parameters 83

FIGURE 3.36. Two-conductor transmission line for computing line parameters

is V0 , the capacitance per unit length of line is then


−
→
ǫ n
· E t dl
Q′ C
C′ = = S2
V0 1
−
→ −
→
Et · d l
S2

The total current on S2 is

 
−
→ −
→ 1 −
→ Q′
I0 = Ht · d l = n

· E t dl =
η ǫη
C2 C2

−  − 
→  →  −
→
since  H t  = η−1  E t  = η−1 n

· E t at the surface S2 and the tangential
electric and normal component of the magnetic fields are separately equal
to zero when the conductor is perfect. The characteristic impedance of the
line is then
V0 V0 ǫ
Z0 = = ′ ǫη = ′ η
I0 Q C

3.9.2 Inductance per unit length on a general two-conductor

line
−
→ −
→
H t and E t are orthogonal and hence the magnetic field lines are coincident
with the equipotentials. All flux lines from the Φ = 0 equipotential to
Φ = V0 /2 in Fig. 3.36 link the current in S2 . Thus, any path P1 S2 or P2 S2
orthogonal to the flux lines coincides with a line of electric force and the
84 3. Theory and applications of multi-conductor transmission lines

flux cutting such a path is then

S2 S2
µ −
→ −
→ µ V0
Ψ = µHt dl = − Et · d l =
η η 2
P1 P1

The inductance of one conductor of the line is hence

magnetic flux linkage Ψ µ V0
L1 = = =
total current I0 2η I0
and the inductance of both conductors then twice this value, i.e.
µ
L′ = Z0
η
from which
µ C′
η= Z0 = Z0
L′ ǫ

3.9.3 Conductance per unit length on a general two-conductor

line
If the dielectric had a complex permittivity ǫ = ǫ′ − jǫ′′ , where ǫ′′ includes
the effects of a non-zero conductivity of the dielectric, the total shunt cur-
rent will consist of displacement and conduction components. The current
leaving S2 per unit length of line is then
  
−
→ −
→ −
→
I = Id + Ic = jωǫ E t · n
dl = jωǫ′ E t · n
dl + ωǫ′′ E t · n
dl
S2 S2 S2

The total shunt admittance is given by

Id + Ic
Yd = jωC ′ + G′ =
V0
or
Ic Ic Id ωǫ′′ σd
G′ = = = ′ C′ = ′ C′
V0 Id V0 ǫ ǫ

3.9.4 Resistance per unit length on a general two-conductor

line
The finite conductivity of the conductors is accounted for by a series resis-
tance per unit length of line R′ as long as this is so chosen that
  2
1 ′ 2 Rm −
→
R I0 =  H t  dl
2 2
S1 +S2
 3.9 ♣Transmission line parameters 85

Thus  − 
→ 2
 H t  dl
Rm C1 +C2
R′ =  2
2   2
−
→
  H t  dl
C2

A further effect of a finite conductivity is to increase the series inductance

of the line since
1+j
Zm = = Rm + jXm = Rm + jωLm
σδ s
The imaginary part of Zm comes from the fact that the magnetic field now
penetrates the conductor. The magnetic energy stored in Xm is given by
 − 
Xm → 2
Um =  J s  dl
4ω
C1 +C2

or 
2
−  Xm R′ I0 R ′ I0
Xm → 2
Um =  H t  dl = =
4ω 4ω Rm ω 2
S1 +S2

We therefore define a skin-effect or internal inductance Lm per unit length

through
1
Lm I02 = Um
4
so that
ωLm = R′
We see therefore that the series inductive reactance of the line is increased
by an amount equal to the series resistance.

3.9.5 Coaxial line

For a coax line
ln (r/b)
Φ = V0 ,
ln (a/b)
2π 2π 2π
′ −
→ ∂Φ V0 2πǫV0
Q = ǫ r
· E t a dφ = −ǫ a dφ = −ǫ dφ =
∂r ln (a/b) ln (b/a)
0 0 0

Thus
ǫ′ Q′ 2πǫ′
C′ = =
ǫ V0 ln (b/a)
86 3. Theory and applications of multi-conductor transmission lines

The magnetic field is given by

−
→ − →

φ V0
H = H t exp (−jβz) = exp (−jβz)
r η ln (b/a)
and the current
2π
−
→
2πV0
I0 = H t · φa dφ =
η ln (b/a)
0

Hence
V0 ln (b/a)
Z0 = =η
I0 2π
and the flux linking the center conductor is

b b
−
→
V0 dr µ
Ψ = µ H t · φdr =µ = V0
η ln (b/a) r η
a a

Thus


Ψ
′ µηV0 b µ b
L = = ln = ln
I0 2πηV0 a 2π a
and
ωǫ′′ ′ ωǫ′′ 2πǫ′ 2πωǫ′′
G′ = C = =
ǫ′ ǫ′ ln (b/a) ln (b/a)
and the power loss in the conductors


1 ′ 2 Rm πV02 1 1
R I0 = Pcond = 2 +
2 η ln (b/a) a b
so that

Rm 1 1
R′ = +
2π a b
and
R′
Lm =
ω
Hence


′ µ b 1 1 1
L + Lm = ln + +
2π a 2πωσδ s a b

3.9.6 Two-wire transmission line

The capacitance per unit length of a two-wire transmission line with wire
radius a and wire separation D,
πǫ
C′ = −1 , F m−1
cosh D/a
 3.9 ♣Transmission line parameters 87

If D/a >> 1,
D D
cosh−1 ∼ ln
a a
The inductance per unit length is
µ D
L′ = cosh−1 , Hm−1
π a
and
πσd
G′ = , Sm−1
cosh−1 D/a
Assuming a surface current density Js flows in a very thin surface layer of
the conductor, the current in each wire is given by

I = 2πaJs

and

1 Rm
Pc = I 2 , W m−1
2 2πa
and
 
′ Rm 1 πfµc
R =2 − , Ωm−1
2πa πa σc
Note that Js is not uniform over the circumference of the wires in the
strict sense as the proximity of the two wires will tend to distort current
distribution.
88 3. Theory and applications of multi-conductor transmission lines
4
Waveguides

4.1 General wave characteristics along uniform

guiding structures
With no loss of generality, we assume wave propagation in the +z-direction
with an exponential z-dependence and a harmonic time-dependence of the
form exp (jωt − γz). We consider a straight guide in the form of a dielectric-
filled metal tube of an arbitrary but uniform cross-section with its axis lying
along the z-axis as shown in Fig. 4.1. In a charge-free region of the dielec-
tric, the electric and magnetic field intensities satisfy the homogeneous
Helmholtz equations
−
→ −
→
∇2 E + k2 E = 0
and
−
→ −
→
∇2 H + k2 H = 0
where
k2 = ω2 µǫ

Because of the axial symmetry, we decompose the Laplacian and the

fields into axial and transverse components. Noting that
−
→  −→ 
∇2 = ∇ t + z
∇z · ∇ t + z
∇z = ∇2t + ∇2z = ∇2t + γ 2

we can then write

−
→   −
→ 
∇2t E t + z
Ez + γ 2 + k2 E t + z
Ez = 0
90 4. Waveguides

y
x z

FIGURE 4.1. A straight hollow waveguide of arbitrary cross-section

or  −   
−
→ →
∇2t E t + γ 2 + k2 E t + z
∇2t Ez + γ 2 + k2 Ez = 0
or
−
→ −
→
∇2t E t + kc2 E t = 0, kc2 = γ 2 + k2
and
∇2t Ez + kc2 Ez = 0, kc2 = γ 2 + k2
For simple, media, we have already seen that the reduced Maxwell’s curl
equations can be expressed as,
−
→ −
→ −
→ −
→
∇ t × E t = −jωµ

z Hz , ∇ t × H t = jωǫ

z Ez

and

−
→ →
− −
→ −
→ →
− −
→
z
× ∇ t Ez + γ E t = jωµ H t , z
× ∇ t Hz + γ H t = −jωǫ E t

We form the product

 −→ → 
− −
→ →
− −
→
z
× z
× ∇ t Ez + γ E t = − ∇ t Ez + γ E t (

z · z
) = jωµ

z × Ht

Hence after multiplying through by γ, we obtain

−
→ −
→ −
→
γ ∇ t Ez + γ 2 E t = −jωµγ

z × Ht

Also multiplying the curl equation of the magnetic field by −jωµ, we obtain
−
→ −
→ −
→
−jωµγ

z × H t = (−jωµ) (−jωǫ) E t + jωµ

z × ∇ t Hz

Upon subtracting the two equations, we obtain

 2 −→ −
→ −
→
γ + ω2 µǫ E t = jωµ
z × ∇ t Hz − γ ∇ t Ez
 4.1 General wave characteristics along uniform guiding structures 91

or
−
→ jωµ −
→ γ−→
E t = 2 z
× ∇ t Hz − 2 ∇ t Ez
kc kc
kc is a characteristic wavenumber of the guide. In a similar way we can
express the transverse component of the magnetic field intensity as
−
→ jωǫ −
→ γ−→
H t = − 2 z
× ∇ t Ez − 2 ∇ t Hz
kc kc
These equations show that the transverse field components can be deter-
mined from the axial components alone.
Depending on the values of Hz and Ez , we have

• Transverse electromagnetic (TEM) waves if Ez = Hz = 0

• Transverse magnetic (TM or E waves) if Ez = 0 and Hz = 0
• Transverse electric (TE or H waves) if Ez = 0 and Hz = 0
• Hybrid modes if Ez = 0 and Hz = 0

4.1.1 TEM modes

For TEM waves, Ez = Hz = 0 within the guide and a TEM wave will exist
only if we have,
γ 2T EM + k2 = 0
i.e.
√
γ T EM = jk = jω µǫ
These waves will propagate through the guide with a phase velocity
ω 1
vp = =√
k µǫ

which is the same as the phase velocity in the unbounded medium filling
the guide. The intrinsic impedance or wave impedance for TEM waves is
given by 
jωµ γ µ
η T EM = = T EM = =η
γ T EM jωǫ ǫ
just like for the unbounded medium. Thus,
−
→
−
→ jωǫ −
→ z
× E t
Ht = z
× E t =
γ T EM ηT EM
Since magnetic field lines always close on themselves, TEM waves exist in
−
→ −
→
a waveguide only if the field lines of B , hence of H form closed loops in
the transverse plane.
92 4. Waveguides

From the general Ampere’s circuital law, the line integral of the magnetic
field intensity in any closed contour in a transverse plane must be equal
to the sum of the axial conduction and displacement currents enclosed by
the path of integration. In the absence of an inner conductor in a guide,
no axial current can exist inside a waveguide. Hence no closed magnetic
loops will exist in any transverse plane. Therefore, hollow waveguides of
any shape do not support the TEM mode of propagation.

4.1.2 TM modes or E modes

For TM modes, Ez = 0, Hz = 0. Thus, we have
−
→ γ −
→
E t = − T2M ∇ t Ez
kc
−
→
Taking the transverse divergence of E t , we obtain
−
→ − → γ
∇ t · E t = − T2M ∇2t Ez = γ T M Ez
kc
or
∇2t Ez + kc2 Ez = 0
With Hz = 0, we have
−
→ jωǫ −
→
H t = − 2 z
× ∇ t Ez
kc

and since
−
→ k2 −→
∇ t Ez = − c E t
γT M
then
−
→
−
→ jωǫ −
→ z
× E t
Ht = z
× E t =
γT M ηT M
where
γT M
ηT M =
jωǫ
is the wave impedance for the TM wave.
Since we have

kc2 = γ 2T M + ω 2 µǫ ⇒ γ T M = kc2 − ω 2 µǫ

We define kc2 = ω 2c µǫ or

kc kc c
fc = √ =
2π µǫ 2π
 4.1 General wave characteristics along uniform guiding structures 93

where
1
c= √
µǫ
is the phase velocity of a plane wave in the unbounded medium. At an
operating frequency f = fc , γ T M = 0 and the wave does not propagate in
the waveguide. fc is hence called the cut-off frequency of the guide. The
value of fc for a given waveguide mode depends on the eigenvalue of that
mode, i.e. the particular value of kc for the mode. Thus,

2
fc
γ T M = jk 1 −
f

and we note that

• For (fc /f )2 < 1, γ T M is imaginary. Thus


2 
2
kc fc
γ T M = jβ = jk 1 − = jk 1 −
k f
√
where k = ω µǫ. Such a mode will propagate down the guide with a
propagation constant

2
fc
β =k 1− , rads−1
f

The wavelength associated with β is called the guide wavelength and

given the symbol λg , i.e.
2π λ
λg = = >λ
β
1 − (fc /f)2

where λ = c/f is the wavelength of a plane wave of the same frequency

propagating in the unbounded medium. If we let λc = c/fc , we can
then write,
1 1 1
= 2+ 2
λ2 λg λc

The phase velocity of the mode propagating in the guide is then given
by
ω c λg
vp = =  = c>c
β λ
1 − (fc /f )2
The phase velocity in the guide is hence always greater than c in
the unbounded medium and frequency-dependent as well through the
ratio λg /λ. Waveguides are therefore dispersive.
94 4. Waveguides

ηTM/η

evanescent
region

f/fc
1

FIGURE 4.2. Variation of the normalised intrinsic impedance of TM modes with

normalised frequency

The group velocity of a propagating mode in the guide is given by


2
1 fc λ
vg = =c 1− = c<c
dβ/dω f λg
and we note that
vg vp = c2
In a lossless guide, signal propagation or velocity of energy transport
is equal to the group velocity and we note that

2
fc
ηT M = η 1 −
f
is purely resistive for a propagating mode. The variation of η T M nor-
malised to η and frequency normalised to fc for the TM modes in a
waveguide is shown in Fig. 4.2.
• If (fc /f)2 > 1, γ T M is real and

2
fc
γT M = α = k −1
f
The wave decays rapidly with increasing z and is said to be evanes-
cent. The intrinsic impedance of evanescent modes is given by


2
k fc
ηT M = −1
jωǫ f
which is purely reactive.
 4.1 General wave characteristics along uniform guiding structures 95

4.1.3 TE modes or H modes

For TE modes, Ez = 0, Hz = 0 and
−
→ γ − →
H t = − T2E ∇ t Hz
kc
−
→
Thus, taking the divergence of H t , we obtain
−
→ − → γ
∇ t · H t = − T2E ∇2t Hz = γ T E Hz
kc
or
∇2t Hz + kc2 Hz = 0

and with
−
→ k2 −→
∇ t Hz = − c H t
γT E
we have
−
→ jωµ −
→ jωµ −
→ −
→
E t = 2 z
× ∇ t Hz = − z
× H t = −η T E z
× H t
kc γT E

where
jωµ
ηT E =
γT E
is the wave impedance of the TE waves.
From 
2
 fc
γ T E = kc2 − ω 2 µǫ = jk 1 −
f
we note,

• For (fc /f ) < 1, γ T E is imaginary and wave propagation in the guide

is possible with a propagation constant

2
fc
γ T E = jβ = jk 1−
f

and the wave impedance

η
ηT E =  >η
1 − (fc /f )2

is purely resistive.
96 4. Waveguides

ηTE/η

f/fc
1

FIGURE 4.3. Variation of the normalised intrinsic impedance of TE modes with

normalised frequency

• For (fc /f )2 > 1, γ T E is real and


2
fc
γT E = α = k −1
f

and
jωµ
ηT E = 
k (fc /f)2 − 1
which is purely reactive. The variation of the ηT E normalised to η
with the frequency normalised to fc is shown in Fig. 4.3.

Example:

(a) Determine the wave impedance and guide wavelength at an operating

frequency equal to twice the cut-off frequency in a waveguide for TM
and TE modes

(b) Repeat for a frequency equal to one half the cut-off frequency

(c) What are the wave impedance and guide wavelength for the TEM mode?

(a) At f = 2fc , we have

 √

2
2
fc 1 fc 3
= ⇒ 1− = = 0, 866
f 4 f 2
 4.2 Wave velocities and dispersion in guiding structures 97

Thus
ηT M = 0, 866η < η, λT M = 1, 155λ > λ
ηT E = 1, 155η > η, λT E = 1, 155λ > λ
(b) At f = 0, 5fc no propagating mode exists and the guide wavelength has
no physical meaning.

2
kc f 0, 276
ηT M = 1− = −j kc
jωǫ fc fǫ

2 −1
jωµ f 3, 63fc µ
ηT E = 1− = −j
kc fc kc

(c) TEM modes exhibit no cut-off characteristics and

ηT EM = η, λT EM = λ

For propagating modes, γ = jβ and the variation of β with frequency de-

termines the properties of the wave along the guide. Fig. 4.4 is a qualitative
illustration of the ω-β dispersion or Brillouin diagram. For the propagating
modes, we have
βc βc
ω= =
2
1 − (ω c /ω) 1 − (fc /f )2
Hence, the slope of the line connecting the origin to any point on the ω-β
curve is a measure of the phase velocity vp of a propagating mode with a
cut-off frequency fc for a guide operated at a frequency f .
Example: Obtain a graph showing the relationship between the attenua-
tion constant α and the operating frequency f for evanescent modes in a
waveguide.
For evanescent modes, f < fc and thus

2
f
α + f 2 = fc2
kc
which is a circle of radius fc centred at (0, 0). The value of α can be
obtained from the quarter circle bounded by the positive axes as shown in
Fig. 4.5.

4.2 Wave velocities and dispersion in guiding

structures
√
Let c = 1/ µǫ denote the phase velocity with which constant-phase points
on the wavefront of a TEM wave in a boundless homogeneous medium
98 4. Waveguides

propagating TE
and TM modes

P TEM mode

ωc

FIGURE 4.4. Variation of β with waveguide operating frequency

α/kc

f/fc

FIGURE 4.5. Variation of α/kc with normalised frequency

 4.3 Group velocity 99

characterised by µ and ǫ propagate. In general, we denote the velocity at

which a point on a constant-phase wavefront would propagate regardless
of the wave type by the symbol vp and note that vp and c are identical in
TEM propagation.
If the phase velocity and attenuation in a guide are constant with fre-
quency, a signal composed of more than one frequency component would
propagate down the guide without distortion since all frequency compo-
nents will preserve their phases relative to each other. However, whenever
phase velocity is a function of frequency, distortion will result as the higher
frequency components will lead the slower ones at a point further down
the guide. This effect is known as dispersion. Think of a hiking group up a
mountain. The more energetic hikers will be further up the mountain after
a short time even though the group started out together at the base. When
dispersion exists, a single phase velocity cannot be defined for the wave.
Dispersion is however not severe in most practical waveguides especially
in the usual narrowband operation of most waveguides. We can therefore
define a group velocity for these signals.

4.3 Group velocity

Group velocity can be defined as the velocity with which a narrow-band
signal appears to propagate. Consider a narrow-band signal f (t) that is
Fourier transformable and has the Fourier transform F (ω). f (t) and F (ω)
are related through the Fourier transform pair

∞
F (ω) = f (t) exp (−jωt) dt
−∞

and
∞
1
f (t) = F (ω) exp (jωt) dω
2π
−∞

We regard the guide in which this signal propagates as a linear system with
an impulse response H (ω) ⇋ h (t). At the receiving end (load side) of the
guide, we can express the output signal as

Fo (ω) = H (ω) F (ω)

and if the guide is lossless and matched

H (ω) = A exp (−jβz) = |H (ω)| exp {−jφ (ω)}

100 4. Waveguides

where β is the propagation constant of the guide and A is an amplitude

constant. Thus
∞
1
fo (t) = F (ω) A exp {j (ωt − φ (ω))} dω
2π
−∞

If φ (ω) is linear in ω, say φ (ω) = ωτ, then

∞
1
fo (t) = F (ω) A exp {jω (t − τ )} dω = f (t − τ )
2π
−∞

which is a replica of f (t) but delayed by a time τ . Thus, a guide with a

transfer function of the form
H (ω) = A exp (−jωτ )
does not distort the input signal.
In a lossless TEM propagation,
√ ω
β = ω µǫ =
c
Thus, as long as µ and ǫ are independent of frequency no distortion will
occur. If the guide is lossy however, attenuation will be a function of fre-
quency in general and distortion will occur.
Consider a narrow-band signal mathematically described by
s (t) = f (t) cos ω 0 t = Re [f (t) exp (jω 0 t) t]
i.e. an amplitude modulated signal. We assume the highest frequency in
f (t) is ωmax << ω 0 . Thus
∞ ∞
S (ω) = {f (t) exp (jω 0 t)} exp (−jωt) dt = f (t) exp {−j (ω − ω0 ) t} dt = F (ω − ω 0 )
−∞ −∞

The output signal is then

So (ω) = AF (ω − ω 0 ) exp {−jβ (ω) z}
so that
∞
1
so (t) = A F (ω − ω 0 ) exp {j (ωt − β (ω) z)} dω
2π
−∞

Suppose that β (ω) can be expanded in a Taylor’s series about ω0 , valid for
narrow-band signals with ω max << ω0 . Then,
dβ (ω) 1 d2 β (ω)
β (ω) ∼ β (ω 0 ) + |ω=ω0 (ω − ω0 ) + |ω=ω0 (ω − ω 0 )2 + . . .
dω 2 dω 2
 4.3 Group velocity 101

and keeping only the first two terms, we can write,

β (ω) ∼ β 0 + β ′0 (ω − ω0 )

where
dβ (ω)
β 0 = β (ω 0 ) , β ′0 = |ω=ω0
dω
Letting Ω = ω − ω 0 , we obtain
 
ωm
1   
so (t) = A Re exp {j (ω 0 t − β 0 z)} F (Ω) exp j t − β ′0 z Ω dΩ
2π
−ωm
'  (
= A Re f t − β ′0 z exp {j (ω 0 t − β 0 z)}

= Af t − β ′0 z cos (ω0 t − β 0 z)

This is a time shifted replica of the original modulated signal envelope and
the velocity of the envelope, i.e., velocity of a constant amplitude point on
the wavefront, is defined as the group velocity,
 −1
1 dβ
vg = = |ω=ω0
β ′0 dω

Example: Calculate the group velocity for a waveguide mode propagating

in an air-filled guide and compare this velocity to the general phase velocity
vp and c.  

2 ω 2
β = k0 − kc2 = − kc2
c
thus 
dβ ω/c2 k0
= =
dω cβ
(ω/c)2 − kc2
and


cβ ω k0 c
vg = < c < vp = =
k0 β β
102 4. Waveguides
5
Waveguide modes in the
parallel-plate transmission line

5.1 Waveguide modes in a parallel-plate

transmission line
5.1.1 TM modes
Consider the parallel-plate structure of Fig. 5.1. The plates are assumed
to be perfectly conducting and of sufficient width in the x-direction such
that the fields have no variation with the x dimension and edge effects are
negligible.
Suppose TM waves are propagating in the structure in the +z-direction
with an exponential z-dependence. We can then write for the electric field
intensity,
Ez (y, z) = Ez (y) exp (−γz)
where Ez (y) satisfies the homogeneous Helmholtz equation

d2
Ez (y) + kc2 Ez (y) = 0
dy 2

subject to the boundary conditions

Ez (y) = 0 @ y = 0, b

Thus,
nπ
Ez (y) = An sin y, n = 0, 1, 2, . . .
b
104 5. Waveguide modes in the parallel-plate transmission line

x z

FIGURE 5.1. A parallel-plate waveguide

where
nπ
kc = , n = 0, 1, 2, . . .
b
Since
−
→ γ− →
E t = − 2 ∇ t Ez
kc
and
−
→ jωǫ −
→
H t = 2 z
× ∇ t Ez
kc
then the only other non-zero components are,
jωǫ nπ
Hx (y) = An cos y, n = 0, 1, 2, . . .
kc b
and
γ nπ
Ey (y) = − A cos y, n = 0, 1, 2, . . .
kc b
since
∂Ez
=0
∂x
The propagation constant is expressed as
 
nπ 2
γT M = − ω 2 µǫ
b
and the cut-off frequencies for the different TM modes are
n nc
fc,T M,n = √ =
2b µǫ 2b
We note that TM0 for n = 0 exhibits no cut-off characteristics and is
therefore TEM as a consequence of the two conductors in the structure
 5.1 Waveguide modes in a parallel-plate transmission line 105

which allow an axial current flow. The mode having the lowest cut-off
frequency is called the dominant mode and is the TEM mode in this case.
The first TM waveguide mode in the parallel-plate guide is the TM1 mode
and starts to propagate when the operating frequency is f > c/2b.
Example:

(a) Write the instantaneous field expressions for the TM 1 mode in a parallel-
plate waveguide.
(b) Sketch the electric and magnetic field lines in the yz-plane.
(a) We have
 π  π
Ez (y, z, t) = Re A1 sin y exp (−jωt − jβz) = A1 sin y cos (ωt − βz)
b b
b π
Ey (y, z, t) = β A1 cos y sin (ωt − βz)
π b
b π
Hx (y, z, t) = ωǫ A1 cos y sin (ωt − βz)
π b
where   π 2
β = ω 2 µǫ −
b
−
→
(b) In the yz-plane, E has both y- and z-components. At a given time t,
the slope of the electric field lines is given by
dy Ey
=
dz Ez
At t = 0 say,
dy Ey (y, z, 0) b π
= = −β cot y tan βz
dz Ez (y, z, 0) π b
We can re-write this in the form
sin (πy/b) b sin βz
dy = −β dz
cos (πy/b) π cos βz
or
d (cos (πy/b)) d (cos βz)
− = dz
cos (πy/b) cos βz
which on integration yields
 π 
− ln cos y = ln [cos βz] + const
b
or
π
cos y cos βz = const, 0 ≤ y ≤ b
b
106 5. Waveguide modes in the parallel-plate transmission line

y/b

βz
x π/2 π 3π/2

FIGURE 5.2. Fields lines for the TM 1 mode in a parallel-plate waveguide

−
→
Since H has only an x-component, the magnetic field lines are every-
where normal to the yz-plane. For the TM 1 mode therefore, we have
ωǫb π
Hx (y, z, 0) = A1 cos y sin βz
π b
The density of the Hx lines varies as cos (πy/b) in the y-direction
and as sin βz in the z-direction. At the conducting plates, there are
surface currents due to the discontinuity in the tangential magnetic
field and surface charges due to the presence of a normal electric field.
Fig. 5.2 illustrates the field lines.
Example: Show that the field solution of a propagating TM 1 mode in a
parallel-plate waveguide can be interpreted as the superposition of two plane
waves bouncing back and forth obliquely between the two plates.
We have for n = 1,
π
Ez (y, z) = A1 sin y exp (−jβz)
b
A1   π   π 
= exp j y − exp −j y exp (−jβz)
j2 b b
A1    π    π 
= exp −j βz − y − exp −j βz + y
j2 b b
which is a superposition of two waves propagating in the +z- and ±y-
directions with phase constants β in the z- and π/b in the ±y-directions.

5.1.2 TE modes
We have Ez , Hz = 0. We can then write
Hz (y, z) = Hz (y) exp (−γz)
 5.1 Waveguide modes in a parallel-plate transmission line 107

which obeys the homogeneous Helmholtz equation

d2
Hz (y) + kc2 Hz (y) = 0
dy 2
Since
jωµ dHz
Ex =
γ dy
and must vanish at the conducting plates, we must then have
dHz
=0
dy
at the plates. Thus,
nπ
Hz (y) = Bn cos y
b
and noting that ∂Hz (y) /∂x = 0, then
γ nπ
Hy (y) = Bn sin y, n = 1, 2, . . .
kc b
and
jωµ nπ
Ex (y) = Bn sin y, n = 1, 2, . . .
kc b
The expressions for the propagation constant γ and cut-off frequencies are
the same as for the TMn modes since fc = 0 in the parallel-plate waveguide.
For n = 0, both Ex and Hy are zero and the TE0 mode does not exist in
a parallel-plate waveguide.
Example: Write the instantaneous field expressions for the TE 1 mode in
a parallel-plate waveguide and sketch the electric and magnetic field lines
in the yz-plane.
π
Hz (y, z, t) = B1 cos y cos (ωt − βz)
b
βb π
Hy (y, z, t) = B1 sin y sin (ωt − βz)
π b
ωµb π
Ex (y, z, t) = − B1 sin y sin (ωt − βz)
π b
The electric field has only an x-component and is therefore everywhere
normal to the yz-plane. At t = 0, we have
ωµb π
Ex (y, z, 0) = B1 sin y sin (βz)
π b
The magnetic field has both y- and z-components and the equation of the
magnetic field lines is thus from
dy Hy (y, z, 0) βb π
= = tan y tan βz
dz Hz (y, z, 0) π b
108 5. Waveguide modes in the parallel-plate transmission line

y/b

βz
x
π/2 π 3π/2

FIGURE 5.3. Field lines for the TE1 mode

or
π
sin y cos βz = const, 0 ≤ y ≤ b
b
with the constant lying between ∓1. Fig. 5.3 illustrates the field lines.

5.2 Energy transport velocity

Group velocity is defined as the velocity of the envelope of a narrowband
signal and loses meaning in the case of a broadband signal such as a pulse
of short duration. This statement is justified since the low frequency com-
ponents in the pulse may be below the cut-off frequency of the waveguide,
hence evanescent. The high frequency components propagate at different
velocities hence strongly distorting wideband signals. No single value of
group velocity can therefore be defined for these signals.
The velocity at which energy is transported down the waveguide is de-
fined as the energy transport velocity. Consider a lossless waveguide. We
can define the energy transport velocity ven as the ratio of the time-average
propagated power to the time-average stored energy per unit guide length,
i.e.
Pz ,ave
ven = ′
, ms−1
Uave
where 
−
→ −
→
Pz,ave = S P,ave · d S
S

and 
′
Uave = {ue + um } dS
S

with ue , um the time-average energy densities stored in the electric and

′
magnetic fields respectively. Pz,ave and Uave are calculated for a given prop-
agating mode to obtain the energy transport velocity.
 5.2 Energy transport velocity 109

Example: Determine the energy transport velocity of the TM n modes in

a lossless parallel-plate waveguide.
We have
−
→ 1 −
→ − →  1
P ave = Re E × H ∗ = Re [−

z Ey Hx∗ + y
Ez Hx∗ ]
2 2

and
−
→ 1 ωǫβ 2 nπ
z
· P ave = − Re [Ey Hx∗ ] = 2
An cos2 y
2 2kc b

For a unit width of the parallel-plate guide,

b
−
→ ωǫβb 2
Pz,ave = P ave · z
dy = A
4kc2 n
0

−  
ǫ → − →  ǫ nπ β2 nπ
ue = Re E · E ∗ = A2n sin2 y + 2 cos2 y
4 4 b kc b

and
−

µ → − →  µ ω2 ǫ2 nπ
um = Re H · H ∗ = A2n cos2 y
4 4 kc2 b

Hence

b b  
ǫ 2 2 nπ β2 2 nπ
ue dy = A sin y + 2 cos y dy
4 n b kc b
0 0

2 
2
ǫb 2 β ǫb k
= An 1 + = A2n
8 kc 8 kc

and

b b

µ ω 2 ǫ2 nπ
um dy = A2n cos2 ydy
4 kc2 b
0 0

2
µb  2 2 2 ǫb k
= 2
ω ǫ An = A2n
8kc 8 kc

Thus

 
2
ωβ  ω  β fc
ven = 2 = = vp 1 −
k k k f
110 5. Waveguide modes in the parallel-plate transmission line

5.3 Attenuation in the parallel-plate the guide

Losses in the guide arise from the finite conductivity of the plates and lossy
dielectric. Losses modify the electric and magnetic field distributions inside
the waveguide but in practical guides, losses are usually low so that we may
assume the transverse field distributions remain essentially unaffected by
these losses. The propagation constant of the propagating modes now has
a real part which we express as
α = αd + αc
where αd is due to a lossy the dielectric and αc from the finite conductivity
of the plates.

5.3.1 TEM modes

We have 
G′ Z0 σd µ σd
αd = = = η, N pm−1
2 2 ǫ 2
If we represent losses in the dielectric by ǫ′′ , σd = ωǫ′′ and
ωǫ′′
αd ∼ η, N pm−1
2
We also have 
R′ 1 πf ǫc
αc = = , N pm−1
2Z0 b σc

5.3.2 TM modes
We have,


 nπ 2 ǫ′′
γ = − ωµǫ′
1−j ′
b ǫ
  σd   nπ 2
= j ω 2 µǫ′ 1 − j ′ −
ωǫ b
  
 nπ 2  nπ 2 −1
∼ j ω 2 µǫ′ − 1 − jωµσd ω2 µǫ′ −
b b
or    −1 
 nπ 2 jωµσ d 2 ′  nπ 2
γ∼j ω 2 µǫ′ − 1− ω µǫ −
b 2 b
where we have assumed
 nπ 2
ωµσ d << ω2 µǫ′ −
b
 5.3 Attenuation in the parallel-plate the guide 111

We have
nπ √
= 2πfc µǫ
b
Hence 
  nπ 2
2
 fc
ω 2 µǫ′ − = ω µǫ′ 1−
b f
Thus
 
2
σd µ 1  fc
γ = αd + jβ = ′
 + jω µǫ′ 1−
2 ǫ 2 f
1 − (fc /f)

or
σdη
αd =  , N pm−1
2
2 1 − (fc /f )
and 
2
 fc
β = ω µǫ′ 1− , rads−1
f
Thus, for TM modes, αd decreases with increasing frequency.
From the principle of energy conservation,

Pl (z)
αc =
2P (z)

where P (z) is the time-average power flowing through a cross-section say

of width W and Pl (z) the time-average power loss per unit length in the
two plates due to Joule heating. For TM modes,

b
1
P (z) = W − Ey Hx∗ dy
2
0

2 b
2
ωǫ′ β bAn nπ bAn
= W cos2 ydy = W ωǫ′ β
2 nπ b 2nπ
0

On the lower plate, we have

−  →  ωǫ′ b
→  − 
 J s,z  =  H x (y = 0) = An
nπ
The total power loss per unit length in the two plates of width W is hence

 
′ 2
→ 2
1 − ωǫ b
Pl (z) = 2W  J s,z  Rm = W An Rm
2 nπ
112 5. Waveguide modes in the parallel-plate transmission line

and hence
2ωǫ′ Rm 2Rm
αc = = 
βb
ηb 1 − (fc /f )2
With 
πf µc
Rm =
σc
then
 
2 −1/2
2 πf µc fc
αc = 1−
ηb σc f
  
2 −1/2
2 πfc µc fc fc
= 1−
ηb σc f f

Thus, increases with increasing frequency.

5.3.3 TE modes
The propagation constant is the same as for the TM modes so that αd has
the same value. For αc , we note that

b
1
P (z) = W Ex Hy∗ dy
2
0

2 b
2
W ωµβ bBn nπ bBn
= sin2 ydy = W ωµβ
2 nπ b 2nπ
0

and


1
Pl (z) = 2W |Js,x |2 Rm = W |Hz (y = 0)|2 Rm = W Bn2 Rm
2

Thus,

2
Pl (z) 2Rm  nπ 2 2Rm kc 2Rm (fc /f)2
αc = = = = 
P (z) ωµβb b ηb k
ηb 1 − (fc /f)2

since ωµβ = k2 η and hence αc → 0 as f → ∞ for TE modes.

6
Rectangular section waveguides

6.1 Rectangular section waveguides

In practice, the plates of a parallel-plate transmission line have finite width.
Since for the wide plate the fields vary sinusoidally in the x-direction, we
can place short-circuiting plates at positions in the x-direction where the
electric field is zero without affecting the field distribution between the
plates. The resulting structure is the hollow rectangular waveguide shown
in Fig. 6.1. The hollow rectangular waveguide is the most widely used
waveguide in the frequency range 320 M Hz−330 GHz. The cross-sectional
dimensions for typical waveguides in the field are listed below.

Waveguide App. frequency a (cm) b (cm)

WR-2300 320 MHz 148, 387 73, 914
WR-3 333 GHz 0, 0864 0, 0432
WR-90 8, 2 − 12, 5 GHz 2, 286 1, 016

The WR-90 waveguide is standard for applications in the X-band.

6.1.1 TM or E Modes
We assume the guide walls are perfectly conducting and that the enclosed
dielectric has parameters ǫ, µ. For a wave propagating in the +z-direction
with a propagation factor of the from exp (jωt − γz), the axial component
of the electric field in a charge-free region in the dielectric satisfies the
114 6. Rectangular section waveguides

b
y

x z

FIGURE 6.1. Cross-section of a rectangular waveguide

Helmholtz wave equation,


∇2t Ez + γ 2 + k2 Ez = 0

or
∇2t Ez + kc2 Ez = 0, kc2 = γ 2 + k2
with the boundary conditions that the tangential electric field components
vanish at the conducting walls.
We assume that Ez can be expressed as

Ez = X (x) Y (y) = 0

Hence
1 d2 X (x) 1 d2 Y (y)
+ = −kc2
X (x) dx2 Y (y) dy 2
This means that each of the terms must be separately equal to a constant,
say
1 d2 X (x) 2 1 d2 Y (x)
= −k x , = −ky2
X (x) dx2 Y (x) dx2
such that the separation constants kx and ky satisfy

kc2 = kx2 + ky2

We can then write

d2 X (x)
+ kx2 X (x) = 0
dx2
and
d2 Y (x)
+ ky2 Y (y) = 0
dx2
 6.1 Rectangular section waveguides 115

In the x-direction, we have Ez (0, y) = Ez (a, y) = 0 and in the y-

direction, Ez (x, 0) = Ez (x, b) = 0. Therefore, we must have
mπ
X (x) = Am sin x, m = 1, 2, . . .
a
and
nπ
Y (y) = Bn sin y, n = 1, 2, . . .
b
so that
mπ nπ
Ez (x, y) = Am Bn sin x sin y, m = 1, 2, . . . , n = 1, 2, . . .
a b
or
mπ nπ
Ez (x, y) = Emn sin x sin y, m = 1, 2, . . . , n = 1, 2, . . .
a b
where
Emn = Am Bn
and  mπ 2  nπ 2
kc2 = +
a b
The transverse components are hence from
−
→ γ− → γ ∂ γ ∂
E t = − 2 ∇ t Ez = − 2 x

Ez − 2 y
Ez
kc kc ∂x kc ∂y
and
−
→ jωǫ −
→ jωǫ ∂ jωǫ ∂
H t = − 2 z
× ∇ t Ez = 2 x

Ez − 2 y
Ez
kc kc ∂y kc ∂x
we obtain
γ mπ mπ nπ
Ex = − Emn cos x sin y
kc2 a a b
γ nπ mπ nπ
Ey = − 2 Emn sin x cos y
kc b a b
jωǫ nπ mπ nπ
Hx = 2 Emn sin x cos y
kc b a b
jωǫ mπ mπ nπ
Hy = − 2 Emn cos x sin y
kc a a b
where
 
 mπ 2  nπ 2
γ = jβ = j ω 2 µǫ − kc2 = j ω 2 µǫ − +
a b

for the propagating modes. For a given TMmn mode, the cut-off frequency
is given by
 
c  mπ 2  nπ 2 c  m 2  n 2
fc,mn = + = +
2π a b 2 a b
116 6. Rectangular section waveguides

or alternatively,
2
λc,mn = 
π (m/a)2 + (n/b)2
Neither m nor n can be zero and the dominant TM mode is hence the
TM11 .
Example: Write the instantaneous field expressions for the TM 11 mode
hence sketch the electric and magnetic field lines in a typical xy-plane and
a typical yz-plane in a waveguide with cross-section dimensions a, b.
We have
β π π π
Ex (x, y, z, t) = E11 cos x sin y sin (ωt − βz)
kc2 a a b
β π π π
Ey (x, y, z, t) = 2
E11 sin x cos y sin (ωt − βz)
kc b a b
π π
Ez (x, y, z, t) = E11 sin x sin y cos (ωt − βz)
a b
ωǫ π π π
Hx (x, y, z, t) = − 2 E11 sin x cos y sin (ωt − βz)
kc b a b
ωǫ π π π
Hy (x, y, z, t) = 2 E11 cos x sin y sin (ωt − βz)
kc a a b
where 
  mπ 2  nπ 2
β = k2 − kc2 = ω2 µǫ − +
a b
In a typical xy-plane, the slopes of the electric and magnetic field lines
are given by

dy a π π
= tan x cot y
dx E b a b


dy b π π
= − cot x tan y
dx H a a b
and we note that


dy dy
= −1
dx E dx H
−
→ −
→
i.e. the E and H lines are orthogonal everywhere in the xy-plane and the
−
→ −
→
E lines are normal and the H lines parallel to the conducting guide walls.
In a typical yz plane, say at x = a/2 or sin (πx/a) = 1 and cos (πx/a) =
0, we have

dy β π π
= 2 cot x tan (ωt − βz)
dz E kc b a
−
→
H has only an x-component. Fig. ?? illustrates the field lines of the TM 11
mode.
 6.1 Rectangular section waveguides 117

y/b y/b

x/a βz
π/2 π 3π/2 2π
x/a

FIGURE 6.2. TM 11 electric and magnetic field lines in a rectangular waveguide

6.1.2 TE or H modes
Here, we have 
∇2t Hz + γ 2 + k2 Hz = 0
with boundary conditions
∂Hz
= 0, @ x = 0, a ⇒ Ey = 0 @ x = 0, a
∂x
and
∂Hz
= 0, @ y = 0, b ⇒ Ex = 0 @ y = 0, b
∂y
−
→
since tangential components of E on the walls must vanish.
A solution of the homogeneous Helmholtz equation is hence readily seen
to be
mπ nπ
Hz = Hmn cos x cos y, m = 0, 1, 2, . . . , n = 0, 1, 2, . . .
a b
The transverse field components are from
−
→ γ− → γ ∂ γ ∂
H t = − 2 ∇ t Hz = − 2 x

Hz − 2 y
Hz
kc kc ∂x kc ∂y
and
−
→ jωµ −
→ jωµ ∂ jωµ ∂
E t = 2 z
× ∇ t Hz = − 2 x

Hz + 2 y
Hz
kc kc ∂y kc ∂x
Hence,
jωµ nπ mπ nπ
Ex = 2
Hmn cos x sin y
kc b a b
jωµ mπ mπ nπ
Ey = − 2
Hmn sin x cos y
kc a a b
γ mπ mπ nπ
Hx (x, y) = 2 Hmn sin x cos y
kc a a b
γ nπ mπ nπ
Hy (x, y) = 2 Hmn cos x sin y
kc b a b
118 6. Rectangular section waveguides

with γ given by the same expression as for the TM modes.

The lowest cut-off frequency for a waveguide with a > b is for the case
m = 1, n = 0 or the fc,10 . Thus
1 c
fc,10 = √ =
2a µǫ 2a
or
λc,10 = 2a
This is the TE10 which is hence, the dominant mode in a rectangular
waveguide. Since TE10 mode has the lowest attenuation of all propagat-
−
→
ing modes in a rectangular waveguide and has E polarised in only one
direction, it is of particular engineering importance.
Example: Write the instantaneous expressions for the TE 10 mode hence
sketch the electric and magnetic field lines in a typical xy-, yz- and xz-
planes. Proceed to sketch the surface currents on the guide walls.
π
Hz (x, y, t) = H10 cos x cos (ωt − βz)
a

Ex (x, y, z, t) = 0
ωµ π π
Ey (x, y, t) = 2 H10 sin x sin (ωt − βz)
kc a a
β π π
Hx (x, y, t) = − 2
H10 sin x sin (ωt − βz)
kc a a
Hy (x, y, t) = 0
where   π 2
β= ω 2 µǫ −
a
Only Ey , Hx , Hz are non-zero with Ey and Hx independent of the y-
dimension. At x = a/2 say, sin (πx/a) = 1 and cos (πx/a) = 0 so that
only Ey and Hx varying sinusoidally with z are non-zero. Hence, at t = 0,
the electric and magnetic field lines in the xy-plane are as shown in Fig.
6.3.
In the xz-plane, all the field components are non-zero and the slope of
−
→ −
→
the H and E lines are governed by


dx β π π
= 2 tan x tan βz
dz H kc a a

Thus, the field lines are independent of y. The surface current density on
−
→
the guide walls is related to H by
−
→ −
→
Js=n

×H
 6.1 Rectangular section waveguides 119

y/b y/b

x/a βz
π/2 π 3π/2
x/a

βz
π/2 π 3π/2 2π

FIGURE 6.3. The TE10 electric and magnetic field lines in a rectangular
waveguide
−
→
where n
is the outward normal from the inner surfaces of the walls and H
the field intensity at the walls.
Hence, at t = 0,
−
→
J s (x = 0) = −
yHz (0, y, z, 0) = −

y H10 cos βz
−
→ −
→
J s (x = a) = y
Hz (a, y, z, 0) = −

y H10 cos βz = J s (x = 0)
−
→
J s (y = 0) = x

Hz (x, 0, z, 0) − z
Hx (x, 0, z, 0)
π β π π

H10 cos x cos βz − z
2 H10 sin x sin βz
= x
a kc a a
and
−
→ −
→
J s (y = b) = − J s (y = 0)
Fig. 6.4 shows the surface currents on the guide walls for the TE 10 in a
rectangular waveguide.
Example: Standard air-filled waveguides have been designed for the radar
bands. One type designated the WG-16 is suitable for X-band applications
and has dimensions, a = 2, 29 cm, b = 1, 02 cm. If it is desired that a
WG-16 guide operate only in the dominant TE 10 mode and the operating
frequency be at least 25% above the cut-off frequency of this mode but no
higher than 95% of the next higher cut-off frequency, what is the allowable
operating frequency range?
120 6. Rectangular section waveguides

βz
π/2 π 3π/2 2π

FIGURE 6.4. Surface current distribution for the TE10 mode in a rectangular
waveguide

c 30 GHzcm
fc,10 = = = 6, 55 GHz
2a 2 × 2, 29 cm
30 GHzcm
fc,20 = = 13, 10 GHz
2, 29 cm
Thus, the range is
1, 25fc,10 ≤ f ≤ 0, 95fc,20
or
8, 19 GHz ≤ f ≤ 12, 45 GHz

6.1.3 Power transport in rectangular waveguides

Consider a propagating T Emn mode. The power or the rate of energy flow
in the +z-direction is given by the time-average Poynting’s vector
 a b 
 
1 −
→ −
→
Pmn = Re  E × H ∗ · z
dx dy
2
0 0
 a b 
 
1 
= Re  Ex Hy∗ − Ey Hx∗ dx dy 
2
0 0
 
a b
1 
= Re ηT Emn Hy Hy∗ − Hx Hx∗ dx dy 
2
0 0

Now
a b a b
2 mπ nπ mπ nπ
sin x cos2 y dx dy = cos2 x sin2 y dx dy
a b a b
0 0 0 0
ab/4, m=0, n=0
= {ab/2, m or n=0
 6.1 Rectangular section waveguides 121

Thus

2 
2 ab β mn mπ 2  nπ 2
Pmn = |Hmn | 2
ηT Emn +
ǫ0m ǫ0n kc,mn a b

2
2 ab β mn
= |Hmn | ηT Emn
2ǫ0m ǫ0n kc,mn

where ǫ0m , ǫ0n are the Neumann factors, with

ǫ0m = 1, m = 0
= 2, m > 0

and

ǫ0n = 1, n = 0
= 2, n > 0

Because of the orthogonality of the eigenfunctions, the total power when

two modes T Emn and T Elp are simultaneously present is the sum of the
powers in each mode, i.e.

P = Pmn + Plp

since
a
mπ lπ
sin x sin x dx = 0, l = m
a a
0

For low-loss waveguides, the power orthogonality will still be valid except
for the case when two or more degenerate modes are present. The presence
of low losses may also result in strong coupling between the degenerate
modes.

6.1.4 Attenuation in rectangular waveguides

Losses are usually low and we can assume the transverse field patterns to be
largely unaffected. The attenuation constant due to losses in the dielectric
can be accounted for using a complex electric permittivity ǫd = ǫ + σ d /jω
such that
σd
αd =  η
2 1 − (fc /f )2
This decreases monotonically with increasing frequency towards a value
σd
αd = η
2
122 6. Rectangular section waveguides

For the conduction losses, we note that only Ey , Hx and Hz are non-zero
for the TE10 mode so that

b a
1
P (z) = − Ey Hx∗ dxdy
2
0 0

 a 2 b a
1 π
= ωµβ H10 sin2 xdxdy
2 π a
0 0
a 2
= ωµβ H10 ab
2π

The power loss Pl (z) is computed by noting that we have four walls to
consider with
−
→ −
→
J s (x = 0) = J s (x = a) = −

yHz (x = 0) = −

y H10

and
−
→ −
→ π βa π
J s (y = 0) = − J s (y = b) = x

H10 cos x − z
H10 sin x
a π a
Thus
Pl (z) = 2Pl (z) |x=0 + 2Pl (z) |y=0

where
b  2
1 −
→  b 2
Pl (z) |x=0 =  J s (x = 0) Rs dy = H10 Rm
2 2
0

a  2 − 2
1 −→  → 
Pl (z) |y=0 =  J s,x (y = 0) +  J s,x (y = 0) Rm dx
2
0

2 
a βa 2
= 1+ H10 Rm
4 π

Thus,

2 
a βa 2
Pl (z) = b+ 1+ H10 Rm
2 π
 
2 
a f 2
= b+ 1+ H10 Rm
2 fc
 6.1 Rectangular section waveguides 123

and
1 + (2b/a) (fc /f )
αc,T E10 =  Rm
ηb 1 − (fc /f)2
 
2 
1 πµc 2b f
=   1+
ηb σc 1 − (fc /f )2 π fc

Hence αc → ∞ as f → fc but decreases towards a minimum as frequency

increases before starting to increase steadily again.
For a given guide with a given a, αc,T E10 decreases with increasing b
but this lowers fc for the next higher mode thus narrowing the available
bandwidth of the usable TE10 mode. Usually b/a ∼ 1/2 is a good trade-off.
For the TM11 mode,
b/a2 + a/b2
αc,T E10 = 2Rm 
ηab 1 − (fc /f )2 (1/a2 + 1/b2 )

with fc,11 = 16, 10 GHz and fc,10 = 6, 55 GHz. The curves for αc,T M11
and αc,T E10 show a broad minimum in the operating range f > fc with
αc,T E10 < αc,T M11 .
Theoretical formulae for attenuation are in good agreement with mea-
sured values for frequencies up to ∼ 5 GHz. At higher frequencies however,
measured values are generally higher depending on the surface smoothness
of the guide walls. When the roughness is on the order of the skin depth at
the operating frequency, the effective surface area is increased thus, increas-
ing the losses. This may explain the higher measured attenuation values.
Example: A TE 10 wave at 10 GHz propagates in a brass rectangular
waveguide with inner dimensions a = 1, 5 cm, b = 0, 6 cm. If the dielectric
is polyethylene with ǫ = 2, 25, µr = 1 and loss tangent 4 × 10−4 , determine

(a) The phase constant and guide wavelength

(b) The phase velocity and wave impedance
(c) The attenuation constant due to loss in the dielectric
(d) The attenuation constant due to loss in the guide

Take the conductivity of brass as 1, 57 × 107 Sm−1

Wavelength in unbounded polyethylene is given by
c 30 GHzcm
λ= √ = √ = 2 cm
f ǫr 2, 25 × 10 GHz
20 GHzcm
fc,10 = √ = 6, 67 GHz
2a ǫr
124 6. Rectangular section waveguides

(a) 
2
ω fc
β= 1− = 74, 5π = 234 radm−1
v f
λ
λg =  = 0, 0268 m
1 − (fc /f )2

(b)
v
vp =  = 2, 68 × 108 ms−1
2
1 − (fc /f )

1 µ
ηT E10 =  = 337, 4 Ω
ǫ
1 − (fc /f)2

(c) αd due to dielectric, we use tan δ = ǫ′′ /ǫ′ or σd /ωǫ, we have

σd = ωǫ tan δ = 5 × 10−4
thus
σd
αd = η = 0, 084 N pm−1 = 0, 73 dBm−1
2 TE

πf µc
Rm = = 0, 0501 Ω
σc
Thus
αc = 0, 0605 N pm−1 = 0, 526 dBm−1

Example: A length of an X-band air-filled copper waveguide has dimen-

sions a = 2, 286 cm, b = 1, 016 cm. Find the cut-off frequencies of the first
four propagating modes hence determine the attenuation in dB of a 1 m
section of guide operated at 10 GHz?

c  mπ 2  nπ 2
fc,mn = +
2π a b
Thus
mode m n fc,mn (GHz)
TE 1 0 6, 562
TE 2 0 13, 123
TE 0 1 14, 764
TE, TM 1 1 16, 156
TE, TM 1 2 30, 243
TE, TM 2 1 19, 753
Thus
 

 π 2 2  π 2
2πf
β 10 = k2 − = − = 158, 05 m−1
a c a
 6.1 Rectangular section waveguides 125

b d jBC b jBL
d

a a
(a) CAPACITIVE IRIS (b) INDUCTIVE IRIS

d d

b b

a a
(c) PROTRUDING POST (d) TUNING POST OR SCREW

FIGURE 6.5. Applications of discontuities in a rectangular waveguide

σ Cu = 5, 8 × 107 S/m

ωµ
Rm = = 0, 026 Ω
2σCu
Rm 
α= 3  µ = 26π 2 + a3 k2 = 0, 0125 N p/m
a bβk ǫ
attenuation
−20 log10 exp (−αl) = 0, 11 dB

6.1.5 Discontinuities in rectangular waveguides

Shunt susceptance can be introduced in a waveguide in the form of metal
diaphragms with an iris as shown in Fig. 6.5(a) and (b). With an iris,
the fields must satisfy further boundary conditions on the metal surface.
If the waveguide is operated in the TE10 mode, the additional boundary
conditions require the presence of all higher order modes but the guide is
usually designed such that only the dominant mode propagates away from
the discontinuity.
The higher-order modes are referred to as cut-off modes and are localised
−
→
near the iris. For a capacitive iris, the E lines are in the y-direction and
reduction of b to the gap width d can be expected to enhance this field as
well as the stored energy in the local electric field. Thus,
4b 1
bC = BC η T E10 = ln
λg sin (πd/2b)
126 6. Rectangular section waveguides

which is accurate to 5% over the normal range of frequencies.

The inductive iris provides additional current paths in the y-direction
causing a new longitudinal H-field in the iris opening, which increases the
energy stored in the local magnetic field. We thus have

λg πd
bL = BL ηT E10 = − cot2
a 2π
Another discontinuity that provides shunt susceptance is a conducting
post protruding into the waveguide on the broad face as shown in Fig.
6.5(c). If d is small, the susceptance is capacitive. When d becomes appre-
ciable compared to b, considerable current flows along the post causing an
inductive effect. Resonance occurs near d = 3b/4 and for d longer than this,
the susceptance becomes inductive.
In practice, the post is a screw installed in an axial slit in the centre of
the broad face where H is minimum and does not therefore disturb the
field pattern significantly. By adjusting d and the position along the guide,
impedance matching similar to stub matching or tuning can be provided.
Example: Measurements on a WG-10 S-band waveguide ( a = 7, 21 cm,
b = 3, 40 cm) feeding a horn antenna show a VSWR of 2, 0 at the operating
frequency of 3 GHz and existence of an electric field maximum 12 cm from
the neck of the horn. Find the location and dimensions of a symmetrical
inductive iris necessary to achieve a perfect match. Assume the guide to be
lossless.
c
fc,10 = = 2, 08 GHz
2a
λ c
λg =  = = 13, 9 cm
1 − (fc /f )2 f − fc2
2

The measured electric field maximum is located at

12 cm
= 0, 863λg
13, 9 cm

from the neck of the horn.

At this location, the normalised effective load resistance is

RL
rL = =S
Z0

Hence
YL 1
gL = = = 0, 5
Y0 S
Proceed as in the single-stub tuning scheme to obtain d = 4, 72 cm located
at (0, 863 − 0, 348) λg = 7, 16 cm.
 6.2 Power capacity of guiding structures 127

6.2 Power capacity of guiding structures

For an air-filled transmission line or waveguide, power handling capacity
is limited by the breakdown voltage of air which occurs at a field strength
Ed = 3 × 106 V/m at room temperature and sea-level pressure. For an
air-filled coaxial line with an inner conductor of radius a at a potential V
w.r.t. to the grounded outer conductor of inner radius b, the radial electric
field intensity for the TEM mode of operation varies with radial distance
r from the axis of the line as,
V
Er =
r ln (b/a)

with the maximum field at r = a. Thus, the maximum peak voltage,


b
Vmax = Ed a ln
a

and the maximum power then

2


Vmax πa2 b
Pmax = = Ed ln
2Z0 µ0 /ǫ0 a

A higher Pmax requires the use of a larger coaxial line while keeping the
ratio b/a fixed to ensure the same characteristic impedance. However a
and b determine the highest frequency at which the line can be operated
mode-free. Thus, for a given maximum operating frequency fmax ,

2
2
0, 025 cEd Ed
Pmax =  = 5, 8 × 1012
µ0 /ǫ0 fmax fmax

For a rectangular waveguide of cross-section a × b, a > b, the electric

field intensity of the fundamental T E10 mode varies as
 πx 
Ey = E0 sin
a
with the maximum occurring at x = a/2. Thus,

ab E02 ab Ed2
Pmax = =
4 Z 4 Z
where Z is the wave impedance of the guide. Many rectangular waveguides
are constructed for single-mode operation with a ∼ 2b to avoid propagation
of the TE20 , which is the next higher mode. Thus, we must have
c
a<
fmax
128 6. Rectangular section waveguides

so that
2
2
0, 11 cEd 13 Ed
Pmax = = 2, 6 × 10
µ0 /ǫ0 fmax fmax
Arcing and voltage breakdown are high frequency processes so that Pmax
represents the peak value. Hence, in constructing a guiding structure, pro-
vide at least a safety factor of 2 so that practical Pmax values are approxi-
mately half the theoretical ones. In case the reflection coefficient becomes
unity, the maximum guide voltage will be twice the value in a reflection-
free case. Hence Pmax allowable is reduced by a factor of 4. Power handling
capacity can be improved by using pressurised air, inert gases or using a
dielectric with a higher dielectric strength.

6.3 Some useful ideas for waveguide analysis

6.3.1 Transmission line analogy of waveguides
−
→ −
→
There is an exact analogy between the E and H fields of the TM and TE
waves and the voltage and current waves on a suitably loaded transmission
line. This analogy enables us to construct equivalent circuit diagrams for the
waveguide which are useful in analysing some unfamiliar electromagnetic
problems using familiar transmission line ideas. We saw that we can express
the fields in the TE case as


∂ jωǫ ∂
E z = I (z) = −jωǫV (z)
∂z kc2 ∂z
and 
∂ kc2
V (z) = − jωµ + I (z)
∂z jωǫ
where
jωǫ
I (z) = Ez , V (z) = Φe
kc2
These have the same form as the differential equations for a lossless trans-
mission line with series impedance per unit length given by

kc2
Z ′ = jωµ +
jωǫ

and shunt admittance per unit length

Y ′ = jωǫ

This can be represented by the high-pass equivalent circuit of Fig. 6.6.

 6.3 Some useful ideas for waveguide analysis 129

k2c
jωµ jωε

jωε

FIGURE 6.6. Transmission line equivalent circuit for TM modes in a rectangular

waveguide

For TE waves, Ez = 0 and

∂
V (z) = −jωµI (z) = −Z ′ I (z)
∂z
and 
∂ kc2
I (z) = − + jωǫ V (z) = −Y ′ V (z)
∂z jωµ
where
jωµ
Vz = Hz , Iz = Φm
kc2
Z ′ = jωµ
and
kc2
Y′ = + jωǫ
jωµ
Fig. 6.7 illustrates the transmission line equivalent circuit for these expres-
sions.
The loaded transmission line circuits of Figs. 6.6 and 6.7 have high-pass
characteristics. In the TM case, cutoff occurs at the frequency for which
the series impedance goes to zero. In the case of TE modes, this is when
the shunt susceptance vanishes. Both conditions require that

kc2 = ω2c µǫ

where ωc is the radian cutoff frequency from wave theory. The characteristic
impedance for the TM equivalent circuit is given by
   
Z′ jωµ + kc2 /jωǫ µ  ω 2
c
Z0 = = = 1 − = ZT M
Y′ jωǫ ǫ ω
130 6. Rectangular section waveguides

jωµ

k2c
jωε
jωµ

FIGURE 6.7. Transmission equivalent circuit for TE modes in a rectangular

waveguide

while for the TE equivalent circuit

   
Z′ jωǫ µ  ω 2 −1/2
c
Z0 = = = 1− = ZT E
Y′ jωǫ + kc2 /jωµ ǫ ω
The transmission line analogy is a powerful analysis tool for solving many
waveguide problems. An example is the iris problem with edges normal or
−
→
parallel to E and the problem of a waveguide with a discontinuity in its
dimensions. These problems are illustrated in Fig. 6.8. Recall that an iris
in a waveguide behaves like a shunt reactance on the equivalent line. The
reactance will be inductive when the edges of the iris are parallel to the elec-
tric field and capacitive in the case of edges normal to the field. An abrupt
change in dimensions is represented by a junction between two transmis-
sion lines of different characteristic impedances with an appropriate shunt
reactance connected at the junction as shown in Fig. 6.8.
Determination of the actual value of the shunt reactance is a field problem
which can be solved through matching the solutions at the junction. The
field at the discontinuity is represented by the sum of the dominant mode
and the higher order modes whose relative amplitudes are determined from
matching the tangential components of the electric and magnetic fields at
the discontinuity. Although the higher order modes are required by the
boundary conditions, these decay rapidly away from the region of the dis-
continuity since their cutoff frequencies will generally be much higher than
the operating frequency.
The load and generator terminations are assumed to be sufficiently far
away from the discontinuity so that they are unaffected by the fields of the
higher order modes. This is the reason why the transmission line analogy
applies since the effect of the discontinuity on the dominant mode alone
suffices. The higher order modes do not contribute significantly to the volt-
age at the discontinuity but their contribution to the current is substantial.
Although the current is continuous across the discontinuity as required by
 6.3 Some useful ideas for waveguide analysis 131

Z0 Z0

Iris with edges normal to E

Z0 Z0

Iris with edges parallel to E

Z01, Z02,
γ1 γ2

Change of guide dimensions

FIGURE 6.8. Example field problems where transmission line analogy proves
useful
132 6. Rectangular section waveguides

ε ε0

a x
t
Partially filled waveguide

FIGURE 6.9. A partially loaded rectangular waveguide

the boundary conditions, the current associated with the dominant mode is
discontinuous by the amount contributed by the higher order modes. This
is what is accounted for by the shunt reactance connected at the disconti-
nuity.

6.3.2 TEm0 modes of a partially loaded rectangular waveguide

Partially loaded waveguide sections are often used for impedance match-
ing and phase shifting. Consider a partially loaded waveguide with the
cross-section shown in Fig. 6.9. Since the waveguide geometry is uniform
in the y-direction and n = 0 for the TEm0 modes, these modes have no
y-dependence. Thus, the Helmholtz equation satisfied by Hz ,

2 
∂
+ kc Hz = 0, kc2 = γ 2 + ω2 µǫ,
2
∂x2
can be written separately for the dielectric and air regions as

2 
∂ 2
+ k d Hz = 0, 0 ≤ x ≤ t
∂x2

2 
∂ 2
+ ka Hz = 0, t ≤ x ≤ a
∂x2
where kd and ka are the cutoff wavenumbers for the dielectric and air
regions respectively.
For the propagating modes, The propagation constant for the propagat-
ing modes is γ = jβ and must be the same in both regions to ensure phase
matching along the dielectric-air interface at x = t. Thus, the dispersion
relations are of the form
β2 = ω 2 µǫ − kd2 , 0 ≤ x ≤ t
= ω 2 µ0 ǫ0 − ka2 , t ≤ x ≤ a
 6.3 Some useful ideas for waveguide analysis 133

and the solution for Hz is hence of the form

Hz = A cos kd x + B sin kd x, 0 ≤ x ≤ t
= C cos ka (a − x) + D sin ka (a − x) , t ≤ x ≤ a
For the TEm0 modes, Ez = 0 and Hy must be zero as well since ∂/∂y = 0.
Hence
jωµ ∂Hz jωµ
Ey = = {−A sin kd x + B cos kd x} , 0 ≤ x ≤ t
kd2 ∂x kd
jωµ ∂Hz jωµ
= = {C sin ka (a − x) − D cos ka (a − x)} , t ≤ x ≤ a
ka2 ∂x ka
With Ey = 0 at x = 0, a, we must have B = D = 0. Continuity of Ey and
Hx at x = t then yields
A C
− sin kd t − sin ka (a − t) = 0
kd ka
and
A cos kd t − C cos ka (a − t) = 0
The determinant of this set of homogeneous equations must vanishes for a
non-trivial solution. Hence
ka tan kd t + kd tan ka (a − t) = 0

Together with the dispersion relations, this can be solved graphically or

otherwise, for β, kd and ka at the given operating frequency.

6.3.3 Transverse-resonance techniques

A propagating mode in a uniform waveguide structure has a propagation
constant of the form
 
β = ω 2 µǫ − kc2 = ω2 µǫ − kx2 − ky2

When only the propagation constant of the waveguide is of interest, solving

the relevant wave equation to determine kc is not necessary.
The field distributions in the waveguide at cutoff have a standing wave
pattern in the transverse plane which is equivalent to a transmission line
circuit in resonance. On a resonant transmission line, the sum of the input
impedances at any point on the line looking toward the generator and the
load must be zero, i.e.
Zin,g (y) + Zin,l (y) = 0
where Zin,g (y) and Zin,l (y) are the input impedance at x on the reso-
nant line looking toward the generator and load respectively. This is the
134 6. Rectangular section waveguides

y y
ky,ZTM

b b

ε ,µ Zin,l
y
Zin,g
x x
a Transverse-resonance
equivalent circuit

FIGURE 6.10. Transverse-resonance equivalent circuit for a parallel-plate trans-

mission line

transverse-resonance technique which yields information on only the cutoff

frequency of the guide for the given mode.
As an example, we obtain the cutoff frequency for TM modes on a con-
ducting parallel-plane transmission line. The transmission line and the res-
onant transmission line circuit in the transverse plane is shown in Fig.
√
6.10. At cutoff, k = ω µǫ = kc and no propagation down the guide oc-
curs. The field hence forms a standing wave pattern along the y-dimension.
The equivalent circuit of this operation is a transmission line of length b
short-circuited at both ends by the conducting plates at y = 0, b having a
propagation constant ky .
Due to the uniform geometry in the x-direction, kx = 0 and the charac-
teristic impedance of the equivalent line is the wave impedance seen by the
TM mode having β = ky , i.e.
ky
Z0 = ηT M = η
k
where 
µ √
η= , k = ω µǫ
ǫ
At any point 0 ≤ y ≤ b, we have

Zin,g (y) = jη T M tan ky (b − y)

and
Zin,l (y) = jηT M tan ky y
Thus
jηT M {tan ky (b − y) + tan ky y} = 0
or
sin ky b
jηT M =0
cos ky (b − y) cos ky y
 6.3 Some useful ideas for waveguide analysis 135

y y

b b
ky,a,Z0,a
ε0
t t
ε ky,d,Z0,d
x x
Partially filled waveguide Transverse-resonance
equivalent circuit

FIGURE 6.11. Transverse-resonance equivalent circuit for TE0n modes in a par-

tially filled rectangular waveguide

Hence
nπ
kc = ky = , n = 0, 1, 2, . . .
b
For a partially loaded rectangular waveguide, the situation for the TE0n
modes can be represented as shown in Fig. 6.11. In the region 0 < y < t,
we have β = ky,d and

k k0
Z0,d = η = η0
ky,d ky,d

where

√ µ0
k0 = ω µ0 ǫ0 , η 0 =
ǫ0

In the region t < y < b, we have β = ky,a and

k0
Z0,a = η0
ky,a

Thus
ky,a {tan ky,d t + ky,d tan ky,a (b − t)} = 0

and together with the phase matching condition of the tangential field
components
β 2 = k2 − ky,d
2
= k02 − ky,a
2

yields solutions for ky,a and ky,d .

136 6. Rectangular section waveguides

6.3.4 Mode properties in rectangular waveguides

TE modes TM modes
Hz cos mπx nπx
a cos b exp (−jβ mn z) 0
jβ mn m
Hx ak2 sin a cos nπx
mπx
b exp (−jβ mn z) −Ey /ηT M,mn
c,mn
jβ mn n
Hy 2
bkc,mn cos mπx nπx
a sin b exp (−jβ mn z) Ex /ηT M,mn
Ez 0 sin mπx nπx
a sin b exp (−jβ mn z)
jβ mn m
Ex η T E,mn Hy − ak2 cos mπx nπx
a sin b exp (−jβ mn z)
c,mn

Ey −ηT E,mn Hx − jβ
bk2
mn n
sin mπx nπx
a cos b exp (−jβ mn z)
c,mn

k0
ηT E,mn β mn Z0
β mn
ηT M,mn k0 Z0
  2
m 2
kc,mn π a + nb
λc,mn √ 2ab 2 2
(mb) +(na)
β mn 2 2
k0 − kc,mn

The attenuation coefficient for non-degenerate modes


 2   
2
2Rm b kc,mn b ǫ0m kc,mn m2 ab + n2 a2
αT E =  1+ + −
2
bZ0 1 − kc,mn /k02 a k02 a 2 k02 m2 b2 + n2 a2

2Rm m2 b3 + n2 a3
αT M =  2 2
2
bZ0 1 − kc,mn /k0 m b a + n2 a3
2

where 
ωµ0
Rm =
2σ
and

ǫ0m = 1, m = 0
= 2, m > 2

6.4 Rectangular waveguide cavity

Ordinary lumped elements such as resistors, inductors and capacitors are
difficult to realise above UHF frequencies ( 300 M Hz). This is because
circuits of the the same order of magnitude as the wavelength of the signals
are efficient radiators and thus interfere with other circuits and systems.
Skin effect and radiation increase the effective resistance of conventional
wires.
 6.4 Rectangular waveguide cavity 137

probe excitation loop excitation

y
y
z
x

z
d

FIGURE 6.12. Rectangular resonant cavity showing two methods of excitation

Consider the rectangular cavity of Fig. 6.12. We choose the z-axis as the
reference direction of propagation but because of the conducting termina-
tion at z = 0, d, only a standing wave pattern exists in the cavity. We use
three indices m, n, p to denote the TM or TE standing wave pattern in the
cavity.

6.4.1 TMmnp modes

In a rectangular waveguide, we assumed that the axial variation of the
field is exp (−jβz) for a mode propagating in the +z-direction. The wave
is reflected back and propagates in the −z-direction with a propagation
factor exp (jβz) on reaching the termination at z = d.
Consider the Ey (x, y, z) component of the electric field. Boundary con-
ditions at z = 0, d require that Ey = 0 which implies that Ey must vary
with z as sin βz where β = pπ/d, where p is an integer. This must hold true
for Ex (x, y, z) as well. Since the factor −jβ appears due to the ∂/∂z com-
−
→
ponent of ∇, the other components must have a cos βz variation whenever
this factor is missing. Thus, for the TMmnp modes, we have
mπ nπ pπ
Ez = Emn sin x sin y sin z
a b d
1 mπ mπ nπ pπ
Ex = − Emn cos x sin y sin z
kc2 a a b d
1 nπ mπ nπ pπ
Ey = − 2
Emn sin x cos y sin z
kc b a b d
jωǫ nπ mπ nπ pπ
Hx = − Emn sin x cos y sin z
kc2 b a b d
138 6. Rectangular section waveguides

1 mπ mπ nπ pπ
Ey = − Emn cos x sin y sin z
kc2 a a b d
where  mπ 2  nπ 2
kc2 = +
a b
For the TMmnp modes, resonance frequency of the cavity is given by
 
1  mπ 2  nπ 2  pπ 2
ω mnp = + + = 2πfmnp
µǫ a b d

6.4.2 TEmnp modes

The transverse electric field components must vanish at z = 0, d as well
while Ez = 0 everywhere. Thus, the z-variation requires a sin pπz/d. The
factor jβ appears from the −∂/∂z operation and components without this
factor must therefore be multiplied by a factor −pπ/d. Thus
mπ nπ pπ
Hz = Hmn cos x cos y sin z
a b d
1  mπ   pπ  mπ nπ pπ
Hx = − 2
Hmn sin x cos y cos z
kc a d a b d

1 nπ   pπ  mπ nπ pπ
Hy = − 2 Hmn cos x sin y cos z
kc b d a b d

jωµ nπ  mπ nπ pπ
Ex = 2 Hmn cos x sin y sin z
kc b a b d
jωµ mπ mπ nπ pπ
Ey = − Hmn sin x cos y sin z
kc2 a a b d
and we note that fmnp is the same as for the TMmnp modes. Hence, these
modes are degenerate as long as none of the m, n or p is zero. The mode
with the lowest resonance frequency is the dominant mode.
For the TM modes, axial and transverse electric field components are in
time phase but in quadrature phase with the magnetic field components.
Hence, the time-average Poynting vector in any direction is zero. For the
TE101 mode in an a × b × d cavity, the only non-zero components are
π π
Hz = H0 cos x sin z
a d
a π π
Hx = − H0 sin x cos z
d a d
jωµa π π
Ey = − E0 sin x sin z
π a d
which may be excited by a probe inserted in the cavity region at the top
or bottom faces where Ey is maximum or by inserting a loop to couple
 6.4 Rectangular waveguide cavity 139

maximum Hx inside the front or back faces as shown in Fig. 6.12. The
best position for the probe or loop is affected by the impedance matching
requirements of the microwave circuit of which the cavity is part of. A
common method of coupling energy out of the cavity is by introducing a
hole or iris at an appropriate location in the cavity wall. The field at the
hole must have a component favourable for exciting the desired mode in
the cavity.
Example: Determine the dominant modes and their frequencies in an air-
filled rectangular cavity resonator for a > b > d, a > d > b, a = b = d,
where a, b, and d are the x-, y-, z-directed dimensions of the cavity resp.
With the z-axis chosen as the reference direction of propagation, neither
m nor n can be zero for TM modes but TE modes allow either m or n
to be zero but not simultaneously but p cannot be zero. Thus, the lowest
order modes are TM 110 , TE 011 and TE 101 with TM 110 having the lowest
resonance frequency for a > b > d given by

c 1 1
f110 = +
2 a2 b2
TE 101 will have the lowest resonance frequency for a > d > b given by

c 1 1
f101 = +
2 a2 d2
while all three modes will have the same resonance frequency for a = b = d
given by
c
f110 = √
a 2

6.4.3 Quality factor of the rectangular cavity

The walls of a practical cavity have finite conductivity and hence result in
ohmic losses. We define the quality factor or Q of the cavity as

time-average energy stored at resonance

Q = 2π
energy dissipated in one period at resonance

Let U be the time-average energy stored in the cavity resonator. We can

then write
U = Ue + Um
where Ue and Um is the time-average energy stored in the electric and
magnetic fields resp. If Pl is the time-average power dissipated in the cavity
over one period, then the dissipated energy is given by Pl /ω so that

U
Q=ω
Pl
140 6. Rectangular section waveguides

In determining Q at resonance, the loss is customarily assumed to be low

so that the field patterns are essentially the same as in the lossless case.
For an a × b × d cavity carrying the TE101 mode, the time average stored
energy is
 d b a
ǫ0 2 ǫ0 ω 2 µ20 π 2 2 πx 2 πz
Ue = |Ey | dV = H0 sin2 sin dxdydz
4 4kc4 a2 a d
V 0 0 0


ǫ0 ω 2101 µ20 a2 2  a  d 1
= H0 b = ǫ0 µ20 a3 bdf101
2
H02
4π2 2 2 4
where kc2 = π2 /a2 . The time-average energy stored in the magnetic field is
similarly

µ0  2 2

Um = |Hx | + |Hz | dV
4
V
d b a 
µ20 2 π4 πx πz πx 2 πz
= H sin2 cos2 + cos2 sin dxdydz
4 0 4 2
kc a d 2 a d a d
0 0 0



µ20 2  a 2  a  d a d
= H0 b + b
4 d 2 2 2 2

  
µ0 a 2
= abd 1 + H02
16 d
Since 
1 1 1
f101 = √ 2
+ 2
2 µ0 ǫ0 a d
then
 a 2 
µ0
Um = Ue ⇒ U = 2Ue = 2Um = abd 1 + H02
8 d
To compute Pl , we note that power loss per unit area is given by
1 2 1 2
Pave =
|Js | Rm = |Ht | Rm
2 2
Power loss in the back wall is the same as that in the front wall so that
 b a 
   d b 
Pl = Pave dS = Rm |Hx (z = 0)|2 dxdy + |Hz (x = 0)|2 dydz
 
S 0 0 0 0
 d a 
  d a 
2 2
+Rm |Hx | dxdz + |Hz | dxdz
 
0 0 0 0

2 


H2 a b 1 b 1
= Rm 0 a + +d +
2 d d 2 a 2
 6.4 Rectangular waveguide cavity 141

and 
πf101 µ0 abd a2 + d2
Q=
Rm {2b (a3 + d3 ) + ad (a2 + d2 )}
Example: What should be the size of a hollow cubic cavity made of copper
in order for it to have a dominant mode resonance frequency of 10 GHz?
Proceed to determine the Q of the resonator at this frequency.
30 GHzcm
f101 = √ = 10 GHz ⇒ a = 21, 2 mm
2a cm
πf101 µ0 a a
Q= = πf101 µ0 σ Cu = 10700, σCu = 5, 8 × 107 Sm−1
3Rs 3
142 6. Rectangular section waveguides
7
Circular section waveguides

7.1 Circular section waveguides

Circular section waveguides are less common and have diameters in the
range 63, 957 − 0, 607 cm to cover the 800 MHz − 116 GHz range of
frequencies. For a hollow straight waveguide with uniform circular cross-
section of dimensions shown in Fig. 7.1, we break the Laplacian into axial
and transverse components in cylindrical coordinates to take advantage of
the cylindrical symmetry. For the TM modes in a charge-free region of
the dielectric inside the waveguide, Ez is a solution of the homogeneous
Helmholtz equation

∇2t Ez (r, φ) + kc2 Ez (r, φ) = 0

In the case of TE modes on the other hand, Hz is a solution of

∇2t Hz (r, φ) + kc2 Hz (r, φ) = 0

In the cylindrical coordinate system, the radial Laplacian operator is ex-

panded as

1 ∂ ∂ ∂2
∇2t = r + r2 2
r ∂r ∂r ∂φ
The wave equations therefore take the form


1 ∂ ∂ ∂2
r Ez + r2 2 Ez + kc2 Ez = 0
r ∂r ∂r ∂φ
144 7. Circular section waveguides

FIGURE 7.1. Cross-section of a hollow circular waveguide

and

1 ∂ ∂ ∂2
r Hz + r2 2 Hz + kc2 Hz = 0
r ∂r ∂r ∂φ
where
kc2 = γ 2 + ω 2 µǫ

7.1.1 TM modes
Consider the conducting hollow pipe of inner radius a shown in Fig. 7.1.
We propose that Ez (r, φ) can be expressed as a product of analytic one-
dimensional functions as,
Ez (r, φ) = R (r) Φ (φ)
Thus

r d dR 1 d2 Φ
r + kc2 r2 = −
R (r) dr dr Φ (φ) dφ2
Let ν = 0 be a constant such that
1 d2 Φ
= −ν 2
Φ (φ) dφ2
called the angular differential equation and


r d dR
r + kc2 r2 = ν 2
R (r) dr dr
or
d2 R dR  2 2 
r2 2
+r + kc r − ν 2 R (r) = 0
dr dr
 7.1 Circular section waveguides 145

which is called the radial differential equation and we recognise that it is

the Bessel’s differential equation in r.
Let x = kc r. Thus,
dR dR dx dR
= = kc
dr dx dr dx
and


d2 R d dR d dR dx 2
2d R
= = k c = kc
dr2 dr dr dx dx dr dx2
so that, the radial differential equation can be expressed as
d2 R dR  2 
x2 2
+x + x − (ν)2 R (r)
dx dx
We assume that the radial function R (x) can be expressed as a power series
of the form ∞ ∞
/ /
R (x) = xp Cl xl = Cl xp+l , C0 = 0
l=0 l=0

where we have multiplied through by xp , p an arbitrary exponent. Thus,

∞ ∞
dR / d2 R /
= (p + l) Cl xp+l−1 , = (p + l) (p + l − 1) Cl xp+ν−2
dx dx2
l=0 l=0

Substituting these expressions in the differential equation, we obtain

∞
/ ∞
/
 
Cl (p + l) (p + l − 1) + (p + l) − ν 2 xp+l + Cl xp+l+2 = 0
l=0 l=0

or
∞
/   ∞
/
Cl (p + l)2 − ν 2 xp+l + Cl xp+l+2 = 0
l=0 l=0

We can express this in the form,

   /∞   
C0 p2 − ν 2 xp + (p + 1)2 − ν 2 C1 xp+1 + (p + l)2 − ν 2 Cl + Cl−2 xp+l = 0
l=2

Since C0 = 0 by assumption, then we must have p2 − ν 2 = 0 ⇒ p = ±ν

and C1 = 0. The coefficients of xp+l in the series must all vanish or
 
(p + l)2 − ν 2 Cl + Cl−2 = 0, l = 2, 3, 4, . . .

or
Cl−2
Cl = − , l = 2, 3, 4, . . .
(p + l)2 − ν 2
We note that with C1 = 0, Cl = 0 for l odd. We have three possible
solutions depending on the nature of p1 = ν and p2 = −ν,
146 7. Circular section waveguides

(1) p1 and p2 differ by a non-integer value

(2) p1 = p2
(3) p1 − p2 is an integer
Let p1 = ν. We can generate at least one solution of the Bessel’s equation.
Since only even l lead to non-zero coefficients, we let l = 2m, m = 1, 2, 3, . . ..
Thus,
C2m−2
C2m = − , m = 1, 2, 3, . . .
(ν + 2m)2 − ν 2
C2m−2
= − 2 , m = 1, 2, 3, . . .
2 m (m + ν)
A repeated use of the recursion formula for Cl yields
(−1)m C0
C2m = , m = 1, 2, 3, . . .
22m m! (ν + 1) (ν + 2) . . . (ν + m)
(−1)m ν!
= C0 , m = 1, 2, 3, . . .
22m m! (m + ν)!
m
(−1) Γ (ν + 1)
= C0 , m = 1, 2, 3, . . .
22m m!Γ (m + ν + 1)
Since C0 is an arbitrary constant, we can choose C0 in such a way that the
solution is normalised. Therefore we let
1
C0 =
2ν Γ (ν + 1)
so that
(−1)m
C2m = , m = 1, 2, 3, . . .
22m+ν m!Γ (m + ν + 1)
One solution is therefore
∞
/ (−1)m  x 2m+ν
Rν (x) = Jν (x) =
m=0
m!Γ (m + ν + 1) 2

called the Bessel function of the first kind and order ν.

For p2 = −ν, we obtain
∞
/ (−1)m  x 2m−ν
R−ν (x) = J−ν (x) =
m=0
m!Γ (m − ν + 1) 2

If ν is not an integer, Rν (x) and R−ν (x) represent two linearly independent
solutions of the second-order Bessel’s equation. A general solution can the
be expressed as
R (x) = A1 Jν (x) + A2 J−ν (x)
 7.1 Circular section waveguides 147

where A1 and A2 are constants. For small x,

1  x ν
Jν (x) → , x → 0+
Γ (v + 1) 2
and  x −ν
1
J−ν (x) → , x → 0+
Γ (1 − v) 2
Note that J−ν (x) is unbounded at x = 0 for a non-integer ν.
In most Bessel functions encountered in engineering, ν is an integer say
n. In such cases, we have
∞
/ (−1)m  x 2m+n
Jn (x) = , n = 0, 1, 2, . . .
m=0
m!Γ (m + n + 1) 2

Replacing n by −n yields,
∞
/ (−1)m  x 2m−n
J−n (x) = , n = 0, 1, 2, . . .
m=0
m!Γ (m − n + 1) 2
∞
/ (−1)m  x 2m−n
= , n = 0, 1, 2, . . .
m=n
m!Γ (m − n + 1) 2

since
1
= 0, m = 0, 1, 2, . . . n − 1
Γ (m − n + 1)
Thus, putting k = m − n, we obtain
∞
/ (−1)k+n  x 2k+n
J−n (x) = , n = 0, 1, 2, . . .
(k + n)!Γ (k + 1) 2
k=0
∞
/ (−1)k  x 2k+n
= (−1)n , n = 0, 1, 2, . . .
k!Γ (k + n + 1) 2
k
= (−1)n Jn (x)
and now, Jn (x) and J−n (x) are linearly dependent so that we have only
one solution of the second-order differential equation. To generate a second
solution, we define a function
∆ Jv (x) cos vπ − J−v (x)
Yv (x) =
sin vπ
called the Bessel function of the second kind and order v.
Yv (x) is linearly independent of Jv (x) when ν is not an integer. For
integer ν, Yn (x) reduces to the indeterminate ratio 0/0. We thus define
Yn (x) as a limit as
Yn (x) = lim Yν (x) , n = 0, 1, 2, . . .
ν→n
148 7. Circular section waveguides

and use L’Hospital’s rule to evaluate this limit since, for two functions f (x)
and g (x) for which
f (x) 0
lim =
x→x0 g (x) 0
then
f (x) f ′ (x)
lim = lim ′
x→x0 g (x) x→x0 g (x)

For n = 0, we have
 x  2/  
(−1)  x 2k
∞ k
2 1 1 1
Y0 (x) = J0 (x) ln + γ − 1 + + + . . . +
π 2 π
k=1
(k!)2 2 2 3 k

where  
n
/ 1
γ = lim − ln n = 0, 577215 . . .
n→∞ k
k=1

is Euler’s constant.
For small x,
2
ln x, x → 0+
Y0 (x) →
π
The asymptotic behaviour near x = 0 is given by

ν
Γ (ν + 1) 2
Yν (x) ∼ − , ν > 0, x → 0+
π x

and unbounded at the origin. Thus, for problems involving the origin, Yν (x)
is not a valid solution and must be excluded. For arbitrary ν, we can write

R (x) = A1 Jν (x) + A2 Yν (x)

Fig. 7.2 illustrates the Bessel functions of the first kind for the first three
integer orders.

The choice of the value of C0 arose from earlier studies of Bessel func-
tions in which Jn (x) was taken as the coefficient of tn in the expansion of
exp x2 t − 1t , i.e.



n
1  x n
/∞
x 1 1
exp t− = t−
2 t n=0
n! 2 t
/ 1  x n
∞



n 
1 1
= n
C0 tn −n C1 tn−1 + . . . + (−1)n n Cn
n=0
n! 2 t t
x  1

The function exp 2 t− t is known as the generating function of the
first order.
 7.1 Circular section waveguides 149

Jn(kca)

1,0
J0(kca)

J1(kca)
J2(kca)

kc a
0

FIGURE 7.2. Bessel functions of the first kind for the first few orders

Some problems are simplified when Jν (kc r) and Yν (kc r) are linearly
combined to form the Hankel functions of the first and second kind and
order ν defined as

Hν(1) (kc r) = Jν (kc r) + jYν (kc r)

and
Hν(2) (kc r) = Jν (kc r) − jYν (kc r)
respectively. For kc r → ∞,
  
2 π  νπ 
Hν(1) (kc r) → exp j kc r − −
πkc r 4 2

and 
2   π  νπ 
Hν(2) (kc r) → exp −j kc r − −
πkc r 4 2
The large-argument approximations of the Hankel functions are useful in
(1)
the study of radiation and scattering problems. Note that Hν (kc r) for
large kc r is a spherical wave propagating inward toward the origin while
(2)
Hν (kc r) is one propagating outward from the origin.
When kc2 is negative, the solutions take the form

Iν (ξr) = j −ν Jν (jξr)
150 7. Circular section waveguides

and
π
Kν (ξr) = j ν+1 Hν(1) (jξr)
2
where jξ = kc . These are the modified Bessel functions of the first and
second kind and order ν. For ξr → ∞, these have the asymptotic approxi-
mations 
1
Iν (ξr) → exp (ξr)
2πξr
and 
π
Kν (ξr) → exp (−ξr)
2πξr
Kν (ξr) for large r therefore exhibits an exponential decay which is a char-
acteristic of evanescent waves. This function is useful in the study of surface
waves in rod dielectric waveguides.
For the TM modes, we have

Ez (r, φ, z) = Ez (r, φ) exp (−γz)

where
Ez (r, φ) = R (r) Φ (φ)
Since all field components are periodic in φ with period 2π, the only ad-
missible solution is either sin nφ or cos nφ or a linear combination of these
in the expression of Φ (φ). We may chose either since this only translates
to a shift in the choice of the φ = 0 reference point.
Let Ez (r, φ) = An Jn (kc r) cos nφ. Then
−
→
φ=−γ − →
E t,T M = r
Er + φE ∇ t Ez
kc2
where  
−
→ ∂
∂
φ
∇ t Ez = r
+ Ez
∂r r ∂φ
Thus
jβ
Er = − An Jn′ (kc r) cos nφ
kc
jβn
Eφ = 2 An Jn (kc r) sin nφ
kc r
jωǫn
Hr = − An Jn (kc r) sin nφ
kc2 r
jωǫ
Hφ = − An Jn′ (kc r) cos nφ
kc
where
d
Jn′ (kc r) = Jn (kc r)
dr
 7.1 Circular section waveguides 151

TABLE 7.1. The first few zeros of the Bessel’s function of the first kind and order
n
n p1n p2n p3n
0 2, 405 5, 520 8, 654
1 3, 832 7, 016 10, 174
2 5, 135 8, 417 11, 620

Now, Ez must vanish at r = a. Hence, we must choose kc a such that

it corresponds to the zeros of Jn (kc a). Let us designate the n-th root of
Jn (kc a) = 0 as pmn . The allowed eigenvalues of kc are then
pmn
kc =
a
A choice of m and n then specifies a particular TMmn mode. m is related
to the number of circumferential variations of the field and n the radial
variations. For the mn-th propagating mode, the propagation coefficient is
  p 2
mn
β mn = ω2 µǫ −
a
and the cut-off wavelength
2πa
λc,mn =
pmn
The wave impedance is
β mn
ηT M,mn =
ωǫ
The first few zeros of Jn (kc a) are listed in Table 7.1
From Table 7.1, for the TM01 mode, (kc a)01 = 2, 405 so that

2, 405 kc,T M01 0, 383

kc,T M01 = ⇒ fc,T M01 = √ = √
a 2π µǫ a µǫ

For this mode, only Ez , Er , and Hφ are non-zero. In circular waveguides,

T M01 is analogous to the TM11 mode in rectangular waveguides. However
the TM01 does not exist in rectangular waveguides. Fig. 7.3 illustrates the
electric and magnetic field lines for the TM01 mode in a circular waveguide.

7.1.2 TE modes
Hz (r, φ, z) = Hz (r, φ) exp (−γz)
where
∇2t Hz + kc2 Hz = 0
152 7. Circular section waveguides

FIGURE 7.3. TM01 electric and magnetic field lines in a hollow circular waveguide

We can thus write

Hz = Bn Jn (kc r) cos nφ
and
jβ
Hr = − Bn Jn′ (kc r) cos nφ
kc
jβn
Hφ = Bn Jn (kc r) sin nφ
kc2 r
jωµn
Er = Bn Jn (kc r) sin nφ
kc2 r
jωµ
Eφ = Bn Jn′ (kc r) cos nφ
kc
We require that ∂Hz /∂r vanish at r = a. Thus

Hz (r, φ) = B1 Jn (kc r) cos nφ

with
∂
Jn (kc r) |r=a = Jn′ (kc a) = 0
∂r
Let the roots of this equation be p′mn . Thus

p′mn
kc,mn =
a
and the first few roots are presented in Table 7.2
Now
d
J0 (x) = −J1 (x)
dx
thus p′0n = p1n and T E0n and T M1n are therefore degenerate.
 7.1 Circular section waveguides 153

TABLE 7.2. The first few zeros of the derivative of Bessel functions of the first
kind and order n
m p′m1 p′m2 p′m3
0 3, 832 7, 016 10, 174
1 1, 841 5, 331 8, 536
2 3, 054 6, 706 9, 970

The first propagating TE mode is the TE11 with

2πa 2π
λc,11 = = a = 3, 41a
p′11 1, 841
The TE0n modes are a special case in terms of their attenuation, which is
given by
Rm f2
α=  c,0n
aZ0 f f2 − f2 c,0n

and hence decreases as f 3/2 at high frequencies, since Rm increases as

f 1/2 . This property is unique for TE0n modes in circular waveguides and
makes it possible to construct long lossless waveguide communication links.
However, operating the waveguide at frequencies for which the TE01 mode
propagates means that the waveguide operation is no longer single mode.
Any small irregularities in the guide walls will lead to power conversion from
one mode to another (mode conversion) with two serious consequences.

• Loss of power from the desired TE01 mode

• Conversion of power back and forth between the desired TE01 mode
and the other modes. Due to the different modal phase velocities, the
resulting dispersion may lead to serious signal distortion.

Currents associated with TE0n modes are in the circumferential direction

only and this property makes it possible to construct mode filters that will
suppress modes with substantial axial currents. A small-pitch helical guide
is one structure with these mode suppression properties.
The attenuation constant due to a finite conductivity of the guide wall
can be determined in the same way as for the rectangular waveguide and
it is found that the attenuation constant of the TE0n modes decreases
monotonically with increasing frequency.
Example: A 10 GHz signal is to be transmitted inside a hollow circular
conducting pipe. Determine the inside diameter of the pipe such that its
lowest cut-off frequency is 20% below this signal frequency.
If the pipe is to operate at 15 GHz, what waveguide modes can propagate
in the pipe?
0, 293c 0, 0879
fc,T E11 = = GHz = 8 GHz ⇒ 2a = 2, 2 cm
a a
154 7. Circular section waveguides

The inner radius of the pipe is a = 1, 1 cm. Thus

p01 2, 405
fc,T M01 = fc,T E11 = = 10, 45 GHz
p′11 1, 841
where p and p′ are the zeros of Jn (kc a) and Jn′ (kc a) resp.
p01 3, 405
fc,T E21 = fc,T E11 = = 13, 27 GHz
p′11 1, 841
and all other higher modes have fc > 15 GHz.
Example: Find the cut-off frequency of the first two propagating modes
of a circular waveguide with a = 0, 5 cm and ǫr = 2, 25. If the waveguide
is silver plated and tan δ = 0, 001, calculate the attenuation in dB for a
50 cm length of guide operated at 13, 0 GHz.
TE 11 and TM 01 are the first propagating modes.

p′11 c 1, 841 3 × 108
T E11 : fc = √ = √ = 11, 72 GHz
2πa ǫr 2π (0, 005) 2, 25

p01 c 2, 405 3 × 108
T M01 : fc = √ = √ = 15, 31 GHz
2πa ǫr 2π (0, 005) 2, 25
At 13, 0 GHz, only the TE 11 mode propagates. Thus
√  √
2πf ǫr 2π 13 × 109 2, 25
k= = = 408, 4 radm−1
c 3 × 108
and for the TE 11 mode

2
p′11
β= k2 − = 176, 7 radm−1
a

Attenuation due to dielectric

k2 tan δ
αd = = 0, 47 N p/m
2β
σAg = 6, 17 × 107 ⇒ 0, 029 Ω. Thus attenuation due to the conductor


Rm k2
αc =  kc2 + ′2 = 0, 066 N p/m
ak µ/ǫ p11 − 1

We see that the dielectric loss dominates. The total attenuation is then

α = αc + αd = 0, 547 N p/m

and loss
−20 log10 [exp (−αl)] = 2, 38 dB
 7.1 Circular section waveguides 155

7.1.3 Mode Properties in Circular Waveguides

TE (H) Modes
TE(H) Modes
′
Hz Jn pmn a
r
exp (−jβ mn z) (cos nφ or sin nφ)
 ′ 
jβ mn p′mn ′
Hr − ak2 Jn pmna
r
exp (−jβ mn z) (cos nφ or sin nφ)
c,mn  
jβ mn p′mn p′mn r
Hφ − rk2 nJn a exp (−jβ mn z) (cos nφ or sin nφ)
c,mn
Ez 0
Er ηT Emn Hφ
Eφ −ηT Emn Hr
k0
ηT Emn β η0
mn
p′mn
kc,mn a
2πa
λc,mn p′mn
η0 k0 β mn π  ′2
power 4
2kc,mn ǫom pmn − m2 Jn2 (p′mn )
 k2 
√ 2Rm2 c,mn m2
α 2 k 2 + (p ′ 2
) −m2
aZ0 1−kc,mn /k0 0 mn

TM (E) Modes

TM (E) Modes
Hz 0
Hr −Eφ /ηT M,mn
Hφ Er /η
 T M,mn −jβ z
Ez Jn pmn a
r
e mn (cos nφ or sin nφ)
jβ mn pmn ′  pmn r −jβ mn z
Er − ak2 Jn a e (cos nφ or sin nφ)
c,mn
jβ mn pmn  pmn r −jβ z
Eφ − rk2 nJn a e mn (cos nφ or sin nφ)
c,mn
β mn
ηT M,mn k0 η 0
pmn
kc,mn a
2πa
λc,mn pmn
k0 β mn π 2 ′2 ′
power 4
2η0 kc,mn ǫom pmn Jm (pmn )
α √ 2Rm2
aZ0 1−kc,mn /k02

where  
ωµ0 µ0
Rm = , η0 =
2σ ǫ0
and

ǫ0m = 1, m = 0
= 2, m > 2
156 7. Circular section waveguides

7.2 Waveguide Modes in a Coaxial Transmission

Line
The coaxial line can support TE and TM modes but these are usually
evanescent at the usual operating frequencies of the coax line. The domi-
nant waveguide mode in the coaxial line sets the high frequency limit of the
line, hence the dominant TE11 in a coaxial line is of special interest. Since
the region of interest a ≤ r ≤ b does not include the origin, the Neumann
function is valid as well and the axial magnetic field intensity is then given
by
Hz (r, φ) = (A1 cos nφ + A2 sin nφ) {B1 Jn (kc r) + B2 Yn (kc r)}
We require Eφ (r, φ) = 0 @ r = a, b and since
jωµ
Eφ = (A1 cos nφ + A2 sin nφ) {B1 Jn′ (kc r) + B2 Yn′ (kc r)}
kc
we require
B1 Jn′ (kc a) + B2 Yn′ (kc a) = 0
and
B1 Jn′ (kc b) + B2 Yn′ (kc b) = 0
which for a non-trivial solution, requires that the determinant vanishes.
Hence
Jn′ (kc a) Yn′ (kc b) = Jn′ (kc b) Yn′ (kc a)
and kc then defines the TEmn modes of the coax line. For n = 1,
2
kc ∼
a+b
and once we know kc , the propagation constant or the cutoff frequency can
be calculated. Fig. 7.4 shows the field distribution for the TE11 mode in a
coax line.
Example: A piece of RG-142 coax cable has a = 0, 035′′ and b = 0, 116′′
and ǫr = 2, 2. What is the highest frequency before the TE 11 can propagate?
b 0, 116
= = 3, 3
a 0, 035
The curve of kc a vs b/a can be used to obtain kc . However, using the
approximate expression
2
kc a ∼ = 0, 465
1 + b/a
which gives
ckc
fc,11 = √ = 17 GHz
2π ǫr
 7.2 Waveguide Modes in a Coaxial Transmission Line 157

FIGURE 7.4. Field distribution of the TE11 mode in a coaxial line

In practice, a 5% safety margin is advisable and the highest useful frequency

is then
fmax = 0, 95 × 17 GHz = 16 GHz

7.2.1 Coaxial connectors for microwave systems

Most coaxial lines and connectors used in practice are constructed for a
characteristic impedance of 50 Ω except for the 75 Ω line in TV systems.
The rationale behind this is that the attenuation for an air-filled coax line
is minimum for a characteristic impedance of 77 Ω while maximum power
handling capacity peaks at 30 Ω. The 50 Ω characteristic impedance which
is roughly the geometric mean between these two impedances, is hence a
compromise between power handling capacity and attenuation. High power
handling capacity is more important for microwave systems since powers
encountered are large. For TV systems, the signals from the antenna are
small and the attenuation characteristics of the line are then of greater
importance.
Connectors are used in pairs and must possess certain characteristics
such as
• Low SWR
• mode-free operation at high frequencies
• High repeatability after a connect-disconnect cycle
• High mechanical strength
Among the most common coaxial connectors are
1. Type N (1942 after P. Neill)
Male and female connectors thread together.
The female connector has an outer diameter ∼ 0, 625′′ hence, the
connector operates up to 11 − 18 GHz with an SWR∼ 1, 07
158 7. Circular section waveguides

2. SMA (subminiature type-A, early 1960’s)

Fabrication motivated by a requirement for light, compact connec-
tors. Female connector has outer diameter of ∼ 0, 250′′ so that fmax ∼
25 GHz. This connector is the most widely used at microwave fre-
quencies.
Due to the great demand for this type of connector, two popular
variants exist

• K-connector (Maury) fmax ∼ 40 GHz

• 2, 4 mm connector (HP or Agilent) fmax ∼ 50 GHz

3. SSMA (scaled SMA)

Similar to the SMA in design but of smaller size with the female
connector having an outer diameter of ∼ 0, 192′′ , fmax ∼ 38 GHz

4. APC-7 (Amphenol precision connector)

Usable repeatably with an SWR < 1, 04 @ f up to 18 GHz
This is a sexless connector and uses butt contacts for both the inner
and outer conductors

5. BNC (baby N connector) and TNC (threaded BNC)

Common use is at RF and IF frequencies, never at microwave fre-
quencies. (Recall the low frequency oscilloscope and network coax
connectors.)

7.3 Circular waveguide cavity

For simplicity, we consider the TM01 mode in a circular waveguide of radius
a at cutoff so that there are no variations in the z-direction. The ends of
the cavity are shorted by conducting plates at a distance d < 2a apart.
Fields inside the cavity are hence


2, 405
Ez = C0 J0 (kc r) = C0 J0 r
a


C0 ′ C0 2, 405
Hφ = −j J0 (kc r) = j J1 r
η0 η0 a
where for n = 0, J0′ (kc r) = J1 (kc r) .
The electric and magnetic fields are hence in quadrature time phase,
resulting in no power loss in the cavity wall. For a circular cylindrical
 7.3 Circular waveguide cavity 159

cavity of radius a and length d, the energy stored in the cavity for the
TM010 mode is given by
 a

ǫ0 2 ǫ0 C02 2, 405
U = 2Ue = |Ez | dV = (2πd) J02 r rdr
2 2 a
V 0
 2
a 2
= πǫ0 dC02 J (2, 405)
2 1

The time-average power loss due to dissipation in the wall arising from a
finite conductivity is given by
1 1
Pave = |Js |2 Rm = |Hφ |2 Rm
2 2
−
→
There are radial currents J r on the face plates and uniform axial surface
−
→
currents J z on the inside of the cylindrical wall. Thus,
 a 
  
Rm
Pl = 2 |Jr |2 (2πr) dr + (2πad) |Jz |2
2  
0
 a 
  
2 2
= πRm 2 |Hφ | rdr + ad |Hφ (r = a)|
 
0
 a 


πRm 2  2 2, 405 2

= C 0 2 J1 r rdr + adJ1 (2, 405)
η20  a 
0

or
πRm 2
Pl = C (a + d) J12 (2, 405)
η20 0
so that

η0 2, 405
Q=
Rs 2 (1 + a/d)
Note that
    
r2 n2
Jn2 (kc r) rdr = ′
Jn (kc r) + 1 − 2
Jn (kc r)
2 (kc r)2

1
J1′ (kc r) = J0 (kc r) − J1 (kc r) , J0 (kc a) = 0
kc r
With 
πf µc
Rm =
σ
160 7. Circular section waveguides

2, 405 0, 115
fT M,010 = √ = GHz
2πa µ0 ǫ0 a
Example: A hollow circular cylindrical cavity resonator is constructed of
copper such that its length d equals its diameter 2a. Determine a and d
for a resonance frequency of 10 GHz for the TM 010 mode hence estimate
its Q.
We have
2, 405 0, 115
fT M,010 = √ = GHz = 10 GHz
2πa µ0 ǫ0 a

Hence
a = 1, 15 cm, d = 2a = 2, 30 cm

πfµ0
Rm = = 2, 61 × 10−2 Ω
σ
so that

η0 2, 405
Q= = 11580
Rs 2 (1 + a/d)
Comparing with rectangular and circular resonators of comparable size at
the same frequency, we note

circular cavity rectangular cavity

resonant mode at freq TM010 @ 10 GHz TE101 @ 10 GHz
dimensions radius=1, 15 cm, length= d = 2, 3 cm a = b = d = 2, 12 cm
vol. πa2 d = 9, 56 cm3 abd = 9, 53 cm3
tot. surf. area of walls 2πa2 + 2πad = 24, 93 cm2 6a3 = 26, 97 cm2
Q 11580 10700

Thus, these have approximately the same volume but the total surface area
of the rectangular cavity is 8, 2% larger leading to a lower Q. The Q of the
circular cavity is ∼ 8, 2% higher than that of the rectangular cavity.
8
Dielectric waveguides

8.1 Planar dielectric slab in free space

Consider a dielectric slab of thickness d with constitutive parameters ǫd ,
µd and σ d = 0 in free space as shown in Fig. 8.1. For simplicity, we assume
that there are no variations of the fields in the x-direction, i.e. the dielectric
slab is of infinite extent in this direction and wave propagation is in the
+z-direction with an exponential z-dependence of the form exp (−γz).

8.1.1 TM modes
We have
d2
Ez (y) + kc2 Ez (y) = 0
dy 2
where 
kc = γ 2 + ω 2 µǫ
with γ = jβ inside the slab. Hence, we can write

d
Ez (y) = Eo sin ky y + Ee cos ky y, |y| ≤
2
where Eo and Ee are amplitudes of the odd and even field components
resp. and

ky = ω2 µd ǫd − β 2
162 8. Dielectric waveguides

d/2 µ0,ε0

z
0

µd,εd

-d/2 µ0,ε0

FIGURE 8.1. A longitudinal cross-section of a dielectric-slab waveguide

is the wavenumber inside the slab. This expression for ky is referred to as

the dispersion relation.
In the free space regions, i.e., |y| > d/2, the waves must decay exponen-
tially for the wave to be guided along the dielectric slab and not radiate
away from it. Hence, we define the attenuation constant with the the proper
sign for fields decaying with distance from the air-dielectric interface as,
α2 = β 2 − ω 2 µ0 ǫ0
Hence, in the air region, we require


d d
Ez (y) = CA exp −α y − , y>
2 2


d d
= CB exp α y + , y<−
2 2
We consider the three region separately

Odd TM modes
(i) In the dielectric region,
Ez (y) = Eo sin ky y
jβ
Ey = − Eo cos ky y
ky
jωǫd
Hx (y) = Eo cos ky y
ky
(ii) In the region y > d/2,


d d
Ez (y) = Eo sin ky exp −α y −
2 2
 8.1 Planar dielectric slab in free space 163


jβ d d
Ey = − Eo sin ky exp −α y −
α 2 2


jωǫ0 d d
Hx (y) = Eo sin ky exp −α y −
α 2 2

(iii) In the region y < −d/2



d d
Ez (y) = −Eo sin ky exp α y +
2 2


jβ d d
Ey = Eo sin ky exp α y +
α 2 2


jωǫ0 d d
Hx (y) = − Eo sin ky exp α y +
α 2 2

Continuity of Hx at the dielectric surfaces requires that Hx (d/2) have

the same value in both free space and the dielectric. Thus

α ǫ0 ky d
= tan
ky ǫd 2

and
α2 + ky2 = ω 2 {µd ǫd − µ0 ǫ0 }

or 
α= ω 2 {µd ǫd − µ0 ǫ0 } − ky2

We combine these expressions to obtain

 ǫ0 ky d
ω 2 {µd ǫd − µ0 ǫ0 } − ky2 = ky tan
ǫd 2

Note that at y = 0, Ez (y) = 0. Hence, we can introduce a conducting plane

here without affecting the field pattern. Thus, odd TM modes propagating
in a dielectric of thickness d are the same as those that would propagate in
a slab of thickness d/2 backed by a conducting plane.
The surface impedance looking down on the slab surface is given by

Ez α
Zm = − =j
Hx ωǫ0

which is inductive and implies that an inductive surface can support a

surface wave.
164 8. Dielectric waveguides

Even TM modes
We have
d
Ez (y) = Ee cos ky y, |y| ≤
2
and invoking the continuity conditions at the dielectric surface, we obtain

α ǫ0 d
= − cot ky
ky ǫd 2

The phase constant of the propagating TM even modes lies between the
intrinsic phase constant of free space and that of the dielectric, i.e.
√ √
ω µ0 ǫ0 < β < ω µd ǫd
√
As β → ω µ0 ǫ0 , α → 0 and the wave is no longer confined within the
dielectric. The limiting frequencies under this condition are known as the
cut-off frequencies of the dielectric waveguide. At cutoff, we have for odd
TM modes, 
ω co d 
tan µd ǫd − µ0 ǫ0 = 0
2

πfco d µd ǫd − µ0 ǫ0 = (n − 1) π, n = 1, 2, 3, . . .
n−1
fco = √
d µd ǫd − µ0 ǫ0
and 
ω ce d 
cot µd ǫd − µ0 ǫ0 = 0
2


 1
πfce d µd ǫd − µ0 ǫ0 = n − π, n = 1, 2, 3, . . .
2
n − 1/2
fce = √
d µd ǫd − µ0 ǫ0
for even TM modes. Note that fc,o = 0 for n = 1. Hence, the dominant
mode on a slab of any thickness is an odd TM mode.

8.1.2 TE modes
We have
d2
Hz (y) + kc2 Hz (y) = 0
dy 2
with a solution
d
Hz = Ho sin ky y + He cos ky y, |y| ≤
2
 8.1 Planar dielectric slab in free space 165

and


d d
Hz (y) = CA exp −α y − , y>
2 2


d d
= CB exp α y + , y<−
2 2

Odd TE modes
We have

(i)
jβ
Hz (y) = Ho sin ky y, Hy (y) = − Ho cos ky y
ky
jωµd
Ex (y) = − Ho cos ky y
ky
inside the dielectric
(ii) For y ≥ d/2


ky d d
Hz (y) = Ho sin exp −α y −
2 2


jβ ky d d
Hy (y) = − Ho sin exp −α y −
α 2 2


jωµd ky d d
Ex (y) = − Ho sin exp −α y −
α 2 2

(iii) For y ≤ −d/2



ky d d
Hz (y) = −Ho sin exp α y +
2 2


jβ ky d d
Hy (y) = Ho sin exp α y +
α 2 2


jωµd ky d d
Ex (y) = Ho sin exp α y +
α 2 2

Thus, at y = d/2,
α µ ky d
= 0 tan
ky µd 2
Ex jωµ0
Zm = =−
Hz α
which is capacitive but increases with increasing frequency. Hence, a TE
surface wave can be supported by a capacitive surface.
166 8. Dielectric waveguides

Even TE modes
d
Hz (y) = He cos ky y, |y| ≤
2
and
α µ ky d
= 0 cot
ky µd 2
Therefore, the lowest order odd TE mode just like the lowest order odd
TM mode has no cutoff frequency.
Example: A dielectric slab waveguide with constitutive parameters µd =
µ0 , ǫd = 2, 50ǫ0 is situated in free space. Determine the minimum thickness
of the slab so that a TM or TE wave of the even type propagates along the
guide at 20 GHz.
Lowest even TM and TE modes have the same cut-off frequency

n − 1/2
fc = √
d µd ǫd − µ0 ǫ0

Thus, for n = 1,
c c
fc =  ⇒ dmin =  = 6, 12 cm
2d ǫd /ǫ0 − 1 2fc ǫd /ǫ0 − 1

Example:

(a) Obtain an approximate expression for the decaying rate of the dominant
TM surface wave outside a very thin dielectric slab waveguide.

(b) Determine the time-average power per unit slab width transmitted along
the guide

(c) What is the time-average power transmitted in the transverse direc-

tion?

(a) fco = 0 for n = 1 for all d. If ky d/2 << 1, tan ky d/2 ∼ ky d/2. Thus,
for
d ǫd
α <<
2 ǫ0
then
1 ǫ0 2 1 ǫ0 2
α= ky d ∼ ω {µd ǫd − µ0 ǫ0 } d
2 ǫd 2 ǫd

(b)
−
→ 1
P ave = Re [−

y Ey × x

Hx ]
2
 8.1 Planar dielectric slab in free space 167

µ0,ε0

θi

θi µd,εd

µ0,ε0

FIGURE 8.2. Bouncing-wave interpretation of propagating waves along a dielec-

tric waveguide

Thus

d/2 d/2
−→  ωǫd β 2
Pave = 2  P ave  dy = 2 Eo cos2 ky ydy
ky
0 0

ωǫd β 2 1
= E d+ sin ky d , W m−1
4ky2 o ky
 √
ky ∼ ω µd ǫd − µ0 ǫ0 , β ∼ ω µ0 ǫ0

(c) Since Hx is π/2 out of phase with Ez , then

−
→ 1
P ave = Re [

z Ez × x

Hx ] = 0
2

Consider the dielectric slab shown in Fig. 8.2. There will be total internal
reflection of a wave incident obliquely on the lower surface of the slab when
ǫd > ǫ0 and θi > θ i,c , where

ǫ0
θi,c = sin−1
ǫd

An evanescent wave exists on the surface as well.

Since the reflected wave will be incident on the upper surface at θi ,
the wave will suffer total internal reflection here as well. The two plane
waves incident on the top and lower surfaces hence interfere and the inter-
ference pattern is the mode pattern of the propagating wave. In circular
waveguides, TM and TE separation requires that the fields have circular
symmetry, i.e. independent of φ. If a φ-dependence exists, we must consider
the so called hybrid modes.
168 8. Dielectric waveguides

FIGURE 8.3. Dielectric-coated conducting plane

8.1.3 A grounded dielectric slab

In many practical situations, we have a dielectric slab backed by a grounded
conducting plane. Suppose the dielectric slab shown in Fig. 8.3 has a thick-
ness a = d/2 and is of infinite extent in the xz-plane as is the perfectly
conducting ground plane. TM modes will the be solutions of
∂2
Ez + kc2 Ez = 0
∂y2
where we have again assumed Ez does not vary in the x-coordinate. From
the previous discussion, we anticipate a surface-wave mode of propagation
and denote
kc = kd , 0 < y < a
and
kc = jka , y > a
For Ez to vanish at y = 0 and y = ∞, we see that Ez must be of the form

Ez = A1 sin kd y, 0 < y < a

and
Ez = A2 exp {−ka (y − a)} , y > a

The tangential electric and magnetic fields are continuous at y = a, a

condition that requires that the propagation coefficient β be the same in
both regions. Hence,
β 2 = k2 − kd2 = k02 + ka2
or
√
(ǫr − 1) k02 = kd2 + ka2 , k = k0 ǫr
also
A1 sin kd a = A2
 8.1 Planar dielectric slab in free space 169

and since

β k −
→ 
Hx = −j z
× ∇ t Ez · x


kc2 βη
where
k k0
= ǫr = ωǫ, 0 < y < a
η η0
k0
= = ωǫ0 , y > a
η0

then at y = a
ǫr ∂Ez 1 ∂Ez
2 =− 2
kd ∂y ka ∂y
i.e.
ǫr 1
A1 cos kd a = A2
kd ka
or dividing the two transcendental equations and multiplying through by
a, we have
kd a tan kd a = ǫr ka a
When a solution for real ka exists, this will be a surface wave since its
field will decay exponentially in the y-direction while propagating in the
z-direction.

8.1.4 Finding a solution for the TM surface wave modes by

reverse resonance technique
The eigenvalues for the solution can be obtained more directly through
the transverse-resonance technique which is an application of transmission
line theory to the equivalent circuit of a transverse section of the guiding
structure applicable in a waveguide that is uniform in the direction of
propagation. Thus, the surface-wave guide can be viewed along the y-axis
as shown in Fig. 8.4. The propagation coefficient in the y-direction is jkd
in the dielectric and ka in air. Thus, the wave impedance for the wave
propagating in the y-direction is then

kd
η T M,y = η
k
in the dielectric and
ka
ηT M,y = η
jk0 0
in air. The equivalent circuit is then a short-circuited transmission line of
length a and impedance −jka η0 /k0 . At the reference plane, the resonance
170 8. Dielectric waveguides

FIGURE 8.4. The transmission line equivalence of the surface-wave propagation

mode

condition requires that the sum of the impedances seen looking toward the
short circuit and the input to the infinite line vanish. Hence
kd η ka η0
j tan kd a − j =0
k k0
or
kd tan kd a = ka ǫr
since η/k = η 0 / (k0 ǫr ).
The T M0 mode surface-wave does not have a low-frequency cut-off fre-
quency but this is not a characteristic of all surface-wave guiding structures.
However, all will have only a finite number of propagating modes at any
given frequency. When the mode exists, its phase velocity will be less than
the plane wave velocity in air since
k0 k0
vp = c=  2 c<c
β k0 + ka2

For this reason, a surface-wave is also referred to as a slow wave. The field
components are then
Ez = A sin (kd y) e−jβz
jβ
Ey = − A cos (kd y) e−jβz
kd

k0
Hx = −ǫr A cos (kd y) e−jβz
kd η 0
ωǫ
= − A cos (kd y) e−jβz
kd
 8.2 ♣The cylindrical optical fibre 171

in the dielectric and

Ez = A sin (kd a) e−ka (y−a)−jβz

jβ
Ey = − A sin (kd a) e−ka (y−a)−jβz
ka

k0
Hx = − A sin (kd a) e−ka (y−a)−jβz
ka η0
ωǫ0
= − A sin (kd a) e−ka (y−a)−jβz
ka
in air.

8.2 ♣The cylindrical optical fibre

Consider a circularly symmetrical TM mode in a circular dielectric rod of
radius a and permittivity ǫd situated in air. The transverse distribution of
the axial component of the electric field intensity Ez in the rod r ≤ a is

Ez,i = C0 J0 (kc r) , r ≤ a

where
kc2 = γ 2 + kd2 = ω2 µ0 ǫd − β 2
The corresponding Hφ,i is given by
jωǫd
Hφ,i = − C0 J0′ (kc r)
kc
Outside the rod, the fields must be evanescent and we may thus choose

Ez = K0 (ξr)

the modified Bessel function of the second kind and order zero. We write

Ez,0 = D0 K0 (ξr) , r ≥ a

where
ξ 2 = β 2 − k02 = β 2 − ω2 µ0 ǫ0
and
jωǫ0
Hφ,0 = D0 K0′ (ξr) , r ≥ a
ξ
Ez and Hφ must be continuous across the boundary at r = a. Hence

C0 J0 (kc a) = D0 K0 (ξr)
172 8. Dielectric waveguides

FIGURE 8.5. A step index cylindrical dielectric waveguide (optical fibre)

and
ǫd ǫ0
C0 J0′ (kc r) = D0 K0′ (ξr)
kc ξ
and combining,
J0 (kc a) ǫd ξK0 (ξr)
′ =
J0 (kc r) ǫ0 kc K0′ (ξr)
with
kc2 + ξ 2 = ω2 µ0 {ǫd − ǫ0 }
Commercially available optical fibres are of two types however,

• Step-index fibre with a homogeneous central core dielectric and an

outer sheath of a material of a lower dielectric constant

• Graded-index fiber with a central core with an inhomogeneous refrac-

tive index profile.

8.2.1 Step-index cylindrical waveguide

A typical fibre consists of a core guiding region with a higher dielectric con-
stant (high refractive index) surrounded by a low refractive index cladding
region as illustrated in Fig. 8.5. Within either the core or the cladding re-
gion, the propagation constant must be uniform so that the axial electric
and magnetic fields satisfy

∇2 Ez + ω 2 µ0 ǫ0 ǫr − β 2 Ez = 0

or 
∇2 Ez + k02 n2 − β 2 Ez = 0
where
√
n= ǫr
is the refractive index of a non-magnetic dielectric medium.
 8.2 ♣The cylindrical optical fibre 173

An intuitively important point is that Ez does not couple to the other two
components as the electromagnetic field propagates even after reflection at
the cylindrical surfaces. Therefore, since Ez couples only to itself, the wave
equation can be written in cylindrical coordinates as

∂2 1 ∂ 1 ∂2 
2
Ez + Ez + 2 2 Ez + k02 n2 − β 2 Ez = 0
∂r r ∂r r ∂φ

We assume an Ez of the form

Ez (r, φ) = Ez (r) exp (jmφ)

so that the radial function is a solution of the Bessel’s equation


∂2 1 ∂ 2 2 2 m2
Ez + Ez + k0 n − β − 2 Ez = 0
∂r2 r ∂r r

Thus, as long as
kc = k02 n2 − β 2 > 0
valid solutions are the Bessel functions of the first kind and order m since
the fields must remain finite at the centre of the core.
When
k02 n2 − β 2 < 0
the valid solution will be the modified Bessel function of the second kind,
Km (ξr) since the fields must vanish at infinity, where

ξ 2 = β 2 − k02 m2

For the guided mode, β must be such that

k0 ncore > β > k0 ncladd

Therefore, within the guiding layer (core)

Ez (r, φ, z) = A1 Jm (kc r) exp {j (mφ − βz)} ,

Hz (r, φ, z) = A2 Jm (kc r) exp {j (mφ − βz)} , r < a

and outside

Ez (r, φ, z) = B1 Km (ξr) exp {j (mφ − βz)} ,

Hz (r, φ, z) = B2 Km (ξr) exp {j (mφ − βz)} , r > a

The coefficients A1 , A2 , B1 , B2 are determined from the continuity condi-

tion for the tangential fields. From
−
→ − → −
→
∇ × E = −jωµ0 H
174 8. Dielectric waveguides

we can write


j β ∂Ez ∂Hz
Eφ = − − ωµ0
k02 n2 − β 2 r ∂φ ∂r


j ωµ0 ∂Hz ∂Ez
Er = − 2 2 +β
k0 n − β 2 r ∂φ ∂r


j ∂Ez β ∂Hz
Hφ = − 2 ωǫ +
k0 n2 − β 2 ∂r r ∂φ


j ∂Hz ωǫ ∂Ez
Hr = − 2 2 β −
k0 n − β 2 ∂r r ∂φ
In the core region
 
jβ ′ jmωµ0
Er = − 2 A1 Jm (kc r) + A2 Jm (kc r) exp {j (mφ − βz)}
kc βr
 
jβ jm jωµ0 ′
Eφ = − 2 A1 Jm (kc r) − A2 kJm (kc r) exp {j (mφ − βz)}
kc r β
 
jβ ′ jmωǫcore
Hr = − 2 A2 kJm (kc r) + A1 Jm (kc r) exp {j (mφ − βz)}
kc βr
 
jβ jm jmωǫcore ′
Hφ = − 2 A2 Jm (kc r) + A1 kJm (kc r) exp {j (mφ − βz)}
kc r β
and in the cladding
 
jβ ′ jmωµ0
Er = 2 B1 γKm (ξr) + B2 Kn (ξr) exp {j (mφ − βz)}
ξ βr
 
jβ jn jωµ0 ′
Eφ = 2 B1 Km (ξr) − B2 γKm (ξr) exp {j (mφ − βz)}
ξ r β
 
jβ ′ jmωǫcladd
Hr = 2 B2 γKm (ξr) + B1 Km (ξr) exp {j (mφ − βz)}
ξ βr
 
jβ jm jmωǫcladd ′
Hφ = 2 B2 Km (ξr) + B1 γKm (ξr) exp {j (mφ − βz)}
ξ r β
and on matching the boundary conditions
  
Jm (kc a) 0 −Km (γa) 0 A1
 0 J (k a) 0 −K m (ξa)

 m c   A2 
=0
 β ωµ0 ′ βn ωµ0 ′ 
 akc2 Jm (kc a) j kc Jm (kc a) K (ξa)
aξ2 m
j ξ Km (ξa)  B1 
ωǫcore ′ β ωǫcladd ′ βm B2
−j kc Jm (kc a) ak2 Jm (kc a) −j ξ Km (ξa) aξ2 Km (ξa)
c
 8.3 ♣Dielectric-coated conducting wire 175

FIGURE 8.6. Dielectric-coated wire

with a characteristic equation

 2  ′   2 2 
β 2 m2 1 1 Jm (kc a) ′
Km (ξa) ′
k0 ncore Jm (kc a) k02 n2cladd Km
′
(ξa)
+ = + + +
a2 ξ 2 kc2 kc Jm (kc a) ξKm (ξa) kc Jm (kc a) ξKm (ξa)
from which β can be determined. We then have
Jm (kc a)
B1 = A1
ξKm (ξa)
and
Jm (kc a)
B2 = A2
ξKm (ξa)
and on using the continuity of the electric field
  ′ ′
−1
jβm 1 1 Jm (kc a) Km (ξa)
A2 = + + A1
ωµ0 a ξ 2 kc2 kc Jm (kc a) ξKm (ξa)
or on using the continuity of the magnetic field
  −1
jωa n2core Jm
′
(kc a) n2cladd Km
′
(ξa) 1 1
A2 = + + A1
βm kc Jm (kc a) ξKm (ξa) ξ 2 kc2

8.3 ♣Dielectric-coated conducting wire

The cross-section of a dielectric-coated wire is shown in Fig. 8.6. For circu-
larly symmetric E modes, we let kc = kd in the dielectric and kc = jka in
the air region surrounding the structure. Since ∂/∂φ = 0 in this case, Ez
is a solution of the Bessel equation
∂ 2 Ez 1 ∂Ez
2
+ + kc2 Ez = 0
∂r r ∂r
176 8. Dielectric waveguides

In the region a < r < b, Ez is generally a linear combination of the

zeroth order Bessel functions of the first and second kinds since the origin
is not included in the region of interest in this case. Thus

Ez = A1 J0 (kd r) + A2 Y0 (kd r) , a < r < b

In the region r > b, the field must decay with increasing r to correspond to a
surface-wave. We recall that for large ka r we can make the approximations
 
2 π
J0 (jka r) ∼ cos jka r −
jπka r 4

and 
2  π
Y0 (jka r) ∼ sin jka r −
jπka r 4
Hence,
  
2 π 
J0 (jka r) + jY0 (jka r) ∼ 2 exp j jka r −
jπka r 4

This is a Hankel’s function of the first kind and order zero.

The Bessel functions with imaginary arguments can be replaced by the
modified Bessel functions, In (y) and Kn (y) defined as

In (y) = j −n Jn (jy)

and
π
Kn (y) = j n+1 {Jn (jy) + jYn (jy)}
2
We note that Kn (y) decays exponentially with increasing y making it a
suitable solution for Ez in the region r > b for n = 0. In (y) on the other
hand, increases exponentially and is therefore not valid for physical fields.
Thus,
Ez = A3 K0 (ka r) , r > b
Since Ez = 0 at r = a, (electric wall), then it is necessary that

A1 J0 (kd a) + A2 Y0 (kd a) = 0

or
J0 (kd a)
A2 = −A1
Y0 (kd a)
At r = b, Ez is continuous so that

J0 (kd a)
A3 K0 (ka b) = A1 J0 (kd b) − Y0 (kd b)
Y0 (kd a)
 8.3 ♣Dielectric-coated conducting wire 177

From the dispersion relation

β 2 = ǫr k02 − kd2 = k02 + ka2

we can write
kd2 + ka2 = (ǫr − 1) k02
We invoke the continuity condition of the tangential magnetic field at r = b
so that, in terms of Ez , we have
−
→ jβ −
→
E t = − 2 ∇ t Ez
kc

With
−
→ ∂
∂
φ ∂
∇ t = r
+ = r

∂r r ∂φ ∂r
we have
−
→ jβ ∂Ez
E t = r
Er = −

r 2
kc ∂r
and
−
→
−
→
φ = z
× E t , ηT M = β η
H t = φH
ηT M k
Noting that
k ǫr k0
kc = kd , = = ωǫ
η η0
in dielectric and
k k0
kc = jka , = = ωǫ0
η η0
in air, the transverse fields are found to be

jβ
Er = {kd A1 J0′ (kd r) + kd A2 Y0′ (kd r)}
kd2

and
ǫr k0 ωǫ
Hφ = Er = Er
βη0 β
in the region a < r < b and

jβ k0 ωǫ0
Er = ka A3 K0′ (ka r) , Hφ = Er = Er
ka2 βη0 β

in the region r > b. Invoking the continuity of Hφ at r = b, we obtain

k0 ǫr k0
j A3 K0′ (ka b) = −j {A1 J0′ (kd b) + A2 Y0′ (kd b)}
ka η 0 kd η0
178 8. Dielectric waveguides

or 
J0 (kd a)
A3 kd K0′ (ka b) ′ ′
= −ǫr ka A1 J0 (kd b) + Y0 (kd b)
Y0 (kd a)
Hence, we can write
kd K0′ (ka b) J ′ (kd b) Y0 (kd a) − Y0′ (kd b) J0 (kd a)
= −ǫr ka 0
K0 (ka b) J0 (kd b) Y0 (kd a) − Y0 (kd b) J0 (kd a)

When d = b − a << a, J0 (kd b) can be expanded about kd a in a Taylor’s

series as
 
d
J0 (kd b) ∼ J0 (kd a) + (J0 (kd b)) kd (b − a)
d (ka b) kd a

and since a << λ0 in a typical surface-wave guide, then ka b is small for

small d. Hence,
1
K0 (ka b) ∼ − ln (0, 89ka b) , K0′ (ka b) = K1 (ka b) ∼
ka b
and
ǫr ka2 b ln (0, 89ka b) = − (ǫr − 1) k02 d + ka2 d
As an example, a copper wire with a = 0, 09 cm, b = 0, 1 cm, d =
0, 01 cm, ǫr = 2, 56 and λ0 = 3, 14 cm,

ka ∼ 0, 258 N p/cm, k0 ∼ 2 ⇒ ka2 << k02

and
β = 2, 02 rad/cm
For large ka r, 
π
K0 (ka r) ∼ exp (−ka r)
2ka r
Thus at r ∼ 2λ0 , the field is down a factor 0, 052 from its value at r = b.
The effective waveguide is thus about 4λ0 . Therefore surface-wave guides
are practical only at high frequencies.

8.4 ♣Inhomogeneously-filled parallel-plate

transmission line
Fig. 8.7 shows the cross-section of an inhomogeneously-filled parallel-plate
transmission line. The plate separation is b and the thickness of the dielec-
tric is a with the dielectric resting intimately on the bottom conducting
plate referred to as the ground plane. The plates are assumed to be per-
fectly conducting and of infinite extent.
 8.4 ♣Inhomogeneously-filled parallel-plate transmission line 179

FIGURE 8.7. An inhomogeneously-filled parallel-plate transmission line

At low frequencies, the transmission line has similar properties to the

two-conductor
√ lines discussed earlier with a phase propagation constant
β = ω L′ C ′ . However, at higher frequencies, β increases faster than ω and
hence, the line exhibits dispersion. As we shall see, there exists a surface-
wave mode of propagation in the vicinity of the air/dielectric interface.
In analysing the transmission line, we apply the magnetic wall concept,
which as we can recall is nothing more than a mathematical artefact. Thus
we require that at the magnetic wall
−
→
n

×H =0

and
−
→
n

· E =0
We can insert these walls safely along planes for which the tangential mag-
netic field is zero without disturbing the field. Such planes correspond to
certain symmetry planes in a given field problem.
We assume that the electric field does not vary in the x-coordinate and is
therefore a function of y and z only. Thus, only Ey , Ez and Hx are present.
This then allows us to place a magnetic wall along any x = const. plane
without disturbing the field. In our present case, we choose the planes
x = ±W as shown in Fig. 8.7. The problem can now be treated as a
waveguide structure with

• Electric walls at the conducting plates at y = 0, b

• Magnetic walls at x = ±W

In an ideal transmission line, the propagation constant is equal to that

of a plane or TEM wave in the surrounding medium. Since the situation in
the present problem is non-uniform, then we can anticipate β to have an
180 8. Dielectric waveguides

intermediate value between its value in air and the dielectric. i.e.,
√
k0 = ω µ0 ǫ0 < β < k

Ez is a solution of the Helmholtz wave equation,

∂ 2 Ez
+ kc2 Ez = 0
∂y 2
where

kc2 = k02 − β 2 , in the air region

= k2 − β 2 , in the dielectric region

We note that β must of necessity be the same in both air and the dielec-
tric regions at the interface since the tangential components of the fields
are continuous across the air/dielectric interface for all z. Thus, let kc = kd
in the dielectric and kc = ka in air. Thus,

ka2 − k02 = kd2 − k2

or
(ǫr − 1) k02 = kd2 − ka2
In the two regions,
∂ 2 Ez
+ kd2 Ez = 0, 0 ≤ y ≤ a
∂y 2
and
∂ 2 Ez
+ ka2 Ez = 0, a ≤ y ≤ b
∂y2
with boundary conditions

Ez (y) = 0, y = 0, b

and continuous at y = a and continuity of the magnetic field at y = a

yields,
ǫr ∂Ez 1 ∂Ez
2 = 2 , @y=a
kd ∂y ka ∂y
Thus
β ∂Ez
Ey = −j
kc2 ∂y
and
1 β ∂Ez
Hx = j
η e kc2 ∂y
where
β
ηe = η
k
 8.4 ♣Inhomogeneously-filled parallel-plate transmission line 181

in the dielectric and

β
ηe = η
k0 0
in air. For the boundary conditions at y = 0, a, b to be satisfied, we require

Ez = A1 sin kd y, 0 ≤ y ≤ a
= A2 sin {ka (b − y)} , a ≤ y ≤ b

Hence
A1 sin kd a = A2 sin ka d
and continuity of Hx at y = a requires
ǫr 1
A1 cos kd a = − A2 cos ka d
kd ka
Dividing the two equations we obtain

kd tan kd a = −ǫr ka tan ka d

and recall that

kd2 − ka2 = (ǫr − 1) k02
Solutions for kd and ka are an infinite number so that an infinite number
of E modes are possible. Since
 
β = k02 − ka2 = k2 − kd2

most of the modes will be evanescent since for real β, we must have ka < k0
and kd < k. Thus, for β lying between k0 and k, ka must be imaginary, say
ka = −jka,0 . We assume a corresponding value kd = kd,0 and hence

kd,0 tan kd,0 a = ǫr ka,0 tanh ka,0 d

and
2 2
kd,0 + ka,0 = (ǫr − 1) k02
The problem can be attacked separately at low and high frequencies.

8.4.1 Low frequency solution

At very low frequencies, k02 is very small and both kd,0 and ka,0 are as a
consequence very small. Thus

tan kd,0 a ∼ kd,0 a

and
tanh ka,0 d ∼ ka,0 d
182 8. Dielectric waveguides

Thus, we have
2
kd,0 a = ǫ2r ka,0 d
or
2 (ǫr − 1) k02 a
ka,0 =
a + ǫr d
and 
 ǫr b
2 2 √
β = k0 + ka,0 = k0 = k0 ǫef f
a + ǫr d
where ǫef f is an effective dielectric constant.
If the current density Jz on the inner surface of the upper plate is uniform
then the magnetic field between the plates is given by Hx = Jz (electric
wall). Thus, the time-average energy stored in the magnetic field is given
by
b W
µ0 µ 1
Um = Hx2 dx dy = 0 W bJz2 = L′ Iz2
4 2 4
0 −W

where
Iz = 2W Jz
and
µ0 b
L′ =
2W
The capacitance per unit length can be regarded as two series capacitances,
′ ′
one for the air region, Cair and the other for the dielectric, Cdiel , where

′ 2ǫ0 W ′ 2ǫr ǫ0 W
Cair = , Cdiel =
d a
so that  −1
a 2ǫr ǫ0 W
C ′ = 2ǫ0 W d+ =
ǫr a + ǫr d
and
ǫr ǫ0 µ0 b ǫr
L′ C ′ = = ǫ0 µ0 b
a + ǫr d a + ǫr d
At low frequencies, we can of course write
sin kd,0 a kd,0 a
A2 = A1 ∼ −jA1
j sinh ka,0 d ka,0 d
In the dielectric, 0 ≤ y ≤ a, we then have

Ez = A1 kd,0 y

jβ b
Ey = − A1 = −jA1
kd,0 (ǫr − 1) d
 8.4 ♣Inhomogeneously-filled parallel-plate transmission line 183

jǫr k0 A1 ǫr (ǫr d + a)
Hx = A1 = j
kd,0 η 0 η0 (ǫr − 1) d
and in the air region, a ≤ y ≤ b,
kd,0 a
Ez = A1 (b − y)
d

jβkd,0 a jβ b
Ey = − 2 A1 = −j ǫr A1 = jA1 ǫr
ka,0 kd,0 (ǫr − 1) d

jk0 A1 ǫr (ǫr d + a)
Hx = A1 = j
kd,0 η 0 η0 (ǫr − 1) d
We note therefore Ez → 0 as k0 , kd,0 → 0 while Ey and Hx remain constant,
hence a TEM situation. Thus, letting

b 
b
V = − Ey dy = jA1 (a + ǫr d)
(ǫr − 1) d
0

the total z-directed current Iz = 2W Jz = 2W Hx so that

 
V η0 (ǫr d + a) L′
ηe = = b=
Iz W ǫr C′

like in the TEM case.

8.4.2 High-frequency solution

When the frequency is high, k0 , kd,0 , ka,0 are large so that ka,0 d is also large,
hence tanh ka,0 d → 1. We therefore have

kd,0 tan kd,0 a = ǫr ka,0 = ǫr (ǫr − 1) k02 − kd,02

which can be seen to be independent of b, the plate separation. The top

plate can thus be placed at ∞ without affecting the high frequency solution.
In this case

Ez (y) = A1 sin kd,0 y, 0 ≤ y ≤ a

sinh {ka,0 (b − y)}
= jA2 sinh {ka,0 (b − y)} = A1 sin kd,0 a
sinh ka,0 d
eka,0 (b−y)
∼ A1 sin kd,0 a , a≤y≤b
eka,0 d
∼ A1 sin (kd,0 a) e−ka,0 (y−a) , a ≤ y ≤ b
184 8. Dielectric waveguides

FIGURE 8.8. Graphical solution for the TM modes in a grounded dielectric slab

Thus Ez decays exponentially away from the air-dielectric interface and

is independent of the plate separation as long as ka,0 d = ka,0 (b − a) is
large. Hence with b → ∞, the field is guided by the dielectric sheet and is
referred to as a surface wave since it is confined to the region close to the
dielectric surface. The first root for kd,0 occurs when
π
kd,0 a <
2
2
As k0 → ∞, kd,0 remains bounded but ka,0 (ǫr − 1) k02 − kd,0
2
→ ∞. Hence
2 2
at a high enough frequency, kd,0 << k0 and β ∼ k. Thus, β varies from a
√ √
low-frequency value k0 ǫeff to an asymptotic high-frequency value k0 ǫr
and since β is now not linear with ω, this line exhibits dispersion. A second
surface wave mode occurs for
3π √
π < kd,0 a < , f or k0 ǫr − 1 > π
2
Note that there are other solutions with a real ka = nπ/b where n is an
integer. In this case, β is imaginary and these modes are hence evanescent
so long as π/b > k0 . Cutoff occurs when ka = k0 so that β = 0. Thus
√ √
kd tan kd a = −ǫr k0 tan k0 d = k0 ǫr tan (k0 a ǫr )

since kd = k @ β = 0,
√ √
tan (k0 a ǫr ) = −k0 ǫr tan k0 d

Fig. 8.8 shows a graphical solution for this equation.

9
Antenna theory

Radiation of electromagnetic energy from an electric circuit may be

• Undesirable for example,

— Reduces resonator Q through radiation losses

— Produces interference in operation of other electronic equipment
(electromagnetic interference or EMI)

• Desirable in launching electromagnetic energy into space e.g., wireless

communication through radio waves.

The main consideration of the present topic is this desirable aspect of

radiation and we will look at structures that maximise the radiation ef-
ficiency, i.e. antennas. A number of considerations are of concern in the
design of antennas. These include,

• Relative radiation strength in a given direction described in terms of

the antenna radiation pattern
• Total radiated power for a given excitation
• Input impedance presented by the antenna to the drive element for
the purposes of impedance matching
• Bandwidth of the antenna
• Radiation efficiency and radiation impedance of the antenna
186 9. Antenna theory

FIGURE 9.1. A short current element as a radiating system

• Maximum field strength at given key locations in air/dielectrics for

avoidance of corona discharges and dielectric breakdown.

Antenna analysis may be treated as a field problem, hence involves so-

lution of Maxwell’s equations with the appropriate boundary conditions at
the antenna and at infinity. This is however complicated for most practical
antenna structures and approximate solutions are usually employed.

9.1 Radiation from elemental dipoles

9.1.1 The Hertzian dipole
Fig. 9.1 shows a short linear current element with the current assumed to
have a sinusoidal spatial distribution over the length l. The current element
I is in the +z-direction and assumed to be at the origin of a spherical
coordinate system. The length l of the current element is assumed to be
much smaller than the wavelength so that the current distribution over
the length of the wire can be regarded as uniform. Continuity condition
requires equal and opposite charges at the two ends of the conductor and
for this reason, this structure is an electric dipole referred to as the Hertzian
dipole.
Let the current in the wire be

i (t) = Re [I (z) exp (jωt)]

 9.1 Radiation from elemental dipoles 187

Since the current vanishes at the ends of the wire, charges must be deposited
there. We have
d
i (t) = ± Q (t)
dt
where
Q (t) = Re [Q (z) exp (jωt)]
and
I (z) = ±jωQ (z)
or
I (z)
Q (z) = ±
jω
The pair of positive and negative charges separated by a short distance
constitutes an electric dipole of dipole moment
−
→
p = z
Q (z) l

At a point P a distance r from the current element, the magnetic vector

−
→
potential A is entirely in the z-direction. Hence
µIl  ωr  µIl
Az = exp −j = exp (−jkr)
4πr v 4πr
where v is the wave phase velocity. We can then express this in spherical
coordinates as
µIl
Ar = Az cos θ = exp (−jkr) cos θ
4πr
µIl
Aθ = −Az sin θ = − exp (−jkr) sin θ
4πr
Aφ = 0
Due to symmetry about the axis
∂
=0
∂φ
We therefore have

 
−
→ 1−→ − →
1 ∂ ∂Aθ ∂Ar
H = ∇×A =φ r −
µ µr ∂r ∂r ∂θ

2 
= −φ
Il k2 sin θ 1 + 1
exp (−jkr)
4π jkr jkr

and
 
−
→ 1 −→ −
→ 1 1 ∂

θ ∂
E = ∇×H = r
(Hφ sin θ) − (rHφ )
jωǫ jωǫ r sin θ ∂θ r ∂r
188 9. Antenna theory

or 
2
3 
Il 2 1 1
Er = − ηk cos θ + exp (−jkr)
2π jkr jkr

2
3 
Il 2 1 1 1
Eθ = − ηk sin θ + + exp (−jkr)
4π jkr jkr jkr
and
Eφ = 0

Near field
In the region near the Hertzian dipole (near zone),
2πr
kr = << 1
λ
so that,
1 1
exp (−jkr) ∼ 1 + (−jkr) + (−jkr)2 + (−jkr)3 + . . . ∼ 1
2! 3!
Thus, the term
Idl
Hφ ∼ sin θ
4πr2
dominates and
Il η
Er ∼ 2 cos θ
4πr3 jk
Il η
Eθ ∼ sin θ
4πr3 jk
These expressions describe quasi-static fields and we see that Hφ is in phase
with the current near the current element and can be identified with the
induced field of Ampere’s law while the near-field electric field is identified
with the field obtained from an electric dipole. The H- and E-field compo-
nents are 90◦ out of phase and represent no net energy flow according to
Poynting’s theorem. This zone is referred to as the reactive zone.

Far field
In the far field, we have
2πr
kr = >> 1
λ
Hence
Il
Hφ ∼ j sin θ exp (−jkr)
4πr
Il
Eθ ∼ j ηk sin θ exp (−jkr)
4πr
 9.1 Radiation from elemental dipoles 189

and Eθ and Hφ are in space quadrature but in time phase. We note


Eθ µ
η= = ∼ 120π Ω
Hφ ǫ

in free space. The Poynting vector is radial and the time-average radiated
power is hence

1 −
→ → 
− k 2 I 2 l2
Pr = Re E θ × H ∗φ = η0 sin2 θ
2 32π2 r2
The total power over a spherical surface at r from the current element is
 π π
2
−
→ − →  k2 I 2 l2 3 l
PT = P ·d S = Pr 2πr2 sin θ dθ = η 0 sin θ dθ = 40π I2 2
W
16π2 λ
S 0 0

since
π π π
3 2

sin θ dθ = sin θ sin θ dθ = 1 − cos2 θ sin θ dθ
0 0 0
1 ' 3 (π
= [− cos θ]π0 + cos θ 0
3
4
=
3
The far field zone is also called the Fraunhofer zone.
In the intermediate zone called the Fresnel zone, both radiative and
stored energy exists. At a distance rf z from the current element such that

2
1 1 1 λ λ
− = 0 ⇒ rf z = = ∼
krf z krf z k 2π 6

the two components are equal. rf z therefore marks the boundary between
the near and far zones.

9.1.2 Elemental magnetic dipole

We consider the small current loop depicted in Fig. 9.2. The small filamen-
tary loop is assumed to carry a uniform time-harmonic current

i (t) = I cos ωt

hence, constitutes an elemental magnetic dipole with a phasor representa-

tion
−
→
m = z
Iπb2 = z
m
190 9. Antenna theory

We can then derive the vector potential as


→ µI exp (−jkr′′ ) ′
−
A= dl
4π r′′
where r′′ changes with the location of dl′ . For a small loop, we can rewrite
exp (−jkr′′ ) = exp (−jkr) exp {−jk (r′′ − r)}

and expand this as

exp (−jkr′′ ) ∼ exp (−jkr) {1 − jk (r′′ − r)}

since r′′ − r is very small at a large distance from the current loop. Thus,
  ′ 
−
→ µI dl ′
A= exp (−jkr) (1 − jkr) − jk dl
4π r′′

where we recognise that jk dl′ = 0. We also have

−
→
d l ′ = −

x sin φ + y
cos φbdφ

For every current element Idl′ there is another diametrically opposite,

−
→
contributing to A , an equal but oppositely directed component in the y-
direction and an x-directed component. Hence, we can write
2π 2π
−
→ µI b sin φ′ ′
µI b sin φ′ ′
A = −

x exp (−jkr) (1 − jkr) dφ = φ exp (−jkr) (1 − jkr) dφ
4π r′′ 4π r′′
0 0

The law of cosines gives

r′′2 = r2 + b2 − 2br cos ϕ

where r cos ϕ is the projection of r on the radius OP ′ of the loop. This is

the same length as the projection on the radius OP ′ of the projection of r
on the y-axis, i.e. OP ′′ = r sin θ. Thus,

r′′2 = r2 + b2 − 2br sin θ sin φ′

With r >> b, we have


1 1 b ′
′′
∼ 1 + sin θ sin φ
r r r
Hence
2π

−
→
µIb b
A =φ exp (−jkr) (1 − jkr) 1 + sin θ sin φ sin φ′ dφ′
′
4πr r
0
 9.1 Radiation from elemental dipoles 191

z P(r,θ,π/2)

r
r’
b
ϕ
I O y
φ’ P’ P’’
b
dl’
x

FIGURE 9.2. A magnetic dipole radiator

−
→
µm 2
A =φ β (1 + jkr) exp (−jkr) sin θ, m = Iπb2
4πr2

The electric and magnetic field intensities are then


2 
jωµm 2 1 1
Eφ = k sin θ + exp (−jkr)
4π jkr jkr

2
3 
jωµm 2 1 1
Hr = − k cos θ + exp (−jkr)
2πη jkr jkr

2
3 
jωµm 2 1 1 1
Hθ = − k sin θ + + exp (−jkr)
4πη jkr jkr jkr
or 
2 
jωµIb2 2 1 1
Eφ = k sin θ + exp (−jkr)
4 jkr jkr

2
3 
jωµIb2 2 1 1
Hr = − k cos θ + exp (−jkr)
2η jkr jkr

2
3 
jωµIb2 2 1 1 1
Hθ = − k sin θ + + exp (−jkr)
4η jkr jkr jkr
which is thus the dual of the Hertzian dipole.
Recall that the source-free Maxwell’s curl equations are expressed as
−
→ − → −
→
∇ × E = −jωµ H
192 9. Antenna theory

and
−
→ − → −
→
∇ × H = jωǫ E
−
→ −
→ − → −
→
Thus, exchanging H for E , − E for H , µ for ǫ and ǫ for µ in the two
equations leaves the equations unchanged so that the case of an electric
source can be adapted to one with a magnetic source. We can therefore
represent a current loop of radius b and current I by a magnetic dipole
m = πb2 I which has a magnetic field of the same form as the electric field
of an electric dipole p = Ql. From continuity, the electric charge on one
end of the dipole is
I I
Q= ⇒p= l
jω jω
Replacing Il/jω by µm, the magnetic fields for a magnetic dipole can be
found from those of the Hertzian dipole.
−
→ −
→
If E e and H e are the electric and magnetic field intensities of the
−
→ −
→
Hertzian dipole and E m and H m those of the magnetic dipole, then
−
→
−
→ −
→ −
→ Em
E e = η H m, H e = −
η
as long as the electric and magnetic moments are related as
√
Il = jkm, k = ω µǫ

The far-zone fields are hence

ωµm exp (−jkr)
Eφ ∼ k sin θ
4π r
and
ωµm exp (−jkr)
Hθ ∼ − k sin θ
4πη r
−
→ −
→
E φ and E θ of the magnetic and electric dipoles resp. are in space and time
phase quadrature and they can be combined to create circularly polarised
radiation.

9.2 Thin linear antennas

Consider the dipole of length 2l shown in Fig. 9.3. Since the dipole is centre-
fed, currents in the two halves are symmetrical and go to zero at the dipole
ends. We assume a sinusoidal current distribution in the dipole of the form

I (z) = Im sin k (l − z) , z > 0

= Im sin k (l + z) , z < 0
 9.2 Thin linear antennas 193

The far-field contribution of the current element I (z) dz to the field

measured at the field point is given by
I (z) dz
dEθ = ηdHφ = j ηk sin θ exp (−jβr′ )
4πr′
where 
r′ = r2 + z 2 − 2rz cos θ ∼ r − z cos θ
The magnitude difference between 1/r′ and 1/r is insignificant though and
we can therefore write
l
Im
Eθ = ηHφ = j ηk sin θ exp (−jkr) sin k (l − z) exp (jkz cos θ) dz
4πr
−l

or
l
Im
Eθ = j ηk sin θ exp (−jkr) sin k (l − z) cos (kz cos θ) dz
2πr
0
60πIm
= j exp (−jkr) F (θ)
r
where
l
cos (kl cos θ) − cos kl
F (θ) = k sin θ sin k (l − z) cos (kz cos θ) dz =
sin θ
0

is the E-plane pattern function of the linear dipole antenna and is sym-
metrical about the plane θ = π/2.
Note that we can determine the fields alternatively by recognising that,
since ϕ = θ anywhere along the line, then
l
−
→
N = z
I exp (jkr′ cos θ) dz ′
−l
0
= z
Im sin k (l + z) exp (jkr′ cos θ) dz ′
−l
l
z Im sin k (l − z) exp (jkr′ cos θ) dz ′
+

0

and with

exp (αx)
exp (αx) sin (bx + c) dx = 2 {a sin (bx + c) − b cos (bx + c)}
a + b2
194 9. Antenna theory

z r’
θ

r
dz
l ϕ
z

Im
l

FIGURE 9.3. A 2l thin linear dipole antenna

−
→ 2Im
N = z
{cos (kl cos θ) − cos (kl)}
k sin2 θ
and
Nθ = −Nz sin θ
We therefore have
 
jηIm cos (kl cos θ) − cos (kl)
Eθ = exp (−jkr)
2πr sin θ

and
Eθ
Hφ =
η
as before

9.2.1 The half-wave dipole

If 2l = λ/2, we have
2π π
kl = l=
λ 2
and π
cos θ cos 2
F (θ) =
sin θ
which has a maximum at θ = π/2 and nulls at θ = 0◦ and 180◦ . The
far-field phasors are then given by

exp (−jkr) cos π2 cos θ
Eθ = j60πIm
r sin θ
 9.3 Antenna patterns and parameters 195

jIm cos π2 cos θ exp (−jkr)
Hφ =
2π sin θ r
Hence 
−  π 2
→  1   15Im 2 cos cos θ
 P ave  = Eθ Hφ∗  = 2
2
2 πr sin θ
and the radiation intensity
2
 2
ηIm cos (kl cos θ) − cos kl
K=
8π2 sin θ
which for the half-wave dipole reduces to
  2
15Im2 cos π2 cos θ
K= W/sr
π sin θ

The total radiated power

2ππ  π
−→  2 2 1 π 
Pr =  P ave  r sin θdθdφ = 30Im cos2 cos θ dθ
sin θ 2
0 0 0
 2 2
∼ 1, 218 30Im = 36, 54Im W

and
2Pr
Rr = 2
∼ 73, 1 Ω
Im
The input reactance of the dipole is small and inductive and can be elimi-
nated by making 2l slightly less than λ/2. The half-dipole is a near-resonant
structure for thin wires for which the open-circuit uniform transmission line
approximation is valid. The resonant length and resonant terminal resis-
tance decrease with increasing wire radius to wavelength ratio.

9.3 Antenna patterns and parameters

The main purpose of an antenna is to provide impedance matching between
the wave guiding structure and free space as well as to radiate the energy
in the desired direction. For the 2l-dipole, the radiation intensity is given
by
 2  2
ηI 2 cos (kl cos θ) − cos kl ηI 2 cos kl − cos kl
K = m2 ∼ m2 =0
8π sin θ 8π θ
for small θ. Hence, the z-axis is a null and K is symmetrical about the
z-axis. No physical antenna radiates uniformly in all directions in space
therefore.
196 9. Antenna theory

z y

90o x x
z

(a) E-plane (b) H-plane

FIGURE 9.4. The E-and H-plane radiation pattern of an isolated Hertzian dipole

The radiation pattern or antenna pattern is a graph of the relative far-

zone (Fraunhofer zone) field strength against direction at a fixed distance
from the antenna. While the pattern is 3-D, we can plot the magnitude
of the normalised field strength versus the elevation angle θ for constant
azimuthal angle φ, the so-called E-plane pattern and a corresponding one
for constant θ = π/2 against φ (H-pattern).
Example: Plot the E-plane and H-plane antenna patterns for the Hertzian
dipole.
Eθ and Hφ are proportional in the Fraunhofer zone. For a given r, Eθ
is independent of φ so that the normalised E-field

|Eθ |n = |sin θ|

Thus, the E-plane pattern is a pair of circles forming a figure-8 as shown

in Fig. 9.4(a). At θ = π/2, |Eθ |n = 1 and the H-plane pattern is therefore
a circle of unit radius centred at the z-directed dipole. This is shown in Fig.
9.4(b).
Important parameters which describe the antenna pattern are

• The width of the main lobe,

• The level of any side lobes
• The directivity of the antenna

Directivity is a measure of the ability of the antenna to direct radiated

power in a given direction. Recall that the radiation intensity K (W/sr)
is the time-average power per unit solid angle of the antenna radiation
pattern used to determine the directive gain of an antenna. Since a unit
spherical surface of area r2 subtends a unit solid angle at the origin, the
radiation intensity is numerically equal to r2 times the time-average power
per unit area, i.e.,
−
→
K = r2 P ave
 9.3 Antenna patterns and parameters 197

The total radiated time-average power is hence

 
−
→ −
→
Pr = P ave · d S = KdΩ, (W )
S Ω

where dΩ = sin θdφ. The directive gain of the antenna is hence defined as
K (θ, φ) K (θ, φ)
GD (θ, φ) = = 4π 
Pr /4π
KdΩ
Ω

The maximum directive gain is defined as the directivity of the antenna and
is the ratio of the maximum radiation intensity to the average radiation
intensity and usually denoted by
Kmax Kmax
D = GD,max = = 4π
Kave Pr
This can be expressed as

4π |Emax |2
D=
2ππ
|E (θ, φ)|2 sin θdθdφ
0 0

For the half-wave dipole, we have

  2
15Im2 cos π2 cos θ
K= W/sr
π sin θ

which gives the directivity of the half-wave dipole as

4πKmax
D= = 1, 64
Pr
where −  π  15I 2
→  m
Kmax = r2  P ave =
2 π
The directivity of the half-wave dipole is therefore 2, 15 dBi, i.e. referred
to an omnidirectional or isotropic antenna. The half-power beam-width of
the pattern is the angle between the solutions of the pattern factor
 
cos π2 cos θ 1
= , 0<θ<π
sin θ 2
and is 78◦ for the half-wave dipole making this antenna only slightly more
directive than the Hertzian dipole which has a directivity 1, 76 dB and π/2
beam-width.
198 9. Antenna theory

FIGURE 9.5. The radiation pattern of a half-wave and the Hertzian dipoles

Example: Determine the directive gain and directivity of a Hertzian di-

pole.
For the Hertzian dipole,
−
→ 1 −
→ − →  1
P ave = Re E × H ∗ = |Eθ Hφ |
2 2
Thus
(Il)2
K (θ, φ) = η k2 sin2 θ
32π2 0
and
4π sin2 θ 3
GD (θ, φ) = = sin2 θ
2ππ 2
 2
sin θ sin θdθdφ
0 0
or π 
D = GD,max = GD , φ = 1, 5 = 1, 76 dBi
2
Fig. 9.5 illustrates a comparison of the radiation intensity (radiation pat-
tern) as a function of θ for a half-wave dipole with that of the Hertzian
dipole.
A measure of the antenna radiation efficiency is given by the power gain
Gp . This is the ratio of the antenna maximum radiation intensity to the
radiation intensity of a lossless isotropic source with the same power input.
Because of power loss in the antenna and nearby lossy structures which
include ground, the radiated power Pr is less than the total input power
Pi . Thus,
Pi = Pr + Pl
From
4πKmax
Gp =
Pi
 9.3 Antenna patterns and parameters 199

the radiation efficiency is the ratio of the power gain to the directivity,
Gp Pr
ηr = =
D Pi
The amount of power radiated can be given in terms of an equivalent
radiation resistance Rr which is a fictitious resistance that would dissipate
the same amount of power as Pr when the current through this resistance
is the maximum current along the antenna. The radiation resistance can
be obtained from the input impedance of the antenna regarded as a series
frequency-dependent impedance of the form

Zi = R + jX

where R and X can be calculated from a field consideration of the antenna

with the appropriate boundary conditions. Zi is important for optimum
matching into the antenna.
We note that R is a measure of the total effect of the radiated power
and any resistive losses and can be considered to be two series connected
resistances. If I is the magnitude of the sinusoidally distributed current on
the antenna, then
∆ 2Prad
Rr =
I2
and
2Pl
Rloss = 2
I
Rr is the radiation resistance which for the Hertzian dipole, is

2
2 l
Rr = 80π
λ
and for the half-wave dipole, Rr ∼ 73 Ω.
The power loss for the same Hertzian element is given by
1 − 
→ 2
Pl = Rm  J s  A
2
where
I
Js =
2πa
is the surface current and A = 2πal is the surface area of the conductor
and Rm the surface resistance. Hence
Rm l 2
Pl = I
4πa
or
Rm l
Rl =
2πa
200 9. Antenna theory

The radiation efficiency is then

Rr 80 (l/λ)2 π 2
ηr = =
Rr + Rl 80 (l/λ)2 π2 + lRm / (2πa)
80 (l/λ)
=
80 (l/λ) + Rm λ/ (2π 3 a)
We note therefore that the radiation efficiency decreases with decreasing
l/λ.
For an antenna with l << λ and with no end loading of the dipole, the
current decreases linearly from the point of excitation to the conductor end.
I 2 then has an average uniform value along the antenna that is a quarter
2
of the current amplitude Im . The radiated power is then a quarter of the
value of the uniform-current excitation and

2
2Prad 2 l
Rr = 2
= 20π
Im λ
and average losses are also half those of the uniform-current case. For a
lossless filamentary half-wave dipole,
π 
2Prad η0 cos2 π2 cos θ
Rr = 2
= dθ ∼ 73, 09 Ω
Im 2π sin θ
0

For a small loop antenna,

4
2 circumf erence
Rr = 20π
λ
Example: Determine the radiation resistance of the Hertzian dipole.
Assuming little ohmic losses in the dipole, the time-average power ra-
diated by the Hertzian dipole for an input time-harmonic current with an
amplitude I is given by
2ππ
1 ' (
Pr = Re Eθ Hφ∗ r2 sin θdθdφ
2
0 0
2ππ
(Il)2 2  2
= ηk sin θ sin θdθdφ
32π2
0 0

2 
2
(Il) I2 l
= ηk2 = 80π 2
12π 2 λ

and since the current along the dipole is uniform,

2
l
Rr = 80π
λ
 9.3 Antenna patterns and parameters 201

Example: Determine the radiation efficiency of an isolated Hertzian di-

pole made of a metal wire of radius a and a length l and conductivity σ.
We let I be the current in the wire dipole having a loss resistance Rl .
The ohmic loss is hence
1
Pl = I 2 Rl
2
and the radiated power
1
Pr = I 2 Rr
2
Thus,
Pr Rr 1
ηr = = =
Pr + Pl Rr + Rl 1 + Rl /Rr
Now,

l πf µ0 l
Rl = Rm =
2πa σ 2πa
Hence 


Rm λ λ
ηr = 1 + 2
160π a l
For a = 1, 8 mm, l = 2 cm, f = 1, 5 M Hz, σ = 5, 8×107 Sm−1 , ηr ∼ 58%.
Example: A thin quarter-wave vertical antenna over a perfectly conduct-
ing ground plane is excited by a sinusoidal source at its base. Determine its
radiation pattern, radiation resistance and directivity.
We use the method of images as shown in Fig. 9.6. The electromagnetic
field in the upper half-space due to the quarter-wave antenna is the same as
that of a half-wave dipole with
− the E-plane pattern function applying only
→ 
in the range 0 ≤ θ ≤ π/2.  P ave  for the quarter-wave dipole applies for
0 ≤ θ ≤ π/2 so its radiated power is half that of the λ/2 antenna. Thus
2
Pr = 18, 27Im

2Pr
Rr = 2
= 36, 54 Ω
Im
and
Kmax Kmax
D= = = 1, 64
Kave Pr /4π
the same as the λ/2 antenna.

9.3.1 Antenna effective length

For thin linear antennas with a given current distribution, we can define an
effective length le to which the far-field is proportional. For the half-wave
202 9. Antenna theory
z z

λ/4 λ/4
I

λ/4
I

FIGURE 9.6. A quarter-wave monopole

dipole carrying a current with a general phasor distribution I (z),

l
exp (−jkr)
Eθ = η0 Hφ = j30k sin θ I (z) exp (jkz cos θ) dz
r
−l

Let I (0) be the input current at the feed point of the antenna. Then,

exp (−jkr)
Eθ = j30I (0) k le (θ)
r
where
l
sin θ
le (θ) = I (z) exp (jkz cos θ) dz
I (0)
−l

is the effective length of the transmitting antenna. Thus, le measures the

effectiveness of the antenna as a radiator. In most cases, the important
value of le is at θ = π/2 where

l
1
le = I (z) dz
I (0)
−l

Example: Assume a sinusoidal current distribution on a centre-fed thin

straight half-wave dipole. Determine its effective length. What is its maxi-
mum value?

λ/4

λ
le (θ) = sin θ sin k − |z| exp (jkz cos θ) dz
4
−λ/4
 9.4 Antenna array theory 203

or  π 
2 cos 2 cos θ λ λ
le (θ) = ⇒ le = le,max (θ) = <
k sin θ π 2

Note that le (θ) is only meaningful for relatively short antennas with
maximum current at the feed point. For a receiving antenna, the effective
length is defined as the ratio of the open-circuit voltage Voc induced at the
antenna terminals to the effective field intensity |Ei | at the antenna that
induces it,
Voc
le (θ) = −
Ei
This is the same as when the antenna is used for transmitting. If Ei is not
parallel to the dipole, there is a polarisation mismatch and the magnitude
of the open-circuit voltage will be
 − 
 →
|Voc | = le E i 

9.4 Antenna array theory

In array theory, the usual problem is the determination of the radiation
pattern for a group of similar radiating elements with similar current or field
distributions although the magnitudes and the phasing may be different
between individual radiators. Like in the far field radiation problem of a
single radiator of finite dimensions, we assume that the distance differences
from individual radiators in an array observed from a distant field point
are significant only as far as the phase is concerned.
Consider the two-element array shown in Fig. 9.7 consisting of two iden-
tical radiating elements a distance d apart. We assume that the far-field of
each element is in the θ-direction and the antennas are aligned along the
x-axis. The elements are excited by equal-amplitude currents but with the
current on element 1 leading that of element 0 by an angle ϕ. The electric
field intensities of the two elements are then given by

exp (−jkr0 )
E0 = Em F (θ, φ)
r0

exp (−jkr1 )
E1 = Em F (θ, φ)
r1
and the electric field intensity at P is hence

exp (−jkr0 ) exp (−jkr1 )
E = E0 + E1 = Em F (θ, φ) +
r0 r1
204 9. Antenna theory

r0
r1

φ
0 1
x
d

FIGURE 9.7. A two-element array

In the far-field, r0 >> d/2 and the difference between r0 and r1 is suffi-
ciently small such that r1 ∼ r0 in the magnitude expression. In the phase
factor, we retain this difference however since k may be large. Hence,

r1 ∼ r0 − d sin θ cos φ

and
exp (−jkr0 )
E = Em F (θ, φ) [1 + exp {j (kd sin θ cos φ + ϕ)}]
r0
exp (−jkr0 )  ϕ
ψ

= Em F (θ, φ) exp j 2 cos
r0 2 2
where
ψ = kd sin θ cos φ + ϕ
and  
2Em  ψ
|E| = |F (θ, φ)| cos 
r0 2
 
 
|F (θ, φ)| may be called the element factor and cos ψ2  the space factor of
the array or array factor.
The pattern function of an array of identical elements is described by
the product of the element and array factors (principle of pattern mul-
tiplication). For two parallel z-directed half-wave dipoles, the total field
magnitude is hence,
    
2Em  cos π2 cos θ   ψ 
|E| =   cos 
r0  sin θ  2

and since ψ is also a function of θ, the E-plane pattern is not the same
as that of a single half-wave dipole except for ψ = ±π/2. In the H-plane,
 
θ = π/2 and the pattern is determined by cos ψ2 .
 9.4 Antenna array theory 205

FIGURE 9.8. A two-element broadside array

Example: Plot the H-plane radiation patterns of two parallel dipoles for
the following cases

(a) d = λ/2, ϕ = 0

(b) d = λ/4, ϕ = −π/2

Let the dipoles be z-directed and placed along the x-axis. In the H-plane
where θ = π/2, each dipole is omnidirectional and the normalised pattern
function is equal to the normalised array factor |A (φ)|. Thus,
  

 ψ  kd 
|A (φ)| = cos  = cos cos φ + ϕ 
2 2

(a) d = λ/2, kd = π, ϕ = 0
 π 
 
|A (φ)| = cos cos φ 
2
The pattern has a maximum at φ0 = ±π/2, i.e. broadside direction.
Fig. 9.8.

(b) d = λ/4, kd = π/2, ϕ = −π/2

 π 
 
|A (φ)| = cos (cos φ − 1) 
4
with a maximum at φ = 0◦ and vanishes at φ = π. The direction of
the pattern maximum is along the line of the array i.e. endfire array
as shown in Fig. 9.9.
206 9. Antenna theory

FIGURE 9.9. A two-element endfire array

Example: Discuss the radiation pattern of a linear array of three isotropic

sources λ/2 apart. The excitations are in-phase and have amplitudes in the
ratio 1 : 2 : 1.
Fig. 9.10(a) illustrates the situation. This is equivalent to two two-element
arrays displaced λ/2 from each other as shown in Fig. 9.10(b). Each can
be considered as a radiating source with an element factor
 π 
 
|A (φ)| = cos cos φ 
2
and an array factor of the same value.
By the principle of pattern multiplication,
4Em   π 2

|E| = cos cos φ 
r0 2
and exhibits no side lobes, a characteristic of binomial arrays in which d is
restricted to λ/2.

9.4.1 General uniform linear arrays

Consider an array of N identical antennas equally spaced along a straight
line as shown in Fig. 9.11. The antennas are fed with currents of equal
magnitude and are assumed to have a uniform progressive phase shift along
the line, the so called uniform line array. Since the elements are identical,
the array pattern factor is the product of the element and the array factors.
We seek to establish how the array factor depends on kd = 2πd/λ and the
progressive phase shift ϕ between neighbouring elements. In the xy-plane,
the normalised array factor is given by
1
|A (φ)| = |1 + exp (jψ) + exp (j2ψ) + . . . + exp (j (N − 1) ψ)|
N
 9.4 Antenna array theory 207

1 2 1
x
λ/2 λ/2
φ
(a) 1 2
1
0o

λ/2
1 1
1
x
1

(b)
(c)

FIGURE 9.10. A three isotropic elements array

0 φ 1 2 N-1
x
d

FIGURE 9.11. A uniform linear array

208 9. Antenna theory

where
ψ = kd cos φ + ϕ
Thus  
1  1 − exp (jN ψ)  1 sin Nψ/2
|A (φ)| =   =
N 1 − exp (jψ) N sin ψ/2
As φ varies from 0 to 2π, ψ varies from kd + ϕ to −kd + ϕ covering a
range of 2kd or 4π/λ and defines the visible range of the radiation pattern.
Thus,

(1) the maximum radiation is in the direction for which ψ = 0, i.e.,

kd cos φ0 + ϕ = 0

or
ϕ
cos φ0 = −
kd
The values of ϕ lead to

(1) (a) A broadside array in which maximum radiation occurs in a direc-

tion normal to the axis of the array i.e. for φ0 = ±π/2. This
requires that ϕ = 0 which implies that all elements are excited
in phase.
(b) An endfire array in which maximum radiation occurs in the axial
direction, i.e. for φ0 = 0 or ϕ = −kd cos φ0 = −kd.

(2) A pattern with nulls when |A (φ)| = 0 or

ψ
N = ±nπ, n = 1, 2, . . .
2
Note that the locations of the nulls in φ are different for broadside and
endfire arrays due to the different values of ϕ for the null condition.

(3) An angular width of the main beam between the first nulls for large
N given by
ψ 2π
N 01 = ±π ⇒ ψ01 = ±
2 N
where ψ01 denotes the value of ψ at the first nulls.

9.4.2 The broadside array, (ϕ = 0, φ0 = π/2)

We have ψ = kd cos φ and the first null occurs at φ01 . Thus, the width of
the main beam between the first nulls is

2∆φ = 2 (φ01 − φ0 )
 9.4 Antenna array theory 209

At φ01 , we have
ψ01
cos φ01 = cos (φ0 + ∆φ) =
kd
which for φ0 = π/2, gives
π  2π
cos + ∆φ = − sin ∆φ = −
2 N kd
or
λ λ λ
∆φ = sin−1 ∼ =
Nd Nd l
where l = N d is the length of the array. Thus, for a long uniform broadside
array, the beam width of the main lobe is approximately twice the reciprocal
of the array length in wavelengths.

9.4.3 Endfire array (ϕ = −βd, φ0 = 0)

We have
ψ = kd (cos φ − 1)
or
ψ01 2π λ λ
cos φ01 − 1 = =− =− =−
kd N kd Nd l
For small ∆φ, we have

∆φ2
cos φ01 = cos ∆φ ∼ 1 −
2
Hence  
∆φ2 λ 2λ λ
∼ ⇒ ∆φ ∼ = 2
2 Nd Nd l
which is larger than that of the broadside array.
The side lobes of the radiation occur where the numerator |sin Nψ/2| = 1
or
ψ π
N = ± (2m + 1) , m = 1, 2, . . .
2 2
The first side lobes occur when
ψ π
N = ±3 , m = 1
2 2
where we note that
ψ π
N =± , m=0
2 2
does not represent a location of side lobes because these points are still
within the main lobe region.
210 9. Antenna theory

The first side lobe is usually the strongest and we have

   
1  1   
 ∼ 1  2N  = 2 = 0, 212

N sin (3π/2N )  N 3π  3π


for large N . Thus, for a long uniform array, the first side lobes are 13, 5 dB
below the main lobe. The level of the side lobe can be lowered by tapering
the current distribution in the array elements.
Example: Determine the array factor and plot the normalised radiation
pattern of a linear array of five isotropic elements spaced λ/2 apart and
having excitation amplitude ratio 1 : 2 : 3 : 2 : 1. Compare the level of the
first side-lobe with that of a five-element uniform array.
The normalised array factor is given by

1
|A (φ)| = |1 + 2 exp (jψ) + 3 exp (j2ψ) + 2 exp (j3ψ) + exp (j4ψ)|
9
1
= |exp (j2ψ) {3 + 2 exp (jψ) + exp (j2ψ)}|
9
1
= |{3 + 4 cos ψ + 2 cos 2ψ}|
9
Fig. 9.12 illustrates the radiation pattern of a non-uniform array. This
expression holds for a general ψ = kd cos φ+ϕ. Broadside radiation requires
ϕ = 0, ψ = kd cos φ and for an element spacing d = λ/2, ψ = π cos φ. The
normalised array factor is hence given by

1
|A (φ)| = |{3 + 4 cos (π cos φ) + 2 cos (2π cos φ)}|
9

9.4.4 Phase-scanning of Arrays

From the consideration of the broadside or end-fire arrays, we see that
there will be a particular phase delay in the excitations that gives a main
lobe pointing in a direction intermediate between those of the broadside
and end-fire arrays. If the phase delay is electronically controlled (using
ferrite or semiconductor devices), the direction of the main lobe can be
scanned without having to physically turn the antenna. This is especially
important in large installations requiring instantaneous direction scanning
such as surveillance radar systems at airports.
Let the phase delay between the elements be ∆ϕ. Then for the end-fire
array excited by currents of unequal amplitude and progressive phase ∆ϕ
described by

an = a0 exp (−jn∆ϕ) , n = 0, 1, 2, . . . , N − 1
 9.4 Antenna array theory 211

π 0o

FIGURE 9.12. A five-element non-uniform broadside array with d = λ/2

the space factor of the array is given by

sin {N (kd cos φ − ∆ϕ) /2}

|A (φ)| = a0
sin {(kd cos φ − ∆ϕ) /2}

and the direction of the principal maximum is then

(kd cos φ0 = ∆ϕ)

which is variable if ∆ϕ is variable.

9.4.5 Arrays with unequally spaced array elements

For the linear array with equal element spacing, the radiation can be ex-
pressed in terms of polynomials with the zeros of the polynomial in the com-
plex plane providing useful information about the overall radiation pattern
of the array. Let z represent the complex variable

z = exp (jψ) = exp (jkd cos φ)

Thus
N−1
/
|A (φ)| = an z −n
n=0
212 9. Antenna theory

FIGURE 9.13. Super gain array showing the zero positioning array synthesis
technique

Real φ then corresponds to points on the unit circle in the complex plane
with phase angles between −kd and kd. When the zeros of |A (φ)| lie on
this part of the unit circle, they correspond to the zeros of the pattern or
the cones of silence.
The zeros are uniformly distributed over the entire unit circle for the
broadside array except for the missing one at φ = 0, where the main lobe
builds up. This suggests a technique for array synthesis by positioning zeros
on the unit circle so that they lie close together in the region where the
pattern is to have minimum magnitude and wider apart where it is to
have maximum magnitude. For an element spacing of less than λ/2, real φ
covers only part of the unit circle and since the zeros of the uniform array
are spread over the entire unit circle, some of these are invisible because
they fall on the part of the circle where φ is imaginary.
An arbitrarily high gain antenna can be synthesised by moving the in-
visible zeros to the visible part of the circle but this then creates a very
large lobe in the position vacated by the zeros. While this may seem incon-
sequential, it represents reactive energy which can be a problem especially
with broadside arrays. Uniform arrays with in-phase currents have very
high gain and synthesised arrays with gains higher than these are known
as supergain antennas. Fig. 9.13 shows the zero pattern for an example of
a supergain antenna.
We note that the half-wave dipole can be analysed as an array of Hertzian
dipoles. A common array antenna is the Yagi-Uda array used for TV re-
ceptions in many homes.
 9.5 Antennas for reception 213

A r B
1

FIGURE 9.14. Antenna reciprocity

9.5 Antennas for reception

The same antenna useful as a transmitting antenna works as well when used
as a receiving one. The radiation pattern, directivity and input impedance
play the same important role in antenna design for reception.
In transmission, the source is connected to localised terminals and the
wave is radiated into space in the form of spherical wavefronts. In reception,
the antenna sees a portion of a plane wavefront from a distant transmitting
antenna so that the electric field set up is very different from that of the
localised source in the transmitting case. This implies that the induced cur-
rents must be different for the total field to satisfy the boundary conditions
at the antenna.
The currents set up on the receiving antenna by the plane wave will
convey useful power to the load as well as produce a component of energy
reradiated back into space. Radiation and reception characteristics of the
antenna are related through reciprocity theorems such as
• Antenna pattern for transmission and reception are identical
• The antenna input impedance is the internal impedance of the equiv-
alent generator that represents a receiving system
• An effective area (aperture) for the receiving antenna can be defined
and reciprocity related to directivity
Consider the antenna system of Fig. 9.14. Antenna A transmits electro-
magnetic energy which is absorbed by antenna B placed in the far-zone
of antenna A. Antenna B moves about A at a constant distance r and is
oriented so as to always receive maximum power.
The two antennas and the intervening space can be modelled as a two-
port network as shown in Fig. 9.15. The voltage and current pairs V1 , I1 and
V2 , I2 model the terminal characteristics of the two antennas respectively.
We can therefore write,
V1 = Z11 I1 + Z12 I2
214 9. Antenna theory

Z11-Z12 Z22-Z12
I1 I2

V1 Z12 V2 ZL

FIGURE 9.15. A Z-parameter equivalent circuit representation of coupled anten-

nas

V2 = Z21 I1 + Z22 I2

A bilateral space is such that reciprocity holds i.e., Z21 = Z12 . In practice,
A and B will be separated by large distances such that Z12 = Z21 ∼ 0 in
the limit as r → ∞. Hence

V1 ∼ Z11 I1 = ZA IA

V2 ∼ Z22 I2 = ZB IB
However, the coupling from the transmitting to the receiving antenna can-
not be neglected as it is by this means that energy arrives at B. We can
modify the equivalent two-port as shown in Fig. 9.16 and apply Thevenin’s
theorem at the load ZL looking toward antenna A, whereby the open-circuit
voltage is then seen to be
Z12
Voc = V1
Z11
and the source impedance as

Zg = Z22 − Z12 + (Z11 − Z12 ) //Z12

Z12 (Z11 − Z12 ) Z2
= Z22 − Z12 + = Z22 − 12
Z11 Z11
.Since Z12 is small, the equivalent source impedance for antenna B operated
in receiving mode is approximately its impedance in the transmitting mode
of operation, i.e.
Zg ∼ Z22 = ZB
When B is receiving,
Z21
V2 = −I2 ZL ⇒ I2 = − I1
Z22 + ZL
and the time-average power absorbed in ZL therefore
 2
1 |I1 |2  Z21 
Pr = Re [−V2 I2∗ ] = Re [ZL ]
2 2  Z22 + ZL 
 9.5 Antennas for reception 215

FIGURE 9.16. The modified transceiver equivalent circuit

For two successive positions of B, the ratio of the power absorbed in ZL

is given by  2
Pr (θ1 , φ1 )  Z21 (θ1 , φ1 ) 
=
Pr (θ2 , φ1 ) Z21 (θ2 , φ2 ) 
If we now transmit using B and receive using A, the absorbed power in an
identical load ZL terminating A for the two positions of B would be the
same since Z12 = Z21 for a reciprocal medium. Hence we conclude that, the
radiation pattern of an antenna in reception mode of operation is identical
to the pattern with the antenna in a transmission mode of operation.

9.5.1 Effective aperture or receiving cross-section

For a receiving antenna, we define the effective aperture Ae as the ratio
of the time-average power delivered to a matched load to the time-average
power density of the incident electromagnetic wave at the receiving an-
tenna. We can then write
Pr = Ae Pave
where Pave is the maximum average power transferred to the load under
conditions of maximum power transfer with the receiving antenna properly
oriented w.r.t. the polarisation of the incident wave. For matched condi-
tions, we have
ZL = Zg∗ ∼ RB − jXB
and
|I1 Z21 |2
Pr =
8RB
If we let RA be the radiation resistance of the transmitting antenna A, then
the transmitted power is given by
1
Pt = |I1 |2 RA
2
so that
Pr |Z12 |2
=
Pt 4RA RB
216 9. Antenna theory

FIGURE 9.17. Calculating effect aperture of an antenna

When B is receiving, the time-average power density at B depends on

the directive gain of the transmitting antenna in the direction of B, i.e.

Pt
Pave = GD,A
4πr2
Thus
Pr Ae,B GD,A
=
Pt 4πr2
so that
RA RB Ae,B GD,A
|Z21 |2 =
πr2
If B transmits instead while A receives, then

RA RB Ae,A GD,B
|Z12 |2 =
πr2
Hence,
GD,A GD,B
=
Ae,A Ae,B
The ratio of the directive gain to the effective aperture of an antenna is
the universal constant determined by the directive gain and effective area
of any antenna.
Example: Determine the effective area, Ae (θ) of a Hertzian dipole of
length 2l << λ used to receive an incident plane electromagnetic wave of
wavelength λ polarised as shown in Fig. 9.17.
Let the amplitude of the electric field intensity at the dipole be Ei . Thus,

Ei2
Pave =
2η0
 9.5 Antennas for reception 217

The average power delivered to a matched load ZL = Zg∗ is

 2 2
1  Ei l sin θ  (Ei l)
Pr =   Rr = sin2 θ
2 Zg + Zg∗ 8Rr

where
2
πl
Rr = 80 , (Ω)
λ
The effective aperture is hence


Pr η 3 λ2
Ae (θ) = = (l)2 sin2 θ = sin2 θ
Pave 4Rr 2 4π
which is independent of the dipole length.
For an elemental electric dipole,
3 2
GD (θ) = sin θ
2
Thus
4π
GD (θ) = Ae (θ)
λ2
and
4π
GD (θ, φ) = Ae (θ, φ)
λ2
The universal constant is hence given by
GD (θ, φ) 4π
= 2
Ae (θ, φ) λ
In a thin linear antenna of effective length le (θ), power available to the
antenna under matched conditions is
2
Voc (−le Ei )2
Pr = =
8Rr 8Rr
Hence
30π 2
Ae (θ) = l (θ)
Rr e

9.5.2 Back scatter cross-section

In cases where the incident wave impinges on a passive object whose pres-
ence creates a scattered field, we define a back scatter cross-section or
radar cross-section as the equivalent area that would intercept that amount
of incident power in order to produce the same scattered power density at
the location of the receiver if the object is assumed to scatter the radiation
uniformly in all directions i.e. isotropically.
218 9. Antenna theory

Let Pi be the time-average incident power density at the object, Ps be

the time-average scattered power density at the receiver site, σbs the back
scatter cross-section and r the distance of the scatterer from the receiver.
Then
σbs Pi
= Ps
4πr2
or
Ps
σbs = 4πr2 , m2
Pi
The back scatterer cross-section is a measure of the detectability of the
object (target) by a radar. This depends on the geometry, orientation and
constitutive parameters of the object and the frequency and polarisation
of the incident wave.
−
→
Example: A uniform plane wave with an electric field intensity E =
z
Ei impinges on a small dielectric sphere of radius b << λ and dielectric
constant ǫr . Assuming the resultant polarisation in the sphere is the same
−
→
as the polarisation in a uniform static electric field intensity E and to be
given by

−
→ −
→ ǫr − 1
P = ǫ0 (ǫr − 1) E = z
3ǫ0 Ei
ǫr + 2

(a) Find the back scatter cross-section σ bs

(b) Determine σ bs for a spherical rain drop of diameter 3 mm at 15 GHz

assuming that the dielectric constant of water is 55 at this frequency.
−
→
The induced polarisation vector P is the volume density of electric dipole
moment and is constant within the sphere. The total electric dipole moment
induced in the sphere is hence


−
→ 4π 3 −
→ ǫr − 1
p = b P = z
4πb3 ǫ0 Ei , Cm
3 ǫr + 2

Thus, the sphere acts electromagnetically like an elemental electric dipole

of moment −→p . The scattered electric field intensity in the far-zone is then

−
→ ωp exp (−jkr)
Es =

θEs = −
θ η0 β sin θ
4π r


2 3 exp (−jkr) ǫr − 1
= −θk b Ei sin θ
r ǫr + 2

The time-average back scattered power density is

2
1 b2 (kb)4 ǫr − 1
Ps = |Es |2θ=π/2 = Ei2
2η0 2η 0 r2 ǫr + 2
 9.5 Antennas for reception 219

and the time-average incident power density

Ei2
Pi =
2η0
Hence
 4
2 ǫr − 1
σbs = 4πb kb , m2
ǫr + 2
At f = 15 GHz, λ = 20 mm, the radius of the rain drop is given by
3/2 mm = 1, 5 mm << λ. Thus

σbs = 1, 25 × 10−6 m2 = 1, 25 mm2

We see that
σ bs
= 0, 177
πb2
where πb2 is the geometrical cross-section area of the drop.

9.5.3 Transmit-receive systems

Consider a communication circuit between stations 1 and 2 with antennas
of effective apertures Ae,t and Ae,r resp. The stations are r apart and we
wish to determine the relationship between the transmitted and received
power. We let Pr and Pt be the received and transmitted powers resp. Thus




Pr Ae,r Ae,r 4π Ae,t Ae,r
= G D,t = 2 Ae,t =
Pt 4πr 2 4πr 2
λ λ2 r2
This is known as the Friis transmission formula. Alternatively, we can ex-
press the Friis formula as

2
Pr λ
= GD,t GD,r
Pt 4πr
Note that matched conditions are assumed and power dissipation in the
antenna itself is ignored.
Now, consider a radar system using the same antenna to transmit and
receive short pulses of time-harmonic radiation. This is called the monosta-
tic mode of ranging. The ranging is referred to as bistatic if the receiving
and transmitting antennas are different. For a transmitted power Pt , the
power density at a target, a distance r is given by
Pt
PT = GD (θ, φ)
4πr2
where GD (θ, φ) is the directive gain of the antenna in the direction of the
target. Denoting the backscatter or radar cross-section of the target by σ bs ,
220 9. Antenna theory

the equivalent power scattered isotropically is σ bs PT resulting in a power

density σ bs PT /4πr2 at the antenna. If Ae is the antenna effective aperture,
then
PT PT
Pr = Ae σ bs = Ae σbs 2 GD (θ, φ)
4πr2 (4πr2 )
is the power received by the antenna. In other words,

2
Pr σ bs λ2 2 σ bs Ae
= 3 GD (θ, φ) =
PT (4π) r4 4π λr2

which is known as the radar equation.

Example: Assume that 50 kW is fed into an antenna of a radar system
operating at 3 GHz. The antenna has an effective area of 4 m2 and a
radiation efficiency of 90%. The minimum detectable signal power (over
noise inherent in the receiving system and from the environment) is 1, 5 W
and the power reflection coefficient for the antenna receiving is 0, 05. De-
termine the maximum usable range of the radar for detecting a target with
a backscatter cross-section of 1 m2 .
At f = 3 GHz, λ = 0, 1 m, Ae = 4 m2 ,

Pt = 0, 90 × 5 × 104 W = 4, 5 × 104 W


−12 1
Pr = 1, 5 × 10 = 1, 58 × 10−12 W
1 − 0, 05
σ bs = 1 m2


σbs A2e Pt
r4 = ⇒ r = 4, 36 × 104 m = 43, 6 km
4πλ2 PL
In communication systems using geostationary satellites, the geosynchro-
nous orbit has a radius of 42300 km. The Earth’s radius is ∼ 6380 km
and satellites are therefore ∼ 36000 km from the Earth’s surface.
Now consider the situation shown in Fig. 9.18, where Ae is the effective
scattering aperture of the object. For a large and flat conducting sheet
oriented for specular reflection, Ae = Ap where Ap is the physical aperture.
For most objects, Ae < Ap since scattering is isotropic. The radar equation
can now be modified as
Pr A2 σ bs
= e2 4
Pt 4πλ r
where
scattered power (assumed isotropic) 4πr2 Psc
σ bs = =
incident power density Pi
is the radar cross-section, Psc the scattered power density and Pi the inci-
dent power.
 9.5 Antennas for reception 221

FIGURE 9.18. Reflection of elemagnetic wave from a distant reflecting object-

the radar equation

We note that

Pr = scattered power × fraction of sphere subtended by rec. antenna

= power dens. inc. on obj. × σ bs × fraction of sphere subtended by
rec. antenna
Pt 4πAe Ae Pt A2e σ bs
= × σ bs × =
4πr2 λ2 4πr2 4πλ2 r4
Alternatively, if the aperture of the object is set equal to the radar cross-
section as regards the incident power and the aperture is set to the aperture
of an isotropic antenna as regards scattered power

λ2
A′e =
4π
then
σ bs λ2
A2e = σbs A′e =
4π
For a large perfectly conducting sphere of radius a, σbs = πa2 , the physical
cross-section area. It will be smaller for spheres of a finite conductivity
however.
The noise power per unit bandwidth available at the terminals of a resis-
tance R at an absolute temperature T is given by the Johnson noise power
formula
Pn = kT
T is then referred to as the noise temperature of the resistor. Replacing the
resistor by a lossless antenna, with a radiation resistance Rr = R, the input
222 9. Antenna theory

impedance looking into the terminals remains unchanged but the noise will
be different unless the region from which the antenna receives power is at
the same temperature T . In radio telescopes, the radiated beam is pointed
to a region of the sky whose temperature is close to 0 K and the antenna
temperature is then this value and not the physical antenna temperature.
Let the antenna temperature be designated as TA , then the noise power is
given by
Pn = kTA W/Hz
The radio telescope can thus be regarded as a temperature measuring
device (radiometer). This assumes that the antenna has negligible thermal
losses and little side and back lobes. Reception with the telescope is then
conveniently described in terms of received power flux and the flux density
found as
Pn kTA
S= = W m−2 Hz −1
Ae Ae
In astronomy, the flux unit (f u) = 10−26 W m−2 Hz −1 is used. This ex-
pression gives the flux density for a celestial source of extent Ω that is
small compared with the antenna beam solid angle ΩA and the antenna
aligned with the source. Hence the expression applies to point sources. The
actual flux density for other objects is twice this value since the antenna
will receive only half of the radiated unpolarised power.
For small angular size sources of known magnitude, the source tempera-
ture can be determined from
ΩS
TA = TS
ΩA
where ΩS is the source solid angle, ΩA the antenna beam solid angle and
TS the source temperature.

9.6 Wave propagation near the Earth’s surface

9.6.1 A thin-wire dipole above a perfectly conducting planar
Earth
Consider transmitting and receiving antennas A and B at heights h1 and
h2 resp. above an assumed flat Earth’s surface as shown in Fig. 9.19. The
separation between the antennas is d. We assume A is an elemental electric
dipole so that the electric field intensity at B is the vector sum of the direct
−
→ −
→
ray E θ1 from A and an indirect contribution E θ2 after reflection at C, i.e.
−
→ − → −
→
E = E θ1 + E θ2
where
K
Eθ1 = exp (−jkr) sin θ1
r
 9.6 Wave propagation near the Earth’s surface 223

B
z r Eθ2
Eθ1
θ1
A

r’ h2
h1

C
d
h1
θ2

A’

FIGURE 9.19. Transmit-receive system near the earth’s surface

and
K
Eθ2 = exp (−jkr′ ) sin θ 2
r′
where
Il
K=j ηk
4π
and assuming a perfectly conducting Earth.
−
→ −
→
In general, E θ1 and E θ2 are not parallel but for d >> h1 and h2 ,
θ1 ∼ θ2 = θ so that

K
Eθ ∼

θ F sin θ exp (−jkr)
r
where
F = 1 + exp {−jk (r′ − r)}

We then have

2

2 2 h2 − h1 (h2 − h1 )2
r = d + (h1 − h2 ) = d 1 + ∼d+
d 2d

and

(h2 + h1 )2
r′ = d2 + (h2 + h1 )2 ∼ d +
2d
so that
(h2 + h1 )2 − (h2 − h1 )2 h1 h2
r′ − r ∼ =2
2d d
and 

 h1 h2 
|F | = 1 + exp −j2k 
d
224 9. Antenna theory

FIGURE 9.20. Antennas above ground showing reflection into the plane

similar to an array factor of a two-element array. This can be expressed as




 
 h1 h2 h1 h2 h1 h2 
|F | = exp −jk exp jk + exp −jk 

d d d
 
 2πh1 h2 
= 2 cos
λd 

Thus, as h1 /d is varied, Eθ at B will have maxima and nulls while |F |

varies from 0 to 2 and is known as the path-gain factor.

9.6.2 A thin-wire antenna over a finite conductivity flat Earth

For a 2l-thin wire antenna above a large conducting plane compared to its
largest dimension and of a sufficiently high conductivity such that the plane
can be considered perfect, the effect of the ground plane can be accounted
for by reflecting the antenna into it. Fig. 9.20 illustrates this technique.
The current direction in the antenna image is such that the vertical
components are in the same direction and the horizontal ones in opposite
sense at any instant. For a long-wire vertical antenna above a conducting
plane and excited at its base, the image reduces the problem to that of
a dipole with the field strength in the antenna for maximum Im having
the same expression as that of the dipole considered previously for points
 9.7 Other types of practical antenna systems 225

FIGURE 9.21. Separation of radiation field into a skywave and ground wave for
a finite σ ground plane

above the plane i.e. for 0 ≤ θ ≤ π/2. Thus,

π/2
ηI 2 [cos (kl cos θ) − cos (kl)]2
PT = m dθ
4π sin θ
0

When the conductor is not perfect as will be the case when the ground
plane is the Earth’s surface, for which the displacement and conduction
currents are approximately equal at f ∼ 5 MHz, the radiated field splits
into a space or sky wave and a ground or surface wave as shown in Fig.
9.21 for the Hertzian dipole. The space wave varies as 1/r with distance
from the antenna while the surface wave varies as 1/r2 .

9.7 Other types of practical antenna systems

9.7.1 Traveling-wave antennas
If the antenna is several wavelengths long and properly terminated such
that no reflections exist on the antenna,the antenna behaves like a trans-
mission line terminated in its characteristic impedance. Consider the trans-
mission line section of length L shown in Fig. 9.22. The current distribution
on the antenna is given by

I (z) = Im exp (−jkz)

226 9. Antenna theory

θ
I
excitation terminating
source h impedance
z

ground plane

FIGURE 9.22. A travelling-wave antenna

If we ignore the effect of the nearby ground, the far-field will have an electric
field intensity given by

L
jηk exp (−jkr)
Eθ = sin θ I (z) exp (jkz cos θ) dz
4π r
0
L
jηIm exp (−jkr)
= sin θ I (z) exp {−jkz (1 − cos θ)} dz
4π r
0

or 

j60Im L
Eθ = exp −jk r + (1 − cos θ) F (θ)
r 2
where
sin θ sin (kL (1 − cos θ))
F (θ) =
1 − cos θ
is the pattern factor of an isolated travelling-wave antenna of length L.
The pattern is not symmetrical about the θ = π/2 plane unlike in resonant
antennas such as the λ/2 dipole. The main lobes tilt in the direction of the
current wave with the tilt angle becoming greater with increasing antenna
length. The side lobes of this antenna are also large.
The effect of ground can be accounted for using the method of images as
shown in Fig. 9.23. This is then an array of two travelling-wave antennas
carrying equal but opposite currents. Using the principle of pattern multi-
plication, the resulting pattern factor is F (θ) |cos ψ/2| of the two-element
array with ϕ = π. Thus,
    
 
cos ψ  = cos kh sin θ cos φ + π  = |sin (kh sin θ cos φ)|
 2 2
 9.7 Other types of practical antenna systems 227

2h
h ground plane

I end-view
(two-element array)

FIGURE 9.23. Method of images analysis of a travelling-wave antenna

9.7.2 Helical antennas

Radiation from a linear antenna or a small loop is linearly polarised, i.e.
the direction of the electric field phasor does not change with time. The
receiving antenna must then be oriented parallel to the electric field for effi-
cient reception. In some cases, the polarisation of the desired signal will be
unknown or like in satellite communication, the radiation first propagates
through the ionosphere where it suffers Faraday rotation whose magni-
tude depends on the electron density of the ionosphere, the strength of the
Earth’s magnetic field and the path of propagation, factors that are largely
unpredictable. Such a situation therefore requires antennas with circular
polarisation because this can intercept radiation polarised in any direction
not normal to the plane of circular polarisation. One such antenna is the
turnstile antenna consisting of two crossed dipole elements excited by cur-
rents of equal magnitude but in time-phase quadrature. Another example
is the helical antenna of Fig. 9.24(a).
Depending on the dimensions of the helix, the helical antenna has two
distinct modes of operation. With dimensions much smaller than the oper-
ating wavelength, its radiation pattern is like that of an elemental electric
dipole with maximum radiation occurring in the plane normal to the axis of
the helix which is referred to as the normal mode of operation. To analyse
the antenna in the normal mode, two assumptions are made

• The helix can be replaced by a combination of elemental loops and

elemental electric dipoles as shown in Fig. 9.24(b).

• Current along the helix is uniform in both amplitude and phase

The far-field for an N -turn helical antenna is then a combination of the

results of the elemental electrical and magnetic dipoles, i.e.

−
→
 

φ = N kI η exp (−jkr) j

E = θEθ + φE θs +

φkπb 2
sin θ
4π 0 r
228 9. Antenna theory
z

coax ground plane 2b (c) E-plane radiation pattern

(b) combination of elemental

(a) helical antenna electric and magnetic dipoles

FIGURE 9.24. Helical antenna, its equivalent visualisation and E-plane radiation
pattern

θ- and φ-components are in space- and time-phase quadrature resulting in

elliptical polarisation. Circular polarisation then requires that

2 1 sλ
s = kπb ⇒ b =
π 2

Maximum radiation occurs in the broadside direction with the doughnut

pattern of dipole antennas of zero inner diameter. In practice, normal-mode
helical antennas are seldom used due to their low radiation efficiency and
directive gain. If both the turn circumference and spacing between turns
are now comparable to the wavelength, radiation is endfire as shown in Fig.
9.25. This is referred to as the axial mode. The radiation in the main lobe
is elliptically polarised.

9.7.3 The Yagi-Uda antenna

Multi-element arrays present a problem of excitation since each element
must be excited with a carefully controlled feed current. In the Yagi-Uda
array, only one element referred to as the active element is excited. The in-
duced currents in the other elements (passive elements) are used to provide
extra contribution to the total radiated field. Correct phasing is achieved
by adjusting the size and positions of the passive elements relative to the
active element.
In the transmitting mode, the Yagi is an array of linear antennas with
only one active element. In receiving mode, on the other hand, an incident
electromagnetic wave impinges on all the elements but is collected only
from the active element. The feed structure is very simple accounting for
the popularity of the Yagi antenna in broadcast system such as TV and
police communication.
Consider the two-element Yagi of Fig. 9.26. From the linearity property
of Maxwell’s equations, we can write a set of equations relating the voltages
 9.7 Other types of practical antenna systems 229

FIGURE 9.25. Axial-mode helix antenna showing the main lobe

at the centre of the antenna to the currents at the feed points as

V1 = I1 Z11 + I2 Z12

V2 = I1 Z21 + I2 Z22
where the Zij are constants dependent on the element length l1 and l2 as
well as the separation d. Since the voltage at the drive point of the parasitic
element is zero, I2 is determined by I1 through induction. We can therefore
write
Z21
I2 = − I1
Z22
and the array factor then given by

A (θ) = a0 + a1 exp (jkd cos θ)


I2
= I1 1 + exp (jkd cos θ)
I1


Z21
= I1 exp 1 − jkd cos θ
Z22

We see that A (θ) depends on Z21 and Z22 .

From field analysis, Z21 can be shown to be relatively insensitive to the
element length 2l1 and 2l2 when these are approximately λ/2. Hence, the
phase of I2 depends on the relative phase of I1 and I2 . If the array is to
maximise radiation in the −z-direction, element 2 is called a reflector and
230 9. Antenna theory

FIGURE 9.26. A two-element Yagi-Uda array

l2 must then be slightly longer than l1 . Maximum radiation occurs with

d ∼ 0, 16λ

and
2l2
0, 51 < < 0, 52
λ
with 2l1 ∼ λ/2 in this case.
When the intention is to have maximum radiation in the +z-direction,
element 2 is known as a director and has a slightly smaller length than
the active element. The optimum length and separation from the active
element with 2l1 ∼ λ/2 are in this case

d ∼ 0, 11λ

and
2l2
0, 38 < < 0, 48
λ
The Yagi-Uda array in typical use has both a director and reflector and
the three elements complicate the analysis somewhat. However, conclusions
drawn from the two element array remain fairly valid. Practical Yagis use
a longer reflector and shorter lengths resp. than those given for maximum
radiation. The directivity of the array then turns out to be insensitive
to the director and reflector spacing from the active element. The input
impedance of the array is then given by



I2 I3
Zin = Z22 + Z21 + Z32
I1 I2
 9.7 Other types of practical antenna systems 231
z

2a

b34
h3

2 3 4 5
1

FIGURE 9.27. The Yagi-Uda passive array

FIGURE 9.28. Radiation pattern of a Yagi-Uda array

and due to the current phasing, Zin tends to be rather low. If the driven
element is a folded dipole, Z22 is four times that of a single dipole and
the overall input impedance can be raised. Additional directors improve
antenna characteristics and directivities in the range 10 − 100 are common
depending on the number of directors in the array. No significant improve-
ment is achievable from providing a second reflector, though. Fig. 9.27
illustrates a five-element Yagi-Uda antenna.
Since all the elements in the Yagi-Uda antenna are coupled, the current
distribution on each element depends on the length and spacing of all the
other elements. From experience, little advantage accrues from using more
than one reflector but directivity improves with the number of director
elements. The Yagi-Uda is an endfire array with the main lobe pointing
away from the reflector as shown in Fig. 9.28 for a three element array. A
good Yagi-Uda antenna should have

• High directivity
• Narrow beam width
232 9. Antenna theory

TABLE 9.1. Performance parameters of an unoptimised Yagi-Uda antenna

Element length, λ 2h1 = 0, 510 2h2 = 0, 490 2h3 = 2h4 = 2h5
= 2h6 = 0, 430
Element spacing, λ b12 = 0, 250 b23 = b34 = b45
= b56 = 0, 310
Direct. (w.r.t. λ/2 dip) 7, 54 (8, 77 dB)
Half-pwr beam width 45◦
First side lobes −7, 2 dB
Front-back ratio 9, 52 dB

TABLE 9.2. Performance parameters of an optimised Yagi-Uda antenna

Elem. length 2h1 2h2 2h3 2h4 2h5 2h6
= 0, 476λ = 0, 452λ = 0, 436λ = 0, 430λ = 0, 434λ = 0, 430λ
Elem. spacing, b12 b23 b34 b45 b56
= 0, 250λ = 0, 289λ = 0, 406λ = 0, 323λ = 0, 422λ
Directivity 13, 36
(rel. λ/2 dipol) (12, 58 dB)
HPBW 37◦
1st side lobes −10, 9 dB
Front-back ratio 10, 04 dB

• Low side-lobes

• High front-to-back ratio

For a Yagi with a = 0, 003369λ, (ln λ/2a = 5), a typical well dimensioned
six-element Yagi with four directors of equal length and equal spacing may
have the parameters shown in Table 9.1. The directivity of the λ/2 dipole
used as reference is of course 1, 64 or 2, 15 dBi referenced to an omnidirec-
tional antenna.
Uniformly spaced directors of equal length do not translate into an opti-
mal design. An optimally designed Yagi with dipole radius a = 0, 003369λ
would have the parameters shown in Table ??.

9.7.4 The log-periodic array

The determining factor for the radiation pattern and impedance of a loss-
less antenna is the antenna dimensions-to-wavelength ratio. Therefore, an
antenna must have all dimensions scaled to the operating frequency. If the
structural form of the antenna is such that a multiplication of all dimensions
with the wavelength ratio at two frequencies leaves the antenna unchanged,
the antenna will be independent of frequency, a desirable property. This is
equivalent to requiring that the antenna dimensions be scaled with respect
to the wavelength. One radiating structure with this characteristic is the
 9.7 Other types of practical antenna systems 233

FIGURE 9.29. A planar two-arm equiangular spiral antenna

equiangular spiral defined by

r = r0 exp {a (φ − ∆)}
where r and φ are the usual polar coordinates and r0 , ∆ and a are design
constants. The spiral shown in Fig. 9.29 is equiangular in the sense that
it makes the same angle with the radial vector at all points on the curve.
This structure is also referred to as the log-periodic spiral since the polar
angle is related to the radial distance as
1 r
φ−∆= ln
a r0
These three design parameters determine the size of the termination region,
which is the reciprocal of the rate of the spiral, a and the arm width ∆.
The four edges of the planar equiangular spiral are defined by
r1 = r0 exp (aφ)
r2 = r0 exp {a (φ − ∆)}
r3 = r0 exp {a (φ − π)}
r4 = r0 exp {a (φ − π − ∆)}
The excitation voltage source at the terminals drives the current flow.
One view of how this structure radiates is that these currents flow out-
wards until they get to the point on the spiral where most of the radiation
occurs. Another view is that the electric field vector between the spiral
arms travels outwards until the space between the arms is ∼ λ/2 at the
operating frequency. Resonance in this region then causes strong radiation
to occur. Beyond the resonance region, the current decays rapidly such that
truncating the spiral is of little consequence. An increase or decrease in fre-
quency only shifts the resonance region along the spiral. As the operating
frequency is changed, the radiation pattern rotates about the axis normal
to the spiral. The two-arm planar equiangular spiral radiates circularly po-
larised fields with a broadside main lobe on both sides of the plane which
234 9. Antenna theory

rn+1
rn

FIGURE 9.30. A planar log-periodic antenna

may be a disadvantage. In the strict sense, the spiral must be infinite for
the antenna to be truly frequency-independent.
The log-periodic antenna of Fig. 9.30 is another example of a frequency-
independent radiative structure. The teeth present discontinuities that tend
to localise the region of maximum radiation and cause the current to dimin-
ish rapidly beyond the region. Tooth lengths are determined by the angles
between the lines from the origin and the region of strongest radiation is
where the tooth length is ∼ λ/4.
The spacing between successive edges of the teeth follow the same rule
governing the distance between neighbouring conductors in the equiangular
spiral, i.e.

rn+1 r0 exp {a (φ − ∆)}

= = exp (−2πa) = τ
rn r0 exp {a (φ − 2π − ∆)}

For an infinite structure, antenna characteristics would be identical at a

discrete number of frequencies related to τ by

fn = τ fn+1

or
1
ln fn+1 = ln fn + ln
τ
Wire or tubular log-periodic antennas may be constructed when the
tubes outline the cut-out design of the planar log-periodic antenna. The-
oretically, sheet thickness and wire diameter should increase linearly with
distance from the feed point in accordance to τ and this becomes important
when the bandwidth demand is severe. The planar log-periodic antenna is
also bidirectional i.e. the main lobe is broadside on either side. However,
 9.7 Other types of practical antenna systems 235

rn

rn+1 dn

ln
ln+1

FIGURE 9.31. A log-periodic dipole array

folding the two arms into a wedge-like structure will make the antenna uni-
directional. Without the teeth, we have the bowtie antenna with limited
broadband characteristics and is useful in UHF television reception.
The log-periodic dipole array of Fig. 9.31 is frequency-independent as
well. The dipoles are unequally spaced and have lengths given by
ln+1 rn+1
= =τ
ln rn
where τ is a design parameter. Since the element spacings are related to
the distances to an imaginary apex point and apex angle α, i.e.

dn = rn − rn+1 = rn (1 − τ )

then
dn+1
=τ
dn
as well.
The angle α or spacing factor κ given by
dn
κ=
2ln
is another necessary design parameter. The relationship between τ , α and
κ is
α 1 ln ln (1 − τ ) 1−τ
tan = = =
2 2 rn dn 4κ
236 9. Antenna theory

FIGURE 9.32. Feed system for a log-periodic array

Strictly, it is also necessary to scale the diameter of the wire according to

an+1
=τ
an
This array operates in endfire with the main beam radiating towards the
shorter dipoles.
To understand the operation of the log-periodic array, we consider the
n-th element resonant at a frequency fn . The two adjacent neighbours then
play the role of director and reflector as in the Yagi-Uda array with the
radiation in the direction of the shorter element.
In a practical log-periodic array, the feed point is at the small end of
the array and the element arms connected on alternate conductors of the
feed transmission line as shown in Fig. 9.32 and the element spacing one-
half the element length at that point in the array. The current phasing in
adjacent elements is such that in the resonant region, maximum radiation
occurs when the current in the n + 1-th element leads that in the n-th
element by π/2. Since the n-th element is assumed to be the resonant one,
then ln = λ/4 and the n + 1-th element is then λ/4 away with its radiation
towards the small end of the array and in phase with the radiation of the
n-th element.
If the feed point is at the long elements-end of the array, then the λ/2 res-
onance for some frequency lying towards the small end would encounter res-
onant dipoles with lengths nλ/2 first. These would depend on the frequency
which implies that the pattern as well as impedance become a function of
frequency, what this antenna seeks to avoid. Feeding from the small-end of
the array alleviates this problem because the fundamental resonant dipole
is encountered first. The alternating connection to the transmission line is
a consequence of feeding the array at the short element-end of the array.
 9.8 Radiating apertures 237

Plane of aperture

Actual
source Aperture
(S)

FIGURE 9.33. Radiation from apertures

9.8 Radiating apertures

In many cases, the electromagnetic radiation can be viewed as emanating
from an opening in a conducting enclosure, e.g. slots, horns, reflectors,
lenses. Consider the aperture of Fig. 9.33. We assume that the plane is
absorbing so that fields exist only in the aperture. We also assume that
we know the tangential electric and magnetic fields E (x′ , y ′ ) and H (x′ , y ′ )
in the aperture and these fields may be regarded as being generated by
equivalent sources located in the plane of the aperture. We recall that the
tangential magnetic field intensity is related to the surface current density
as
−
→ −
→
Js=n
×H
Similarly, we regard the tangential electric field intensity as related to a
−
→
surface magnetic current density M s which is however fictitious. Thus, we
can write
−
→ −
→
M s = −
n× E
where these relations assume tangential fields to the left of the plane of the
aperture vanish.
We therefore express the augmented Maxwell’s equations as
−
→ − →
∇ · D = ρe
−
→ − →
∇ · B = ρm
−
→
−
→ − → −
→ ∂B
∇ × E = −M −
∂t
−
→
−
→ −→ − → ∂D
∇×H = J +
∂t
238 9. Antenna theory

−
→
Similar to the vector potential A associated with electric sources, we can
−
→
define a vector potential F associated with the magnetic sources and write,
 −
→
−
→ µ J s exp (−jkr)
A= dS ′
4π r
S′

and  −
→
−
→ ǫ M s exp (−jkr)
F = dS ′
4π r
S′

The fields are then given by

−
→ → jω −
− → −
→ −→ 1 −
→ −→
E = −jω A − 2 ∇ ∇ · A − ∇ × F
k ǫ
and
−
→ −
→ jω − → −
→ −→ 1 −
→ −→
H = −jω F − 2 ∇ ∇ · F − ∇ × A
k µ
If the plane of the aperture is conductive, any currents flowing on the right
−
→
side surface of the plane must be added in the calculation of A but the
correction is usually insignificant.
In the far-field, we can write
−
→ µ exp (−jkr) −
→
A= N
4πr
and
→ ǫ exp (−jkr) −
− →
F = L
4πr
−
→ −
→
where L is the radiation vector just as N is. Thus, we have

−
→ −
→
N = J s exp (jkr′ cos ϕ) dS ′
S′

and 
−
→ −
→
L = M s exp (jkr′ cos ϕ) dS ′
S′
′
where r and ϕ have the previously defined meaning. The only components
of the field decreasing no faster than 1/r are

exp (−jkr)
Eθ = ηHφ = −j {ηNθ + Lφ }
2λr
exp (−jkr)
Eφ = −ηHθ = j {−ηNφ + Lθ }
2λr
 9.8 Radiating apertures 239

FIGURE 9.34. Plane wave sources

and the radiation intensity is then

 2  2 
η  Lφ   Lθ 
K = 2 Nθ +  + Nφ −
 

8λ η η 

which is the expression for Fraunhofer diffraction in optics. The near-field

(Fresnel) radiation would require a more elaborate approximation to r.

9.8.1 Plane wave sources

If on a plane wavefront
−
→ −
→ E0
E =x

E0 , H = y
H0 = y

η
the equivalent current sheets are
E0
Jx = −H0 = − , My = −E0
η
If the differential area on the plane wavefront is small compared to the
wavelength,
E0
Nx = − dS, Ly = −Ex dS
η
which are shown in Fig. 9.34.
Since the source elements are at the origin of the coordinate system, then
in spherical coordinates
E0 E0
Nθ = − dS cos φ cos θ, Nφ = dS sin φ
η η
240 9. Antenna theory

and
Lθ = −E0 dS sin φ cos θ, Lφ = −E0 dS cos φ
and the far-field radiation is then
jE0 dS
Eθ = exp (−jkr) (1 + cos θ) cos φ
2λr
jE0 dS
Eφ = − exp (−jkr) (1 + cos θ) sin φ
2λr
and

E02 (dS)2  2 2

K = (− cos φ cos θ − cos φ) + (sin φ + sin φ cos θ)
8ηλ2
2
E02 (dS) θ
= cos4
2ηλ2 2

For small values of θ, (paraxial approximation), cos θ ∼ 1, so that

Ex = Eθ cos θ cos φ − Eφ sin φ ∼ Eθ cos φ − Eφ sin φ

Ey = Eθ cos θ sin φ + Eφ cos φ ∼ Eθ sin φ + Eφ cos φ

and
E0
Ex = j dS, Ey ∼ 0
λr
In the paraxial approximation, the radiation field is in the direction of
the source field. We can thus write

j exp (−jkR) ′ ′
E (x, y, z) = E (x′ , y′ ) dx dy
λ R
S′

where 
(xx′ − yy ′ )
R= (x − x′ )2 + (y − y ′ )2 + z 2 ∼ r −
r
and since the difference between r and R is important only in the phase
consideration, then
 
j exp (−jkr) ′ ′ (xx′ + yy ′ )
E (x, y, z) = E (x , y ) exp jk dx′ dy′
λr r
S′

which is the standard Fraunhofer diffraction equation. We see that, this

equation is a 2-D spatial Fourier integral and implies that in the paraxial
approximation as shown in Fig. 9.35, the far-field is the spatial Fourier
transform of the aperture field.
 9.8 Radiating apertures 241

FIGURE 9.35. Paraxial approximation

9.8.2 Apertures with plane wave excitation

The following assumptions will be made
−
→ −
→
• E and H are related as in a plane wave everywhere in the aperture.
The strength may vary over the aperture though

• Contributions from induced currents outside the aperture are negli-

gible

• Only small angles about the polar axis are considered so that the
paraxial approximation is valid

These assumptions are satisfied if the aperture is large in comparison

with the wavelength. Therefore, consider a rectangular aperture with sides
a and b and uniform illumination E0 over the aperture. We can then write,

a/2 b/2


exp (−jkr) xx′ yy ′
E (x, y, z) = j E0 exp jk exp jk dx′ dy ′
λr r r
−a/2−b/2
exp (−jkr) sin (kax/2r) sin (kby/2r)
= j E0 ab
λr (kax/2r) (kby/2r)



exp (−jkr) kax kby
= j E0 ab sinc sinc
λr 2r 2r

and the pattern in the plane y = 0 has nulls at

x mλ
θ null ∼ = , m = 1, 2, 3, . . .
r null a
242 9. Antenna theory

In the plane x = 0 the null angles have a similar form with b and y replacing
a and x respectively. The main lobe hence becomes narrower as the aperture
dimensions increase.
The directivity of the rectangular aperture is then
Piso
D=
P
where
|Emax |2 E 2 (ab)2
Piso = 4πr2 = 2π 0 2
2Z0 λ0 η
is the isotropically radiated power. The actual power radiated, neglecting
currents in the surrounding aperture plane is
E02
P = ab
2η
and
4π 4π
D= ab = 2 A
λ2 λ
where A = ab is the aperture area.

9.8.3 Radiation from a rectangular aperture

For a horn antenna, the aperture field is derived from the waveguide mode
propagating into a horn of infinite extent while the aperture fields of the
reflector and lens are found by methods of geometrical optics from the
reflection and refraction rays emanating from the primary feed. For the
TE10 mode excitation, the field in a plane aperture is approximately linearly
polarised, hence analysis results deviate little from those obtained using
geometric optics. With a nearly uniform phase over the aperture, the far-
field is a 2-D Fourier integral of the field distribution in the aperture as we
have seen.
Let the electric field distribution in the rectangular aperture of Fig. 9.36
be linearly polarised in the x-direction and exhibit no phase variation, i.e.,
−
→
Ea = x

Ea

If the aperture dimensions are much larger than the wavelength, almost all
the energy of the radiated field will be contained in a small region about
the z-axis (paraxial field). Thus
−
→
Ep = x

Ep

where 
j
Ep = Ea (x′ , y′ ) exp (−jkr) dx′ dy ′
λr0
 9.8 Radiating apertures 243

a
b

FIGURE 9.36. An aperture antenna

and for kr >> 1

r xx′ + y
y ′ ) · (

= r0 − (
x sin θ cos φ + y
sin θ sin φ)
= r0 − (x sin θ cos φ + y ′ sin θ sin φ)
′

Thus,
j
Ep = exp (−jkr0 ) F (θ, φ)
λr0

where
 
F (θ, φ) = Ea (x′ , y′ ) exp {jk sin θ (x′ cos φ + y ′ sin φ)} dx′ dy ′
apert.

is the pattern function of the aperture.If the aperture has dimensions a × b

and the field distributions are separable, we have

Ea (x′ , y ′ ) = f1 (x′ ) f2 (y ′ )

and

a/2 b/2 a/2

′ ′ ′
F (θ, φ) = f1 (x ) exp (jk sin θx cos φ) dx + f2 (y ′ ) exp (jk sin θy′ sin φ) dy ′
−a/2 −b/2−a/2

Only the patterns in the principal planes are of interest. Hence,

244 9. Antenna theory

(a) In the xz-plane where φ = 0,

 
b/2 a/2
 ′ ′
Fxz (θ) =  f2 (y ) dy  f1 (x′ ) exp (jkx′ sin θ) dx′
−b/2 −a/2

a/2
= C1 f1 (x′ ) exp (jkx′ sin θ) dx′
−a/2

where C1 is a constant.
(b) In the yz-plane where φ = π/2,
 
a/2 b/2
 ′ ′
Fyz (θ) =  f1 (x ) dx  f2 (y ′ ) exp (jky ′ sin θ) dy ′
−a/2 −b/2

a/2
= C2 f2 (y ′ ) exp (jky ′ sin θ) dy ′
−a/2

with C2 another constant.

The directivity of the rectangular aperture is given by
4πKmax
D=
Pr
where
 2
  
1 2 1  
Kmax = r |Ep |2max = ′ ′ ′ ′
Ea (x , y ) dx dy 
2η 0 2ηλ2  
apert.

and
Pr = total power radiated
 
1 2
= |Ea (x′ , y ′ )| dx′ dy ′
2η
apert.

Thus  2
  
 
 E (x ′ ′
, y ) dx ′
dy ′
 a 
4π  apert. 
D= 2 
λ
|Ea (x′ , y ′ )|2 dx′ dy′
apert.
 9.8 Radiating apertures 245

FIGURE 9.37. The electromagnetic horn

D is maximum for constant Ea (x′ , y′ ) (uniform aperture field distribution)

and is 4π/λ2 times the area of the aperture.
An important rectangular aperture antenna is the electromagnetic horn
shown in Fig. 9.37. This horn is used in practice as a feed system for the
reflector or directive lens antenna systems. The flare of the horn provides
impedance matching between free space and the feed waveguide as well
as directivity of the antenna. For the horn, the fields across the aperture
are not uniform and their relationship is no longer the same as in a plane
wave. In addition, there are some effects of currents induced on the exterior
surfaces and supporting structures.
Assume that the horn is excited in the TE10 mode. The field in the
aperture is then fairly approximated by

′
πx
E (x′ y′ ) = E0 sin
a
where a is the dimension of the long side of the rectangular aperture of the
horn. Thus
b/2 a/2


exp (−jkr) πx′ xx′ + yy′
E (x, y, z) = j E0 sin exp jk dx′ dy′
λr a r
−b/2−a/2
 

E0 ab exp (−jkr) cos (kax/2r) kby
= 2 sinc
π λr 1 − (kax/πr)2 2r

and the directivity

2


 2E0 ab 4Z0 4π 8ab
D = 4πr2 = 2
πλr E02 ab λ π2
The factor 8ab/π2 can be regarded as an equivalent cross-section area of
the horn.
246 9. Antenna theory

Example: For an a × b rectangular aperture with a uniform field distrib-

ution, determine

(a) The pattern function in a principal plane and the half-power beam
width
(b) The location of the first nulls and the level of the first side lobes

Let Ea (x′ , y ′ ) = 1 for simplicity.

(a) In the xz-plane ( φ = 0),

a/2

′ ′ sin ψ
Fxz (θ) = b exp (jkx sin θ) dx = ab
ψ
−a/2

where
πa
ψ=sin θ
λ
Fyz (θ) has the same form, where b replaces a. Note that the pattern
function is similar to the array factor of a uniform array for small
ψ.
The half-power points are obtained by setting
sin ψ1/2 1
=√
ψ1/2 2

Hence,
πa
ψ1/2 = sin θ1/2 = 1, 39
λ
or
λ
sin θ1/2 = 0, 0442
a
For a sufficiently large aperture, sin θ 1/2 ∼ θ1/2 and

λ λ
2θ1/2 ∼ 0, 88 rad ∼ 50◦
a a

(b) The location of the first side lobes is found by setting


∂ sin ψ
=0
∂ψ ψ
which requires
tan ψ1 = ψ1
or
ψ1 = ±1, 43π
 9.8 Radiating apertures 247

Hence    
 sin ψ1   sin 1, 43π 
 = 
 ψ   1, 43π  = 0, 217
1

When referred to unity at ψ = 0, the first side lobes are 20 log10 (1/0, 217) =
13, 3 dB below the level of maximum radiation.

9.8.4 Radiation from a long slit

In the case of a slot antenna, the dominant mode is the TE10 and the field
distribution is usually assumed to be a half-sinusoid with the maximum
at the slot centre and tapering to zero at the edges. With b very long,
the expected pattern in the plane x = 0 should be very thin which would
be the case when the illumination is coherent over the length of the slit.
This is not the case in practice however and an interference pattern is seen
only over the shorter dimension along x. The degree of coherence for the
illumination is defined in terms of the coherence length and we assume that
the slit width is shorter than this length and the slit length longer. The
resulting radiation intensity is then proportional to |E|2 using only x.
The slit has symmetry only in two dimensions and we use cylindrical
coordinates in this case with φ measured about the y-axis and φ = 0 along
the x-axis. Hence
x
∼ cos φ
r
and


2 ka
K = A sinc cos φ
2

where A is a constant of the aperture illumination.

An important slot antenna is the resonant slot antenna of Fig. 9.38. For a
perfectly conducting plane of infinite extent, it is only necessary to consider
the E-field in the aperture since the tangential electric field intensity is zero
along the conductor and dies off at ∞.
We assume the electric field in the aperture is uniform in the x direction
and to have a half-sine distribution in the z-direction with the maximum
at the centre of the slot as shown in Fig. 9.38. The magnetic current sheet
equivalent to this lies in the z-direction and is equal to Ex .

Mz = Ex = Em cos kz

With Ex assumed zero behind Mz , the gap can be regarded as covered

by an electric conductor. The infinite conductor can be accounted for by
replacing it with the image of Mz which is in the same direction as Mz and
hence the free-space value must be doubled.
248 9. Antenna theory

FIGURE 9.38. The resonant slot antenna

For a small gap width g, the magnetic radiation vector is given by

λ/4
Lz = 2gEm cos kz ′ exp (jkz ′ cos θ) dz ′
−λ/4

cos π2 cos θ
= 4gEm
k sin2 θ
and 
j exp (−jkr) cos π2 cos θ
Eφ = −ηHθ = gEm
πr sin θ
−
→ −
→
which is similar to the case of the resonant dipole except that H and E
are interchanged (i.e. polarisations are orthogonal).
The radiated power is given by

π π
πgEm cos2 2 cos θ
Pr = dθ
2πη0 sin θ
0

which may be interpreted in terms of a radiation conductance of the slot

given by
π 2  π
2Pr 1 cos 2 cos θ
Gr,slot = = dθ
(gEm )2 πη sin θ
0

or
2Rr,dipole
Gr,slot = ∼ 0, 00103
η2
 9.8 Radiating apertures 249

The radiation conductance is twice this value if radiation is on both sides

of the conductive plane. If the slot radiates to both sides, then we have

η2
Zstrip Zslot =
4

9.8.5 Circular aperture with uniform illumination

In polar coordinates, we have

x = r sin θ cos φ, y = r sin θ sin φ, x′ = r′ cos φ′ , y′ = r′ sin φ′

Thus
2πa
exp (−jkr)    
E (r, θ, φ) = j E r′ , φ′ exp jkr′ sin θ cos φ − φ′ dr′ dφ′
λr
0 0

where a is the aperture radius. If E r′ , φ′ is independent of φ′ , E (r, θ, φ)
will be also and φ = 0 in the integral. We use the integration result
2π
exp (jx cos ϕ) dϕ = 2πJ0 (x)
0

with J0 the Bessel function of the first kind and order 0. Thus
a
2π exp (−jkr)
E (r, θ) = j E (r′ ) J0 (kr sin θ) dr′
λr
0

If E (r′ ) = E0 , constant over the aperture, we can show that

2π exp (−jkr) J1 (ka sin θ)

E (r, θ) = j E0 a
λr ka sin θ
and the first null is at


3, 83λ λ
θ null = sin−1 ∼ 0, 61
2πa a
The directivity is calculated using isotropic radiation equal to that at θ = 0
and the actual plane wave power through the aperture. Hence


4πr2 2πE0 a2 η 4π
D= = 2A
2η 2πa πa2 E02 λ
as for the rectangular aperture. The pattern again has an infinite number
of nulls.
250 9. Antenna theory

If the illumination were Gaussian, i.e.

′2 
′2 
 ′ ′ r x + y′2
E r , φ = E0 exp − 2 = E0 exp −
w w2
−
→ −
→
we assume E and H to still keep their relationship as in a plane wave with
w the beam radius at the 1/e point in the field.
If we assume that w is small in comparison with the aperture radius a,
then a can be assumed infinite and the field will then be a function of r′
alone. In this case, reverting to Cartesian coordinates for convenience,
∞ ∞ 
′2 
′2 
exp (−jkr) x xx′ y yy ′
E (x, y, z) = j E0 exp − − jk − − jk dx′ dy ′
λr w2 r w2 r
−∞−∞

which can be written as

∞ ∞   ′
   ′
 
4 exp (−jkr) x2 kxx′ y2 kyy′
E (x, y, z) = j E0 exp − 2 cos exp − 2 cos dx′ dy ′
λr w r w r
−∞−∞

since the imaginary parts are odd functions and integrate to zero. We can
thus write
  
exp (−jkr) 2 k 2 x2 + y 2 w 2
E (x, y, z) = j E0 πw exp −
λr r2

2 2 2 
exp (−jkr) k w sin θ
= j E0 πw2 exp −
λr 4

which is tabulated. The angle, θ0 where the radiation is 1/e of the value at
θ = 0, is given by


−1 2 2 λ
θ 0 = sin ∼ =
kw kw πw

Tapered illumination hence eliminates the side lobes.

−
→
Example: A linearly polarised uniform electric field E a = x

E0 exists in
a circular aperture of radius b in a conducting plane at z = 0. Assuming
b >> λ,

(a) Derive an expression for the far-field

(b) Determine the width of the main lobe between the first nulls

We use polar coordinates and write

x′ = r′ cos φ′ , y ′ = r′ sin φ′
 9.8 Radiating apertures 251

and
  
x′ cos φ + y ′ sin φ = r′ cos φ cos φ′ + sin φ sin φ′ = r′ cos φ − φ′

Thus

b 2π
  
F (θ, φ) = E0 exp jkr′ sin θ cos φ − φ′ r′ dφ′ dr′
0 0
b
 J1 (kb sin θ)
= E0 2πJ0 (kr′ sin θ) r′ dr′ = E0 2πb2
kb sin θ
0

Thus
−
→ 2πb2 J1 (u)
Ep = x

jE0 exp (−jkr0 )
λr0 u
where
2πb
u = kb sin θ = sin θ
λ
The first null occurs at the first zero of J1 (u), u11 say, where

u11 = 3, 832

corresponding to an angle

3, 832λ 3, 832λ λ
θ 1 = sin−1 ∼ = 1, 22
2πb 2πb D
where D = 2b diameter of the aperture. Thus, the width between the first
nulls is
2, 44λ
2θ1 = rad
D
This is wider than that of the rectangular aperture having a = D.
The level of the first side lobes for the circular aperture is 0, 13 ( 17, 7 dB)
down from the maximum which is better than that of the rectangular aper-
ture with a = D. Note that
2π

exp jw cos φ′ dφ′ = 2πJ0 (w)
0

and 
wJ0 (w) dw = wJ1 (w)
252 9. Antenna theory

FIGURE 9.39. Parabolic dish for the case with focal length equal to dish depth

9.8.6 Parabolic reflector

A practical application of the circular aperture is to be found in parabolic
reflectors. In practical cases, the illumination is not uniform due to practical
difficulties as well as the desirable effect of reduced side lobes with a non-
uniform illumination. For most primary radiators e.g. a small horn or dipole
placed at the focal point of the paraboloid, illumination of the reflector in
two orthogonal axes will not be the same. A typical design value for the
half-power beam direction for a paraboloid of radius a is


◦ λ
2θ0,half −power ∼ (70 )
2a
Recall that the equation of a parabola is given by

x2
y=
4p
where x is the dish radius or half the dish aperture (diameter). The focus
of the dish shown in Fig. 9.39 is then at
D2 f D
f= ⇒ =
16d D 16d
where D is the diameter of the aperture and d the depth of the dish. If
f = d, then f /D = 0, 25.
A dish may be deep or shallow depending on the paraboloid slice envis-
aged during manufacture. Depending on the location of the focal point, we
can have the following situations

• Focus inside the aperture

 9.8 Radiating apertures 253

f 1
<
D 4
Part of the dish is not illuminated i.e. we have under illumination. This
may shield the feed system from spurious radiation.

• Focus flush with aperture

f 1
=
D 4
Illumination of the reflector is more uniform.

• Focus outside aperture

f
=1
D
We have over illumination and some signal may miss the dish altogether.
The gain of the dish antenna depends on the size of the dish and the
wavelength and can be written as

(πd)2 η
Gd = 10 ln
λ2
where d is the reflector aperture, λ the wavelength and η the efficiency,
usually 0, 6-0, 8. Deep dishes with low f/D ratios have higher efficiencies
and are more shielded from noise. From manufacturability considerations,
0, 25 ≤ f/D < 1. Shallow dishes are easier to manufacture and transport
while it is also difficult to illuminate over 180◦ . Beam width design may be
used to counter the effects of over illumination.
According to the feed system location, we can have prime focus with
the focus at the centre of the dish or offset focus antennas with the focus
off centre. Large dishes use prime focus because of the greater mechanical
stability offered by the structure. Medium to high power antennas operating
above 10 GHz employ the offset focus configuration. This is because, for
small dishes, the shadow cast by the feed element is a problem. The offset
focus antenna is installed with the dish inclined towards the cooler sky
hence reducing antenna temperature and allowing for debris to slide off the
dish easily.
According to the Rayleigh limit encountered in optical mirrors, there
is little gain in making the reflector more accurate than λ/8. Hence we
can get away with quite some hefty deviations in the focal length without
compromising performance. Furthermore, the dish need not be made from
a continuous metal sheet but may be out of a mesh as long as the mesh is
small compared to the wavelength at the operating frequency. This allows
for lighter dish construction.
254 9. Antenna theory

FIGURE 9.40. Cassegrain and Gregorian dish antennas

T C
A



In the Cassegrain antenna, the rays reflecting off the parabolic dish to-
ward the dish focus are trapped before they reach the focus by a hyperbolic
sub-reflector. The feed element is now at the back end of the dish (back-fire
configuration). The antenna then has a slimmer profile than the traditional
front end feed system allowing for better shielding of the element from noise.
Due to the extra hyperbolic sub-reflector, the antenna system will be more
expensive however. Since the sub-reflector also blocks signal rays, the main
dish must have an aperture larger than 100λ. This explains why parabolic
antennas are to be found only in the microwave range of operation.
The sub-reflector is installed with one of its focus coinciding with the
focus of the parabola. Note that a hyperbola is described by
 y 2  x 2
+ =4
a b

where
a2 + b2 = c2

and 2c is the distance between the two foci of the hyperbola.

In the back-fire configuration, the feed head is placed near the sub-
reflector so that a shorter focal length gives the system a wide angle view.
The feed head is then fed via a long waveguide. Note that in a mirror tele-
scope, magnifying power increases with reduced angle of view hence longer
focal length mirrors.
G 
A



The hyperbolic sub-reflector is replaced by an elliptical one which then
allows small reflectors, hence higher efficiency at longer wavelengths. The
elliptical reflector is governed by the same equation as the hyperbolic one
with the foci at

y = ± a2 + b2

The Cassegrain and Gregorian antennas are illustrated in Fig. 9.40.

 9.9 Feed techniques for wire antennas 255

FIGURE 9.41. Feed technique for a vertical monopole antenna

9.9 Feed techniques for wire antennas

A common feed situation is that of feeding a wire perpendicular to a con-
ducting plane and that of an isolated dipole using a balanced two-conductor
transmission line.

9.9.1 Monopole (whip antenna) perpendicular to a conducting

plane
Fig. 9.41 shows a simple monopole made out of a coaxial line with the
centre conductor forming the monopole and the outer one shorted to the
conducting plane. If the antenna is resonant, the load on the coax is re-
sistive. For l ∼ λ/4 (a Marconi antenna), the resistance at resonance is
∼ 37, 5 Ω. However deviations from this value arise due to the changing
separation of the inner and outer conductors at the ground connection.
For a air-coaxial line, the ratio of the outer-to-inner conductor diameters
b/a ∼ 1, 868 for Z0 = 37, 5 Ω. Empirically, it has been shown that the
resistance at resonance is ∼ 36, 8 Ω for a/λ = 0, 00159 and l/λ ∼ 0, 236.

9.9.2 The folded dipole

TV and FM broadcast use coupling transmission lines that are balanced
with respect to ground. Such lines usually have a characteristic impedance
typically in the range 300 − 600 Ω. A folded dipole λ/2 long with parallel
conductors of the same diameter shorted at the ends, has an input im-
pedance that is 4 times that of a single dipole for a mode with currents
flowing in the same direction in the parallel conductors. These currents are
of the same magnitude so that if the terminal current is I, then the effective
current is 2I. Hence the radiated power is 4 times that of the single dipole.
256 9. Antenna theory

FIGURE 9.42. The folded dipole antenna

At resonance,

1
P = I 2R
2

where R is the input resistance at resonance. This implies that for the
folded dipole, R is four times that of the single dipole so that for the folded
half-wave dipole of Fig. 9.42, R ∼ 292 Ω, constant over a wider range of
frequencies than in the single half-wave dipole.

9.9.3 BALUN
A dipole in isolation or one in parallel with a ground plane is balanced with
respect to ground, since the dipole has equal currents in its two arms. To
avoid unbalancing the radiating structure, the dipole should be fed using
a balanced transmission line such as a two-conductor line. For frequencies
above ∼ 200 M Hz, the feed line of preference is a coaxial line which is an
unbalanced line and will therefore lead to distortion of the radiation pat-
tern. The coaxial line must therefore be connected to the antenna through
a structure that adapts the unbalanced to the balanced structure. Such
a device is known as a BALanced-to-UNbalanced (BALUN) impedance
transformer. In TV systems with a 75 Ω input impedance and using a
folded dipole antenna, the balun is usually a 4 : 1 impedance transformer.
The balun may take various forms the most common ones being the
transmission line transformer of Fig. 9.43 and the structure shown in Fig.
9.44. For the balun in Fig. 9.44, the centre conductor is connected in parallel
with the short-circuited outer conductor of a coax line a distance λ/4 away
from the feed point. The short-circuit is thus seen as an open-circuit at the
feed point.
 9.9 Feed techniques for wire antennas 257

FIGURE 9.43. A balun as a 4 : 1 step-down impedance transformer

FIGURE 9.44. A balun feed to a dipole

258 9. Antenna theory

9.10 Circuits of a size comparable to signal

wavelength
In a typical circuit, we can express the electric field intensity at a given
point in the circuit path as
−
→
−
→ J
E =
σ
where the quantities carry their usual meaning. The electric field intensity
−
→ −
→
will in general, consist of an applied component E a and a component E i
induced by the charges and the currents in the circuit itself. We therefore
−
→
express E as
−
→ −
→
−
→ − → −
→ −
→ −
→ ∂A J
E = E a + E i = E a − ∇Φ − =
∂t σ
−
→ −
→
The terms ∇Φ and ∂ A /∂t are due to the charge and circuit currents respec-
−
→
tively. Note that with this definition, E is indefinite at a non-conducting
point in the circuit path where σ → 0. The conductivity is itself indeter-
minate inside a localised source i.e. an ideal voltage source. Hence carrying
out the integration only over portions of the circuit path where the con-
ductivity is nonzero but finite, we can write
  −
→  −→ 
−
→ −
→ J −
→ ∂A − → −
→ −
→
Ea · d l − ·d l − · d l − ∇Φ · d l = 0
σ ∂t
We deduce that the term  −
→
J −
→
·d l
σ
represents the total voltage drop across the internal impedance,
 −
→
∂A − →
·d l
∂t
accounts for the inductive effects in the circuit and

−
→ −
→
∇Φ · d l

for the capacitive effects. The term


−
→ −
→
Ea · d l

is the applied voltage.

The applied electric field intensity integrated over the circuit path rep-
resents the applied voltage. In a circuit such as a receiving antenna, the
 9.10 Circuits of a size comparable to signal wavelength 259

EM wave
dl
I

2 2 3
I R C
Va E 4
1 1

dl wire loop
(a) Distributed source (b) Localised source

FIGURE 9.45. Wire loop showing distributed and localised EM sources

applied field is distributed over the entire circuit through the mechanism
of the incident electromagnetic waves. In this case, the integration path is
the closed curve C enclosing the circuit and we can therefore write,

−
→ −
→
Va = E a · d l
C

When the sources are localised, the integral path is across the source. The
applied voltage is in this case independent of the path of integration as
shown in see Fig. 9.45 where
1 2
−
→ −
→ −
→ −
→
Va = Ea · d l = − Ea · d l
2 1

The voltage drop across the internal impedance is the complex contribu-
tion of the internal resistance and reactance. The internal impedance per
unit length of the circuit Zi′ is related to the tangential surface electric field
intensity Em
Em
Zi′ =
I
where I is the total current through the circuit. The total internal im-
pedance is then obtained as
 −→  
1 J −
→ 1 − → −
→
Zi I = ·d l = E m · d l = Zi′ dl
I σ I
where the integral is carried out only over the conductive portions of the
circuit path.
If we select the path of integration to include only the contributions from
the magnetic flux external to the conductor and ignore gaps such as those
at the localised source and capacitors, we can use a closed-path formulation
with negligible error as shown in Fig. 9.46. Thus,
 − → 
∂A − → ∂ − → − →
·d l = A ·d l
∂t ∂t
260 9. Antenna theory

integration path

V +
-
wire loop

FIGURE 9.46. Integration path for determination of external inductance term

which from Stokes’ theorem can be re-expressed as

   
∂ → −
− → ∂ −
→ − → − → ∂ − → − → ∂I
A ·d l = ∇ × A · dS = B · dS = L
∂t ∂t ∂t ∂t
S S

This is the contribution from the flux threading the chosen circuit path
and is valid for a circuit of a size much smaller than the signal wavelength.
−
→
In Fig. 9.45, we integrate ∇Φ over the conductive portion of the circuit
path between points 3 and 4 and assume charges due to the capacitance are
negligible at the source. Recalling that the integral of a gradient function
around a closed path is zero, then the voltage across the capacitor is related
to the charge Q on the capacitor as

Q
V =
C
Hence
3 
−
→ −
→ Q 1
∇Φ · d l = Φ3 − Φ4 = = I dt
C C
4

For a circuit of a size comparable to the wavelength, we must use retarded

potentials. Consider the case of a receiving antenna having a current as-
sumed to flow through a thin wire. Assuming that the resistance in the
circuit is small and that the circuit contains no capacitors, we can write
 
−
→ −
→ → −
− →
E a · d l − jω A · d l = 0

where harmonic time dependence has been assumed as well. For a filamen-
tary current,

−
→ µI exp (−jkr) −→
A= dl ′=0
4πr
 9.10 Circuits of a size comparable to signal wavelength 261

where r is the distance of the observation point from the current element
under consideration. Recalling that
−
→ − → − →
B =∇×A
and
−
→ ω−→ −→ − → −
→
E = −j 2 ∇ ∇ · A − jω A
k
we can therefore write
  
−
→ −
→ µI cos kr − j sin kr −
→ − →
E a · d l − jω d l ·d l ′ =0
4π r
Thus, even with a current that is in phase throughout the circuit, a finite
value of kr leads to a real and imaginary contribution. The real part of the
contribution is related to energy radiation and the imaginary part to an
inductive reactance.
We therefore see that, in a circuit of a size comparable to the signal
wavelength, the retardation effect from one part of the circuit to another,
causes a phase shift which leads to induced effects with real and imaginary
parts unlike the purely reactive effects in a circuit that is small relative to
a wavelength.

9.10.1 Fields in a radiating structure

For a harmonic time variation in a simple medium, we can write
−
→ 1− → −→
H = ∇×A
µ
−
→ −
→ −
→
E = − ∇Φ − jω A
where the phasor retarded potentials are

−
→ µ − → exp (−jkr) ′
A= J dV
4π r
V′

1 exp (−jkr) ′
Φ= ρ dV
4πǫ r
V′
√
with k = ω µǫ = 2π/λ.
We also have
−
→ − →
∇ · J = −jωρ
and
−
→ − →
∇ · A = jωǫΦ
the Lorentz gauge. Hence, there is usually no call to evaluate the integrals
−
→ −
→
involved in the definition of the sources since E and H are related through
a spatial differentiation. We can therefore determine
262 9. Antenna theory

FIGURE 9.47. The resonant dipole and monopole antennas

−
→ −
→
(i) A from J
−
→ −
→
(ii) H from A
−
→ −
→
(iii) E from H

9.11 Overview of practical radiating systems

9.11.1 Dipole antennas
These are the most common radiating structures and encountered in most
communication installations. The dimensioning of the antenna is such that
it is resonant at the radiation frequency and hence has a total length of
λ/2 which is often known as a Hertz radiator operated at frequencies above
2 MHz. If only one arm of the resonant dipole is used, this is a monopole
often referred to as a Marconi antenna and is operated usually at frequen-
cies below 2 M Hz. The dipole and the monopole are shown in Fig. 9.47.

9.11.2 Loop antennas

The loop antenna is shown in Fig. 9.48. When the loop size is small, the
radiation pattern is similar to that of a small dipole but with orthogonal
polarisation to the dipole, i.e. electric and magnetic fields interchanged.
The loop is referred to as a magnetic dipole.

9.11.3 Traveling-wave antenna

If a long transmission line is properly terminated such that there are no
reflections on the line, an electromagnetic field can be launched into space
with two lobes at an angle to the wire. If two wires are thus arranged into
a V-form, we have the V-antenna. A derivative of the V-antenna is a the
rhombic antenna which uses two V-antennas as shown in Fig. 9.49. The
 9.11 Overview of practical radiating systems 263

FIGURE 9.48. The loop antenna

FIGURE 9.49. Rhombic and dielectric radiating structures

arms of the rhombus are usually 2-7λ long and the antenna placed 1-2λ
above ground.
A wave propagating in a waveguide operated near cut-off, has a phase
velocity approximately equal to the velocity in the unbounded medium. The
surface waves propagating on a dielectric guide may then excite appreciable
fields in space. This structure is also shown in Fig. 9.49.

9.11.4 Slot and aperture antennas

Fields in an aperture may excite radiation in space. When the apertures
are small, they must be resonant so as to excite appreciable radiated power,
however larger ones need not be resonant. A typical aperture antenna is
the electromagnetic horn and is to be found as the feed element in many
parabolic dish antenna systems.
A thin resonant (λ/2 in length) slot shares many similarities with the λ/2
dipole but with the electric and magnetic fields interchanged. An example
of such a radiating system is the leaky waveguide. The leaky waveguide
and the electromagnetic horn are shown in Fig. 9.50.
264 9. Antenna theory

FIGURE 9.50. The electromagnetic horn and leaky waveguide radiating struc-
tures

9.11.5 Reflectors and lenses

Most common reflector is the parabolic dish for microwave radiation. The
corner reflector is also to be seen in many TV antenna systems in the UHF
bands. For the parabolic dish, the beam becomes narrower with increasing
aperture size e.g. for an aperture of ∼ 70λ diameter, the half-power beam
width is ∼ 1◦ .
Lens systems serve a similar purpose. These may have
• A solid dielectric lens
• Artificial dielectric made from metal particles dispersed in a support-
ing dielectric
• Metal waveguide with appropriate phasing

9.11.6 Integrated-circuit (patch) antennas

An example of these antennas is shown in Fig. 9.51.Other newer structures
are the fractal antennas that promise wideband operation.

9.12 Antenna analysis by field computation

Calculation of the radiating fields and power from radiating structures re-
quires the exact knowledge of the current distribution on the antenna al-
lowing the use of retarded potentials. In many practical structures however,
the current distribution cannot be readily determined. Fortunately, every
antenna can be regarded as made up of many infinitesimal current elements
regarded as short electric dipoles. Superposition of the radiated fields from
the individual short dipoles is then the total field from the whole antenna.
 9.12 Antenna analysis by field computation 265

FIGURE 9.51. A microstrip patch antenna

The antenna field can be divided into the near- or Fresnel field and the
far- or Fraunhofer field. The far-field is important in wireless communica-
tion and we will confine ourselves to the study of this field. In the far-field
region, we assume
• The differences in the lengths of the radial vector − →r at different
points on the antenna as observed from an observation point P in
the far-field are insignificant in describing the magnitude of the field
contributions. This implies that the wavefront far away from the ra-
diating structure can be regarded as planar.
• The differences in the lengths of the radial vector are important when
regarding the phase differences between different points on the an-
tenna as observed from the far-field zone.
• All field components decreasing with distance from the radiating
structure faster than 1/r are insignificant in the far-field (Sommerfeld
radiation condition).
Consider the volume V ′ enclosing a point source a short distance r′ away
from the origin of the spherical coordinate system as shown in Fig. 9.52.
The vector potential at P at r >> r′ from the source is given by
 −
→
−
→ J exp (−jkR) ′
A =µ dV
4πR
V′

where R is the distance of P from the point source. Applying the cosine
rule, 
R = r2 + r′2 − rr′ cos ϕ ∼ r − r′ cos ϕ
where ϕ is the direction angle between −
→
r and −
→r ′ . Hence

−
→ µ exp (−jkr) −
→
A= J exp (jkr′ cos ϕ) dV ′
4πr
V′
266 9. Antenna theory

FIGURE 9.52. Far-field calculation for a point source

−
→
This integral is a function of the assumed current distribution J and
the direction angle alone and can be defined as a radiation vector

−
→ −
→
N = J exp (jkr′ cos ϕ) dV ′
V′

so that
−
→ µ exp (−jkr) −
→
A= N
4πr
−
→
Generally, N may have components in any direction so that
→ µ exp (−jkr) 
− 
A= r
Nr +

φ
θNθ + φN
4πr
In spherical coordinates, we have
 
 
−
→ −
→ − → r
∂ ∂Aθ θ 1 ∂Ar ∂
B = ∇×A = (Aφ sin θ) − + − (rAφ )
r sin θ ∂θ ∂φ r sin θ ∂φ ∂r

∂
φ ∂Ar

+ (rAθ ) −
r ∂r ∂θ
Field components not decreasing faster than 1/r are thus,
1 ∂ jk
Hθ = − (rAφ ) = Nφ exp (−jkr)
µr ∂r 4πr
and
1 ∂ jk
Hφ = (rAθ ) = − Nθ exp (−jkr)
µr ∂r 4πr
−
→ − 
→ −
→ − → −
→
Since E = ∇ ∇ · A − jω A , then only

jωµ
Eθ = − Nθ exp (−jkr)
4πr
 9.12 Antenna analysis by field computation 267

and
jωµ
Eφ = − Nφ exp (−jkr)
4πr
decrease at a rate not faster than 1/r. Thus, there are no radial electric
fields.
The time-average Poynting vector is then
1 ' ( η  2 2

Pr = Re Eθ Hφ∗ − Eφ Hθ∗ = 20 2 |Nθ | + |Nφ |
2 8λ r
where
Eθ Eφ
η0 = =−
Hφ Hθ
and the time-average radiated power is
π 2π π 2π 
η0
PT = 2
Pr r sin θ dθ dφ = 2 |Nθ |2 + |Nφ |2 sin θ dθ dφ
8λ
0 0 0 0

−
→
The Poynting vector P gives the actual power density at any point. We
define a radiation intensity K independent of the radial distance r from the
source as the power radiated in a given direction per unit solid angle and is
hence the time-average Poynting vector on a unit-radius sphere. Therefore,
we can write
η  
K = 02 |Nθ |2 + |Nφ |2
8λ
so that
π 2π
PT = K sin θ dθ dφ
0 0

If all the currents are in one direction which may be taken as the axis of a
−
→ −
→
set of spherical coordinates, then N and hence A has only one component,
say Nz . Thus
Nφ = 0, Nθ = −Nz sin θ
and
η0 2 2
K= 2 |Nz | sin θ
8λ
in this case. With the currents directed azimuthally about an axis however,
the axis could be taken as the axis of a spherical coordinate system. If
symmetry in φ exists,
η
K = 02 |Nφ |2
8λ
and
π
PT = 2π K sin θ dθ
0
268 9. Antenna theory

dSθ

dSφ dSr
rdθ

dθ dr
θ
y
φ
rsinθ
rsinθdφ
x

FIGURE 9.53. The spherical coordinate system

Note that as shown in Fig. 9.53, the Cartesian coordinates (x, y, z) are
expressed in terms of the spherical coordinates as

x = r cos φ sin θ

y = r sin φ sin θ

z = r cos θ

Thus, in Cartesian coordinates

Nθ = (Nx cos φ + Ny sin φ) cos θ − Nz sin θ

and
Nφ = −Nx sin φ + Ny cos φ

If (θ, φ) are the angular coordinates of P and θ′ , φ′ those of the source,
then

cos ϕ = cos θ cos θ ′ + sin θ sin θ′ cos φ − φ′
 9.13 Waves guided by biconical guide 269

FIGURE 9.54. The biconical waveguide

9.13 Waves guided by biconical guide

Waves on a biconical guide are important for understanding waves prop-
agating along dipole antennas and some cavity resonators. An important
mode propagates along the cones at the velocity of light and has only radial
field components with the wave symmetrical about the cone axis. In the
biconical guide of Fig. 9.54, an independent set of solutions has Eθ , Er and
Hφ components only with

1 ∂ 1 ∂Er
(rEθ ) − + jωµHφ = 0
r ∂r r ∂r
 
1 ∂
(sin θHφ ) − jωǫEr = 0
r sin θ ∂θ
1 ∂
− (rHφ ) − jωǫEθ = 0
r ∂r
and
Er = 0
η
rEθ = {A exp (−jkr) + B exp (jkr)}
sin θ
1
rHφ = {A exp (−jkr) − B exp (jkr)}
sin θ
where
Eθ
η=
Hφ
and ∂/∂φ = 0 because of the φ-symmetry.
270 9. Antenna theory

The voltage between the two cones is given by

 0
π−θ  0
π−θ
1
V = Eθ rdθ = −η {A exp (−jkr) + B exp (jkr)} dθ
sin θ
θ0 θ0


θ0
= 2η ln cot {A exp (−jkr) + B exp (jkr)}
2

from which we see that V is independent of r except through the propaga-

tion factor exp (∓jkr).
The azimuthal magnetic field intensity corresponds to a current flow in
the cones at θ = θ0 , i.e.

I = 2πrHφ sin θ 0 = 2π {A exp (−jkr) − B exp (jkr)}

which is independent of r as well. At any radius r, the currents on the two

cones are of opposite phase just like in a TEM line. Thus, the biconical
guide is characterised by a characteristic impedance

ln cot θ20
Z0 = η
π
If the cone angle θ0 is small, then the cones can be regarded as cylindrical
wires and the current distribution along the cone length is then sinusoidal
to a first order. This forms the basis of a wire dipole antenna.
10
Fundamentals of electromagnetic
compatibility

Electromagnetic compatibility (EMC) has now become a mandatory re-

quirement for electronic equipment. In the past, EMC referred to a variety
of guidelines used in industry to describe the radiated influence and/or
susceptibility of an electrical/electronic appliance in an increasingly rich
electronic environment. Each country had its own requirements for electron-
ics. In 1996, European countries issued an EMC directive (89/336/EEC,
IEC801), which was made binding for the first time.
EMC describes the ability of an electrical piece of equipment to function
satisfactorily in its electromagnetic environment without affecting other
equipment in this environment to an impermissible extent. Manufactur-
ers of electrical and electronic equipment must declare that the equipment
conforms to the protective objectives of the EMC directive through a CE
(Conformitè Europeèn) mark of conformity on all equipment and pack-
aging. While the CE mark does not give full technical qualification, it
indicates that the device meets the fundamental requirements for environ-
mental and safety in Europe. The manufacturer must take full responsi-
bility vis-à-vis their consumers for observing the relevant emission limits
and interference immunity requirements. Since its introduction, the EMC
directive 89/336/EEC has had a sweeping effect on the design of electri-
cal/electronic products worldwide.
Interference immunity requirements are particularly important for equip-
ment operators, installations and systems. Correct functioning is ensured
only if sufficient EMC measures are taken. Constant functionality implies
high availability of installations and systems, hence higher performance
figure for a cost-effective operation of the equipment.
272 10. Fundamentals of electromagnetic compatibility

Electromagnetic interference (EMI) refers to unwanted signals in devices

or systems that make operation difficult or impossible. If the system is
not EMI compatible, it has an EMI problem. An EMI signal must have a
source and a coupling path to the disturbed or victim equipment. Hence,
removing the source and/or the coupling path will solve the EMI problem.
The path need not be a conductive medium since in most case the coupling
mechanism consists of both radiation and conduction. Hence one method
of alleviating the EMI problem is to keep the source strengths across the
coupling path below certain levels. The pertinent levels are covered in the
EMC directive 89/336/EEC and IEC801 applicable to equipment that can
emit disturbance signals or devices that are sensitive to disturbances from
electromagnetic energy such as portable tools, computer peripherals and
industrial equipment. IEC guidelines do not necessarily cover radio and
telecommunication equipment.
To minimise EMI, one must first determine whether the system operates
at high or low frequencies. For EMC purposes, a system is said to operate at
low frequencies if all transmission lines present in the system have lengths
shorter than a tenth of the shortest wavelength of the signals it processes.
If a system has a current flowing from the supply to a load, this current is
subject to change in magnitude at some instant of time. If this change is
not noticeable within a time
l
t=
c

where l is the length of the transmission line and c the speed of light, the
system is said to be a low frequency circuit and the small perturbations in
the currents then referred to as quasi-static phenomena. The low frequency
circuit, can then be modeled with an RLC circuit. Additional issues arise
in high frequency systems such as propagation delays, reflections and radi-
ation, which are no longer negligible.

10.1 Basic information on EMC

As shown in Fig. 10.1, EMC covers both electromagnetic emission (EME)
and electromagnetic susceptibility (EMS). Emission applies in the case of
an EMI source and susceptibility to the victim. Conducted emission (CE)
may be in the form of interference currents or voltages while radiated ones
may be electric or magnetic fields near the source or electromagnetic fields
further away. For cost-effective solutions, all phenomena must be considered
with EMC components for EMI suppression installed either in the source
or the victim.
 10.1 Basic information on EMC 273

EMC
Emission (source) Susceptibility (victim)

EME EMS
CE (conducted) CS (conducted)
RE (radiated) RS (radiated)

FIGURE 10.1. Components of EMI

TABLE 10.1. Examples of discrete and continuous spectrum EMI sources

Discrete spectrum Continuous spectrum
microprocessor and microcontroller systems switchgear (contactors and relays)
RF generators household appliances
medical equipment gas discharge lamps
data processing systems power supplies and battery chargers
microwave equipment frequency converters
ultrasonic equipment welding machines
radio and TV receivers motors with brushes (commutator)
switched-mode power supplies oscillating drives
UPS systems Atmospheric discharges
electronic ballast in gas discharge lamps

10.1.1 EMI sources and victims

An EMI source is an electrical equipment which emits EMI. Examples of
EMI sources are listed in Table 10.1. These can be classified under two
groups
• Sources with discrete spectra e.g. high frequency oscillators and mi-
croprocessor systems which generate narrowband interference (low
energy sine waves)
• Sources with continuous spectra e.g. switch gear and electric motors,
in particular, commutated DC motors in household appliances which
generate broadband interference (impulse, high energy)
The EMI victims are electrical devices, equipment or systems subjected
to interference and which may be affected by it. Victims can be separated
into two groups as well, i.e. those with narrowband or discrete spectra
and those with broadband (continuous) spectra. Some example victims are
listed in Table 10.2.

10.1.2 Propagation of interference

Interference voltages and currents may be grouped into the three types of
interference illustrated in Fig. 10.2 as,
274 10. Fundamentals of electromagnetic compatibility

TABLE 10.2. Examples of discrete and continuous spectrum EMI victims

Discrete spectrum Continuous spectrum
radio and TV receivers Digital and analogue systems
modems data processing systems
data transmission systems process control computers
radio transmission equipment control systems
remote control equipment video transmission systems
microwave equipment frequency converters
cordless and cellular phones interfaces

• Common-mode interference

• Differential-mode interference

• asymmetrical interference.

C
- I


Common-mode interference which is mainly a high frequency (≥ 1 MHz)
phenomenon, occurs between all lines in a cable and the reference potential.
If the potentials on the conductors of a two-conductor transmission line are
V1 and V2 resp., the common-mode potential is defined as

V1 + V2
VCM = − Vref
2
where Vref is the reference potential. Common-mode currents cannot occur
when the paths between the conductors are ideal but this is not possible
in practice.
D 
 - I


Differential-mode interference occurs symmetrically between two lines
and is primarily a low-frequency (below several hundreds kHz) phenom-
enon. If the conductors of a two-conductor transmission line have potential
V1 and V2 resp., the differential-mode potential is defined as

VDM = V1 − V2

A  I


Asymmetrical interference describes the interference between one line
and a reference potential.
In the case of radiated interference, we can have differential radiation
currents which are a natural consequence of the physical layout of the
circuit. Different PCB tracks (traces) may carry different current densities
with differential-mode currents leading to H-field radiation. Common-mode
radiation currents mainly lead to E-field radiation due to losses through
the cables and PCB tracks.
 10.1 Basic information on EMC 275

VDM
VCM V1 V2

Common-mode Differential-mode Asymmetrical

FIGURE 10.2. Types of interference propagation mechanisms

10.1.3 Characteristics of interference

Interference can be classified in terms of

• Mode of propagation
• Coupling mechanism

At low frequencies, propagation is mainly in terms of conduction over

conductive structures with conductive interference for frequencies below a
few hundred kHz primarily differential-mode. At high frequencies, prop-
agation is only by means of EM radiation with the term coupling used
to describe the interference phenomenon in the M Hz range. Interference
at high frequencies is mainly common-mode and increases with frequency
since the coupling factor and effects of parasitic capacitance and inductance
between conductors increase with frequency as well.

10.1.4 Interference measurement

Since interference signals can be either conducted or radiated, propagation
along these paths can be detected by measuring the interference current
and voltage. Interference effects in their immediate vicinity is assessed by
measuring the magnetic and electric fields, commonly referred to as mag-
netic or electric coupling (near-field). At high frequencies, the equipment
dimensions are of the order of the signal wavelength and interference energy
is mainly radiated directly (far-field). Radiated and conducted interference
energy must be taken into account when testing the susceptibility of a
victim equipment.
In measurement, interference sources such as sinusoidal oscillators as
well as pulse generators are usually used to generate known amounts of
interference.
C
- C
 M

Common-mode currents can be measured during the design stage. Emis-
sion of most devices is dependent on the magnitude of the high frequency
currents on the cables and PCB tracks. As a rule of thumb, the maximum
allowed common-mode current must be less than 5 µA at a distance of 3 m
276 10. Fundamentals of electromagnetic compatibility

FIGURE 10.3. Measurement setup for common-mode interference currents

from the common-mode source. In general, this means that the maximum
measured emission must be 30 µV m−1 .
The common-mode currents are related to the electric field as
60ICM
E=
r
where ICM is the common-mode current and r, the distance from the
common-mode source where the field is measured. Emission may be due to
capacitive coupling between the lines or from a line to ground. Fig. 10.3
shows a test setup for the common-mode current measurement. Differential-
mode currents could be measured with the same setup when one line is
wrapped around the transformer once.
Common-mode currents cause unwanted radiation and can be suppressed
by placing a choke or clamp such as ferrite beads, in the current path. Dig-
ital ribbon cables can give significant problems with common-mode cur-
rent radiation and clamps may be the more feasible approach to suppress
this radiation. Note that ferrite beads or clamps will have no effect on
differential-mode radiation.
D 
 - C
 M

The far-field H-field is related to the far-field E-field as
E
H=
η
where the symbols carry their usual meaning. The H-field is measured at
0, 1 m from the differential-mode source and the E-field at 3 m, which forms
the legal length of cables supplied with consumer products for operation
below 16-300 MHz band.
 10.1 Basic information on EMC 277

FIGURE 10.4. Measurement setup for the magnetic field

The H-field can be measured using a current loop made out of a coaxial
cable. The outer conductor is removed and a 50 Ω chip resistor connected
between a loop formed by the inner conductor. The H-field induces a volt-
age across the resistor which is measured by the measuring system. The
sensitivity of the probe is high for low frequencies but becomes indepen-
dent of frequency above about 300 M Hz. The reason for this is that while
the voltage increases with frequency, the inductive impedance of the probe
increases in the same direction preventing any further increase in current.
The E-field measured at a distance r = 3 m from the source is related to
the 0, 1 m H-field as
−
→
E = 2, 7 × 10f × H
where f is the frequency in Hz. A measurement setup for the H-field is
shown in Fig. 10.4. Fig. 10.5 shows the general EMI environment.

10.1.5 Propagation of conducted interference

Fig. 10.6 illustrates the mechanism involved in conducted interference. A
floating interference source emits a primarily differential-mode interference
which then propagates along the connecting lines by conduction. The in-
terference current flows from the emitting source toward the victim on one
conductor and returns to the source from the victim through the other
conductor. Differential-mode interference is important mainly at low fre-
quencies.
Parasitic capacitances in interference sources and disturbed equipment
or intended ground connections lead to interference currents in the ground
circuit. These common-mode currents flow toward the disturbed equipment
278 10. Fundamentals of electromagnetic compatibility

FIGURE 10.5. The EMI environment and measurement scheme

Interference source Disturbed equipment

ICM
IDM
Cpara R Cpara
ICM
IDM

ICM

FIGURE 10.6. Propagation of conducted interference

 10.1 Basic information on EMC 279

through the connecting conductors and return through the ground connec-
tion. Since the parasitic capacitive impedances tend toward short-circuits
at high frequencies, the coupling effects in the connecting cables and the
equipment itself will increase with frequency. Hence, common-mode inter-
ference currents become dominant in the M Hz range. Asymmetrical inter-
ference consists of both symmetrical (differential-mode) and common-mode
components.

10.1.6 Cross-talk
When the coupling path between the EMI source and victim is short, we
have cross-talk. Examples are when both the source and victim share a
common impedance which then forms the path for the cross-talk signal as
well as inductive or capacitive coupling through the magnetic or electric
fields resp. Cross-talk must be particularly addressed when designing PCBs
or multi-cable assemblies.
Fig. 10.7 illustrates a network in which the output and input circuits
share a common impedance Zm . The currents in the two circuits flow
through Zm to the reference node which is usually the circuit ground. By
superposition, the voltage established across the load ZL2 by the current
from Vs1 is given by
ZL2 Zm // (ZL2 + Zs2 )
VZL2 = Vs1
ZL2 + Zs2 Zm // (ZL2 + Zs2 ) + ZL1 + Zs1
Zm ZL2
= Vs1
Zm (ZL2 + Zs2 ) + (ZL1 + Zs1 ) (ZL2 + Zs2 )
and when Zm is small, we can write
VZL2 Zm ZL2
=
Vs1 (ZL1 + Zs1 ) (ZL2 + Zs2 )
We therefore see that cross-talk depends on Zm and there will be no cross-
talk if Zm = 0.
The effect of cross-talk on the signal is specified through the distortion
ratio defined as the voltage drop across ZL2 for Zm = 0 to the resulting
drop when Zm = 0. Thus,
[VZL2 ]Zm =0 Zs1 + ZL1 Vs2
=
[VZL2 ]Zm , small Zm Vs1

If for example Zs1 = 10 Ω, ZL1 = 50 Ω, Zm = 0, 1 Ω, Vs2 /Vs1 = 0, 01, the

distortion ratio is ∼ 15, 5 dB.
In order to lower the value of Zm , two conductors rather than one are
used, the lead shortened or low-Zm conductors used, e.g. dedicating a layer
of the PCB as the reference plane. Another solution is to use a galvanic
280 10. Fundamentals of electromagnetic compatibility

Zs1 Vs1 Vs2 Zs2

+ - - +

I1 Zm I2

ZL1 ZL2

FIGURE 10.7. Circuit showing cross-talk through a coupling impedance

FIGURE 10.8. Capacitive cross-talk

separator e.g. an isolation amplifier or an optocoupler between the two

circuits.
Cross-talk can also occur through radiation either capacitively or induc-
tively whereby capacitively coupled cross-talk signals are transmitted in
the form of electric fields. Fig. 10.8 shows a case of capacitive coupling
between two circuits in close proximity as well as a possible circuit model
for the situation. The ratio of the load to the source voltages is given by
VL
= j0, 25ωCAB ZL
VS
We therefore deduce that capacitive coupling
• Behaves like a high-pass filter
• Cross-talk increases with ZL
• Cross-talk depends on VL /VS
• Cross-talk signals are in phase with each other
Two PCB traces at different potentials form a parasitic capacitance be-
tween them whose magnitude depends on the ratio of the length of the two
 10.1 Basic information on EMC 281

FIGURE 10.9. Inductive cross-talk

traces to their separation. To a first order,

ǫ
CAB = W l pF
d
where W ∼ 0, 003 mm is the width of the PCB trace, l is the trace length
and d is the trace separation. Capacitive cross-talk can be minimised by
keeping CAB low, i.e. use a larger trace separation or reducing the surface
area over which coupling occurs. When this is not viable, shielding around
one or both conductors could be used.
Magnetic coupling dominates inductive cross-talk and can be compared
to transformer action, i.e.
VL M
= −j0, 25ω
VS ZL
where M is the mutual inductance between the two coils shown in Fig. 10.9
(b). Again, we see that inductive cross-talk has a high-pass characteristic
with the level of cross-talk increasing with decreasing load impedance for
the same amount of coupling M. The undesirable source and load voltages
VS and VL are however out of phase unlike in the case of capacitive cross-
talk.
Inductive cross-talk can be minimised by increasing conductor separa-
tion or twisting the send and return conductors about each other where
applicable. In such a case, the H-fields counteract hence reducing cross-
talk signal levels. In Fig. 10.9 (a), the magnetic field at X generated by the
282 10. Fundamentals of electromagnetic compatibility

current in the conductors of circuit A is given by

IA D1
H∼ 2
2π rX
where IA is the current in circuit A. It is therefore necessary that

• Each circuit uses its own conductors

• Conductor separation within a circuit be kept as close as is only
possible to reduce loop area
• A small resistor in the signal and/or supply lines is provided
• Only the necessary bandwidth required by the particular application
is used
• Double or multi-layer PCB are used in a PCB design to create con-
ductive shields.

10.1.7 Grounding and referencing

Many EM problems can be traced back to poor circuit grounding. In EMC
jargon, ground or earth-protection is referred to as the system reference
(SR) which defines the central point in a circuit used to check other signals.
The SR will often be defined at the supply which is usually connected to
the earth-protection.
The SR will generally carry many in- and out-going currents making
its location of paramount importance. The impedance offered by the SR
should be as low as possible and any EMI signals in its neighbourhood be
kept low. From an EMC point of view, the location of the SR depends on

• Circuit emission
• Circuit immunity
• Circuit conductors
• Component placement in the circuit

When a system employs multiple PCBs, a star connection of partici-

pating boards is recommended for the supply and system references. This
provides lower impedance SR and eliminates large differential current loops
(ground loops). Short cables should be connected on one side of each board
for best performance.
When using mixed-signal designs, similar components should be grouped
close together so as to share a common ground plane. For example,

• Group analogue components

 10.1 Basic information on EMC 283

• Group high-frequency components close to the connectors to reduce

line inductance

• Group digital components and use the slowest components the appli-
cation can tolerate.

• Separate digital and analogue circuit ground planes where possible.

However, note that ADCs must have the digital parts sharing the
analogue circuit ground plane.

10.1.8 Antenna effect

The, the most important parameter of a cable is usually its characteristic
impedance. Since cables and connectors will act like antennas in a circuit,
then in view of the EMC rule for emission and immunity design,

• Place cables and connectors on one side of the PCB and as close to
each other as possible

• Keep in mind, the effects of parasitics since resistors, inductors and

capacitors behave like RLC networks at high frequencies

• Use as wide a bandwidth as the application requires

• Create a low-impedance reference layer on the side of the board car-

rying the cables and connectors

• Keep in mind that each PCB trace is an antenna for interference

signals

• Use a shield with small holes compared to the shortest wavelength of

interest whenever such a shield is required. Holes required for incom-
ing cables or heat transfer are aperture antennas, hence use many
small ones rather than one big one.

• Keep the straight parts of the traces as short as possible and use 45◦
bends in the traces whenever required. To illustrate, a trace of length
0, 6 mm resonates and hence radiates strongly at ∼ 5 GHz, since
c
f=
l
where l is the trace length, c is the speed of light and f is the oper-
ating frequency. Reducing the longest trace to 0, 18 mm pushes the
resonance frequency to ∼ 16, 6 GHz.
284 10. Fundamentals of electromagnetic compatibility

10.2 The EMC guidelines

The EMC guidelines specify measurement techniques and limiting val-
ues or test severities for both interference emissions and interference im-
munity of electrical equipment, installations and systems. The European
standards are divided into categories for convenient reference for the re-
spective equipment. The basic standards contain information on the in-
terference phenomena and general measurement methods. Equipment not
covered by the basic standards are treated under the generic standards.
Attached to this document are the PDF documents emc_regulations.pdf
and henry_benitez_emc_regulations.pdf for further information on EMC
standards.
The goal of the EMC guidelines, is to design electronic systems that are
electromagnetically compatible with their environment. EMC requirements
exist so that electronic system designers have a set of guidelines that explain
the limits of what is considered electromagnetically compatible. However,
an all-encompassing set of EMC guidelines does not exist and the EMC
guidelines are created by individual product manufacturers and govern-
ments. Requirements set forth by the government are legal requirements
that products must meet. Requirements set by the manufacturer, on the
other hand, are self-imposed and often more stringent than those set by
the government.

10.2.1 Government requirements

Not all countries have the same EMC requirements. Each country is re-
sponsible for enforcing their own set of requirements but this does not
imply that each country has a unique set of EMC requirements. Various
EMC requirements set by all the countries are very similar, with many
countries now moving toward an internationally agreed standard for EMC
requirements known as the CISPR 22 standards. These standards have
been adopted throughout much of Europe and were developed in 1985 by
CISPR (International Special Committee on Radio Interference).
In the United States, the Federal Communications Commission (FCC)
is charged with the regulation of radio and wire communication. Radio
frequency devices are the primary concern in EMC and are defined by the
FCC as devices capable of emitting radio frequency energy by radiation,
conduction or other means whether intentionally or not. FCC defines radio
frequencies as frequencies in the range 9 kHz − 3000 GHz. Some examples
of radio frequency devices include

• Digital computers whose clock signals generate radiated emissions,

• Blenders that have DC motors where arcing at the brushes generates

energy in this frequency range,
 10.2 The EMC guidelines 285

• Television sets that employ digital circuitry.

Practically all digital devices are regarded as radio frequency devices.

With the advent of digital computers and other digital devices becoming
popular, the FCC realized that it was necessary to impose limits on the
electromagnetic emissions of these devices in order to minimize the poten-
tial interference with radio and wire communications. As a result, the FCC
set limits on the radiated and conducted emissions from digital devices.
As defined by the FCC, a digital device is any unintentional radiator
(device or system) that generates and uses timing pulses at a rate in excess
of 9000 pulses or cycles per second and uses digital techniques to process
the signals. All electronic devices with digital circuitry and a clock signal
in excess of 9 kHz are therefore covered under this rule, albeit with a few
exceptions.
The law makes it illegal to market digital devices whose conducted and
radiated emissions have not been characterised and verified to be within
the limits set by the FCC regulations. This means that digital devices
that have not been confirmed as passing these requirements can not be
sold, marketed, shipped, or even be offered for sale. Although penalties
for violating these regulations include fines and or jail time, companies are
usually more concerned with the negative publicity that would ensue once
it became known that they had marketed a product that fails to meet the
FCC regulations. Furthermore, if the product were already made available
to the public, the company would be forced to recall the product. It is
therefore important that every unit that a company manufactures is FCC
compliant.
Although the FCC does not test each and every module, they perform
random tests on products and the entire product line can be recalled if
a single unit fails to comply. The FCC has different sets of regulations
for different types of digital devices. Devices marketed for use in commer-
cial, industrial or business environments are classified as Class A digital
devices. Devices marketed for use in residential environments, notwith-
standing their use in commercial, industrial, or business environments are
classified as Class B. In general the regulations for Class B devices are more
stringent than those for Class A. This is because in general, digital devices
are in closer proximity in residential environments, and the owners of the
devices are less likely to have the skill and or resources to correct potential
problems.
Fig. 10.10 show a comparison of the Class A and Class B conducted
emissions limits, where it can be clearly seen that regulations for Class
B devices are stricter than those for Class A. Personal computers are a
subcategory of Class B devices and are regulated more strictly than other
digital devices. Computer manufacturers must test their devices and submit
their test results to the FCC. No other digital devices require that test data
be sent to the FCC, the manufacturer is expected to test their own devices
286 10. Fundamentals of electromagnetic compatibility

3500

class A digital devices

3000

2500
voltage in µV

2000

1500

1000

1,705
500
250
0
0,01 0,1 1 10 100

frequency in MHz

FIGURE 10.10. A comparison of the FCClimits for conducted emisions in class

A and B digital devices

to be sure they are electromagnetically compatible with the FCC policing

the industry through testing of random product samples.
Conducted emissions are the currents that are passed through the AC
power cord out of the digital unit and placed on the common power grid.
Conducted emissions are undesirable because once these currents are in
the building wiring, they radiate very efficiently as the network of wires
acts like a large antenna. The frequency range of conducted emissions ex-
tends from 450 kHz to 30 M Hz. Devices are tested for compliance with
conducted emissions regulations by inserting a line impedance stabilization
network or LISN into the AC power cord. Current passes through the AC
power line and into the LISN, which then measures the interference cur-
rent and converts this current to a voltage for measurement. The actual
FCC regulations set limits on the voltage from the LISN although it is the
current that is being regulated.
Radiated emissions, on the other hand, are the electric and magnetic
fields radiated by the device that may be received by other devices, and
cause interference in those devices. Although radiated emissions are both
electric and magnetic fields, the FCC and other regulatory agencies only
require that electric fields be measured for certification. The magnitudes
of these fields are measured in dBmV m−1 with the frequency range for
radiated emissions extending from 30 M Hz to 40 GHz. Radiated field
measurements for FCC compliance are carried out either in a semi-anechoic
chamber or at an open field test site. The product under test must be
 10.2 The EMC guidelines 287

rotated so that the maximum radiation will be achieved and measurements

must be made both with the measurement antenna in the vertical and
horizontal polarizations with respect to the ground plane.
The measurement method for radiated emissions depends on the type of
device under test. Class A digital devices must be measured at a distance
of 10 m from the device while Class B devices are to be measured at a
distance of 3 m. This is because of the stricter regulations for Class B
devices. Fig. 10.11 illustrates the radiated emission limits defined by the
FCC for Class A and Class B digital devices. Because the measurement
distances defined by the two requirements are different, we must scale the
measurement distances so that they are both at the same distances in
order to achieve an accurate comparison. One way to do this is with the
inverse distance method, which assumes that emissions fall off linearly with
increasing distance to the measurement antenna. Thus emissions at 3 m are
assumed to be reduced by 3/10 if the antenna is moved out to a distance
of 10 m. Hence, to translate Class A limits from a distance of 10 m to 3 m
, add 20log10 (3/10) = 10, 46dB to the Class A limits. This approximation
is however only valid if the measurements are taken in the far field of the
emitter. We can assume that the far field boundary is three wavelengths
from the emitter, and with the radiated emissions frequency range defined
as 30 MHz to 40 GHz, the maximum distance from the emitter for far
field measurements is 30 m. Therefore, at 10 m, not all measurements will
be in the far field. At 10 m, only frequencies above 90 MHz will be in the
far zone. Hence, for the case of Fig. 10.11, the inverse distance method can
be assumed to be accurate for frequencies above 90 MHz. However, this
comparison still demonstrates that Class B limits tend to be roughly 10 dB
stricter than Class A radiated emission requirements.
Although each country is responsible for its own set of EMC regulations,
CISPR 22 regulations have been adopted by several countries and a compar-
ison with the FCC regulations in the United States may aid understanding
of the standardisation process. CISPR 22 regulations require that radiated
emissions measurements for Class A devices be at a distance of 30 m and
10 m for Class B devices. We scale the measurement limits to a common
distance using the inverse distance method and plot the CISPR 22 and
FCC regulations together for comparison. Although the regulations vary
slightly in different frequency ranges, no significant difference between the
FCC and the CISPR 22 regulations can be established for radiated emis-
sions as shown in Fig. 10.12 for Class A devices and in Fig. 10.13 for Class
B devices.
The differences in the FCC and CISPR-22 regulations are more appar-
ent for the conducted emissions limits with the difference in the regulated
frequency range being the most noticeable. The maximum frequency is
30 M Hz in both regulation but the CISPR-22 regulations extend down
to 150 kHz, while the FCC regulations only extend down to 450 kHz.
The CISPR-22 limit for class B devices rises steadily for frequencies below
288 10. Fundamentals of electromagnetic compatibility

65

60
60
56,5
54 54
electric field in dBµ V/m
55

50 49,5

46
45 43,5

40
40
216

35
88

30
30 100 1000
frequency in MHz

FIGURE 10.11. A comparison of the FCC limits for radiated emissions in Class
A and B digital devices (measurements at 3 m)

45
electric field in dBµV/m

40 40 FCC

37
CISPR-22
35 36,5
34

30
29,5 216
88 230
25
30 100 1000
frequency in MHz

FIGURE 10.12. A comparison of the FCC and CISPR-22 regulations for radiated
emissions in Class A devices (measurement at 30 m)
 10.2 The EMC guidelines 289

45
43,5

electric field in dBµV/m 40

37

35 35,5
33

30
29,5
230
88
216
25
30 100 1000
frequency in MHz

FIGURE 10.13. A comparison of the FCC and CISPR-22 regulations for radiated
emissions in Class B devices (measurement at 10 m)

500 kHz. The rationale behind this extension is the increasing importance
of switching power supplies over their linear analogues occasioned by their
higher efficiency and light weight. The CISPR-22 regulations for conducted
emissions are specified using a quasi-peak detector (QP) and an average
detector (AV). FCC conducted and CISPR-22 and FCC radiated emissions
limits require the use of a quasi-peak detector. Figs. 10.14 and 10.15 illus-
trate these differences for Class A and B devices.
Because of the great importance of EMC issues in military applications
to ensure that missions will not be compromised, the military has developed
military EMC regulations. In addition to regulating both conducted and
radiated emissions, the military regulates susceptibility which is very im-
portant in military applications since military equipment must be immune
to outside interference. The military EMC regulations are more stringent
than the FCC or CISPR regulations and cover a much wider frequency
range with several subdivisions within this range. In addition, the military
may require that EMC requirements be waived for certain applications if
this is deemed necessary to mission success. On the other hand, CISPR
and FCC regulations cannot be waived for commercial products.

10.2.2 Radiated emissions measurement

To ensure testing for radiated emissions is accurate, the FCC and CISPR
have developed testing standards that explain how the testing must be car-
ried out to verify compliance. This ensures testing that is both accurate and
290 10. Fundamentals of electromagnetic compatibility

80 79,5

75
73
CISPR-22 (QP)
voltage in dBµV

70 69,5
FCC

66
65

60
60 0,15 CISPR-22 (AV)

0,5 1,705
55
0,1 0,45 1 10 30
frequency in MHz

FIGURE 10.14. A comparison of the FCC and CISPR-22 regulations for con-
ducted emissions in Class A devices

70

66
65
voltage in dBµV

60 CISPR-22 (QP)

56
55

50 CISPR-22 (AV)
48 FCC
46
45
0,1 0,15 0,45 1 5 10 30
0,5
frequency in MHz

FIGURE 10.15. A comparison of the FCC and CISPR-22 regulations for con-
ducted emissions in Class B devices
 10.2 The EMC guidelines 291

repeatable. For radiated emissions, the FCC specifies that the measurement
of radiated and conducted emissions be performed on the complete system.
All cables interconnecting to peripheral equipment must be connected and
the system must be in a typical configuration and both the cable and the
system be configured in a way that maximises the emissions. For example,
if the unit under test has interior wire harnesses, these harnesses must be
configured in such a way that of all the possible ways the unit can be as-
sembled with those wire harnesses, this way causes the maximum radiated
emissions. This ensures that for a mass production unit, the worst-case
scenario is taken into consideration.
Testing standards set forth by the FCC for radiated emissions testing
are very specific and difficult to automate. Radiated emissions are to be
measured at 10 m for Class A and at 3 m for Class B devices. These
measurements are to be made over a ground plane using a tuned dipole
antenna at an open field test site. In addition, the tests are to be made with
the measurement antenna in both the vertical and horizontal polarisation
positions. During product development, most companies test their products
in a semi-anechoic chamber, a shielded room with radio frequency absorbing
cones on the walls and ceiling. This semi-anechoic chamber shown in Fig.
10.16 simulates an open field test site by minimising reflections from the
walls and eliminates any ambient signals that may be present in an open
field environment.
Companies sometimes simplify the FCC test procedure using a broad-
band antenna such as a log-periodic or discone antenna. Such antennas
are desirable since, unlike a tuned dipole, their length does not need to
be adjusted with each frequency change allowing companies to test their
products using a frequency sweep rather than measuring at each frequency
separately and adjusting the dipole lengths with each measurement. An
additional test requirement for radiated emissions testing is that the band-
width of the measurement receiver must be at least 100 kHz. A wide re-
ceiver bandwidth ensures that the test picks up emissions from broadband
sources such as the arcing at the brushes of a DC motor in addition to the
intended narrowband signals such as clock signals.
The FCC and CISPR test procedures require that the measurement re-
ceiver use a quasi-peak detector while typical spectrum analysers use peak
detectors. The quasi-peak detectors ensure that fast changing, momentary
signals such as randomly occurring spikes will not charge up the detector to
as high a level as periodic signals would. The FCC is concerned primarily
with the more significant and frequent emissions that would cause inter-
ference with radio and wire communications rather than with randomly
occurring one-time signals.
292 10. Fundamentals of electromagnetic compatibility

shielded room

DUT

(3 or 10 m)

scan height
1 - 4 m for
both vertical
& horizontal
polarisations

ground plane
spectrum analyser
or receiver

FIGURE 10.16. A semi-anechoic chamber for emissions measurements

 10.2 The EMC guidelines 293

10.2.3 Measurement requirements for conducted emissions

The intent of conducted emissions limits is to prevent noise currents enter-
ing the common power net of the installation through the AC power cord of
the device. The common power net can be regarded as a large antenna and
any noise currents placed onto the power net will consequently radiate very
efficiently. An example is the interference occurring on a television or radio
set when a blender is switched on. Arcing at the commutator of the DC
motor in the blender causes noise currents that conduct through its power
cord into the common power net of the house. The wiring in the house acts
as an antenna and radiates this noise, which is then picked up as interfer-
ence by the television and radio sets. Hence, conducted emissions limits are
concerned with these noise currents. The FCC and CISPR-22 conducted
emission limits specify voltages because the LISN used to measure con-
ducted emissions converts the noise currents to a voltage. To understand
the function of the LISN, understanding the standard AC power distribu-
tion system is essential. In the United States, AC voltage is distributed to
residential and business premises at 60 Hz and 120 V, rms while in Kenya,
this is at 50 Hz, 240 V, rms. The power distribution in the home is through
a three-wire cable consisting of a phase wire (brown), a neutral wire (blue),
and the green or green/yellow-stripped wire. Both the phase and neutral
wires carry the 50 Hz power and the potential between each wire and
ground is hence 240 V . The currents that must be measured for conducted
emissions tests are the currents occurring on the phase and neutral wires.
Fig. 10.17 is an illustration of the LISN used for the FCC conducted
emissions tests. A similar LISN is used for CISPR-22 conducted emissions
testing, but component values are different due to the different frequency
range defined by CISPR for conducted emissions testing. The two functions
of the LISN are

• To isolate external noise from the common AC net from contaminat-

ing the measurement.

• To present a constant impedance in frequency from site to site to the

product between phase and ground and between neutral and ground.

In operation, one of the 50 Ω resistors represents the input impedance

of the spectrum analyzer while the other 50 Ω resistor is a dummy load.
The 100 µF capacitors C1 block DC so as not to overload the test receiver.
Resistors R1 are rated 1 kW and provide a DC path for C1 to discharge
if the 50 Ω resistors become disconnected. The device under test should
operate normally at 60 Hz power frequencies, hence the capacitors are
dimensioned to make them appear as open circuits at the power frequency
while inductors appear as short circuits. The equivalent circuit of the LISN
at 60 Hz hence has the form shown in Fig. 10.18. In the frequency range of
conducted emissions (450 kHz-30 M Hz) however, the capacitors appear as
294 10. Fundamentals of electromagnetic compatibility

LISN

IP L1
C1 C1
IP L1
L IN IN
product N to AC
under net
G
test C2
50 Ω 50 Ω C2
R1 R1
VP VN

green wire

FIGURE 10.17. The FCC LISN

L
product N to AC
under net
G
test

green wire

FIGURE 10.18. Equivalent circuit of an LISN at power frequencies

short circuits while the inductors appear as open circuits and the equivalent
circuit is as shown in Fig. 10.19. The currents on the neutral and phase
lines can therefore be isolated and measured at the 50 Ω resistors. The
currents on the phase and neutral lines have no path onto the AC power
net.

10.2.4 Additional product requirements

Recall that the FCC and CISPR-22 regulations are set by law to regulate
digital devices. Individual companies, however, impose their own set of
regulations on their products, which are often much more stringent than the
required regulations. The automobile industry is for example, exempt from
FCC requirements, yet have self-imposed regulations that far exceed those
set by the FCC for normal digital devices. This is because companies face
a far greater financial loss as a result of a faulty or poorly designed product
than they would by investing to make sure their product is safe and well
designed. People put their lives in the hands of automotive manufacturers
 10.2 The EMC guidelines 295

LISN

L
product N to AC
under net
G
test C2
50 Ω 50 Ω
R1 R1
VP VN

green wire

FIGURE 10.19. The high frequency equivalent circuit of the LISN

every time they drive a vehicle, hence automobile manufacturers cannot

afford to have lax standards.
Apart from imposing stricter versions of government regulations on them-
selves, many companies also impose design constraints on their products
that ensure enhanced radiated and conducted immunity as well as immu-
nity against electrostatic discharge (ESD). The FCC does not regulate these
areas because they do not pose a threat to radio or wire communications.
Hence, individual manufacturers are left to create their own standards.
Each of these categories is associated with the ability of the device to func-
tion in the presence of external interference. Therefore these are of utmost
importance for manufacturers to guard against.
Radiated immunity is the ability of the product to operate in the presence
of high power transmitters such as AM and FM transmitters and airport
surveillance radar systems. Manufacturers therefore test their products by
illuminating them with typical waveforms and signal strengths that sim-
ulate worst-case exposure the product could encounter in its operational
environment. Conducted immunity, on the other hand, is the ability of a
product to operate despite a variety of interferences that enter the device
through the AC power cord. An example of such interference is the power
surge caused by a lightning strike. Manufacturers must design tests that
simulate the effect of lightning-induced transients and design the product
to resist such interference. Electrostatic discharge describes the discharge of
the static charge build-up on the human body or furniture through a prod-
uct when the person or furniture come in contact with the product. Such
static voltages can rise to as high as 25 kV . When the discharge through
a product occurs, large currents momentarily course through the product
and may cause machines to reset, IC memories to clear, etc. Manufacturers
test their products by subjecting them to controlled ESD events and design
their product to operate successfully in the event of such ESD occurrences.
296 10. Fundamentals of electromagnetic compatibility
11
Standards and standardisation

11.1 Introduction
A standard is a document established by consensus and approved by a recog-
nised body which provides for common and repeated use, rules, guidelines or
characteristics for activities or their results aimed at achieving an optimum
degree of order in a given context.
Standards should be based on consolidated results of science, technology
and experience with the aim of promoting optimum community benefits.
While voluntary documentary standards are crucial in facilitating and reg-
ulating industry and commerce, this role is often grossly overlooked. Stan-
dards do not dictate how products are designed or manufactured but may
be used to influence safety requirements through incorporation into regu-
lations. It is important to remember that standards derive their legitimacy
from the voluntary and consensual nature of their development process
however.
Most standards provide utility, i.e. when one uses a standard, it is often
because it provides a reliable though not necessarily the only or the best
solution to a given problem. In research, standards are one of the most
important tools for taking new technology to market and in transform-
ing research findings into guidance documents thereby providing a bridge
between research and industry. This is a crucial link for successful com-
mercialisation of research outcomes. By providing this bridge, standards
are invaluable as a tool for promoting innovation and commercialisation
through
298 11. Standards and standardisation

• Efficient dissemination of new ideas and good practice

• Validation of new measurement tools and methods

• Implementation of new processes and procedures

As mentioned, standards development require a consensual approach.

But what is consensus? Roughly, a consensus is a general agreement that
is characterised by

• The absence of sustained opposition to substantial issues by any im-

portant part of concerned interests

• A process that seeks to take into account views from all parties con-
cerned and reconciliation of any conflicting arguments

Note however, that consensus does not imply unanimity.

Standards have as their main functions

• Interoperability and or compatibility leading to efficient production

• Ensuring safety, quality and reliability of products, processes and

services, i.e. fitness for purpose or safety

• Variety reduction and optimisation based on best practices thereby

leading to mass production and lowered unit price through competi-
tion

• Information and measurement i.e., test and measurement methods

for describing, quantifying and evaluating product attributes such as
materials, processes and functions thereby playing a supporting role
to regulation

Typically, standards contain normative and informative elements. Nor-

mative elements are parts of the standard that shall be complied with in
order to demonstrate compliance with the standard and are characterised
by the use of the legalistic imperative "shall". Informative elements on the
other hand, provide clarification or additional information and are usually
to be found in notes or appendices. Wording in the informative parts of the
standard may not contain requirements, hence "shall" is not used.
We have seen that standardisation can support dissemination and im-
plementation of suitable results from research and development work by
making them readily accessible to potential users in a consistent format as
well as aiding identification and addressing any outstanding issues.
 11.2 Status of standards 299

11.1.1 Types of standards

(1) Formal standards developed by

• — National standards bodies (NSB) such as AFNOR (France), BSI

(Britain), DIN (Germany), JISC (Japan), KEBS (Kenya), etc.
— Regional standards bodies such as CEN, CENELEC, ESTI (Eu-
rope), PASC (Pacific Area Standards Congress), COPANT (Pan
American Standards Commission), ARSO (African Organisa-
tion for Standardisation), AIDMO (Arabic Industrial Develop-
ment and Mining Organisation), etc.
— International standards bodies such as ISO, IEC, ITU, etc.

(2) Informal standards published by standards development organisations

(SDO) such as ASTM (American Society of Testing Materials) ,
IEEE, SAE (Society of Automotive Engineers), SEMI (Semiconduc-
tor Equipment and Materials International), VDI (Verein Deutscher
Ingeneure), TAPPI (Technical Association of Pulp and Paper Indus-
try), etc.

(3) Private standards developed inhouse by a company or trade organisa-

tion

11.2 Status of standards

Standards are voluntary documents and no compulsion exists on anyone to
use them. However, when agreed to in a contract or referred to in a regu-
lation, there might be a penalty either civil or criminal for not complying
with a standard. Even if standards are voluntary, courts may decide that in
the absence of relevant regulation, noncompliance with a standard to which
a reasonable man might be expected to conform could be sufficient ground
to determine liability. While compliance with the European standards (EN)
is voluntary for example, the "New Approach" directives of the European
Union use compliance with specific standards as a means of demonstrating
compliance with the relevant regulation. Hence, CEN members are required
to adopt ENs at publication and withdraw any conflicting national stan-
dards. Note that a national standard is one adopted by a national body and
made available to the public. Because of their voluntary nature, standards
are different from regulations which have a legal requirement to comply. A
regulation is a document providing binding legislative rules that is adopted
by an authority.
Standards are open-access documents unlike patents and attract no charge
or license fee for their use apart from the cost of purchase. Expert partic-
ipation in standards development is free and experts waive their rights to
300 11. Standards and standardisation

protection of any intellectual property they contribute. It is unusual that

a standard will use patented material and alternative unpatented means of
complying with a given standard must be provided whenever possible. If for
technical reasons the use of patented material becomes necessary, an under-
taking from the patent holder must be sought indicating willingness to ne-
gotiate world-wide licenses with applicants on reasonable non-disciplinary
terms and conditions. Development of a standard that necessitates the use
of patented information ceases if such an undertaking is not forthcoming.

11.3 How standards are developed

11.3.1 Pre-normative and co-normative research
The technical basis of a new standard is usually established through a
programme of research called the pre-normative research (PNR) which is
research undertaken prior to standardisation or normalisation. The PNR is
used to demonstrate feasibility and reliability of the technique or process
to be standardised and investigate its limitations.
After developing a technique or process and establishing its boundaries,
a pre-standard such as a publicly available standard (PAS) or technical
specification (TS) is usually prepared. This provides a document in a rel-
atively short time frame for evaluation by potential users. Availability of
a pre-standard provides a basis for further research usually referred to as
co-normative research undertaken in conjunction with the standardisation
process to establish a statistical basis for the technique or process. Of par-
ticular emphasis is its reproducibility (same user), repeatability (different
users) and inherent uncertainties.

11.3.2 Standardisation
This is the activity of establishing a standard with regard to the actual po-
tential problems, provisions for common and repeated use aimed at achiev-
ing the optimum degree of order in a given context. In particular, standard-
isation consists of the process of formulating, issuing and implementing the
standard. Important benefits from standardisation are

• Improvement of suitability of products, services and processes for

their intended purpose

• Prevention of barriers to trade

• Facilitation of technological cooperation

 11.3 How standards are developed 301

11.3.3 Formal standards

Formal standards are developed and approved by members of the standards
body. In some jurisdictions, development may be undertaken by a standards
development organisation (SDO) based on consensus. Development of for-
mal standards typically follows the processes and procedures laid out in
the directive Part 1 and Part 2 of the ISO/IEC. Formal standards are de-
veloped by independent experts, nominated by members of the standards
organisation and voting on a voluntary basis within technical committees
(TCs) dealing with standards in a particular area of technology or business.
Approval of standards documents is the responsibility of the membership
of the standards organisation. Depending on the status of the document,
approval will either be given by the membership of the TC in the case of
a PAS, TS or TR (technical report) or by the full membership in the case
of an international standard (IS) or European standard (ES). The process
of developing international standards deliverables from a new work item
proposal (NWIP) to publication of the IS is illustrated in Fig. 11.1. The
process takes 18-30 months to produce a draft IS and 36-48 months to
publish an IS.
Standardisation can support dissemination and implementation of suit-
able outputs from R&D projects through making them more accessible to
stakeholders and other interested parties in an open and consistent format.
For new emerging areas of technology, preparation of preliminary standards
documents such as workshop agreements, PAS or TS provides users with a
structured procedure that they can evaluate and modify in light of further
experience. Although such documents are subject to further development,
users can have a high level of confidence that they represent existing good
practice. When research results are subject to the formal standardisation
process, identifying and addressing outstanding issues not previously appar-
ent is facilitated. However not all outputs from R&D projects are suitable
for the preparation of standards. Even where this is the case, the output
may still provide vital support to new or existing standards through vali-
dation of the test or method, say.

11.3.4 Costs of standardisation

Participation on standardisation activities is almost universally on a volun-
tary basis. The cost of participation in TCs varies from country to country
with some national standards bodies making a charge to cover administra-
tion costs while some do not. In some countries, support for participating
in meetings especially where overseas travel is involved is provided. Most
works on standards development can be undertaken through E-mail or
teleconferencing but it is usual for most projects to have at least one phys-
ical meeting during a standard’s development process. For an IS, this may
mean travelling halfway round the world for a one- or two-hour meeting!
302 11. Standards and standardisation

Working draft
NWIP received NWIP approved submitted by
from member and registered in proposer, further
body database developed by
experts
3 month ballot by
members of relevant
TC.
Approval if at least 5 3 month review Consensus
members agree to and ballot by among
participate and 50% members of TC experts
of those voting support - may take 2 to
proposal 4 months by
agreement with TC

Final committee First committee Final draft

draft draft

Resolution
of comments
PAS & TR published received
if at least simple majority
of voting members approve
document. TS published if
at least 66% of voting members
approve document. Same
requirement for formal route

Formal route draft Final draft IS

IS (enquiry stage) (approval stage)

Resolution of comments received

5 month review and ballot by all
ISO members. Approval if 66% of Consensus among all ISO
P members of responsible committee members. 2 month review
in favour and no more than 25% of ballot by all members. No
total vote against (abstentions & other comments allowed except
negative votes unaccompanied by editorial and for negative votes
technical reasons for vote not counted)

Published as IS

FIGURE 11.1. International standards development process - from NWIP to IS

publication
 11.3 How standards are developed 303

Most organisations such as SMEs, NGOs or consumer groups may find this
untenable.
While physical meetings might provide direct and immediate input to a
project, participation via correspondence and other electronic means can be
more effective since it can be accomplished at a time and place convenient
to the individual.

11.3.5 Decision tree for standards development

Fig. ?? is an illustration of the decision tree for the development of a
standard.

11.3.6 Practical details of standards development

The membership of an ISO TC consists of

• Participating or P members who are the national members of the ISO

wishing to participate in the work of the TC
• Observer or O members who are national members of ISO wishing to
observe the work of the TC without actively participating in it
• Liaison or L members who are members of another TC or external
organisation with an interest in the work

P members can nominate experts to projects and have a responsibility

to vote. Typically, P members have a national mirror committee that

• Makes proposals for new work to be undertaken by the TC

• Identifies and nominates experts to work on projects
• Comments on the documents under ballot
• Recommends the national responses to ballots

While O members can submit NWIP they have no right or responsibility

to vote on TC matters. L members can also submit a NWIP and nominate
experts to work on the projects but have no right to vote on TC matters.
The O members of a TC like all the other members of ISO have an oppor-
tunity to vote on and submit comments on documents at the draft inter-
national standard (DIS) and the final draft international standard (FDIS)
stages in the development of an IS.
NWIPs are typically submitted by P members but this is not always the
case. NWIP submission is usually on the basis of existing national work e.g.
a national standard that a member believes would benefit the international
community. The country or organisation submitting a NWIP is encouraged
304 11. Standards and standardisation

Is the output of a
new and repeatable
technique for
- identification
- characterisation
- manipulation
- preparation No
Ignore output
- verification, etc.
Or
a modification of an
existing technique or
procedure to allow its
use under previously
untested conditions?

Yes

Does the output depend on Yes Ignore output but

the use of specific patents or see standards and
specific pieces of equipment? patent notes

No

Has the output been evaluated Undertake

No
by different investigators using evaluation
different samples, materials, etc?

Yes

Has the output been evaluated No Undertake

to determine its repeatability or evaluation
reproducibility?

Yes

FIGURE 11.2. The decision tree for development of a standard

 11.3 How standards are developed 305

Yes

Have relevant databases

been checked to ensure that No
there are no pre-existing Check databases
standards covering the same
thing?

Yes

Are those involved in

the development prepared No
Ignore output
to work on developing
a standard?

Yes

Identify relevant TC in CEN or ISO,

prepare a draft NWIP for discussion
with appropriate national mirror
committee

FIGURE 11.3. The decision tree for development of a standard (continued)

306 11. Standards and standardisation

to provide a project leader (PL) who will work with experts nominated by
other members to develop the document to the final working draft (FWD)
stage. The country proposing the project is also expected to provide a draft
or outline of the document.
In ISO, approval of a NWIP requires that at least five (5) P members
agree to participate in the work by nominating experts together with at
least a simple majority of those P members voting being in favour of the
development. Abstentions in all ballots are not counted. The approved work
items (AWIs) are then developed by nominated experts acting in their
individual capacity within a project group (PG) under the leadership of
the PL. Project groups covering the same general area of work are usually
grouped together under working groups (WG).
Where a large volume of work is undertaken in an area that does not im-
pact on other parts of the TC’s work and can be managed more effectively
as a separate group, a subcommittee (SC) can be formed to handle the
work. Subcommittees act simultaneously and are essentially independent
of their parent TC. Both the TC and its SC can establish a WG, approve a
NWIP , develop documents to the committee draft (CD) stage etc. Work is
undertaken electronically in general but most PGs have a physical meeting
during the once or twice yearly TC planning meetings.
Once experts reach consensus on the contents of a document, their final
working draft (FWD) is submitted to the membership of the TC or SC
for approval as a CD. This is achieved through a three-month ballot of P
members with recent changes to the ISO/IEC directives permitting two-,
three- or four-month balloting periods depending on the circumstances. In
this ballot, P members are required to provide a national position through
their mirror TCs, i.e. approve, reject or abstain. Where appropriate, this
may be accompanied by comments on the document contents. Comments
on documents under ballot can be of a general, editorial or technical na-
ture and are submitted on a standard comments template. Comments must
be accompanied by a recommendation on how each comment is to be ad-
dressed.
Committee drafts are approved if they garner support of at least two-
thirds of those voting with abstentions not counted. Depending on the type
of document originally proposed, the CD can be published as a PAS, TR
or TS. Only a simple majority of voting P members is required to approve
a PAS or TR. A formal route toward full international standard is via
preparation of a DIS. Whichever route is chosen, it is necessary that any
comments received are satisfactorily resolved before publication or progress
to the DIS stage. This is normally done at a comments resolution meeting
under the leadership of the PL. Once comments have been addressed to
the satisfaction of all commenting members, the document is either sent
for publication or submitted for its DIS ballot. The DIS ballot runs over a
five-month period during which each ISO member has an opportunity to
review and comment on the document. Approval of the document requires
 11.3 How standards are developed 307

at least two-thirds of P members of the TC responsible for the document

to be in favour and no more than a quarter of the total votes against.
Any abstentions or negative votes unaccompanied by technical reasons for
rejection are not counted.
Comments received during the ballot must be resolved before the docu-
ment is ready for its formal final draft international standard (FDIS) vote.
The FDIS stage may be omitted if the document is approved and no tech-
nical comments have been submitted. The FDIS ballot runs for two months
during which time all members of ISO have an opportunity to

• Approve the document with editorial comments where necessary

• Abstain
• Reject the document with technical reasons supplied

Requirements for approval of an FDIS are the same as those for a DIS. In
CEN, all members are automatically members of all TCs. However, to take
part in the work of a given TC, a member must appoint a representative
to the TC in order to access relevant committee documents. Unlike in ISO
which operates on the principle of one member, one vote, CEN operates
a weighted voting system giving larger economies a higher percentage of
the votes. Weighted voting applies to NWIP ballots for TS and EN and to
approve ballots of both types of publications.
Inventive step refers to something that is not obvious to those skilled in
the art, i.e. something that another expert would not do as a matter of
course as a consequence of their training.
All standards have the following, specific to each document

(a) A unique number that includes the source of the document and/or its
status as well as year of approval. For example, ISO 9001:2008, EN
13485:2003, EN 1071-3:2001, ISO TS 80004-1:2010, CWA 14243:2003
(b) Title consisting of up to three elements

(i) Introductory element giving the general field to which the docu-
ment belongs. Often, the title of the committee responsible for
the document.
(ii) Main element indicating the principal subject of the document
(iii) Complementary element giving specific aspect of the main ele-
ment addressed in the document

(c) Scope defining the subject of the document and aspects covered. This
should indicate any limits to applicability of the document or partic-
ular parts of it and should reflect the title while the title reflects the
scope. It is important to agree on the scope as early as possible for
all experts to be fully aware of what they are trying to develop.
308 11. Standards and standardisation

(d) Normative references, i.e. a list of other standards that form or parts
of which form an integral part of the document and without reference
to which, it is impossible to comply with the standard established by
the document.

(e) Terms and definitions - a list may be provided or reference made to

other documents where these terms are defined. If other standards
are referred to where the relevant definitions are to be found, these
should be listed as normative references.

(f ) Symbols and abbreviations

(g) Apparatus

(h) Procedures written in the imperative form and not the narrative

(i) Data analysis

(j) Reporting and results

(k) Annexes

Although TSs require approval through the equivalent of a CD ballot in

which all members of CEN vote, there is no CD ballot stage for documents
expected to be published as full European standards (ENs). The equivalent
to a DIS ballot in CEN is the enquiry ballot and the equivalent to the FDIS
ballot is the formal vote.

11.3.7 Consensus
There are minimum ballot thresholds for a document to be published as a
TS, TR, PAS or IS. In CEN, the minimum requirement for EN or TS is 71%
of the weighted votes of members voting. A TR requires a simple majority of
those members voting. CEN does not have an equivalent to PAS. However,
the overriding objective should always be to reach a consensus amongst all
involved. Remember that by consensus, we mean that there is no sustained
opposition to substantial issues by any important part of the concerned
interests but not necessarily unanimity.
The principle of consensus is vital to standardisation as it imparts legiti-
macy to the process and the documents produced. People and organisations
use standards not through coercion but because they are in general agree-
ment with their content. Even without requiring unanimity, consensus does
require those involved to respond from a position of knowledge not igno-
rance. Members therefore have the responsibility to ensure that they only
support or reject documents when they have sufficient national expertise
to make informed judgement. Members should always abstain when the
necessary expertise does not exist. Note that if a member cannot provide
 11.3 How standards are developed 309

any comments on a document, most likely it is because they do not have

sufficient expertise to review and vote on it. Even where a member has
nominated experts to a project, they still have the responsibility to give
other national stakeholders the opportunity to review and comment on the
document produced and are at liberty to reject the document if the national
view is that it is inaccurate, incomplete or unnecessary.
310 11. Standards and standardisation
12
Liquid- and solid-state insulators

12.1 Liquid insulation

Liquid insulators are principally organic oils and often accompany solid
insulation in the same installation primarily as coolants or impregnation.
Desirable properties a liquid insulator should have include

• High electric and impulse strength

• Low loss factor (loss tangent)

• High specific heat capacity and thermal conductivity

• Good chemical stability and low gas absorbing properties

• Low viscosity, low density, low volatility, low solvent power but high
flash point

• Low flammability and toxicity

Organic oils are mainly used in transformers, switch-gear, bushing, ca-

bles, capacitors and electronic equipment. In transformers, the liquid in-
sulation is changed from time to time but an oil change is not possible in
sealed devices. Examples of liquid-phase insulation include
P 
  
Naturally occurring mixtures that mainly consist of hydrocarbons in the
liquid state. Petroleum mineral oils can be categorised under
312 12. Liquid- and solid-state insulators

• Paraffins
• Asphatics
• Mined

After extraction from the ground, oils are refined to remove metals and
other impurities. In the refined form, oils can have dielectric strengths be-
tween 400 and 500 kV cm−1 which could go as high as 1000 kV cm−1 with
further refining. The dielectric strength will however fall during use due
to a combination of factors which include aging and contamination. Most
insulating oils have an ǫr ∼ 2, 2 and power factors below 0, 0005 at 50 Hz
and 20 ◦ C.
F
   
Fluorinated liquids are characterised by high chemical stability allowing
application at temperatures up to 200 ◦ C. They have low loss tangents
of < 0, 0005, high resistivity 1014 -1018 Ωcm and ǫr ∼ 1, 8. Their cooling
properties are better than those of mineral oils and silicones. These liquids
have therefore found applications in some electronic equipment such as
radio transformers in electronic systems. However, fluorinated liquids suffer
degradation through moisture absorption and are therefore to be found
primarily in sealed equipment.
S
  

Synthetic hydrocarbons have no major advantages over the natural min-
eral oils apart from the relative ease of guaranteeing their chemical compo-
sition. They have found use mainly in smoothing capacitors for DC power
and in high pressure gas-filled cables.
O
 
Esters are useful mainly at high frequencies owing to their relatively low
dielectric losses and high dielectric constants, properties that are desirable
in many high frequency capacitors. Chemically, esters are manufactured
by reacting acids and alcohols but some occur naturally in plants, e.g.
castor oil. Since esters have a high moisture absorption capacity, they are
to be found mainly in sealed devices. Esters have dielectric constants in the
range 2-3, 5 at low frequencies and power factors at 50 Hz of ∼ 0, 05 which
decreases to about 0, 001 at 1 MHz.
V  
Vegetable oils may be classified under

• Drying oils (thickens or hardens on exposure to air) from linseed

• Semi-drying oils from corn, cotton seed, soybean, sesame, sunflower
• Non-drying oils from castor, coconut, palm, olive, peanut

This classification depends on the ease with which the oils form films
resulting from the ease with which they can be oxidised and polymerised
 12.1 Liquid insulation 313

when exposed to air. From a dielectric point of view, drying and non-
drying oils are important with linseed oil finding use in the formulation
of insulation varnishes and castor oil as a plasticiser in insulating resin
compositions employed in structural insulation materials and in coating
compositions in electrical equipment.
A! (
 
)
These are chlorinated aromatic (benzene-based) hydrocarbons. They are
fire-resistant and have been used in transformers and capacitors that re-
quire high dielectric constants. The breakdown voltage is fairly high and
lies between 80 and 180 kV cm−1 and power factors in the range 0, 02-0, 05.
Askarels have resistivities in excess of 10 GΩcm and dielectric constants in
the range 4, 8-5, 3. However, askarels are now recognised as accumulative
persistent ecological contaminants and their use is now largely discontin-
ued.
S  
 
Silicone fluids are polyorganic silicones with a unique combination of
properties. They are clear liquids with an oily consistency and have ǫr ∼
2, 2-2, 8. Their viscosity can vary over a wide range but silicon fluids are
highly stable at high temperatures. Polyorganic silicones have electrical
properties that are largely independent of frequency and temperature mak-
ing them excellent coolants in pulsed radar systems, aircraft and radio
transformers. However, silicones are detrimentally affected by arcing mak-
ing them unsuitable for use in switch-gear.
When choosing a liquid dielectric, several factors such as environmental
factors and space-saving properties must be considered beside their physical
properties. Cost and chemical stability are important factors as well. Hence,
mineral oils remain the dominant liquid insulators in active application to
date.

12.1.1 Breakdown in liquid insulators

Breakdown in liquid insulators is not a well explained phenomenon as yet.
However, some fairly successful theories have been advanced, namely
• Colloidal theory
• Bubble theory
• Breakdown due to liquid globules
• Electronic theory
C  T
When liquids are contaminated with contaminants such as cellulose fi-
bres, these tend to accumulate in regions of maximum stress and may thus
form bridges between the electrodes. This bridges then tend to short-circuit
the electrodes leading to eventual breakdown.
314 12. Liquid- and solid-state insulators

B T
Mechanically pure liquids contain no moisture or other impurities. They
may however contain dissolved gases which tend to collect around the elec-
trodes in the form of bubbles. The gas bubbles may then become ionised
and elongated from polarisation, leading to breakdown.
L   G
 B!&

Similar to bubble theory, liquid globule theory involves a liquid such as

water collecting in the form of a globule around the electrode. The globule
elongates on polarisation leading to breakdown.
E
 T
When the liquids have no dissolved gases, moisture or other impurities,
high electric stresses will lead to ionisation of the molecules. This may lead
to avalanche breakdown as energetic ions lead to further ionisation of the
neutral molecules.

12.1.2 Henry’s law

Henry’s law is a gas law that states: At constant temperature, the concen-
tration of a gas dissolved in a given liquid is directly proportional to the
partial pressure of the gas in equilibrium with the liquid. Mathematically,
a form of Henry’s law may be expressed as

exp (P ) = exp (kH c)

where P is the partial pressure of the solute gas above the solution in atm
of absolute pressure, c is the solute concentration in solution in mol L−1
(molarity) and kH is Henry’s law constant in L atm mol−1 , atm per mole
fraction or P a m3 mol−1 . Thus,

P = kH c

Table 12.1 presents kH values for common solute gases in water at 298 K.
Since there are various formulations of Henry’s law, kH has different di-
mensions in these formulations and some of these are presented in Table
12.1 as well. Note that kH,cp and kH,pc are reciprocals of each other.
where, caq is the concentration of solute in mol L−1 , Lsol is volume of
solution in L, Pgas is the gas partial pressure above solution in atm of
absolute pressure and atm is atmospheres of absolute pressure.
Henry’s law is a limit law and applies only to dilute solutions. The range
of applicable concentrations becomes increasingly narrow with increasing
divergence of the system from ideal behaviour, i.e. roughly as the solute
and solvent become more chemically different. Further, Henry’s law applies
only when the solute and solvent do not chemically react. Therefore, since
CO2 rapidly forms hydrated CO2 and finally carbonic acid (H2 CO3 ) in
water, Henry’s law does not apply in the strict sense.
 12.1 Liquid insulation 315

TABLE 12.1. Some forms of Henry’s law and constants for some common gases
in water at 298 K
P c Pgas caq
equation kH,pc = cgas kH,cp = Pgas
aq
kH,pn = naq kH,cc = cgas
 aq    
Lsol ·atm molgas atm·molsol
dimension molgas Lsol ·atm molgas dimensionless
−3 4
O2 769, 23 1, 3 × 10 4, 259 × 10 3, 180 × 10−2
CO2 29, 41 7, 8 × 10−4 7, 099 × 104 1, 907 × 10−2
H2 1282, 05 3, 4 × 10−2 0, 163 × 104 0, 8317
N2 1639, 34 6, 1 × 10−4 9, 077 × 104 1, 492 × 10−2
He 2702, 7 3, 7 × 10−4 14, 97 × 104 9, 051 × 10−3
Ne 2222, 22 4, 5 × 10−4 12, 30 × 104 1, 101 × 10−2
Ar 714, 28 1, 4 × 10−3 3, 955 × 104 3, 425 × 10−2
CO 1052, 63 9, 5 × 10−4 5, 828 × 104 2, 324 × 10−2

TABLE 12.2. C values in the temperature dependence of Henry’s law coefficient

equation for different gases
Gas O2 CO2 H2 N2 He Ne Ar CO
C (K) 1700 2400 500 1300 230 490 1300 1300

If the temperature of the system is changed, Henry’s law constant changes

as well. For this reason, this constant is often referred to as Henry’s law
coefficient. A number of formulations describe this temperature dependence
of Henry’s law constant. An example is van’t Hoff equation of the form

1 1
kH,cp (T ) = kH,cp (T0 ) exp −C −
T T0
where T is the absolute temperature of the system, T0 = 298 K is the refer-
ence temperature, kH,cp (T0 ) is the solubility form of Henry’s law constant
at the reference temperature and C is a constant with units of temperature.
Table 12.2 presents values of C for some common gases.
Since gas solubility generally decreases with increasing temperature, the
partial pressure at a given gas concentration in a liquid must increase with
temperature. For example, on heating water saturated in N2 from 25◦ C to
95◦ C, the solubility of nitrogen gas decreases to ∼ 43% of its initial value.
On the other hand, the solubility of CO2 in seawater doubles every 16 K
in temperature. The constant C may be expressed as
∆sol H d
C= =− ln [kH,cp ]
R d (1/T )
where ∆sol H is the enthalpy of the solution and R = 8, 314 kJmol−1 is the
universal gas constant.
In geophysics, a version of Henry’s law that applies to the solubility of a
noble gas in a silicate melt has the form,
ρmelt   
= exp −β µmelt,E − µgas,E
ρgas
316 12. Liquid- and solid-state insulators

where ρmelt and ρgas are the densities of the solute gas in the melt and gas
phases resp., µmelt,E and µgas,E are the excess chemical potentials of the
solute gas in the melt and gas phases and
1
β=
kT
where k = 1, 38 × 10−23 JK −1 is Boltzmann’s constant.
Raoult’s law is similar to Henry’s law and states that the partial pressure
of a component in a solution is proportional to the concentration of that
component in the solution. Using the molar fraction n = N/NA , where
N is the number of molecules of the component in solution and NA =
6, 02 × 1023 mol−1 is Avogadro’s number, to express the concentration of
the solute in solution, we can express Henry’s law in the form

P = kH n

Raoult’s law is of the form

P = P ∗n
where P ∗ is the vapour pressure of the pure substance. Hence, Raoult’s law
can be considered a special case of Henry’s law where kH = P ∗ .
This holds for pairs of closely related substances such as benzene and
toluene which obey Raoult’s law over the entire range of compositions.
Such mixtures are called ideal mixtures. Both Raoult’s and Henry’s laws
are limit laws in general and apply at the opposite ends of the composition
range.
The vapour pressure of the component with the highest molar concen-
tration such as the solvent in a dilute solution is proportional to its mole
fraction in the system. The constant of proportionality is the vapour pres-
sure of the pure substance (Raoult’s law). In a similar way, the vapour
pressure of the component with the lowest molar concentration e.g. the
solute in a dilute solution is proportional to its mole fraction in the solu-
tion with Henry’s law constant as the constant of proportionality (Henry’s
law). This must be determined experimentally. Thus,
P
lim = P ∗ , Raoult’s law
n
n→1

and
P
lim = kH , Henry’s law
n
n→0

12.2 Solid-state insulating materials

A wide variety of solids with insulating properties can be understood under
 12.2 Solid-state insulating materials 317

TABLE 12.3. The seven classes of solid-state insulators

class limiting temperature ◦ C material in class
Y 90 cotton, silk, paper, rubber, PVC
A 105 impregnated paper, silk, cotton, polyimide resin
E 120 enamel on polyvinyl base, polymethane, epoxy
B 130 mica, fibre glass, asbestos
F 155 polyester epoxide varnish, heat resistant varnish
H 180 mica, fibre glass, asbestos
C 255 mica, ceramics, glass, Teflon

• Organic - paper, cotton, rubber

• Inorganic - mica, glass, silicon dioxide, silicon nitride, plastic

Solid-state insulators may also be broadly classified under

• Paper and pressboards

• Fibrous materials which may be oil impregnated or not

• Resins or polymers

• Filling or bonding materials

• Natural or synthetic materials

• Composite materials

Materials will usually experience electric, thermal and mechanical stress

and their ability to withstand electrical stress may depend on how well they
can withstand other types of stress. The limiting temperatures at which
these materials can be used as insulators must therefore be established.
Solid-state insulators are therefore classified in terms of the temperature
range over which they can be used, i.e. thermal stability, into seven cate-
gories as shown in Table 12.3.

12.2.1 Organic solid-state insulators

P( 
 P
These insulators are made from cellulose fibre of a high degree of pu-
rity with wood fibre often constituting the main ingredient. Some of the
important parameters of these insulators include

• Thickness

• Apparent density, which affects dielectric strength, loss and perme-

ability
318 12. Liquid- and solid-state insulators

• Finish
• Porosity
• Tensile strength
• Tearing resistance

Paper insulators are prone to water absorption and are therefore usually
impregnated with mineral or vegetable oil. This treatment lends these insu-
lators an ǫr of ∼ 3, power factor in the range 0, 0017-0, 0025 at densities of
0, 75 - 1, 7 gcm−3 and a conductivity of 10−17 Scm−1 at 25 ◦ C. The impulse
breakdown voltage in paper insulators is approximately 3000 kV cm−1 . The
usual application areas of these insulators are hence

• Winding and cable conductor insulation

• Primary dielectric in paper capacitors and transformers

F  M 
Fibrous materials combine strength and durability and therefore are to
be found in many applications. Fibrous materials include cotton, silk and
synthetic fibres such as jute, viscous rayons, nylon and fibre glass. They are
flexible, cheap, possess high mechanical strengths and are easy to process.
They however have low dielectric strengths and are highly hygroscopic.
However, these materials can have improved electrical properties with an
oil impregnation.
The dielectric constants of the fibrous materials range from 3-8 in a
perfectly dry condition but many of their electrical properties are frequency
dependent. For example, ǫr decreases with increasing frequency and shows
a fair amount of anisotropy. Fibrous insulators are affected by moisture and
their resistivity will decrease with increasing moisture content.
These insulators find such applications as conductor insulation, winding
and coil insulation, backing for mica etc. Inorganic fibres such as asbestos
and fibre glass withstand high temperatures but have low elasticity and
machinability.
I(
, C
, F 
 
 B

 M 
These materials can be classified as waxes, insulating varnishes, filling
and bonding materials. Waxes are soft in nature and have low mechanical
strength but can be used as impregnating materials. Artificial waxes such as
paraffins, ceresine and halowax have excellent insulating properties. How-
ever, natural ones such as beeswax enjoy limited insulation applications.
Insulating varnishes are solutions of resins, bitumen, etc. and are mainly
used as impregnating, coating or adhesive materials. Bonding materials are
used to seal off separately insulated parts while adhesive tapes are to be
found extensively in wiring.
R

 12.2 Solid-state insulating materials 319

Resins are mainly natural and synthetic polymers (plastics) and can be
classified on the basis of their behaviour under thermal stress as thermo-
plastic or thermosetting resins. Thermoplastic resins can be formed into
fibres and will retain their properties after melting and solidification on
cooling. However, they have low mechanical strength. Thermosetting resins
on the other hand, lose their properties after a heat treatment and have
poor insulating properties but good mechanical properties. Natural resins
find extensive application and include such materials as shellac, resins,
copal and amber. Shellac is obtained from tropical trees and has good ad-
hesive properties. Synthetic resins include polystyrene, polyethylene, PVC,
acrylic, polyimide and resins derived from cellulose, polyester epoxy resins
etc.
N 
 S
  R
Natural rubber is a polymer of hydrocarbon isoprene and has poor ther-
mal properties and is prone to oxidation in air. It has a low loss tangent,
∼ 0, 01-0, 03 and an ǫr of ∼ 2, 5-5. Synthetic rubber includes butadiene
rubber, butyl rubber, polychloroprene rubber and silicon rubber and find
use in wiring and aircraft cabling.

12.2.2 Inorganic solid-state insulators

These insulators include glass, ceramics, mica, and asbestos. They are brit-
tle, hence have poor mechanical strength but can withstand extreme arcing
conditions and high temperatures.

12.2.3 Composite insulating materials

These are a combination of different materials to take advantage of the
positive attributes of the constituent materials. One of the constituent ma-
terials is laminated with the other of superior pertinent properties. Ex-
amples include asbestos polyethylene fibre and asbestos-paper-polystyrene
resin. In general, composite and organic materials are more widely used as
insulating materials.

12.2.4 Breakdown in solid-state insulators

Breakdown in solids may be caused by several factors which may include

• Defects and material inhomogeneities

• Size and shape of the specimen

• Surface conditions

• Duration of the applied electric stress

320 12. Liquid- and solid-state insulators

FIGURE 12.1. Generated and dissipated heat as a function of temperature for

different field strengths

• Moisture and other contamination levels

There are several breakdown mechanisms which include

I

  B!&

Intrinsic breakdown occurs mainly due to impurities since perfect di-

electrics have no free electrons. In the presence of impurities, the associated
free electrons can cause generation of secondary electrons at high temper-
atures which make the otherwise good insulator to conduct appreciably.
Breakdown occurs at lower voltages as the level of impurities increases.
Hence, the dielectric strength of the insulator decreases with increasing
temperature and degree of contamination.
T B!&

Thermal breakdown occurs due to thermal degradation caused by heat-

ing arising from dielectric losses. Thermal stability is achieved when the
heat generation rate is equal to the rate of heat dissipation to the environ-
ment. A critical temperature can be defined below which thermal stability
is attained and above which the insulator thermally collapses. Dielectric
loss is dependent on the applied voltage stress and hence thermal break-
down voltage is usually specified in the description of the dielectric. Fig.
12.1 shows curves highlighting the critical temperature. Note that thermal
stability is realised only below a field strength E1 .
D  B!&

Discharge breakdown is usually caused by the presence of cavities. Cav-

ities lead to high field intensities which create partial discharges. Consider
the electric field intensity in the cavity shown in Fig. ??. In the cavity
 12.2 Solid-state insulating materials 321

FIGURE 12.2. A solid dielectric capacitor with cavities and its electrical equiva-
lent circuit

region, the voltage Vc , across the cavity can be expressed as

Cm ǫr 1
Vc = Va = Va = Va
Cm + Cc ǫr − (1 − tm /tc ) 1 − (1 − tm /tc ) /ǫr

Hence
Vc 1 Va
Ec = =
tc 1 − (1 − tm /tc ) /ǫr tc
which is significantly higher than Va /tm and hence will cause partial dis-
charge in the cavity. The accumulated effect is to heat up the material
gradually until there is total breakdown. The life of the material with in-
ternal discharges will therefore depend on the number of electrical stress
application cycles.
C  
 E  B!&

Chemical breakdown is related to thermal breakdown. The material grad-

ually changes its chemical composition as the temperature is maintained,
resulting in eventual breakdown. Chemical breakdown is progressive and
occurs even when no stress is applied. The reason for this may lie in ox-
idation, moisture, temperature or contamination. Ions release gases that
contaminate the electrode and could be harmful.
T!

This refers to the formation of a permanent conducting path across the
surface of the insulation. Moisture, carbonaceous dust, industrial deposits
or cellulose fibre can collect on the surface in the form of a film. This film
dries up and separates causing sparking. Tracking can occur at low voltages,
∼ 100 V but can be suppressed by good surface cleaning.