Coll. - Antennas Fundamentals, Design, Measurement, 3rd Edn (Standard) PDF

ANTENNAS: FUNDAMENTALS,
DESIGN, MEASUREMENT
This book is dedicated to my wife Beverly Benson
Long, for our enduring love and her world-wide
work in the promotion of mental health and the
prevention of mental disorders.
ANTENNAS: FUNDAMENTALS,
DESIGN, MEASUREMENT
THIRD EDITION
LAMONT V. BLAKE
MAURICE W. LONG
SciTech Publishing, Inc

Raleigh, NC
scitechpub.com
SciTech Publishing, Inc.,
911 Paverstone Drive, Suite B
Raleigh, NC 27615
(919) 847-2434, fax (919) 847-2568
scitechpublishing.com
Copyright © 2009 by SciTech Publishing

All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form
or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as
permitted under Sections 107 or 108 of the 1976 United Stated Copyright Act, without either the prior
written permission of the Publisher, or authorization through payment of the appropriate per-copy fee
to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400,
fax (978) 646-8600, or on the web at copyright.com. Requests to the Publisher for permission should be
addressed to the Publisher, SciTech Publishing, Inc., 911 Paverstone Drive, Suite B, Raleigh, NC 27615,
(919) 847-2434, fax (919) 847-2568, or email [email protected].
The publisher and the author make no representations or warranties with respect to the accuracy or
completeness of the contents of this work and specifically disclaim all warranties, including without
limitation warranties of fitness for a particular purpose.
Editor: Dudley R. Kay

Production Director: Susan Manning
Production Coordinator: Robert Lawless
Cover Design: Kathy Palmisano
Typesetting: SNP Best-set Typesetter Ltd., Hong Kong
Printer: Hamilton Printing Company
This book is available at special quantity discounts to use as premiums and sales promotions, or for
use in corporate training programs. For more information and quotes, please contact the publisher.
Standard ISBN: 9781891121784

Deluxe with Mathcad ISBN: 9781891121791
LIBRARY OF CONGRESS Cataloging-in-Publication Data

Blake, Lamont V.
Antennas : fundamentals, design, measurement / Lamont Blake and Maurice Long.
p. cm.
ISBN 978-1-891121-78-4 (hardcover : alk. paper)—ISBN 978-1-891121-79-1 (hardcover, deluxe ed. :
alk. paper) 1. Antennas (Electronics) I. Long, Maurice W. II. Title.
TK7871.6.B534 2009
621.382′4—dc22
2008054817
Printed in the United States of America.

10 9 8 7 6 5 4 3 2
Contents
Dedication ii
Preface xv
Chapter 1 Electromagnetic Waves 1
1.1 Characteristics of Electromagnetic Waves 2
1.1.1 Wave Velocity 2
1.1.2 Frequency and Wavelength 3
1.1.3 Space–Time Relationships 4
1.1.4 Polarization 6
1.1.5 Rays and Wavefronts 7
1.1.6 Spherical Waves and the Inverse-Square Law 8
1.1.7 Field Intensity and Power Density 10
1.1.8 Decibel (Logarithmic) Expression of Attenuation 12
1.1.9 Absorption 13
1.2 Radio-Wave Optical Principles 15
1.2.1 Refraction 15
1.2.2 Reflection 18
1.2.3 Principle of Images 21
1.2.4 Interference 22
1.2.5 Huygens’ Principle: Diffraction 25
1.3 Radiation and Reception 26
1.3.1 Requirements for Radiation 28
1.3.2 Reception and Reciprocity 30
1.4 Environmental Wave-Propagation Effects 31
1.4.1 The Earth Environment 31
1.4.2 Beyond-the-Horizon Propagation 32
1.4.3 Reflection from the Earth’s Surface 35
1.4.4 Variations in Earth’s Reflection Coefficient 39
1.4.5 Wave Nomenclature 40
1.4.6 Transionospheric Propagation 41
1.4.7 The Radio-Transmission Equation 42
References 44
Problems and Exercises 44
v
vi Contents
Chapter 2 Transmission Lines 47

2.1 Basic Transmission-Line Concepts 47
2.1.1 Equivalent-Circuit Line Representation 48
2.1.2 The Lossless Infinite Line 50
2.1.3 Reflection and Standing Waves 52
2.2 Transmission-Line Equations 54
2.2.1 Steady-State Lossless Line Equations 54
2.2.2 Some Important Special Line Conditions 55
2.2.3 Impedance-Admittance Relationships 58
2.2.4 Reflection Coefficient and VSWR 60
2.2.5 Standing-Wave Patterns 62
2.2.6 Determination of Load Impedance by Standing-Wave
Measurement 63
2.2.7 Attenuation 64
2.3 Impedance Matching and Power Division 65
2.3.1 The Matching Principle 65
2.3.2 Stub Transformers 66
2.3.3 Power Dividers 67
2.4 Forms of Transmission Lines 69
2.4.1 Characteristic Impedances 71
2.5 Waveguides 72
2.5.1 Phase and Group Velocities 72
2.5.2 Cutoff Frequency 74
2.5.3 Rectangular Waveguides 74
2.5.4 Modes of Propagation 75
2.5.5 Impedance in Waveguides 76
2.5.6 Impedance Matching in Waveguides 76
2.5.7 Transmission-Line-to-Waveguide Coupling 77
2.5.8 Waveguide Junctions 78
2.5.9 Other Forms of Waveguide 78
2.6 Hybrid and Directional Couplers 80
References 83
Chapter 3 Antenna Parameters 87

3.1 Antenna Structures 88
3.1.1 Size 88
3.1.2 Supports 88
3.1.3 Feed Lines 89
3.1.4 Conductors 89
3.1.5 Insulators 90
3.1.6 Weather Protection 90
Contents vii
3.2 Radiation Pattern 91

3.2.1 Coordinate Systems 91
3.2.2 Definition of Pattern 93
3.2.3 Patterns in a Plane 93
3.2.4 Absolute and Relative Patterns 94
3.2.5 Near-Field and Far-Field Patterns 95
3.2.6 Free-Space and Earth-Reflection Patterns 97
3.3 Directivity and Gain 98
3.3.1 Definitions of Directivity and Gain 99
3.3.2 Solid Angle 99
3.3.3 Radiation Intensity 100
3.3.4 Directivity 101
3.3.5 Gain 102
3.4 Effective Area and Friis Transmission Equation 103
3.5 Beamwidth 105
3.5.1 Practical Significance of Beamwidth 105
3.5.2 Beamwidth Definition 106
3.5.3 Principal-Plane Beamwidths 107
3.6 Minor Lobes 107
3.7 Radiation Resistance and Efficiency 108
3.8 Input Impedance 110
3.9 Bandwidth 111
3.10 Polarization 112
3.11 Interdependencies of Gain, Beamwidths, and Aperture
Dimensions 113
References 115
Chapter 4 Basic Radiators and Feed Methods 119

4.1 Short Dipoles 119
4.1.1 Dipole Radiation 120
4.1.2 Pattern of an Elemental Dipole 123
4.1.3 Radiation Resistance 125
4.1.5 Beamwidth 126
4.1.6 Input Impedance 126
4.1.7 Short Dipole with Nonuniform Current
Distribution 127
4.1.8 Short Vertical Antenna with Ground Image 129
4.1.9 Top-Loaded Antenna 130
4.2 Current and Voltage in Longer Antennas 132
4.3 The Half-Wave Dipole 136
viii Contents
4.4 Long-Wire Antennas 138

4.4.1 Patterns of Resonant Antennas 139
4.4.2 Radiation Resistance and Directivity 141
4.4.3 Patterns of Nonresonant Antennas 142
4.4.4 Polarization 142
4.4.5 Effect of the Ground and Other Factors 143
4.4.6 Uses of Long-Wire Antennas 143
4.5 Loop Antennas 144
4.5.1 The Small Loop 144
4.5.2 Other Loop Antennas 146
4.6 Helical Antennas 148
4.7 Horn Radiators 150
4.8 Slot Radiators 155
4.9 Patch or Microstrip Antennas 157
4.10 Surface-Wave and Leaky-Wave Antennas 159
4.11 Basic Feed Methods 159
References 164
Chapter 5 Arrays 169

5.1 Basic Array Theory 169
5.1.1 The Two-Isotropic-Element Array 170
5.1.2 Pattern Versus q and f 174
5.1.3 Parallel-Dipole Two-Element Array 174
5.1.4 Mutual Coupling 177
5.1.5 Input and Mutual Impedances, and Radiation
Resistance 178
5.1.6 Principle of Pattern Multiplication 179
5.1.7 Effects of Element Directivity and Mutual Coupling 181
5.2 Multielement Uniform Linear Arrays 182
5.2.1 Radiation Pattern 183
5.2.2 Broadside Linear Arrays 183
5.2.3 Endfire Arrays 186
5.3 Parasitically Excited Endfire Arrays 188
5.4 Planar and Volume Arrays with Uniform Aperture
Distribution 190
5.4.1 Planar Arrays 190
5.4.2 Pattern Versus q and f 191
5.4.3 Volume Arrays and Plane Reflectors 193
5.5 Linear Array Pattern Calculations 194
5.5.1 General Equations for Array Patterns 194
5.5.2 Linear Array Patterns 196
Contents ix
5.5.3 Broadside Linear Array Patterns 197

5.5.4 Grating Lobes 198
5.5.5 Comparisons: Broadside Array and Continuous
Aperture Antenna 198
5.5.6 Endfire Array Patterns 199
5.6 Array Tapering for Side-Lobe Reduction 201
5.6.1 Introduction 201
5.6.2 Amplitude Tapering of Broadside Arrays: General 202
5.6.3 Dolph-Chebyshev Synthesis 203
5.6.4 Simplified Calculations for Dolph-Chebyshev
Patterns 205
5.6.5 The Taylor Distribution 206
5.6.6 Other Low-Side-Lobe Array Distributions 208
5.7 Planar Arrays: Patterns, Directivity, and Gain 209
5.7.2 Patterns Versus q and f: Equal Element Spacing 210
5.7.3 Separable Aperture Distributions 211
5.7.4 Patterns Versus q and f: Equal Amplitude, Separable
Arrays 212
5.7.5 Summary of Planar Array Patterns 213
5.7.6 Beamwidth and Gain of Broadside Arrays 215
5.8 Some Other Array Types 218
5.8.1 Long-Wire Arrays 218
5.8.2 Horn and Slot Arrays 220
5.8.3 Patch or Microstrip Arrays 220
5.8.4 Phased Arrays for Scanning 221
5.9 General Remarks on Mutual Coupling 221
References 222
Chapter 6 Reflectors and Lenses 227

6.1 Focusing and Collimation 228
6.1.1 Focusing by a Parabolic Reflector 228
6.1.2 Paraboloids and Parabolic Cylinders 230
6.1.3 Space Attenuation for Point and Line Sources 230
6.2 Beamwidth and Directivity 233
6.2.1 Shape of Aperture 233
6.2.2 Primary and Secondary Patterns 233
6.2.3 Beamwidth 234
6.3 Reflector Illumination 234
6.3.1 Primary Pattern Requirements 235
x Contents
6.3.2 Taper and Side Lobes for Elliptical Apertures 237

6.3.3 Optimum Illumination Taper 238
6.3.4 Line-Source Feeds 239
6.3.5 Feed-Positioning Requirements 241
6.3.6 Cassegrain Feeds 244
6.4 Radiation Patterns of Horn Antennas 245
6.4.1 Horn Patterns, in General 245
6.4.2 Shapes of Major Lobes 246
6.4.3 Measured Wide-Beamwidth Horn Patterns 247
6.4.4 Summary 248
6.5 Pattern Calculation and Reflector Antenna Design 248
6.5.1 General Remarks 248
6.5.2 Aperture Distribution and Pattern of a Half
Parabolic Cylinder 250
6.5.3 Steps for Reflector Antenna Design 251
6.6 Reflector Construction 252
6.6.1 Fabrication Methods 253
6.6.2 Reflector Surface Accuracy Requirements 253
6.7 Corner-Reflector Antennas 255
6.8 Lens Antennas 257
6.8.1 Optical Lenses 257
6.8.2 Lens Surface Configurations 258
6.8.3 Lens Refracting Media 260
6.8.4 Variable-Index-of-Refraction Lenses 261
6.8.5 Focusing with Dielectric and Metal Plate Lenses 262
6.8.6 Dielectric Lenses 262
6.8.7 Metal Plate Lenses 264
6.8.8 Geodesic Luneberg Lenses 265
6.9 Beam Steering by Feed Offset 269
6.10 Pattern Calculations for Continuous Apertures 269
6.10.1 Historical Note on Aperture Analyses 270
6.10.2 Continuous Line Source 271
6.10.3 Aperture Analysis: General 274
6.10.4 Apertures Having Separable x and y
Distributions 275
6.10.5 Equivalent Linear Apertures in Principal Planes 277
6.10.6 Circular Apertures 279
6.10.7 Patterns Versus q and f: A Summary 279
6.11 Comparison of Directive Antenna Types 281
References 282
Contents xi
Chapter 7 Antennas with Special Properties 285

7.1 Broad-Band Antennas 285
7.1.1 Basic Broad-Band Principles 286
7.1.2 Broad-Band Dipoles 288
7.1.3 Biconical Antennas 289
7.1.4 Monopoles 290
7.1.5 Folded Dipoles 291
7.1.6 Superturnstile Antenna 292
7.1.7 Miscellaneous Broad-Band Dipoles and Other
Radiators 292
7.2 Frequency-Independent Antennas 293
7.2.1 The Log-Periodic Principle 294
7.2.2 Practical Log-Periodic Designs 297
7.2.3 Equiangular Log-Periodic Antennas 299
7.2.4 Self-Complementary Antennas 300
7.2.5 Spiral Antennas 301
7.3 Antennas for Multiple Polarizations 302
7.3.1 Antennas for Circular Polarization 302
7.3.2 Polarizers and Dual-Mode Transducers 305
7.3.3 Corrugated Horn Feeds 307
7.4 Omnidirectional Antennas 307
7.5 Electrically Small Antennas 308
7.6 Direction-Finding Antennas 310
7.6.1 Low-Frequency Null Systems 311
7.6.2 Beam-Maximum Systems 312
7.6.3 Lobing Antennas 312
7.7 Mechanical Scan Antennas 314
7.7.2 Rapid Mechanical Scan Antennas 315
7.8 Low-Noise Receiving Antennas 320
7.9 Synthetic-Aperture Antennas 324
7.10 Antennas with Extremely Low Side Lobes 326
7.11 Geometrical Theory of Diffraction, Method of Moments,
and Fractals 328
7.11.1 Geometric Theory of Diffraction 328
7.11.2 Method of Moments 329
7.11.3 Fractals and Fractal Antennas 330
References 332
xii Contents
Chapter 8 Electronically Steered Arrays 337

8.1 Phased Array Principles 338
8.1.1 Time Delay and Phase Shift Requirements 339
8.1.2 Effective Aperture Versus Scan Angle 340
8.1.3 Frequency-Scan Antennas 341
8.1.4 Array Bandwidth 342
8.1.5 Phased Array Patterns 344
8.1.6 Grating Lobes 346
8.2 Beam-Steering Technology 347
8.2.1 Phase Shifters 347
8.2.2 Effects of Phase Quantization 349
8.2.3 Effects of Element Mutual Coupling 350
8.2.4 Time Delayers 351
8.3 Phased Array Feed System Technology 351
8.3.1 Feed Systems 352
8.3.2 Active Arrays and T/R Modules 354
8.3.3 Alternative Configurations, Including
Subarrays 355
8.3.4 Butler Matrix 355
8.3.5 Array Sum and Difference Patterns 357
8.3.6 Microstrip and Printed-Circuited Technology 357
8.3.7 Conformal Arrays 357
8.4 Adaptive Array Antennas 358
8.4.1 Background System Developments 358
8.4.2 Adaptive Beam-Forming Arrays 360
8.4.3 Space-Time Adaptive Processing 362
8.5 Examples of Electronically Steered Arrays 364
References 368
Chapter 9 Antenna Measurements 371

9.1 Antenna Patterns, General 371
9.1.1 Radiating Near- and Far-Field Patterns 373
9.1.2 Pattern Statistics: Near- and Far-Field
Statistical Gain 376
9.2 Far-Zone Pattern Measurements 377
9.2.1 Low-Frequency Techniques 379
9.2.2 High-Frequency Techniques 379
9.2.3 Far-Field Ranges 381
9.3 Compact Ranges 382
9.4 Near-Field Antenna Measurements 385
9.4.1 Planar Near-Field Measurements 386
Contents xiii
9.4.2 Cylindrical Near-Field Measurements 387

9.4.3 Spherical Near-Field Measurements 387
9.5 Polarization Measurement 388
9.6 Scale-Model Measurements 389
9.7 Antenna Pattern Measurement Equipment 390
9.8 Directivity and Gain Measurements 393
9.8.1 Absolute-Field-Strength Method 394
9.8.2 Standard Gain Antennas 395
9.8.3 Power-Measurement Methods 396
9.8.4 Gain Measurement by Comparison 396
9.9 Antenna Efficiency 398
9.9.1 Radiation Efficiency 398
9.9.2 Aperture Efficiency 399
9.10 Radiation Resistance 399
9.11 Impedance Measurements 400
9.11.1 Standing-Wave Method 401
9.11.2 Network Analyzers 402
9.12 The Smith Chart 404
9.13 Antenna Noise Measurement 407
9.13.1 Elementary Noise Theory 407
9.13.2 Antenna Noise Temperature 409
9.13.3 Measurement Techniques 410
9.13.4 Low-Noise-Temperature Comparison Sources 413
9.13.5 Measurements Using Ambient Temperature
Noise Sources 413
9.14 System Noise Calculations 414
9.14.1 Noise Figure 414
9.14.2 System Noise Temperature 415
References 417
Answers to Problems 423
Appendices
A. Maxwell’s Equations 427
References 431
B. Polarization Theory 433
B.1 Elliptically Polarized Waves Resolved into Linearly
Polarized Components 433
B.2 Elliptically Polarized Waves Resolved into
Circularly Polarized Components 435
xiv Contents
B.3 Relationships between Linearly and Circularly

Polarized Components 436
References 437
C. Review of Complex-Variable Algebra 439
D. Complex Reflection Coefficients and Multipath Effects 445
D.1 Fresnel’s Equations 445
D.2 Reflection Coefficients for Smooth Land and Sea 447
D.3 Reflection Coefficients for Rough Surfaces 449
D.4 Pattern-Propagation Factor F 450
D.5 Flat Earth Geometry 451
D.6 Multipath Dependencies on Frequency, Polarization,
and Surface Roughness 453
D.7 Spherical Earth Geometry 455
D.8 Divergence Factor 459
References 460
E. Radomes 463
References 466
F. Far-Zone Range-Approximation and Phase Error 467
References 468
G. Radiating Near and Far Fields, and the Obliquity Factor 469
G.1 The Radiating Fields 469
G.2 Principal-Plane Patterns Versus Range 473
References 474
H. Path Length Differences from a Planar Aperture 475
I. Effects of Random Aperture Phase Errors 479
References 482
Index 483
About the Author 503
Supplemental Materials (SM)
The publisher has posted a list of supplemental materials (SM) to an accompanying
website at:
www.scitechpub.com/blakelong3.htm
Within the text of the book you will find references to the specific SM sections that relate
to the material being covered. A computer icon ( ) is used in the margin to further
identify sections that refer to the SM.
SM Table of Contents
SM 1.0 Reflection Coefficients for Flat, Smooth Surfaces
SM 1.1. Reflection Coefficients-Flat, Smooth Sea
SM 1.2. Reflection Coefficients-Flat, Smooth Land
SM 2.0 Spherical Earth Geometry
SM 3.0 Earth Effects on Patterns and Multipath Propagation

SM 3.1. Ground Effects on Elevation Patterns
SM 3.2. Pattern Range Illumination Versus Observation Height
SM 3.3. Multipath Versus Range, Flat Conducting Surface
SM 3.4. Multipath Versus Range, Rough Flat Land
SM 3.5. Multipath Versus Range, Spherical Earth
SM 3.6. Divergence Factor Comparisons
SM 4.0 Antenna Radiation Analyses

SM 4.1. Radiating Near and Far Fields, Uniform Aperture Approximation
SM 4.2. Relative Gain Versus Range
SM 4.3. Random Aperture Phase Errors
SM 4.4. Radiating Near and Far Field Patterns of an Array
SM 4.5. Radiating Near Field along an Aperture
SM 4.6. Linear Array Pattern, Quadratic Phase Distribution
SM 4.7. Steered Array Patterns Versus Phase
SM 4.8. Steered Array Patterns at Two Frequencies
SM 4.9. Pattern of Parabolic Cylinder with Offset Feed
SM 5.0 Three-Dimensional Pattern Construction, by Aaron Loggins

Preface
This is a senior undergraduate or first-year graduate level textbook on antenna funda-

mentals, design, performance analysis, and measurements. In addition to its use as a
formal course textbook, it is well-suited for professional training and self-study by prac-
ticing engineers, scientists, and technologists who desire to expand their knowledge of
antennas. The book provides a broad coverage of antenna types and phenomena, for
operations at very low radio frequencies, as well as frequencies up to those of submilli-
meter wavelengths. Unlike most university-level antenna textbooks, reading it does not
require prior skills in electromagnetic theory, sophisticated mathematics, or computer
programming. An additional feature is the downloadable collection of computer solutions
in both Mathcad® and MATLAB® to numerous antenna radiation examples, which can
be easily implemented and revised by persons not having prior programming experience.
Evolution of the Third Edition

This new edition was prepared for use in a one-semester first-year graduate night
class at the Southern Polytechnic State University located in Marietta, Georgia, where
student backgrounds vary widely. At least half the students are from overseas and hold
a bachelor’s degree in electrical engineering or electronics. Of those from the United
States, about half have degrees in electrical engineering, and the remainder hold degrees
in electrical engineering technology. Generally, the students are older than a typical first-
year graduate student, being 25 to 35 years of age. A few of the students have excellent
backgrounds in vector calculus and the use of Maxwell’s equations, while some of the
older ones may need to refresh their abilities with the phasor calculations of electrical
circuits.
The computer capabilities of the students also vary widely. Many of the entering
graduate students are proficient in use of MATLAB or Mathcad software, but sometimes
they are experienced with neither and rarely both. Additionally, some students who
have been away from academics a few years may have little computer proficiency. Thus,
to expedite the class learning on antennas with computer analyses, Mathcad, which is
easily read without prior experience, is used in classroom lectures. For class assignments,
students are allowed to use whatever computer software they choose. Mathcad software
xviii Preface
is used most often, with MATLAB being the second most popular choice. Often, original
MATLAB users, after switching to Mathcad, have expressed appreciation for their intro-
duction to Mathcad for its relative ease of use and intuitive qualities.
Because of the wide differences in student backgrounds, and after considering the
available textbooks, Blake’s Antennas, Second Edition, was selected for adoption because
of its superior readability. Due to the book’s age, the selection required preparing
and distributing materials for updating and expanding the text and adding appendices.
Consequently, the present book evolved into one that retains the benefits of Blake’s
second edition but expands the subject material suitably for a senior or graduate level
textbook.
Background Assumed
Most antenna textbooks are written for students proficient with vector calculus and begin
with the use of Maxwell’s equations in the development of antenna theory. Such books
often do not meet the needs of many students and practicing engineers who, because of
their backgrounds or personal interests, desire a more direct path for assimilating antenna
fundamentals and their connection to application topics of antenna engineering. Although
antenna theory is founded on Maxwell’s equations, understanding their concepts does
not require advanced mathematics. At the beginning of each antenna course, the revising
author (MWL) uses Appendix A, Maxwell’s Equations, to address the key “postulates”
of Maxwell and provide a brief introduction to, or review of, the essential equations.
Thus, Maxwell equations are discussed with the goal of expressing their meaning in
words. Then, the concepts of displacement current, interdependence of changing electric
and magnetic fields, and wave propagation are described, and thus Maxwell’s equations
are underscored as “the ultimate truth” but thereafter considered outside the scopes of
antenna design, performance analysis, and measurements.
Organization
This book was prepared with the intention of providing a comprehensive antenna text
that can be readily understood by persons with undergraduate educations in engineering,
science, or technology. The chapter titles follow:
Chapter 1. Electromagnetic Waves

Chapter 2. Transmission Lines
Chapter 3. Antenna Parameters
Chapter 4. Basic Radiators and Feed Methods
Chapter 5. Arrays
Chapter 6. Reflectors and Lenses
Chapter 7. Antennas with Special Properties
Chapter 8. Electronically Steered Arrays
Chapter 9. Antenna Measurements
Preface xix
Chapters 1 through 6 cover, generally, the physics and technology of antennas and
include such subjects as wave propagation, reflection, refraction, diffraction, transmission
and reception, basic radiators, antenna arrays, reflector antennas, and lenses.
Chapter 7 discusses antenna properties and analysis techniques not addressed in other
chapters. Its range of topics is wide, and includes techniques for providing wide band-
widths, multiple polarizations, low receiver noise, and extremely low sidelobes. In addi-
tion, direction-finding antennas and mechanical beam scanners are addressed. Finally
discussed are synthetic-aperture antennas, geometrical theory of diffraction (GTD),
method of moments (MoM), and fractals.
Chapter 8 treats electronically steered arrays, whereas Chapter 5 is focused on fixed
beam arrays. In other words, chapter 8 stresses array concepts specific to beam movement
made possible with fast, wide-dynamic-range digital components and cheap computer
memory, along with continued improvements in high-speed switches and phase shifters.
Chapter 9 includes a broad coverage of antenna measurement techniques and equip-
ment. Subjects include radiating near fields as well as far field patterns and pattern sta-
tistics, compact ranges, and near-field measurements. Included also is a comprehensive
treatment of antenna noise, noise temperature, noise figure, and system signal-to-noise
ratios.
There are problems at the end of each chapter, and answers to the odd numbered
problems are included in a section near the book’s end. Appendices provide technical
depth to the chapters, appropriate for a senior or first graduate level antenna course. The
appendix titles follow:
Appendix A. Maxwell’s Equations

Appendix B. Polarization Theory
Appendix C. Review of Complex-Variable Algebra
Appendix D. Complex Reflection Coefficients and Multipath Effects
Appendix E. Radomes
Appendix F. Far-Zone Range-Approximation and Phase Error
Appendix G. Radiating Near and Far Fields, and the Obliquity Factor
Appendix H. Path Length Differences from a Planar Aperture
Appendix I. Effects of Random Aperture Phase Errors
It is to be noted that Appendix C discusses complex-variable algebra. Although its con-

tents will be familiar to most readers, it is included because some may find parts of it
useful for review.
Data files of computer scripts

Where appropriate, the appendices and the downloadable data files are referenced in the
chapters for providing a more complete treatment of antennas. In the Deluxe Edition,
a full-featured copy of Mathcad 14.0 is included so that readers can easily create their
own computer analyses. The downloadable data files provide computer solutions in both
Mathcad and MATLAB to problems in the areas that follow:
xx Preface
– reflection coefficients for surfaces versus dielectric properties, conductivity, polari-

zation, surface roughness, and incidence angle
– earth’s multipath effects on antenna patterns versus surface properties, antenna and
observation heights and separation distance, and polarization for flat and spherical
earth models.
– radiating near and far fields from arrays and continuous aperture antennas, as func-
tions of aperture phase and amplitude distributions and random aperture errors.
The files also include a supplemental chapter in PDF on the creation of antenna radiation
field graphics using Mathcad. It was prepared by student Aaron Loggins as one of three
project assignments in a one-semester antenna course.
Files can be downloaded from the publisher’s web page for this book:
www.scitechpub.com/blakelong3.htm
Acknowledgements
Permissions to use the contents of Antennas, 2nd Edition, by Lamont V. Blake, now
deceased, were provided by Barbara Blake, Lamont Blake’s daughter, and other Blake
family members and are gratefully appreciated. This third edition could not have been
written otherwise because it was built upon an easily read, well-written text based on a
solid technical foundation. Therefore, it could be readily expanded to provide a senior
or graduate level textbook suitable for students with widely different academic back-
grounds, including persons with limited or no computer programming experience.
Two important and closely related tasks were accomplished by Dr. Donald G. Bodnar
in connection with “Chapter 9 – Measurements.” First, he completed a technical review
of an early version of the chapter, and he then wrote Sec. 9.4, a major section titled “Near
Field Antenna Measurements.” That section is copyrighted by MI Technologies, Inc.,
Don Bodnar’s company.
Appreciation is acknowledged to Aaron Loggins for letting me use his classroom
project paper as a PDF file that discusses the creation of 3-D graphics with Mathcad.
A major and generally thankless task of pursuing a penetrating technical edit of each
chapter and appendix was accomplished by Dr. Edward B. Joy, and it was performed
with record-breaking speed. Ed found and corrected not only accidental and careless
errors, but he also underscored and made suggestions for correcting more substantive
oversights.
Special thanks are due to Dr. Anatoliy Boryssenko of the University of Massachusetts
for his expertise in checking the Mathcad files, offering helpful suggestions, and
then rewriting them into MATLAB scripts. He did so under very tight deadlines. Dr.
Boryssenko has also graciously offered additional files from his personal collection to
further enhance the data set of the publisher’s web page.
A major contribution to this book was made by Dr. Randy J. Jost of Utah State Uni-
versity, as an advisor to SciTech Publishing, by reviewing and suggesting additions to
my early book writing plans. One of those suggestions was to include the files that contain
Preface xxi
a number of antenna radiation problems and their computer solutions. Inclusion of the
CD in the Deluxe Edition that contains Mathcad, version 14, software results from the
initiative of Dudley Kay, Founder of SciTech Publishing, and the cooperation of Para-
metric Technology Corporation, the owner of Mathcad. Permission by Parametric Tech-
nology Corporation to use screenshots of computer images from Mathcad software,
included in the file on creating 3-D graphics is gratefully acknowledged.
There have been a number of persons who have made significant editorial improve-
ments and others who have simply expressed an interest in an updated edition of Lamont
Blake’s Antennas becoming available. Some of these include Gerald Oortman of Lock-
heed Martin, Marietta, Georgia; Professor Charles Bachman of the Southern Polytechnic
State University; Dr. Andrew Peterson of the School of Electrical and Computer Engi-
neering, Georgia Institute of Technology; James Gitre of Motorola; Michael Havrilla of
the Air Force Institute of Technology; and Rickey Cotton (deceased), Mark Mitchell, and
Dr. Charles Ryan (retired) of Georgia Tech Research Institute.
I thank Phyllis Hinton of Georgia Tech Research Institute, who has, over the years,
brightened my days when she sketches a figure I need or somehow helps me find my
way through the ends and outs of Microsoft Word.
Appreciation is expressed here to Dudley Kay, Susan Manning, and Robert Lawless
of SciTech Publishing who, during the preparation of this book, have demonstrated an
enthusiasm for producing quality textbooks.
Maurice W. Long
Atlanta, Georgia
[email protected]
CHAPTER 1
Electromagnetic Waves
This book begins with an elementary discussion of electromagnetic wave theory, which
is basic to an understanding of antennas. Readers who are quite familiar with the princi-
ples of electromagnetic waves may prefer to begin with the second (or third) chapter, but
even they may find this introductory chapter a handy review of the subject and a useful
reference source.
Electromagnetic waves in space are the basis of radio transmission over great distances
without direct wire connection between the transmitting and receiving points. At the
transmitting and receiving stations, radio signals exist in the form of high-frequency
alternating currents in conductors and in electronic amplifying devices. Between the
transmitter and receiver they exist as electromagnetic waves in space. Antennas, the
subject of this book, are the devices that act as go-betweens.
At the transmitting station the antenna is energized by the electrical currents generated
in the transmitter, and it converts the energy into the form of an electromagnetic field. It
“launches” the waves into space. At the receiving station the antenna captures energy
from the arriving field, and it converts the field variations into current and voltage replicas
of those at the transmitter (though of much smaller amplitude).
Current and voltage in conductors are always accompanied by electric and magnetic
fields in the adjoining region of space, and in a sense it is incorrect to speak of “convert-
ing” electrical energy from the form of current and voltage into the form of an electro-
magnetic field, and vice versa. In a practical sense, however, the distinction is made. In
the one case the fields are bound to the conductors in which the current flows; in the
other they are “free.”
The picture of electromagnetic waves presented here is considerably simplified and
necessarily leaves unanswered some questions that may be disturbing to the reader with
an inquiring mind. Often these questions can be answered, but only through the use of
rather sophisticated mathematical and physical concepts, based on James Maxwell’s
equations, which he published in 1873 (see Appendix A). Bear in mind that most of the
ideas and principles to be discussed have this background, even though it is not necessary
to look deeply into Maxwell’s equations for the purposes of this book.
1
2 Electromagnetic Waves
1.1. Characteristics of Electromagnetic Waves

A wave is an oscillatory motion of any kind, the most familiar being waves on the surface
of water. Sound waves, another common example, are vibrations of the air or of various
material substances. Both wave types involve mechanical motion. Electromagnetic waves
are electric and magnetic field variations that can occur in empty space as well as in
material substances.
All waves are characterized by the property called propagation. The vibrations at a
particular point in space excite similar vibrations at neighboring points, and thus the wave
travels or propagates itself. This concept is given more specific form as Huygens’ prin-
ciple, to be discussed later in this chapter. The particular substance or space in which a
wave exists is the propagation medium.
1.1.1. Wave Velocity

Waves travel at characteristic speeds, depending on the type of wave and the nature of
the propagation medium. For example, sound travels about 330 meters per second in the
normal atmosphere, but in water the speed is 1,450 meters per second. In both media the
figure varies with the temperature and other factors.
Free space is a term much used in discussion of electromagnetic waves. It implies not
only empty space (a vacuum), but also remoteness from any material substances from
which waves may be reflected. Electromagnetic waves travel exceptionally fast, approxi-
mately 300,000 kilometers per second (3 × 108 m/s) in free space. In other propagation
media their speed may be less, but ordinarily it is very high compared with the speeds
of things observable without special instruments. In the gases of the earth’s normal
atmosphere, in fact, the speed is only slightly less than in empty space (vacuum), and
for practical purposes the difference is negligible except over very long paths. Even then
it is ordinarily permissible to use the approximate free-space velocity figure for calculat-
ing how long it takes a radio wave to travel from one point to another in the
atmosphere.
An important exception to this statement occurs when waves at certain radio frequen-
cies travel in the ionosphere, a layer of charged particles (ions) lying above the earth
between the heights of about 60 and 300 kilometers. At very low radio frequencies
radio waves cannot penetrate the ionosphere; they are reflected from it. At very high
frequencies waves pass through the ionosphere unimpeded at the same speed they would
have in empty space. But in a critical intermediate frequency region, depending on iono-
spheric conditions (which vary considerably from day to night and with the season and
other factors), the wave velocity in the ionosphere may be different than it is in
vacuum.
The speed of electromagnetic propagation in a vacuum is of fundamental importance.
This value, commonly called the “speed of light” in a vacuum, is designated by the
symbol c. (Light waves are actually electromagnetic waves of very high frequency.) The
value of c is 299,793 kilometers per second, rounded off for most purposes to 3 × 108
meters per second.
Characteristics of Electromagnetic Waves 3
1.1.2. Frequency and Wavelength

The oscillations of waves are periodic, or repetitious. They are characterized by a fre-
quency, the rate at which the periodic motion repeats itself, as observed at a particular
point in the propagation medium. Complex waves may contain more than one frequency.
The frequency is expressed in Hertz (cycles per second, a cycle being one full period of
the wave). Hertz is abbreviated as Hz. A single-frequency wave motion has the form of
a sinusoid.
The wavelength of an electromagnetic wave is the spatial separation of two successive
“oscillations,” which is equal to the distance that the wave travels during one sinusoidal
cycle of oscillation. Therefore, if the wave velocity is v meters per second and the fre-
quency is f cycles per second (Hertz), the wavelength in meters is
v
λ= (1–1)
f
As has been noted, v may have different values in different propagation media. When
the wave velocity in free space (vacuum) c is used in (1–1), the resulting value of l is
the free-space wavelength, sometimes denoted by l0.
The electromagnetic spectrum covers an enormous range of frequencies, including
cosmic-ray radiation with frequencies in excess of 1020 Hz. Radio frequencies, their
designations, and free-space wavelengths are given in Table 1-1, that is derived from the
reference IEEE Standard 211-1997. However, the radio waves that may be detected and
amplified, from a practical point of view, are from about 10 kHz to 100 GHz (IEEE
Standard 100–1992, p. 1059).
The International Telecommunications Union (ITU), an organization of the United
Nations, makes the general guidelines for the assignment and use of frequencies.
TABLE 1-1 The Radio Frequency Spectrum
Designation Frequency Free-Space Wavelength

Ultra low frequency (ULF) <3 Hz >108 m
Extremely low frequency (ELF) 3 Hz–3 kHz 108–105 m
Very low frequency (VLF) 3–30 kHz 105–104 m
Low frequency (LF) 30–300 kHz 104–103 m
Medium frequency (MF) 300 kHz–3 MHz 103–102 m
High frequency (HF) 3–30 MHz 102–10 m
Very high frequency (VHF) 30–300 MHz 10–1 m
Ultra high frequency (UHF) 300 MHz to 3 GHz 1 m–10 cm
Super high frequency (SHF) 3–30 GHz 10–1 cm
Extremely high frequency (EHF) 30–300 GHz 1 cm–1 mm
Submillimeter 300 GHz–3 THz 1 mm–0.1 mm
Key to abbreviations. Hz = Hertz, kHz = kilohertz (103 Hz), MHz = megahertz (106 Hz), GHz = gigahertz (109 Hz), THz = terahertz
(1012 Hz), m = meters, cm = 10−2 m, mm = 10−3 m.
Generally, the regulations of individual nations follow those of the ITU. For the United
States, separate authorities for use by federal governmental and nonfederal governmental
frequency usages are assigned to the National Telecommunications and Information
Agency and the Federal Communications Commission, respectively. The ITU Radio
Regulations allocates the frequencies between 9 kHz and 275 GHz into frequency bands,
according to forty-two types of radio services usage—radio amateur, FM broadcast,
television, and so on—and location within three regions of the world.
A governmental operating license, with restrictions on transmit power and waveform,
is ordinarily required for radio transmit equipment, but there are some frequency bands
where operations are permitted for low-power, short-range operations. Ah Yo and Emrick
(2007) is a useful information source on frequency bands available for amateur (“ham”),
commercial, and military.
Some of the better-known frequency allocations within the United States include:
• AM (amplitude modulation) broadcast: 535–1705 kHz
• FM (frequency modulation) broadcast: 88–108 MHz
• Television in 6 MHz bandwidth numbered channels:
• 2–6, 54–60 MHz, . . . , 82–88 MHz
• 7–13, 174–180 MHz, . . . , 210–216 MHz
• 14–36, 470–476 MHz, . . . , 600–608 MHz
Note: Following the current website on U.S. frequency allocations, www.ntia.
doc.gov/osmhome/allocchrt.pdf, the region 614–698 MHz (without channel
numbers) is allocated to TV. The frequencies 608–806 MHz were previously
allocated for TV channels 37–69.
• GPS (global positioning satellite): 1227.6 MHz (military); 1575.42 MHz and
1227.6 (civilian); telemetry on 2227.5 MHz
Radar band designations following IEEE Standard 521-1984, page 8, are given in
Table 1-2. Although standard for radar, these designations are used in a broader electron-
ics community. The region 1 GHz (30 cm) to 30 GHz (1 cm) is usually called the micro-
wave region; and, contrary to Table 1-2, the region 30 GHz (10 mm) to 300 GHz (1 mm)
is usually called the millimeter wavelength region. Although millimeter-wave radars are
becoming more widely used, most radars operate within the microwave region. Some
operate at frequencies of a few MHz and a few even use infrared (8 × 1011–4 × 1014 Hz)
and visible light (4 × 1014–7.5 × 1014 Hz) frequencies.
There are differences between Tables 1-1 and 1-2. In Table 1-1 (which is for nonradar
usage), UHF is the frequency range 300 MHz–3 GHz, and the designations SHF and
EHF cover the range 3–300 GHz. Although Table 1-2 is widely used, there is not uni-
versal agreement on its letter designations and frequencies.
1.1.3. Space–Time Relationships

An electromagnetic wave has two components, an electric field and a magnetic field.
Each component varies sinusoidally in time at a fixed point of space, with time period
T = 1/f seconds, where f is the frequency in Hertz (cycles per second). Also at a fixed
TABLE 1-2 Standard Radar Band Designations
Designation Frequency Free-Space Wavelength

HF 3–30 MHz 100 m–10 m
VHF 30–300 MHz 10 m–1 m
UHF 300–1000 MHz 100 cm–30 cm
L Band 1–2 GHz 30 cm–15 cm
S Band 2–4 GHz 15 cm–7.5 cm
C Band 4–8 GHz 7.5 cm–3.75 cm
X Band 8–12 GHz 3.75 cm–2.50 cm
Ku Band 12–18 GHz 2.50 cm–1.67 cm
K Band 18–27 GHz 1.67 cm–1.11 cm
Ka Band 27–40 GHz 1.11 cm–.75 cm
V Band 40–75 GHz 7.5 mm–4.0 mm
W Band 75–110 GHz 4.0 mm–2.7 mm
mm Band 110–300 GHz 2.7 mm–1.0 mm
instant of time there is a sinusoidal variation in space along the direction of propagation,
with spatial period (wavelength) l = v/f meters, where v is the velocity of propagation
in meters per second [from (1–1)]. In terms of a cartesian coordinate system (rectangular
coordinates x, y, z), if the electric field E of the wave is represented by vectors parallel
to the x-axis and the wave is propagating in free space in a direction parallel to the z-axis,
the magnetic field H will be represented by vectors parallel to the y-axis, as shown in
Fig. 1–1. These space–time relationships for a plane wave are expressed by (1–2) and
(1–3):
2π z
Ex ( z, t ) = E0 sin ⎛⎜ 2π ft − + φ ⎞⎟ (1–2)
⎝ λ ⎠
2π z
H y ( z, t ) = H0 sin ⎛⎜ 2π ft − + φ ⎞⎟ (1–3)
⎝ λ ⎠
The notation Ex(z, t) indicates that Ex is a vector parallel to the x-axis and has a magnitude
that depends on the values of the variables z and t. The parameter E0 is the maximum
value, defined as the amplitude of the wave. Note that E0 is the value that Ex(z, t) attains
when |sin(2pft − 2pz/l + f)| = 1, which in turn will occur periodically at time intervals
of T = 1/2f at a fixed point and at z-intervals of l/2 (half-cycle and half-wavelength
intervals). Sometimes the root-mean-squared (rms) value, rather than the peak value, is
used to characterize the amplitude of a sinusoidal oscillation. The parameter f is the
initial phase angle of the wave; that is, at t = 0 and z = 0, Ex(z, t) has the value E0 sin f.
Similar statements apply to Hy(z, t). Figure 1–1 portrays these relationships
schematically.
As shown, both the electric and magnetic components of the wave are “in phase” in
space, that is, their maxima and minima occur for the same values of z. They are also in
E0
x
Hy
H0
Ex
E(f)
FIGURE 1–1.
Spatial relationships of a plane electromagnetic wave in free space.
phase in time, at a fixed value of z. However, they are both directed at right angles to
each other and to the direction of propagation, a relationship that they always bear to
each other in free-space propagation. The designation plane wave means that the pattern
shown, although described as existing only along the z-axis, actually exists everywhere
in space, the wave vectors at any point (x, y, z) being exactly like those at the point (0,
0, z). At a fixed value of z there is no variation of the field in the x- and y-directions, that
is, in an xy-plane at the point z; hence the name plane wave. (Not all electromagnetic
waves are plane. A plane wave is an idealization never perfectly realized, but in practice
waves may often be considered locally plane, with small error and with great simplifica-
tion of mathematical description.)
The motion of the wave may be visualized by imagining that the entire set of field
vectors, not only those shown but also those at all other values of x and y, is moving in
unison in the positive z-direction at velocity c = 3 × 108 meters per second. An observer
at a fixed point would see a sinusoidal time variation of both E and H. On the other hand,
if he could somehow (magically) “freeze” the motion and take measurements of E and
H along the z-axis he would observe the pattern in Fig. 1–1.
1.1.4. Polarization
The plane wave shown in Fig. 1–1 is linearly polarized; that is, the electric vector has a
particular direction in space for all values of z, in this case the x-axis direction. The wave
is therefore said to be polarized in the x-direction. In actual space above the earth, if the
electric vector is vertical or lies in a vertical plane, the wave is said to be vertically
polarized; if the E-vector lies in a horizontal plane, the wave is said to be horizontally
polarized. (It is conventional to describe polarization in terms of the E-vector.)
The initial polarization of a radio wave is determined by the antenna (and its orienta-
tion) that launches the waves into space. The polarization desired, therefore, is one of
the factors entering into antenna design. In some applications a particular polarization is
preferable; in others it makes little or no difference.
Electromagnetic waves are not always linearly polarized. In circular polarization,
from the viewpoint of a fixed observer, the electric vector appears to be rotating with a
screw motion about the z-axis (direction of propagation), making one full turn for each
rf cycle. In further analogy with a screw thread the rotation may be clockwise or coun-
terclockwise, corresponding to right-hand-circular and left-hand-circular polarizations.
A circularly polarized wave results when two linearly polarized waves are combined—
that is, if they are simultaneously launched in the same direction from the same antenna—
provided that the two linear polarizations are at right angles to each other and their phase
angles [the angle f in (1–2) and (1–3) differ by 90 degrees or p/2 radians]. The right-
hand or left-hand rotation depends on whether the phase difference is plus or minus. For
truly circular polarization it is necessary also that the two linearly polarized components
be of equal amplitude. If they are of different amplitudes, elliptical polarization results
(see Figs. B-1 through B-4 of Appendix B).
The polarization is random when there is no fixed polarization or pattern of polariza-
tion-variation that is repetitive along the z-axis, an effect present in light waves emitted
from an incandescent source (e.g., the sun or an electric light bulb). It is seldom observed
in manmade radio emissions, but such would result if two independently random sources
of radio noise (used in radio and radar military countermeasures, or “jamming”) are con-
nected to right-angle-polarized elements of a single antenna.
Linear polarization is the most commonly employed by far. One application for cir-
cular polarization is in communications between earth and space, to mitigate the effects
of polarization rotation caused by the ionosphere (see sec. 7.3.1).
1.1.5. Rays and Wavefronts

Because the detailed structure of an electromagnetic wave is invisible, its nature can be
determined only by indirect methods. Diagrams such as Fig. 1–1 are not truly pictorial;
they are purely schematic, man-conceived schemes of representing certain aspects of the
waves, namely, the magnitude variations of the E and H components. Another such
scheme utilizes the concept of rays and wavefronts as an aid in illustrating the effect of
variations in the propagation medium (including discontinuities) on the propagation of
the waves.
A ray is a line drawn along the direction of propagation of a wave. The z-axis in
Fig. 1–1 is an example of a ray. Any line drawn parallel to the z-axis in this diagram is
also a ray, since the wave is plane and has the same direction everywhere. Therefore, if
the wave is plane, there is no point in drawing more than one ray, for they are all
alike.
A wavefront is a surface of constant phase of the wave. As mentioned in connection

with Fig. 1–1, such surfaces are planes perpendicular to the direction of propagation
when the wave is plane. As also mentioned, not all waves are plane. In fact, in the vicin-
ity of the source from which waves are emanating (an antenna, for example), rather
complicated wavefronts may exist. Of particular importance, and only slightly more
complicated than a plane wave, is the spherical wave. Any “point” source of waves in
free space will generate a spherical wave, as is readily deduced from the fact that if a
certain part of the wave travels outward from a point, at the same speed in all directions,
it will, after traveling a distance R, define the surface of a sphere of radius R, with its
center at the point of origin of the waves.
If the distance from a source of electromagnetic waves is sufficiently large, compared
to the physical size of the source, the source may be considered equivalent to a point
source. Then, the wavefronts will be spherical. The system of rays and wavefronts gener-
ated by a point source is shown in Fig. 1–2.
It is apparent that the wavefronts here are spherical (appearing as circles in this two-
dimensional drawing) and that all the rays are diverging from the common center or source.
But if a small portion of a spherical wave, at a great distance from its source, is considered,
this small portion will be approximately plane. For example, consider a cubic region of
space, shown dashed in Fig. 1–2 near the midportion of the arc denoted wavefront D. This
is a spherical wavefront. Within the dashed region, however, the small portion of the wave-
front can hardly be distinguished from the plane surface of the cube to which it is tangent.
Moreover, all the ray lines inside this cube are approximately parallel.
1.1.6. Spherical Waves and the Inverse-Square Law

One of the fundamental laws of physics is the Law of Conservation of Energy. An elec-
tromagnetic wave represents a flow of energy in the direction of propagation. The rate
at which energy flows through a unit area of surface in space (energy per unit time per
unit of area) is called the power density of the wave, usually expressed in watts per square
meter. The principle of energy conservation can be applied to a uniform spherical wave
in the following terms, with reference to Fig. 1–2. If the source radiates power at a con-
stant rate uniformly in all directions, the total power flowing through any spherical
surface centered at the source will be uniformly distributed over the surface and must
equal the total power radiated. Such a source is called an isotropic radiator, or
isotrope.
In Fig. 1–2 wavefront B, for example, constitutes a spherical surface. Although only
a portion of it is shown, the complete sphere may be visualized as surrounding the source.
If wavefront B is at a distance RB meters from the source, the total surface area of this
sphere is, from elementary geometry, 4pRB2 square meters. If the source is radiating a
total power Pt watts, since this total power is by hypothesis distributed uniformly over
the spherical surface at distance RB, the power density pB must be
Pt
pB = watts per square meter (1–4)
4π RB2
Wavefront D
Wavefront C
Wavefront B
RC
RB RD
Wavefront A Cubic region

of space
(d << RD)
RA
Source
Rays
d
Circular arc
tangent to
cube face
FIGURE 1–2.
Point-source wavefronts and rays in free space.
[P will denote total power (watts), and lower case p will denote the power density of the
wave (watts per square meter).]
By similar reasoning the power density pC at the greater distance of wavefront C
will be
Pt
pC = watts per square meter (1–5)
4π RC2
This value is obviously smaller than the power density at wavefront B, since RC is greater
than RB. Thus the power density decreases as the distance from the source increases.
What is the law of this decrease? It may be found by dividing (1–4) by (1–5),
2
pB  RC  (1–6)
= 
pC  RB 
which shows that the power density is inversely proportional to the square of the distance
from the source. This is the celebrated inverse-square law of radiation, observed experi-
mentally for electromagnetic waves in free space or in limited regions whose character-
istics approximate the uniformity of free space.
In deriving this result it was assumed that the source radiates isotropically—uniformly
in all directions. Actually this is not a necessary assumption, because the same result is
obtained if the source radiates preferentially in certain directions, as occurs with direc-
tional antennas. It is necessary, however, to assume that the velocity of electromagnetic
propagation is the same in all outward directions from the source. This assumption
permits the distance to the wavefront from the source at any instant to be equal in all
directions, corresponding to the geometrical definition of a sphere.
A propagation medium is called isotropic if the propagation velocity is the same in
all directions. The inverse-square law, therefore, is the result both of the spherical spread-
ing of the wavefronts in an isotropic propagation medium and of the law of conservation
of energy.
The inverse-square law is based on the distance to the wavefront from a point source
at any instant being equal in all directions, corresponding to the geometrical definition
of a sphere. However, this is true only within the far field of an antenna, because an
antenna is of finite size and therefore it is not located solely at a point. The far field exists
beyond a minimum separation distance of an antenna, which depends on antenna dimen-
sions and wavelength. (sec. 3.2.5).
1.1.7. Field Intensity and Power Density

The power density of the field is related to the values of the electric and magnetic inten-
sities in the same way that power in an electric circuit is related to voltage and current;
it is the product of the two. (This assumes the free-space relationship of the field vectors
depicted in Fig. 1–1.) The product of the instantaneous values gives the instantaneous
power, but this quantity is usually of little interest. The average power density over an
rf cycle is ordinarily desired, and, just as in computing power in a-c circuits, it is obtained
by multiplying the effective values of E and H, equal to 1 2 times the amplitudes, or
0.707E0 and 0.707H0. Hence, the average power density p may be expressed as (1–7)
that follows:
p = (0.707 E0 ) × (0.707 H 0 ) = 0.5E0 H 0 (1–7)
where E0 and H0 are the amplitudes as in (1–2) and (1–3). E0 is expressed in volts per
meter, H0 in amperes per meter to give p in watts per square meter.
Just as voltage and current in circuits are related through the resistance by Ohm’s law,
the electric and magnetic intensities are related by the characteristic wave impedance of
space. In a lossless propagation medium this impedance is equal to the square root of
the ratio of its magnetic permeability m to its electric permittivity e, as given by (1–8)
that follows:
Z s = μ / ε ohms (1–8)
In a vacuum m has the value 1.26 × 10−6 henrys per meter, and e is 8.85 × 10−12 farad
per meter. (These values are customarily denoted m0 and e0.) Consequently Zs is about
377 ohms (actually 120p ohms) in free space, a value also applicable in air. Hence in
these media
E2
p= = 377 H 2 watts per square meter (1–9)
377
where E and H are the effective (rms) values, equal to 0.707E0 and 0.707H0, in volts per
meter and amperes per meter, respectively. This also means that the magnetic intensity
can be expressed by (1–10) that follows:
E
H= amperes per meter (1–10)
377
for any wave propagating in free space or air; that is, E and H are related through this
expression, and specifying one of them is equivalent to specifying both. Ordinarily,
therefore, only the electric intensity is specified.
If (1–9) is applied to the inverse-square law, the result is
E B RC
= (1–11)
EC RB
which states that the electric intensity is inversely proportional to the first power of the
distance from the source (subject to the same stipulations that apply to the inverse-square
law in its original form).
Equations (1–6) and (1–11) are different ways of showing how the electromagnetic
wave is attenuated with increasing distance from the source. Equation (1–6) expresses
the attenuation in terms of the power-density ratio, and (1–11) expresses attenuation in
terms of the electric-intensity ratio.
1.1.8. Decibel (Logarithmic) Expression of Attenuation

Wave attenuation is expressed also in terms of the logarithms of the power-density or
electric-intensity ratios, an alternative method widely used in describing signal amplifica-
tion or attenuation in telephone and radio systems. The decibel attenuation is simply ten
times the common (base 10) logarithm of the power-density ratio or twenty times the
logarithm of the electric-intensity ratio. It may be readily verified that these definitions
lead to the same decibel values for a given distance ratio RC/RB when applied to (1–6)
and (1–11).
Example. Suppose that the field strengths are to be compared at two distances from a source,
R1 and R2, such that R2 = 2R1. How much is the field strength reduced if the distance from the
source is doubled? From equations (1–6) and (1–11) the question is answered as follows:
2 2
p1  R2   2R 
=   =  1 = 4
p2  R1   R1 
and
E1 R2
= =2
E2 R1
In terms of decibels the attenuation is
p1
10 log = 10 log 4 = 10 (0.60206 ) = 6 dB
p2
or
E1
20 log = 20 log 2 = 20 (0.30103) = 6 dB
E2
Thus the same result is obtained for the attenuation in decibels from either the power-density
or the electric-intensity ratio.
Attenuation due to the spherical spreading of the wave—that is, as expressed by
the inverse-square law—is sometimes called the space attenuation of the wave. As
can be shown from (1–6) or (1–11) by applying the definition of the decibel, the space
attenuation in decibels is twenty times the logarithm of the distance ratio. As already
shown, if the distance is doubled, the wave is attenuated by 6 dB (20 log 2); if it is tripled,
the attenuation is 20 log 3, or 9.5 dB; if the distance is increased by a factor of 10, the
attenuation is 20 dB; and so on.
Since it is the distance ratio rather than the actual distance change that determines the
space attenuation, a given distance change has a greater effect at points close to a source
than it has far from the source. For example, two points differing by 1 km in distance
from the source will show a 6-dB power-density ratio if they are 1 km and 2 km from
the source, respectively; but if they are 100 km and 101 km from the source, the attenu-
ation is only 0.09 dB, a negligible amount. This fact further supports the validity of
regarding a spherical wave within a limited region of space at a considerable distance
from the source, virtually as a plane wave; for it is a property of a plane wave that the
power density and electric intensity do not change as the wave progresses.
In some antenna considerations, the ratios of the field intensities at different points in
the vicinity of the antenna may be fairly large for relatively short separations of the points
in question. However, at short distances from an antenna the fields may not obey the
inverse-square law power relationship. Therefore, it is to be underscored that the inverse-
square law relationship is applicable only to distances from an antenna that are large
enough to be within the antenna far field (sec. 3.2.5).
1.1.9. Absorption
In addition to space attenuation, which is always present (even though it may be ignored
over short distances far from the source of the waves), there may sometimes be attenua-
tion due to absorption of power by the propagation medium. This does not occur in a
vacuum, but it will occur in a medium that contains material particles that interact with
the waves. At some frequencies, for example, certain gases of the earth’s atmosphere
(oxygen and water vapor) cause absorption. This occurs slightly in the VHF region and
becomes significant over long transmission paths in the UHF region and above. Unlike
space attenuation, attenuation due to absorption does not depend on the distance from
the source but only on the total distance traveled by the wave. That is to say, attenuation
due to absorption is the same over a one-kilometer path whether the path is 1 km or
100 km from the source of the waves.
Absorption in the atmosphere is significant only over appreciable distances, measured
in kilometers, and so it is not a factor in antenna design, except in the very indirect sense
that it may dictate what frequency or frequencies are chosen for a particular application.
The earth’s ionosphere also absorbs waves at some frequencies; this absorption may be
significant in the high HF and low VHF regions. At frequencies above the low UHF the
ionosphere is completely transparent to radio waves ordinarily, and below the HF region
it acts like a reflecting barrier; that is, low-frequency waves do not penetrate the iono-
sphere appreciably.
Certain materials are capable of absorbing radio waves very strongly. Waves traveling
in these materials will be attenuated greatly within a short distance, of the order of cen-
timeters or meters. Sometimes such materials are used in antenna design to suppress
radiation in undesired directions, or to prevent “leakage” of waves from one part of an

antenna to another where they would have an undesirable effect.
Now let us assume a wave propagates in a homogeneous absorbing medium, that is,
one that does not vary from point to point. Then, if the wave propagates in the x-
direction, its amplitude will decrease as e−ax, where e = 2.71828. The term e−ax is the
attenuation factor, and a is the attenuation constant expressed in neper per meter (Np/m).
Thus, if a wave of amplitude E1 travels a distance x of 1 m and a is 1 Np/m, its new
amplitude E2 is E1e−1. Then, the attenuation in decibels is 20 log (E1/E2) = 8.686, and
thus an attenuation of 1 Np/m equals 8.868 dB/m. In other words, if g is the attenuation
factor expressed in dB/m instead of Np/m as it is above for a, g = 8.868 a. Similarly, in
(1–12) attenuation in decibels AdB is related to attenuation in nepers ANP as
AdB = 8.686ANp (1–12)
Nowadays attenuation is commonly expressed in terms of decibels, not nepers.

Attenuation by absorption is now further explained. As before, let g be the attenuation
constant expressed in decibels per meter and R1 and R2 be two distances from the source
of electromagnetic waves. Then, the total absorptive attenuation of the waves in traveling
from R1 to R2 is g(R2 − R1) dB. This implies that the law of absorptive power loss is
p1
= 100.1γ ( R2 − R1 ) (1–13)
p2
The factor 0.1 in the exponent is there because decibels are ten times the logarithm of
the power ratio. Note that the mathematical form of (1–13) for attenuation due to absorp-
tion is different from (1–6), which is for space attenuation.
The total decibel attenuation AdB over this path is the sum of the space attenuation
and the absorption attenuation, which is
 R  
AdB = 20 log  2  + γ ( R2 − R1 ) dB (1–14)
  R1  
Thus the space attenuation depends on the ratio R2/R1, whereas the absorption attenuation
depends on the difference R2 − R1. Therefore close to the source the space attenuation
predominates, but at large distances from the source the absorption becomes more impor-
tant, if g has a nonzero constant value.
In the earth’s atmosphere (troposphere and ionosphere), however, the situation is
somewhat more complicated because g does vary from point to point; it is a function of
altitude (air density) in the atmosphere, and in the ionosphere it is a function of the
electron density. In such cases the absorption attenuation of a wave over the propagation
path from R1 to R2 is expressed mathematically by the following integral:
R2
AdB = ∫ γ ( R ) dR dB (1–15)
R1
Radio-Wave Optical Principles 15
The evaluation of the variation of g in the atmosphere is a complicated problem. Curves

showing the results of calculations for typical paths in a standard troposphere have been
published by Blake (1962).
1.2. Radio-Wave Optical Principles

So far the discussion has applied principally to electromagnetic waves in free space. The
absorption effect is an exception, but it represents a relatively minor departure from “free
space” because it does not alter the ray-wavefront behavior. The rays and wavefronts are
the same in a uniformly absorbing (homogeneous, isotropic) medium as they are in free
space.
In the environment of the earth, or in any “space” that is not empty, effects may occur
that do alter the ray-wavefront behavior. Then Fig. 1–2 no longer represents the actual
behavior. The effects that occur are referred to as optical effects because they were first
observed and studied in connection with the science of optics—the behavior of light
waves. It is, of course, not surprising that these optical phenomena also apply to radio-
wave propagation, for light waves and radio waves are both electromagnetic waves, dif-
fering only in frequency (hence also in wavelength). It is also not surprising that some
of these effects are most strongly observed at the shorter radio wavelengths (higher fre-
quencies); however, they are generally important at all frequencies.
These effects are refraction, reflection, interference, and diffraction. The first two can
be explained basically in terms of rays and wavefronts, the principles of geometric optics.
The second two, however, can be explained only in terms of physical optics, which
requires a more detailed consideration of the electric and magnetic fields. Geometric
optics may be applied when the dimensions of all the significant elements of the situation
are large compared with the wavelength and when no wavelength-dependent effects are
involved; otherwise the problem comes within the purview of physical optics or, more
generally, electromagnetic theory. The significance of these statements will become
clearer in the ensuing discussion.
1.2.1. Refraction
When a wave passes from one region to another in which the wave speed is slower or
faster, refraction occurs. Refraction takes place either when the two propagation regions
are separated by a sharp boundary or when the wave velocity varies gradually and
approximately linearly over a region that is large compared with the wavelength. Refrac-
tion at a plane boundary is illustrated in Fig. 1–3. It is to be noted that reflection also
occurs at the boundary, and it is discussed in the section 1.2.2 that follows.
As shown, a plane wavefront AB in medium I is directed toward the boundary surface.
The direction is defined by the angle of incidence q1, which the rays make with the
normal (perpendicular) to the surface. This wavefront is shown at the exact instant that
its lower edge A has reached the boundary surface.
Original ray direction It is assumed that the wave speed v2 in

NORMAL medium II is slower than the speed v1 in
medium I. The wave proceeding down-
ward from point A will now be traveling
B
Original wavefront in medium II at the slower speed v2,
q1
MEDIUM I
whereas that portion of the wavefront at
B´
B is still traveling in medium I at the
A Refracted wavefront
faster speed v1. Consequently, during the
A´
time that it takes the wave to go from B
MEDIUM II
to B′ in medium I the lower edge of the
q2 wave will go only from A to A′ in medium
II, the distances BB′ and AA′ being in the
New ray direction same ratio as the speeds v1 and v2. Con-
FIGURE 1–3. sequently the wavefront is defined by the
Refraction at a plane boundary between two
line A′B′, which is tilted with respect to
media. AB. (This is the wavefront because it
defines a surface of constant phase of the
wave.)
Since rays are defined as being perpendicular to the wavefront at all points, the rays
have now changed their directions. The direction of the wave has also been changed by
the amount q1 − q2 to a direction defined by the angle q2, the angle of refraction.
By considering triangles ABB′ and B′A′A it can readily be shown, since the distances
BB′ and AA′ have the same ratio as v1 has to v2, that the angles q1 and q2 are related by
the expression
sin θ1 v1
= (1–16)
sin θ 2 v2
This ratio is also commonly expressed in terms of the refractive indexes of the two media.
The refractive index is defined as the ratio of the speed of the wave in vacuum, c = 3 × 108
meters per second, to the speed v in the actual medium. Thus the refractive index of medium
I is n1 = c/v1 and that of medium II is n2 = c/v2. Consequently,
sin θ1 n2
= (1–17)
sin θ 2 n1
In an electrically nonconducting medium, the wave velocity is given by
1
v= (1–18)
µε
where m is the magnetic permeability and e is the electric permittivity. The index of
refraction (i.e., refractive index), therefore, is given by
c με
n= = (1–19)
v μ0 ε 0
where m0 and e0 are the permeability and permittivity of a vacuum. These values are
m0 = 1.26 × 10−6 henry per meter and e0 = 8.85 × 10−12 farad per meter. Then, by using
(1–18) and (1–19), and the numerical values of m0, and e0, c is attained as follows:
1
c= 3 × 108 meters per second
(1.26 × 10 ) (8.85 × 10 )
−6 −12
as previously stated. The permeability of most common dielectrics is the same as that of
a vacuum (i.e., m = m0). Thus, for most lossless dielectrics, the index of refraction n is
the ratio ε ε 0 , where e/e0 is the relative permittivity er. Therefore, from (1–16), one
may usually write
sin θ1 ε
= r2 (1–20)
sin θ 2 ε r1
To account for the conduction current and ohmic losses in a nonideal dielectric, er is
expressed as e′r − je″r, where e′r is the relative permittivity related to the displacement
current (Appendix A) and e″r is the relative permittivity related to the conduction current.
The reader may recognize that er is a complex number (for a review, see Appendix C).
Another name for e′r is dielectric constant, and for low-loss dielectrics (where e″r << e′r )
the dielectric constant (i.e., e′r ) is ordinarily assumed equal to er. Thus, (1–20) is usually
a valid approximation for most dielectrics used with antennas, but it is actually exact
only if er2/er1 is replaced by e′r2/e′r1.
Dielectric constants of typical substances used at radio frequencies range from 1 to
about 10, although a few special materials with higher values are available. Generally
the low-loss materials (low absorption coefficients) tend also to have low dielectric con-
stants. The value er = e′r − je″r = 1 applies to air (approximately) and vacuum (exactly).
When the angle of incidence of a wave is zero (q1 = 0), it is apparent from (1–16) to
(1–18) that the angle of refraction must also be zero (q2 = 0), regardless of the indexes
of refraction. That is, an incident plane wave whose rays are normal to the plane surface
between two propagation media will not be refracted (meaning that its direction of
propagation will not be changed; however, its forward speed will of course be
changed).
In Fig. 1–3 it is assumed that the wave goes from a medium of higher speed into one
of slower speed. If the order of speeds is reversed, the behavior of the rays and wavefronts
is reversed. That is, the directions of the rays in Fig. 1–3 may be reversed and everything
else will remain unchanged, except that q2 becomes the angle of incidence and q1 the
angle of refraction.
Refraction will also occur when a curved (e.g., spherical) wavefront is incident on a
plane interface between two media, when a plane wavefront is incident on a curved
A Original boundary surface, and when curved wave-

ray direction fronts impinge on curved boundaries. The
A¢ New ray
Original analysis of these cases, of course, is more
wavefront direction
complicated, but the principles are the
same. In general, the curvature of the
New
B wavefront wavefront is changed by refraction. This
B¢
is the basis of the design of lenses, both
as used with light waves in telescopes,
FIGURE 1–4. microscopes, spectacles, cameras, and so
Refraction in a medium having gradual forth, and also as used with radio waves
variation of the refractive index. in some types of antenna at very short
wavelengths.
Refraction may also occur when a wave travels in a medium that has a velocity gradi-
ent in the direction perpendicular to the direction of propagation (parallel to the wave-
front); that is, when the index of refraction increases or decreases in a gradual manner
rather than abruptly. Then the rays are bent gradually toward the direction of lower wave
velocity (higher index of refraction), a result that may be deduced by reasoning similar
to that employed for refraction at a plane boundary. In Fig. 1–4 a plane wavefront AB is
traveling horizontally in a medium with a vertical index gradient. The wave velocity is
assumed to be faster in the upper part of the medium than in the lower part; that is, the
refractive index n is greater in the lower part of the medium. In the simplest case the
variation is assumed to be linear. The upper part of the wavefront will travel the distance
AA′ in the same length of time that it takes the lower part to travel the shorter distance
BB′, so that the new wavefront A′B′ is tilted downward. This tilting occurs in a gradual
fashion as the wave progresses, as indicated in Fig. 1–4 by the intermediate wavefronts
shown by dashed lines.
This effect actually occurs for waves traveling approximately horizontally in the
earth’s atmosphere, because the refractive index of air is very slightly greater in the
denser lower atmosphere than it is at very high altitudes, the variation being approxi-
mately linear over a restricted range of altitude. Because the refractive index of the lower
atmosphere is only about 1.0003, compared to 1.0000 for a vacuum, the downward
bending of the rays is very slight indeed, so small as to be unobservable over short
propagation paths. It is significant over very long distances only. A similar type of
bending occurs at some frequencies in the ionosphere, but there the ray curvature is in
general more complicated because of the more complicated behavior of the refractive
index variation.
1.2.2. Reflection
In the discussion of refraction at a plane boundary (Fig. 1–3) it was tacitly implied that
the wave is totally refracted—that all the wave power incident on the surface in medium
I passes into medium II. Actually this is not true. Some of the power is reflected. Reflec-
tion refers to the formation of an additional new wavefront that travels upward from the
surface into medium I so that in general the incident wave is split up into two waves,
one refracted (transmitted through the Incident Reflected

boundary surface) and one reflected. wavefront wavefront
Figure 1–5 represents this situation. The Incident Reflected
ray ray
reflected wave may be thought of, perhaps
rather unscientifically, as one that bounces
off the surface. The reflection of light q1 q3
MEDIUM I
waves (as from a mirror) is a familiar
phenomenon that everyone has observed. MEDIUM II
q2
Radio waves are similarly reflected.
Refracted
Because the reflected wave is traveling at wavefront
the same speed as the incident wave, it Refracted
may be deduced that the angle of reflec- ray
tion q3 is exactly equal to the angle of FIGURE 1–5.

incidence q1. Reflection and refraction at a plane boundary
So far only the relationships between between two propagation media.
the relative directions of the incident,
refracted, and reflected wavefronts have
been considered. Equally important is determination of the relative amounts of the total
power that are reflected and refracted (transmitted). Detailed analysis of this problem is
quite complicated and requires a knowledge of advanced electromagnetic theory. Only
a brief outline is given here.
The ratio of the reflected-wave electric intensity to that of the incident wave is called
the reflection coefficient G (gamma). This coefficient expresses not only the relative
amplitudes but also the phase shift that may occur in the reflection process; that is, G is
a complex number (complex numbers reviewed in Appendix C):
Γ = Γ e jφr (1–21)
where fr is the phase angle of the reflection coefficient. If the incident and reflected
electric intensities are E1 and E3, with individual phase angles f1 and f3, then G is found
as follows:
E3 E3 e jφ3 E
Γ= = jφ1
= 3 e j (φ3 −φ1 ) (1–22)
E1 E1 e E1
Thus |G | in (1–21) is given in (1–22) by the ratio of the amplitudes of the waves, and
fr in (1–21) is given in (1–22) by the difference of their phase angles.
The ratio of the power densities of the reflected and incident waves is given by |G |2,
since the power density is proportional to the square of the electric intensity (1–9).
The fraction of the total incident power transmitted through the surface (refracted) is
expressed by the power transmission coefficient T (usually called the transmission coef-
ficient). Since the Law of Conservation of Energy requires that the total of the reflected
and transmitted power shall equal the incident power, it is readily deduced that
G 2+T =1 (1–23)
Note that T does not allow computing the electric intensity or power density of the
transmitted wave. It expressed a relationship between the total power incident on and
transmitted through the surface, just as |G |2 expresses a similar relationship for the inci-
dent and reflected power.
When the conductivity of medium II (Fig. 1–5) is very high, most of the incident wave
is reflected and very little is transmitted. In fact, for a perfect conductor |G | = 1 and T =
0. For nonconducting dielectric materials both G and T are functions of the angle of
incidence of the waves, and of the polarization, as well as of the dielectric constants of
the two media.
If medium II has some conductivity but is not an excellent conductor, waves penetrat-
ing the surface are absorbed, that is, they set up currents that are converted to heat in
the resistance of the material. In this case it is customary to express the fraction of power
that penetrates the surface in terms of an absorption coefficient, A, rather than a transmis-
sion coefficient. Equation (1–23) holds for this situation, with T replaced by A.
When the reflecting surface is curved rather than plane, the curvature of the reflected
wave is different from that of the incident wave; however, when a curved wavefront is
reflected from a plane surface, the curvature of the reflected wave is the same as that of
the incident wave. Plane and curved metallic reflecting surfaces have important applica-
tions in the design of high-frequency antennas.
Reflection may also occur from a surface that is irregular or rough rather than smooth.
Such a surface may destroy the shape of the wavefront. The reflected wave in this case
is scattered in a random fashion, so that the reflected amplitude or power density in a
given direction is not exactly predictable as it is for smooth surfaces. This phenomenon
is called diffuse reflection, whereas reflection from a smooth surface is said to be specular
(mirror-like).
Diffuse reflection is not characterized by a reflection coefficient containing a phase
angle, as is specular reflection. The random nature of diffuse reflection results in a phase
angle that varies unpredictably at different regions of the surface. Diffuse reflection,
therefore, is expressible only in terms of a power reflection coefficient, Rd.
Surfaces may also be semi-rough. For such surfaces the degree of roughness is not
sufficient to destroy the shape of the reflected wavefront completely; there is a mixture
of diffuse reflection and specular reflection. In such cases a total power reflection coef-
ficient, R, may be expressed in terms of the diffuse-power-reflection coefficient, Rd, and
the specular-reflection coefficient |G |. Since the total reflected power must be the sum
of the diffuse and specular components, the relation for R is
R = Rd + |G |2 (1–24)
When the incident wave is reflected partly specularly and partly diffusely and is also
partly transmitted, R may be substituted for |G |2 in (1–23).
For a given degree of actual roughness of a surface the values of Rd and |G | vary with
the angle of incidence qi of the wave before reflection. At normal incidence (qi = 0°),
Rd has its maximum value and |G | its Incident Specularly

wavefront reflected
minimum value. As qi increases, Rd wavefront
decreases and |G | increases, and for very
large values of qi (near 90°) the semi-
qi
rough surface may reflect almost specu-
larly, that is, with Rd = 0. This effect is d
described by the Rayleigh criterion, which Diffuse

Plane of reflection
states that a semi-rough surface will partial specular
reflect as if it were a smooth surface (i.e., reflection
specularly) when qi is the incidence FIGURE 1–6.
angle Reflection from a semi-rough surface.
λ
d< (1–25) Spherical
8cos θ i wavefronts,
apparent
source S2 P2
where l is the wavelength of the incident Spherical
wave and d is the depth of the surface wavefront,
irregularities, as shown in Fig. 1–6. Here source S1
the specular component of the reflected S1
qi
wave appears to be reflected from a plane h1 qi qr
intermediate between the highest and
lowest portions of the surface, whereas h2 P1
Reflecting surface
the diffuse reflection is indicated by ray qr
lines going off in all directions from each S2 qi = qr
Image wavefront,
point of the surface. from S2 h2 = h1
S1P1 = S2P1
FIGURE 1–7.
1.2.3. Principle of Images The principle of images.
When a spherical wavefront is reflected
from a plane surface, it retains its spherical shape, and its direction is changed as in a
plane wave. The exact behavior of the reflected wavefronts may be predicted by the
principle of images, as illustrated in Fig. 1–7. In that figure, the reflected wavefront
appears to be originating at S2, although it actually originates at S1 and is specularly
reflected from the plane surface. The virtual source S2 is located a distance h2 below
the reflecting surface equal to the actual height h1 of the true source S1 above the
surface and on a line from S1 perpendicular to the surface. Therefore the distance from
the actual source to a point P2, which is S1P1 + P1P2, is the same as the distance
from the virtual source S2P1 + P1P2. The virtual source S2 is called the image of the
actual source S1. This is the familiar optical principle of the mirror, according to which
an observer at P2, looking along the line P2P1, sees an image of an object located at
S1. The object appears to be at S2, although the light waves are actually coming
from S1; they are reflected from the surface in such a way that they seem to be coming
from S2.
1.2.4. Interference
So far the optical principles considered have been in the category of geometric optics,
which means essentially that the discussion has centered on the behavior of rays and
wavefronts. Additional principles are necessary for problems in the realm of physical
optics.
Electromagnetic waves in free space, and in many other media, are subject to the
important principle of linear superposition on the basis of which interference effects can
be analyzed. Interference occurs when two or more electromagnetic waves exist simul-
taneously at the same point in space. The principle of linear superposition states that the
total electric and magnetic intensities at the point are the complex (amplitude and phase)
vector sums of the individual complex wave vectors.
This may seem to be a simple and natural fact. However, linear superposition does
not always hold in every physical situation, and sometimes it does not hold for electro-
magnetic waves. Certain types of propagation media (not commonly encountered, fortu-
nately) have nonlinear properties, so that a field of 2 volts per meter superimposed on a
field of 3 volts per meter (parallel vectors and instantaneous values) does not result in 5
volts per meter. But in ordinary media, including air and vacuum, the principle of linear
superposition does hold. (Even for the medium of air an exception may be noted, when
the voltage exceeds the value at which “breakdown” occurs.)
When the superposition principle does apply, it applies to the instantaneous values of
vectors, added vectorially. The principle of vector addition is illustrated in Fig. 1–8. (The
reader is assumed to have encountered this principle previously, and it is illustrated here
primarily for reference purposes.)
The vectors to be added are E1 and E2, which may be thought of as instantaneous
values of electric intensity at a point in space, described by equations of the type of (1–2).
Their directions in space differ by the angle a, as shown. (Their directions may differ
this much because their polarizations differ by this amount, for example; or one wave-
front may have been tilted the amount a with respect to the other by a refractive or
reflective process.)
The magnitude of the sum of the two vectors E1 and E2 is
E3 = E12 + E22 + 2 E1E2 cos α (1–26)
In this equation the letters E1, E2, E3 denote the magnitudes (lengths) of the vectors E1,
E2, E3. The equation gives no information
concerning the direction of E3, which E1
may be determined from Fig. 1–8 by mea-

E2 a
suring the angles g (gamma) and d (delta). x
E
d 3
It may also be determined by trigonomet- a x E2
ric analysis of Fig. 1–8, which yields g
E1
E2 sin α FIGURE 1–8.
tan γ = (1–27)
E1 + E2 cos α Addition of vectors.
Some special cases are worth noting. Referring to Fig. 1–8, if E1 = E2, it may be seen
that g = a/2 = d. In addition, if E1 = E2, and by using a standard trigonometric formula,
(1–26) becomes
α
E3 = E1 2 + 2 cos α = 2 E1 cos   (1–28)
 2
The specialization a = 0 results in E3 = E1 + E2, and specialization a = 180° gives

E3 = E1 − E2, corresponding to the vector addition of parallel and antiparallel vectors.
This type of complex vector addition of two electric intensities solves the superposi-
tion (interference) problem in which E1 and E2 have the same frequency and the same
time phase but different directions in space at the point under consideration. “The same
time phase” means that the sinusoidal rf variations are perfectly synchronized, with their
maxima and zero values occurring at exactly the same instants. As noted, the case in
which the two intensity variations are exactly opposite in phase can also be handled by
this kind of analysis, because two “parallel” vectors of opposite time phase correspond
to the antiparallel case a = 180°. If the time-phase difference is other than 0 or 180°, a
more complicated analysis is required.
Complex vector addition of two vectors can be made fairly simply, if the spatial direc-
tions of the vectors are the same (a = 0). This is so even if the time-phase difference is
other than 0 or 180°. Now suppose, for example, that the two electric intensities are
2π r
E1 (r , t ) = E01 sin  2π ft − + φ1  (1–29)
 λ 
2π s
E2 ( s, t ) = E02 sin  2π ft − + φ2  (1–30)
 λ 
in accordance with (1–2), except that the letters r and s are used here for distances mea-
sured along the propagation paths. This implies that the two waves may have traveled
different distances before arriving at the point under consideration from the reference
points at which r = 0 and s = 0 and at which the phase angles were f1 and f2. Moreover,
the amplitudes of the two interfering waves E01 and E02 are not necessarily the same.
The interference problem thus described has numerous antenna applications.
To avoid unnecessary complication of the equations in further discussion of this
problem, the following standard abbreviations are used:
2π f = ω (1–31)
2π
=β (1–32)
λ
The quantity w (omega) is known as the angular frequency, and b (beta) is the phase
constant; w expresses the angular rate (in radians) at which the phase angle of the wave
at a fixed point in space changes in time, and b is the rate of phase change with distance
at a fixed instant of time. The customary units for w are radians per second, and for b,
radians per meter; that is, t and l are ordinarily measured in these units. Thus a time
period T = 1/f corresponds to 2p radians, or 360° of phase angle, as does also a distance
x = l. (T = 1 cycle and l = 1 wavelength.)
If this notation is used and the principle of superposition is applied, it is apparent that
when E1 and E2 are parallel vectors at the same point in space the resultant intensity is
E3(t ) = E1(r , t ) + E2 (s, t ) = E01 sin (ω t − β r + φ1 ) + E02 sin (ω t − β s + φ2 ) (1–33)
No distance variable is indicated for E3, since it expresses the intensity at a single point
only. Therefore E3 is a sinusoidally varying electric intensity with an amplitude and phase
angle
E3(t ) = E03 s in(ω t + φ3 ) (1–34)
By algebraic manipulation it is found from (1–33) and (1–34) that
E03 = E01
2
+ E02
2
+ 2 E01E02 cos [φ1 − φ2 − β (r − s )] (1–35)
and
E01 sin (φ1 − βr ) + E02 sin (φ2 − β s )

tan φ3 = (1–36)
E01 cos (φ1 − βr ) + E02 cos (φ2 − β s )
Although these equations appear complicated, they are actually simple in the sense
that if numerical values of all the quantities on the right-hand sides are known numerical
values of E03 and f3 are easily calculated.
It is easily shown that (1–35) and (1–36) are essentially the same as (1–26) and (1–27),
if a is equated to [f1 + f2 − b(r − s)] and if the angle (f1 − br) is arbitrarily taken to be
zero, which is permissible, for it amounts only to choosing a particular origin for the
coordinate system. Therefore, as is well known in a-c circuit theory, the same geometric
construction (Fig. 1–8) can also be applied in this case.
When E1 and E2 are different both in direction and phase (other than 180°) the case
is more complicated and is not treated in detail here. The resultant electric intensity is a
space vector that undergoes an elliptical variation in its direction and amplitude in the
general fashion described for an elliptically polarized wave (see Appendix B), except
that the plane of the ellipse is not perpendicular to the direction of propagation unless
E1 and E2 are propagating in the same direction.
The space–time relationships that have been described are often confusing to
those who are encountering them for the first time. They become clear only after famil-
iarity has been gained by working problems and by thinking about them at some length.
Some helpful exercises are given at the end of the chapter. It is worthwhile to achieve
familiarity with these ideas, because the interference phenomenon is the basic principle
of most directional antennas, and it has other applications as well.
1.2.5. Huygens’ Principle: Diffraction

The discussion of refraction and reflection in terms of rays and wavefronts assumed that
the dimensions of the reflecting or refracting surfaces (or regions, in the case of gradual
refraction) were large in relation to a wavelength. When a wavefront encounters an
obstacle or discontinuity that is not large compared to the wavelength, the simple prin-
ciples of geometric optics cannot be used to analyze the result. The correct analysis is
obtained by application of Maxwell’s equations, which are the basis of electromagnetic
theory.
Huygens’ principle, deducible from Maxwell’s equations, is helpful in solving many
problems of this type. According to the principle, every point on a given wavefront may
be regarded as a radiating source, from which a spherical wavelet is propagated. At
further points in space the resulting field intensity may be found by superposition of all
of the fields due to these wavelets, taking into account their phases and the interference
effects that have just been discussed. This superposition, regarded as the sum of an infi-
nite number of wavelets, is expressed mathematically by an integral.
Diffraction is the process by which the direction of electromagnetic waves is changed
by an obstacle, impediment, or blockage and not attributable to reflection and/or refrac-
tion. Shadowing by a building or focusing by an antenna are examples. Diffraction may
be understood qualitatively with the aid of diagrams. In Fig. 1–9 it is supposed that a
plane wavefront has reached position AA′. Spherical “wavelets” are shown originating
from three arbitrarily chosen points on AA′. In actuality, of course, such wavelets are
assumed by Huygens’ principle to arise at every point on AA′. If now a new surface BB′
is drawn tangent to all of the wavelets with equal radii, the field along BB′ can be con-
sidered as the superposition of fields due to all the wavelets from AA′. For the situation
shown, an exact mathematical solution of the interference integral would show that the
field at a given point on BB′ due to all wavelets originating on AA′ would be exactly the
same as the field at the nearest point of AA′ with its phase retarded by 2pd/l radians,
where d is the distance between AA′ and BB′. Thus the waves appear to propagate along
straight ray paths perpendicular to the wavefronts, but this is actually because the inter-
ference of wavelets from all points on an existing wavefront produces this result. An
analogous result is obtained for the propagation of spherical wavefronts.
This analysis applies exactly only if the plane wavefront extends infinitely in all direc-
tions of the plane. It applies practically if AA′ is of large extent compared to a wavelength;
then the ray-wavefront construction can be applied without appealing directly to Huygens’
principle, although as indicated it is consistent with the principle.
Suppose, however, that the wavefront AA′ encounters an obstacle that has an edge. At
points beyond this obstacle it is apparent that the geometric-optics assumption is violated,
and the wavefront is no longer of infinite extent, a situation described in Fig. 1–10. The
“infinite” wavefront AA′ is propagating to the right and encounters a barrier, or obstacle,
which might be a sheet of metal, for example, or any other substance impenetrable to
A B electromagnetic waves. Therefore, when

the wavefront is passing the edge of the
obstacle, it extends in the upward direc-
tion only to the lower edge of the obstruc-
tion, at B. Beyond the obstacle (to the
right) a semi-infinite wavefront CC′ will
be formed. According to the ray-wave-
front concept, no electromagnetic field
would exist in the “shadow region” above
the dashed line BC. Huygens’ principle,
however, states that a wavelet originating,
for example, at point P on the wavefront
BB′ will propagate into the shadow region
(and so also will wavelets from all other
points on BB′). The existence or nonexis-
tence of a field in the shadow region will
then depend on the interference of all such
wavelets in this region.
A detailed mathematical application of
Huygens’ principle to this problem shows
that, in fact, the field is zero at some points
in the shadow region and nonzero at other
points. This diffraction effect, observable
with light waves under laboratory condi-
tions, is also observable with radio waves
under suitable conditions. It is an impor-
tant effect in analyzing the performance
of some types of antennas—those for
which it is desired to radiate a very narrow
“beam” of waves in a preferred direction
and to minimize radiation in all other
directions. Ideally, it would often be desir-
A' B' able to have absolutely no radiation in
FIGURE 1–9. undesired directions; but this result is
Huygens’ principle applied to propagation of never perfectly achievable because of the
plane waves. diffraction effect. Consequently, an under-
standing of the general principles of dif-
fraction is necessary for an intelligent approach to minimizing the diffraction pattern
(which results in the side lobes and back lobes of a directional antenna pattern).
1.3. Radiation and Reception

So far electromagnetic waves have been discussed as they propagate in space, and
very little has been said about how they originate. The process by which this occurs is
Radiation and Reception 27
A Obstacle
Shadow
region
B
C
A' B' C'

FIGURE 1–10.
Diffraction at the edge of an obstacle.
radiation. Radiation results when high-frequency electric currents flow under suitable
conditions. The detailed mathematical description of the relationship between electric
currents and their associated fields can be derived from Maxwell’s equations. These are
four partial differential equations (as usually formulated), and on them the whole struc-
ture of electromagnetic theory is based. The practical successes of this theory place it
among the most impressive of man’s scientific achievements.
Equations (1–2) and (1–3) are solutions of Maxwell’s equations for the simplest situ-
ation, that of a wave in free space at a great distance from the radiating source—in short,
a plane wave. Equations 1–2 and 1–3 give no information on how radiation occurs, but
Maxwell’s equations do give this information. They state that whenever electric current
flows, a magnetic field is set up in the surrounding space. Any variation of this magnetic
field will result in creation of an electric field. The magnetic field will vary, of course, if
the current varies; therefore, with alternating current the magnetic field will be continu-
ously varying and will therefore continuously generate an electric field. Because the two
fields always exist together—one cannot exist without the other unless it is nonvary-
ing—the combination is called the electromagnetic field. The time variations of the field
components will be related to the nature of the variations of the current; if it is sinusoidal,
the fields will vary sinusoidally at the same frequency.
Moreover, the varying electric field, according to Maxwell’s equations, generates a
magnetic field. This exchange between the two fields provides the mechanism by which
waves may be propagated in space at the “speed of light.” The fields are a form of energy,
and therefore electromagnetic wave propagation represents a transport of energy outward
from the radiating source. The energy is supplied by the electric power source that is the
origin of the current.
Not all of the field surrounding a current-carrying conductor results in propagation of
waves outward into space. Some of the energy of these fields is returned to the conduc-
tor; it is temporarily stored in the fields, which are related to reactive effects—that is,
inductive and capacitive effects. The total field consists of two components—the induc-
tion field and the radiation field. The induction field is confined to a fairly local region
near the conductors; the radiation field may propagate to great distances, although its
strength will decrease with distance, both by spherical spreading of the wavefront
(inverse-square law) and by absorption if there is any. This radiation process will be
described in further detail for the case of a dipole radiator, in sec. 4.1.
1.3.1. Requirements for Radiation

Under some conditions the induction field predominates, and there is little or no radia-
tion. For appreciable radiation to occur, certain requirements must be met. Maxwell’s
equations, or more accurately, conclusions derived from them, establish the conditions
necessary for radiation to occur.
A fundamental requirement is that the current shall be time varying. As might then be
supposed, the more rapid the variation, that is, the higher the frequency of an alternating
current, the greater the amount of radiation that will occur for a given amount of current.
It is not that radiation cannot occur at low frequencies. Instead, it is that conditions are
more favorable for radiation at higher frequencies than at lower ones.
There are also some requirements on the configuration of the current flow. As is well
known, at low frequencies current ordinarily flows in closed circuits, as indicated in Fig.
1–11. As shown, the current I in opposite sides of the circuit at an instant of time is in
opposite directions. Because of this, although each leg of the circuit might by itself result
in an appreciable radiated field, the radiation at distant points due to each leg separately
Radiation and Reception 29
will be exactly of opposite phase and will d1

cancel—that is, destructive interference
L
will occur with resultant zero field I
strength. Consequently, in effect, there
will be no radiation.
d2
This assertion tacitly assumes, however, I E C R I
that the physical separation of the oppo-

site legs of the circuit, denoted by d1 and I
d2, is much smaller than a wavelength at
FIGURE 1–11.
the frequency of the current and also that
the currents in opposite legs of the circuit Current in a closed-loop circuit.
are in phase but oppositely directed. If
either of these assumptions is not correct, the fields will not cancel completely; in fact,
it is readily seen that if phases and separations are just right the interference may be
constructive rather than destructive, and appreciable radiation may result. However, at
the low frequencies of a-c power circuits—60 Hz—and even at “audio” frequencies, the
dimensions of any ordinary circuit will be such that these assumptions are correct. A
wavelength in free space at 60 Hz, for example, is 5,000 km! Consequently, radiation is
negligible, and is neglected in low-frequency a-c circuit theory.
But at radio frequencies it is entirely feasible to make the distances d1 and d2 appre-
ciable compared with a wavelength, and it is also easily possible to produce phase shifts
of the currents in opposite legs of the circuit. Also, it is possible to produce current flow
in a “circuit” that is not (in the usual sense) a closed loop, that is, in a straight wire with
an “open end” or ends. Then appreciable radiation results, and ordinary a-c (low-
frequency) circuit theory is no longer applicable.
Finally, for radiation to occur, the conductor carrying the current must be reasonably
“in the clear,” that is, not enclosed by obstacles that are impenetrable to electromagnetic
waves. This is a fairly obvious requirement, but worth mentioning. It means, for example,
that radiation may be prevented, when desired, by placing circuits inside closed metallic
enclosures; this is the principle of shielding.
A solid conductor carrying current is not necessary for radiation, although most anten-
nas do employ metallic conductors. The element of electric current in conductors is the
electron; the current consists of motion of electrons. Electrons can also exist and move
in empty space (vacuum or near vacuum), as they do, for example, in vacuum tubes, in
the earth’s ionosphere, and in other regions of “outer space.” When such “free electrons”
undergo acceleration, radiation occurs, at a frequency depending on the acceleration.
Radiation of this type from electrons in galactic space is one source of radio “noise” at
high frequencies (and would be at low frequencies as well were it not for the shielding
effect of the earth’s ionosphere). Electrons accelerated in lightning discharges in the
atmosphere also emit radio waves that are heard as noise of the type called “static,” at
frequencies of several MHz and below. (The radio waves radiated by a single lightning
stroke cover a broad band of frequencies.)
Electromagnetic waves are radiated by individual electrons at frequencies in the range
of visible and ultraviolet light and X-rays, when they undergo violent accelerations. Such
accelerations may be due to collision with other particles, interaction with electric and
magnetic fields, or to energy-level transitions of electrons in atomic orbits.
1.3.2. Reception and Reciprocity

The foregoing discussion has dealt explicitly with the radiation aspect of antennas. As is
well known, some antennas are not intended to radiate; their function is to receive. Recep-
tion is precisely the inverse of radiation. Whereas an alternating current flowing in an
antenna will produce radiation, an electromagnetic field impinging on an antenna will
cause current to flow in it. A transmitting antenna “launches” electromagnetic waves into
space. A receiving antenna captures energy from an incoming field and converts it into
electric current. Maxwell’s equations also predict this effect, stating that an electromag-
netic field will cause alternating current to flow in any conductor exposed to it, of the
same frequency and with proportional amplitude and phase variations. A medium or
device is reciprocal if the response at point n to an excitation at point m is the same as
the response at point m to the same excitation at point n. It therefore follows that, if the
medium of wave propagation is reciprocal, the variations of the current in a receiving
antenna will be proportional to the variations in the transmitting antenna that launched
the waves.
Most metals and dielectrics used in making antennas are reciprocal, an exception being
ferrites. Ferrites are used in circulators, some phase shifters, and isolators. An antenna
constructed of only reciprocal materials is reciprocal, that is, its performance features are
identical on transmission and reception. Therefore, any reciprocal antenna can be used
for either transmitting or receiving, the only restriction being that a transmitting antenna
may have to carry heavy currents and withstand high voltages. Antennas intended solely
for receiving may not have heavy enough conductors or large enough insulators for
transmitting. The currents and voltages in receiving antennas are ordinarily so small that
they have to be electronically amplified.
In other respects, however, receiving and transmitting antennas are indistinguishable
and will work equally well for either purpose. For this reason the properties of a particular
antenna may be measured and discussed equally well on either a transmitting or a receiv-
ing basis. This is a very valuable bit of knowledge, because it is often easier to employ
one form of operation than another in experimental work.
Receiving antennas, as sometimes designed, are inefficient. Such a design is accept-
able for reception in a high-field strength region; for example, not very far from a trans-
mitter. It is also permissible with high-gain receivers, at frequencies below a few MHz,
when the noise level is determined primarily by external noise sources (e.g., atmospheric
static). This statement, however, does not contradict what has been said concerning the
interchangeability of the transmitting and receiving functions of an antenna; it merely
stresses that inefficient operation may be acceptable in some receiving applications but
that it is usually unacceptable for transmitting applications. At the higher frequencies,
external noise is weaker and less limiting on reception performance. Then, without
the limitations of receiving external noise, receiving antenna effectiveness is equally
Environmental Wave-Propagation Effects 31
important as transmitting antenna effectiveness, in determining the overall performance

of a complete communications system (see sec. 7.8).
The reciprocal relationships between radiation and reception and between the transmit-
ting and receiving properties of antennas, developed mathematically in advanced texts,
are examples of the reciprocity principle or reciprocity theorem.
1.4. Environmental Wave-Propagation Effects

The propagation of radio waves in free space has been considered in sec. 1.1, and certain
nonfree-space effects such as refraction, reflection, and diffraction have been described
in sec. 1.2. The actual environment in which radio waves are ordinarily propagated may
contain obstacles, discontinuities, and propagation-medium variations that give rise to
some or all of these effects, in ways that have important effects on radio system design,
and thus indirectly at least on antenna design. Sometimes the effect on antenna design
is more direct—for example, where the reflecting properties of the earth are utilized
deliberately (as well as unavoidably perhaps) to produce a desired pattern of radiation
and reception. (Hereafter, in view of the reciprocity relationship that has been described,
properties of antennas may be referred to in terms of either radiation or reception, without
the reciprocal property being expressly mentioned, but it will be assumed that the reader
realizes the application of these properties to either function.)
1.4.1. The Earth Environment

In many respects the concept of free-space propagation is realized in space far from the
earth or any other astronomical body, at least over a large portion of the total propagation
path, to a degree unachievable in the earth’s immediate environs. Even in the earth’s
atmosphere, however, essentially free-space conditions sometimes prevail—for example,
over short paths between highly directional antennas. Such conditions usually exist only
at rather high frequencies. Over long paths, and especially at lower frequencies, the
effects of the earth’s surface, the gases of the lower atmosphere (troposphere), and the
charged particles (electrons) of the iono-
sphere play an important part.
The earth’s surface produces two
important effects. First, being solid and
essentially impenetrable to electromag-
netic waves, it represents an obstacle, in Antenna dh
the sense of the discussion of diffraction Horizon line
(Fig. 1–10). Consequently, it “casts a h
shadow,” as illustrated in Fig. 1–12. Thus, Shadow

Earth
it is apparent that radio waves traveling in region
a straight line from the antenna cannot
penetrate the shadow region. There will FIGURE 1–12.
be some diffraction into the shadow Radio-horizon effect and earth’s shadow.
region, but ordinarily the field strength from this effect is too small to be of practical
value for radio communication. Thus, “line-of-sight” radio propagation from an antenna
of height h, Fig. 1–12, would be limited to the horizon distance, dh, assuming that the
receiving point is close to the earth’s surface.
The spherical earth model is frequently used to estimate the propagation path of waves
over the earth (Appendix D). With this model, the effects of the earth’s curvature and
refraction (bending) in the atmosphere are estimated by assuming the propagation is
along straight lines and that the earth has an “effective” radius of aE. For representing
average or “typical” atmospheric effects, aE is usually taken to be 8,493 km, which is
4/3 times the 6,370 km value geophysicists conventionally assume for the actual earth
radius. With this model, the radar horizon dh, that is, the distance from antenna at height
h to the tangent point on a smooth earth of radius aE, is
dh = 2aE h + h 2 (dh, aE, h in same units) (1–37a)
From (1–37a), by neglecting the h2 term (which is ordinarily much less than 2aEh) and
letting aE equal 8,494 km (the 4/3 earth radius), one gets (1–37b)
dh (km) ≈ 130 h (dh and h in km ) (1–37b)
where both dh and h are in km units.

Distance dh of the equations above is, of course, the distance to the horizon as seen
from height h. Further increase in distance not blocked by the earth is obtained if both
transmitting and receiving antennas are elevated; the total horizon range is then the sum
of the values of d, computed separately for the two antennas.
As already noted, atmospheric gases cause refraction, that is, a bending in the wave
paths along the earth’s surface. Interestingly, atmospheric gases affect the propagation
on radio (or radar) and optical waves somewhat differently. Let dh, do, and dg denote
distances to the radio, optical, and geometric horizons, respectively. Here dg is the dis-
tance to the horizon if the waves were to travel in perfectly straight lines. Then, for a
“standard” or average atmosphere,
dh = 1.07do = 1.15dg
Therefore, under normal conditions, atmospheric refraction extends the distance to the
radio (and radar) horizon dh beyond the geometric horizon dg, and dh is at a slightly
greater distance than the optical horizon do (the one observed visually).
1.4.2. Beyond-the-Horizon Propagation

In 1903, Guglielmo Marconi began a regular transoceanic radio message service, then
called wireless, operating at VLF frequencies (3–30 kHz). This service communicated
between Poldhu, England, and stations he established in Nova Scotia and on Cape Cod
(Kraus, 1988, p. 5). At VLF, the lower edge of the ionosphere behaves virtually like a
perfect conducting surface, and vertically polarized waves at these frequencies may be
propagated around the earth as if they were confined between two perfectly conducting
spherical shells, the inner shell being the earth’s surface. This is known as VLF wave-
guide-mode propagation; and it provides effective, narrow bandwidth, and reliable round-
the-world communication. However, extremely high power and a large transmitting
antenna are required.
By the early 1920s, the use of “wireless” had progressed into the present-day AM
broadcast band (535–1605 kHz). At these higher frequencies vertically polarized waves
may propagate to some distance beyond the line-of-sight horizon by means of a surface
wave. However, this surface wave provides only a moderate extension of the line-of-sight
limitation. Surface wave propagation depends on the fact that a conducting surface
tends to “guide” a wave polarized perpendicularly to it, so that the wavefront remains
approximately perpendicular and thus maintains a propagation direction parallel to the
surface.
However, the guiding effect is only partial when there is but one guiding surface,
rather than two as in the VLF waveguide mode. The surface wave partially escapes from
the guiding surface as it progresses, so that the method is not effective beyond a distance
that depends on the frequency. It also depends on the conductivity of the earth, being
greater over the ocean, for example, than over dry land. Usable field strengths are attain-
able at distances ranging from about 150 kilometers (at the higher end of this frequency
range to thousands of kilometers at the lower end, although here the VLF waveguide
mode is probably partially operating. The surface wave is the principal mechanism of
beyond-the-horizon broadcast-band propagation in the daytime and for propagation to
moderate distances at night. The long-distance reception observed at night is due to iono-
spheric reflection.
The principal method of reaching a point around the curve of the earth consists of two
steps, namely, propagation by a path that first goes upward to reach a point high above
the earth and then downward into what
would be the shadow region for direct-
path propagation. This process is illus- P2
trated in Fig. 1–13. The transmission from
point P1 on the earth’s surface reaches
point P3 by first going to P2, high above Direct-path
the earth, then down again to P3. As shadow region
shown, it is possible to reach points well

below the normal horizon line for direct- P3
path propagation from P1. Several means
P1
of accomplishing this result are now Earth
described.
The original method of long-distance
radio propagation around the curve of FIGURE 1–13.
the earth was by reflection from the Beyond-the-horizon propagation path.
ionosphere. This layer of electrically charged particles (actually, several layer-like regions
at different altitudes) acts as a reflector of radio waves at the lower frequencies—generally
from about 10 MHz downward, although the exact upper limit varies with time of day and
other factors. The ionization is produced by the sun’s radiation and so is most intense in
the daytime; however, it persists to some degree through the night so that ionospheric
propagation is possible throughout the twenty-four hours at some frequencies. For any
particular path there is likely to be an approximately optimum frequency at a particular
time of day. Round-the-world propagation is possible by means of multiple reflections.
The ionosphere extends from about 60 to more than 300 kilometers above the earth. (It is
a reflecting region rather than a reflecting surface; it is in fact more logical to regard the
“reflection” as actually a refraction process, at least at the higher frequencies.)
The ionosphere continues to be a valuable means of long-distance radio propagation;
but as it is subject to the vagaries of solar activity and is restricted to the lower frequen-
cies, it has limited usefulness. Other means have therefore been sought, and found in the
years following World War II.
Tropospheric scatter propagation is useful for distances up to a few hundred kilometers
at frequencies in the VHF and UHF. In this method the waves are “scattered” downward
from irregularities of the atmospheric dielectric constant. The exact nature of these scat-
tering regions is a subject of some controversy. Scattering may be regarded as a diffrac-
tion process, semireflective in effect. Because the scattered field strength is not a very
large percentage of the transmitted field, large antennas and high transmitter powers are
required. Nevertheless, the scatter method has proved to be advantageous for fixed-point
communication links, largely because of the freedom from ionospheric variability effects
and because the distances involved are too short for good ionospheric reflection. It has
the further important advantage of permitting the use of frequencies too high for iono-
spheric reflection, thus relieving the overcrowded conditions in the ionospheric-reflection
portion of the radio spectrum.
Scattering from ionospheric irregularities and from the ionized trails of meteors have
also been found to be practical methods. Ionospheric scatter is limited to the frequency
region of about 25 to 60 MHz; it is effective up to about 1,500 km. Meteor scatter can
be employed from 6 to 75 MHz also for distances up to about 1,500 km.
Because scatter methods are subject to limitations of distance and frequency useage,
National Aeronautical and Space Administration (NASA) began investigating satellite
communications with passive, metallized reflecting spheres (http://msl.jpl.nasa.gov/
QuickLooks/echoQL.html). The first of these satellites, Echo 1, was placed in orbit at
a height of about 1,500 km in 1960; it was a reflecting sphere 100 feet (30.5 m) in
diameter, with an orbital period of about 2 hours. Echo 2, 135 feet (41.1 m) in
diameter, was launched in January 1964. Reflection via these satellites required large
directional antennas that followed the satellite in orbit. Echo 1 reflected 960 and
2,390 MHz signals; and Echo 2 reflected 162 MHz signals. Echo 1 successfully redi-
rected transcontinental and intercontinental telephone, radio, and television signals.
Echo 2 continued the passive communications experiments, and also investigated the
dynamics of large spacecraft and was used for global geometric geodesy. Although NASA
abandonded passive communications in favor of active satellites following Echo 2, the
Echo systems demonstrated several ground station and tracking technologies later used
by active systems.
Active satellites receive signals from the earth, and then retransmit them at a different
frequency to another location. These satellites contain receiving and transmitting equip-
ment, including, of course, receiving and transmitting antennas. There are about 150
communications satellites operational today, and they use geostationary, Molnya, or low
polar circular orbits (http://en.wikipedia.org/wiki/Communications_satellite).
Geostationary satellites appear stationary as viewed from the rotating earth; and they
continue to operate above the equator, at an altitude of about 35,000 km. Although the
path lengths are long, an important advantage is that the earth-based antennas do not
have to track a rapidly moving satellite. However, since geostationary satellites remain
over the equator, they are not well suited to high latitude (far North or far South) cover-
age. Molnya orbits are highly inclined, providing good Northern coverage. A major use
of Molnya satellites is for telephony and TV over Russia.
A low earth orbit typically is circular and about 400 kilometers above the earth’s
surface, and time to revolve around the earth is about 90 minutes. Thus, they change their
position relative to a fixed ground position quickly, and thus a large number of satellites
are needed for uninterrupted connectivity. However, discontinuous coverage is possible
by storing data received while passing over one part of earth and transmitting it later while
passing over another part. Low earth orbiting satellites are closer to the ground than geo-
stationary satellites and consequently their equipment costs are less. Thus, relative costs
involve trades between the number of satellites and their individual costs.
For special purposes reflection from the moon’s surface has been used as a communi-
cation method at VHF. This method has the disadvantage, of course, of being limited to
the times that the moon is above the horizon; it also requires very high power and large
antennas.
At some frequencies, usually VHF and higher, unusual atmospheric conditions some-
times result in abnormally great downward bending of ray paths, owing to an extreme
effect of the type illustrated by Fig. 1–4. Under such circumstances waves may be propa-
gated in a path that follows the curve of the earth, hence into the normal shadow region.
This effect is called superrefraction or sometimes “trapping” or “ducting.” It usually
occurs in warm climates when unusual moisture and temperature conditions prevail and
is not sufficiently reliable or predictable for most beyond-the-horizon propagation appli-
cations. However, under very unusual conditions propagation by this method beyond
1,500 km has been observed, and 150-km paths are not unusual.
1.4.3. Reflection from the Earth’s Surface

Next to its role as a barrier, that is, creating a shadow region; the most important effect
of the earth on the propagation of radio waves is its action as a reflector. Moist earth or
a water surface will reflect electromagnetic waves in the lower part of the radio spectrum
quite well. At the higher frequencies the surface may be too rough to reflect specularly.
However, in accordance with Rayleigh’s criterion, equation (1–25), a rough surface will
reflect specularly at a sufficiently large angle of incidence (small angle between the ray
direction and the surface, called the grazing angle). At larger grazing angles semispecular
reflection may occur, as shown in Fig. 1–6. For smooth earth surfaces, effects of specular
reflection may be observed at frequencies up to at least 35 GHz.
Reflection from a plane surface was illustrated in Fig. 1–5, and the principle of images
for specular reflection was shown in Fig. 1–7. The reflection that occurs from the earth’s
surface may be understood in terms of these concepts. Although the earth is approxi-
mately spherical, the region of reflection may be regarded as virtually plane if the antenna
is located within about a hundred meters of the earth’s surface.
Waves reflected from the earth’s surface are important because they interfere with
waves propagated in a direct path, as illustrated in Fig. 1–14. A transmitting antenna is
assumed to be located at Sl, at a height h above a perfectly reflecting plane surface; and
therefore its image is located at S2, a distance 2 h from Sl. A direct wave travels to a
receiving point at P2 following a straight-line path (except for the slight downward cur-
vature due to atmospheric refraction, which may be disregarded here). A reflected wave
also arrives at P2 via the path S1P1 + P1P2. However, in accordance with the image prin-
ciple, the reflected wave may be regarded as originating at S2 and proceeding to P2 via
the straight-line path S2P1 + P1P2. The elevation angle of point P2 as viewed from Sl is
designated q. Notice that the grazing angles y, whether entering or exiting point P1, are
equal in accordance with the principles of optics (Fig. 1–5).
Since the two waves arriving at P2 are of exactly the same frequency, being from the
same source, they will interfere in accordance with (1–35) and (1–36). Whether the inter-
ference is constructive or destructive depends on the phase difference. Since the lengths of
the two wave paths are not the same, there will in general be a path difference d and a con-
sequent phase difference equal to bd radians (b = 2p/l). At some points in space the two
waves will reinforce each other; at other points they will tend to cancel so that there will
be regions of zero or near-zero field strength. This effect can have serious consequences;
for example, it can create “blind” regions in space where signals cannot be detected.
Figure 1–14 is used to analyze a relatively simple multipath propagation problem that
provides the results of Fig. 1–15. Applicable assumptions follow:
a. The wavelength is long enough that the surface is smooth, in accordance with
the Rayleigh criterion [equation (1–25)]. Ordinarily, even in the presence of
the earth’s surface irregularities and undulations, this assumption is valid for
VHF (30–300 MHz) and lower frequencies.
b. The distance S1P2 between antenna and the observation point is short enough
that the earth’s curvature is neglected.
c. The antenna elevation pattern is sufficiently broad that the amplitude and
phase of the radiated waves are the same at every elevation angle q.
d. The reflection coefficient amplitude and phase delay are constant, correspond-
ing to |G| = 1 and f = −p radians, respectively. This is a valid approximation
for horizontal polarization and a reflecting surface that is sea water or another
good conductor.
e. The paths S1P2 and S2P2 are long enough, relative to 2 h, so that they are
effectively parallel.
P2
th
Direct- wave pa
S1
q
th
e pa
av
d-w
e cte
Refl
y y
2h Reflecting Surface
P1
q1
2hsinq1
S2
FIGURE 1–14.
Geometry of interference between direct-path and reflected-path waves.
In summary, the assumptions above 90°

60°
require that the reflecting surface be a flat
50° Interference-pattern
and smooth conductor, the polarization be field contours
horizontal, and the direct- and reflected- Free-space 40°
wave paths (S1P2 and S2P2) be effectively field contour
parallel. 30°
Vertical distance
Elevation
For purposes of the analysis, we denote angle, q
the electric intensity of the direct-wave 20°
as Ed and the reflected wave as Er.
Then, based on the above assumptions 10°
and Fig. 1–14, the electric field E result-
ing from the phasor sum of the direct-
Horizontal distance
wave and the reflected-wave fields at P2
follows: FIGURE 1–15.
Interference-pattern field-strength contours
E = Ed + Er = Ed + Ed Ge − jβδ relative to free-space contour, for h/l = 1.44
= Ed 1 + G jφ e − jβδ  (1–38) and G = −1.
Here G is the complex (amplitude and phase) reflection coefficient of the earth’s surface,
and bd is the phase delay (in radians) caused by the difference d of the lengths in the
direct- and reflected-wave paths. It can be seen in (1–38) that the amplitude and phase
of G are expressed by |G | and e jf, respectively. Also, recall that b = 2p/l. Thus, the
magnitude of E may be expressed as
E = Ed 1 + G e jφe − jβδ  (1–39)
Note that within the second absolute-value brackets of (1–39) there is a sum of two
phasors, separated in angle by f−bd radians. Consequently, by using the law of cosines
or other means, one finds the magnitude of E to be
E = Ed 1 + Γ 2 + 2 Γ cos(φ − βδ ) (1–40)
Then, by using the assumptions |G | = 1 and f = −p of item d in the list above and a half-
angle formula, (1–40) becomes
βδ
E = 2 Ed 1 + cos(−π − βδ ) = 2 Ed 1 − cos(βδ ) = 2 Ed sin   (1–41)
 2 
where, as previously, b = 2p/l. Note that changes in the path length difference d cause
|E | to vary between 2|Ed| and zero.
From Fig. 1–14 it can be seen that d equals 2hsinq1 requires parallel S1-P2 and S2-P2
paths, but the angle q1 is generally not known. However, in many circumstances, the
path S2-P2 is approximately parallel to path S1-P2 because S1-P2 is very much larger than
height h. Then, q1 is approximately q, where q is the more likely known angle which is
between the direct-wave path and the local horizontal at the antenna. Thus, when q ≈ q1,
the path length difference d is approximately 2hsinq and then
2π
δ = 4π   sin θ
h
βδ = (1–42)
λ  λ
Hence for a given value of h/l (antenna height expressed in wavelengths) the interfer-
ence becomes a function of the elevation angle q only. A final simplification of (1–38)
is then possible from (1–40) and (1–42):
2π h sin θ 
E = 2 Ed sin  (1–43)
 λ 
Application of this formula to the particular value of h/l = 1.44 results in the field-
strength pattern shown in Fig. 1–15.
The contours, or lobes, represent surfaces of constant field strength. The origin of the
coordinate system is the location of the transmitting antenna. Elevation-angle markers
are at 10-degree intervals up to 60 degrees. Field strength is compared to its strength if

the antenna radiated uniformly in all elevation angles and there were no reflected wave
(free-space contour). As shown, there are three lobe maxima between q = 0° and q = 90°,
for the particular h/l assumed, namely, 1.44.
The number of lobes in an interference pattern is approximately 2 h/l. Thus at very
short wavelengths (very high frequencies) there may be many lobes and nulls; for
example, at a frequency of 500 MHz (l = 0.6 m) and with 30 m antenna height, there
are approximately 100 lobes between q = 0° and q = 90°. The pattern shown in Fig. 1–15
may occur with 30 m antenna height, 15 MHz frequency, and horizontal polarization.
Important assumptions in the calculation for Fig. 1–15 are that the reflection coeffi-
cient G is minus one, corresponding to |G | = 1 and f = −p radians (180° delay). These
assumptions are approximately correct for horizontal polarization at all angles of inci-
dence. Thus, in accordance with (1–43), the field strength is zero at elevation angle q =
0°, that is, the horizontal direction. Zero field strength at q = 0° is, of course, undesirable
in numerous applications. A remedy is to bring the first lobe maximum down to as low
an angle as possible by increasing the ratio h/l, by increasing the frequency and/or
increasing the antenna height. (Parenthetically, the elevation angle of the first lobe
maximum is approximately l /4 h radians or about 15 l /h degrees.) However, for long-
distance ionospheric transmission, zero-angle radiation is of less importance than for
non-ionospheric transmission. For long-distance ionospheric transmissions, horizontal
polarization may be used if h/l is equal to at least unity; although h/l = 2 that gives a
maximum field strength at about 7° elevation is better.
1.4.4. Variations in Earth’s Reflection Coefficient

The present section discusses smooth planar surfaces, for which the reflections are specu-
lar. As discussed in Appendix D, these reflections may be reduced by either surface
roughness or the earth’s curvature. Additionally, the accompanying website includes
Mathcad calculations of reflection coefficient and multipath effects versus their contribut-
ing factors.
Recall that, for the analysis pertaining to Fig. 1–15, the paths S1P2 and S2P2 are
assumed parallel, and that causes the elevation angle q and the grazing angle y to be
equal. In fact, multipath propagation is strongly affected by both the amplitude and phase
of the earth’s reflection coefficient, which depend on the grazing angle y. The reflection
coefficient G for dielectrics and conductors is different, and it depends strongly on
whether the polarization is horizontal (H) or vertical (V). Land and sea have different
dielectric properties: dry land is a poor conductor, water improves the conductivity of
land, and seawater is a good conductor. For H and V polarizations, there are major dif-
ferences between G versus grazing angle y. For a given land or sea surface, G also
changes with frequency, but significant differences occur only with major changes in
wavelength.
Land and sea surface reflections for H polarization (H POL) are easier to describe than
those for V POL. This is because for H POL |G | ≈ 1 and f ≈ −p radians, but |G | and f
vary widely for V POL. More specifically, for H POL:
• G changes very little with changes in grazing angle, surface conductivity, or

frequency and
• |G | approaches unity as the grazing angle y approaches zero and it deviates
the most from unity at angles that are near normal incidence, at the higher
frequencies, and for the driest surfaces.
As now discussed, the Brewster angle effect causes large variations in |G | and f. The
Brewster angle is the grazing angle at which the reflection coefficient is zero, and it
occurs only for V POL and a lossless dielectric. Instead, for land and sea, which are
imperfect dielectrics, |G | for V POL has a non-zero minimum at the so-named pseudo-
Brewster angle ypB. Accordingly, |G | is much larger for grazing angles both smaller and
larger than ypB. At grazing angles larger than ypB, f becomes close to 0°, f is approxi-
mately −90° at ypB, and f approaches −180° as y approaches zero. Thus, f will be very
near −180° for very small grazing angles, even with V POL and poorly conducting
ground. Additionally, |G | increases and approaches unity for grazing angles y that
approach zero. Therefore, an interference null exists at small grazing angles for both H
and V polarizations—however, as now discussed, below 1 MHz, the V POL nulls near
the horizontal direction will be increasingly filled as the frequency is reduced.
The pseudo-Brewster angle ypB is at considerably smaller grazing angles for seawater
than for land, and it becomes smaller for both land and sea with increasingly longer
wavelengths. Interesting, the movement of ypB closer to zero grazing permits acceptable
V POL performance (without an interference null) down to very small grazing angles at
low frequencies. For smooth planar surfaces at 1 MHz, ypB is about 0.2° and 3° for sea
and typical land, respectively. At 100 kHz, the corresponding ypB values are about 0.07°
and 1° for sea and land, respectively. Natural surface-slope undulations and other irregu-
larities diminish the destructive interference and thereby further enhance signal ampli-
tudes at the near-zero grazing angles. Therefore, at the lower frequencies it is possible
to place a vertical-polarization radiator close to the ground and obtain acceptably strong
radiation in the horizontal direction. This phenomenon permits the effective use of V
POL for frequencies less than about 1 MHz.
Although the interference of direct and reflected waves creates undesirable minima in
the radiation pattern, it also creates maxima in field strengths. Since the earth is not a
perfect reflector, the amplitude of the reflected wave will be less than that of the direct
wave, with the result that the field strength will be greater than zero at the minima and
less than twice the free-space value at the maxima. As discussed in more detail in Appen-
dix D, the earth’s surface roughness and curvature reduces |G |, and thus the pattern nulls
are partly filled and the maxima reduced. The reader will recall that, in accordance with
the Rayleigh criterion, equation (1–25), surfaces become rougher (electromagnetically)
with increases in the radio wave frequency.
1.4.5. Wave Nomenclature

The terms sky wave and ground wave were introduced many years ago when propagation
at distances beyond the horizon was generally accomplished via ionospheric reflection,
and at shorter distances by means of either

a surface-guided wave or by direct line-
of-sight transmission. These terms are
Ionosphere
still in use, and their meanings are illus-
trated in Fig. 1–16. As shown, the sky
wave is propagated by an ionosphere- Direct wave
Sky wave
reflected path. The ground wave com- Reflected wave
prises the vector sum of (1) a direct-path Transmitting ve
ce wa
antenna Surfa
(line-of-sight) wave; (2) a reflected-path
wave; and (3) a surface-guided wave or,
more simply, surface wave. In other Earth
words, the ground wave is the composite
wave in the vicinity of the earth, that FIGURE 1–16.
would exist in the absence of the Nomenclature of wave propagation (not to
ionosphere. scale).
1.4.6. Transionospheric
Propagation
Now that radio transmission to and from Ionosphere
various outer-space locations is com-
monplace, the effect of the ionosphere on
waves passing through it is of impor-
Antenna Ray path
tance, whereas at one time the ionosphere
was of interest primarily because of its
ability to reflect radio waves back to
earth. Naturally, transmission through
Earth
the ionosphere requires the use of a fre-
quency well above the frequencies that FIGURE 1–17.
are reflected from it; generally, frequen- Refraction in transionospheric propagation.
cies above 10 MHz are necessary.
A wave that is above the critical reflec-
tion frequency may nevertheless be appreciably refracted in passing through the iono-
sphere, as illustrated in Fig. 1–17. As shown, the wave path is at first bent downward as
it enters the ionosphere from below. Then, after reaching the region of densest ionization,
it is bent back upward, and eventually emerges with nearly its original direction. If the
direction of the antenna beam at the earth were used to determine the direction of a source
of incoming waves, this bending effect would cause some error in the determination.
Therefore it is important to be aware of the effect. If the ionosphere electron-density
height profile is known, the magnitude of the effect can be calculated. Generally it is not
an appreciable effect above 100 MHz and is less at night than in the daytime. A similar
statement applies to the absorption losses that occur in the ionosphere.
A more important effect in transionospheric propagation below a few hundred MHz
is the Faraday rotation of the polarization of a wave. This rotation occurs as a result of
the combined effect of the charged particles (electrons) of the ionosphere and of the
earth’s magnetic field. Because of the variability of the effect, it is not feasible to attempt
to predict the exact amount of rotation at low frequencies, where the total rotation may
even be many multiples of 360 degrees. This rotation is important because, if linearly
polarized antennas are used for both transmission and reception, the received polarization
will sometimes be correct and sometimes incorrect for optimum reception. If the polar-
ization of the signal is at right angles to that which the antenna is intended to receive, in
principle no reception will result.
Several remedies are available, however. One is to use either circular polarization, or
two linear polarizations at right angles to each other, so that one or the other of them
will at all times receive a signal. Another remedy is to use a linearly polarized receiving
antenna whose polarization direction can be varied to agree with the polarization of the
incoming wave. The third remedy is to operate at a high enough frequency so that no
appreciable rotation occurs—above 1,000 MHz, or possibly a bit lower. The present
tendency for space communication and other space transmissions is to use frequencies
between about 1 and 10 GHz, although avoidance of Faraday rotation is only one of
several reasons for using this frequency region.
1.4.7. The Radio-Transmission Equation

Equations (1–4) and (1–5) are expressions for the far-field (sec. 3.2.5) power density of
a wave at a given distance from an isotropic radiating source in free space. An expression
of this type is a transmission equation for this special case. A more general transmission
equation will now be derived. Such an equation is needed not only in radio system analy-
sis but also in some antenna measurement applications.
When an antenna radiates nonisotropically, that is, when it radiates more power in
some directions than in others, the power density in the preferred directions will be
greater than it would be for an isotropic antenna. The ratio of the power density in a
given direction to that which would be observed at the same distance if a lossless isotropic
antenna were radiating the same total power is called the gain of the antenna, G. It follows
that if a transmitting antenna has a gain Gt in a certain direction, the received power
density pr at a distance R in that direction will be
Pt Gt
pr = (1–44)
4π R 2
Equation (1–44) is a free-space radio-transmission equation. If the field strength (electric

intensity) rather than the power density is desired, it can be obtained by applying (1–9)
and noting that 377 = 120p, which gives
30 Pt Gt
Er = (1–45)
R
When the path between the transmitter and receiver is not in “free space,” a modifica-
tion of these equations is required. This is done through use of the pattern-propagation
factor F. Following the IEEE definition (IEEE Standard 686–1997, p. 19), F is a scalar
and it is the ratio of the field strength that is actually present at a point in space to that
which would have been present if free-space propagation had occurred with the antenna
beam directed toward the point in question. Accordingly,
Er (actual )
F= (1–46)
Er ( free space)
By merely rearranging this definition, (1–45) which gives Er(free space) can be converted
to a nonfree-space equation, as follows:
F 30 Pt Gt
Er (actual ) = FEr ( free space ) = (1–47)
R
Similarly, the equation for power density pr under nonfree-space conditions becomes
Pt Gt F 2
pr = (1–48)
4π R 2
Ordinarily, the gain of an antenna varies with direction of radiation, but the gain Gt in
(1–47) and (1–48) is the gain in a specific direction of observation. However, for use of
the above IEEE definition of F, one is reminded that, in a nonfree-space environment, the
actual field strength at a point may depend on gain in directions other than that toward the
point in question. For example, in accordance with Fig. 1–14, waves radiated downward
from an antenna may reflect from the earth and be summed at a point in space, both in
phase and amplitude, with waves arriving directly from the antenna. Therefore, with the
definition of F, antenna gain versus direction must be considered when more than one
direction is involved, unless the antenna gain is the same in the various directions.
Effective receive aperture area Ar, and its relationships with maximum receive antenna
gain Gr and received power Pr, is discussed in chapter 3, sec. 3.4. Received power Pr
equals prAr, assuming the receiving antenna and the incident wave are polarization
matched, where pr is the power density incident on the area Ar, and Ar = Grl2/4p. Then,
by using these relationships and (1–48), a useful relationship for received power
follows:
Pt Gr Gt λ 2 F 2
Pr = (1–49)
(4π )2 R 2
The reflection-interference pattern calculation, exemplified by Fig. 1–15, can be used to
determine F, for the case where the antenna gain is independent of the elevation angle.
Accordingly,
E
F= (1–50)
Ed
since Ed is the free-space field strength. With E expressed in terms of (1–43), (1–50)
becomes
2π h sin θ 
F = 2 sin  (1–51)
 λ 
In this case the maximum value of F is obviously 2 (when the sine function has its
maximum numerical values ±1), and the minimum value is zero. In free space, of course,
F = 1.
Appendix D and the accompanying website discuss calculations of the pattern-propa-
gation factor for smooth and rough land and sea surfaces, under assumptions that the
earth is either flat or spherical.
References
Ah Yo, D. M. K., and R. Emrick, “Frequency Bands for Military and Commercial
Applications,” ch. 2 in J. L. Volakis (ed.), Antenna Engineering Handbook, 4th ed.,
McGraw-Hill, 2007.
Blake, L. V., “Tropospheric Absorption Loss and Noise Temperature in the Frequency
Range 100–10,000 Mc”, IRE Transactions on Antennas and Propagation, AP-10,
January 1962, p. 101.
IEEE Standard 100–1992, The New IEEE Standard Dictionary of Electrical and Electronic
Terms, 1992.
IEEE Standard 211-1997, The IEEE Standard Definitions of Terms for Radio Wave
Propagation, December 1997, p. 27.
IEEE Standard 686–1997, IEEE Standard Radar Definitions, 25 March 1998.
Kraus, J. D., Antennas, 2nd ed., 1988.
Problems and Exercises

1. A plane electromagnetic wave in free space has an electric-intensity amplitude
of 10 volts per meter and initial phase angle f of 15 degrees at the time t = 0
and the position z = 0. The frequency is f = 5 × 105 Hz (500 kHz). What is the
instantaneous electric intensity at the position z = 3 meters at time t = 10−8
seconds? (Note: To work this problem in degrees rather than in radians, substitute
the number 360 in place of 2p in Eq. 1–2.)
2. A wave propagating in free space is at a point 20 km from its source (assumed
to be a point source), and the power density is p = 2 × 10−5 watts per square
meter. What is the power density at a distance of 30 km?
3. The rms electric intensity of a wave in free space is 2.7 × 10−3 volt per meter.
(a) What is its power density in watts per square meter? (b) What is the rms
magnetic intensity in amperes per meter?
4. (a) The attenuation constant of a uniformly absorbing propagation medium is

g = 0.01 dB per kilometer. A plane wave propagating in this medium (assumed
to be of infinite extent) has, at an initial point, a power density p1 = 1 × 10−3
watt per square meter. What is the power density p2 at a second point 300 km
away in the direction of travel of the wave?
(b) If the initial point in part (a) is at a distance R1 = 100 km from the source
of the waves, so that the wavefront must now be regarded as spherical (and
spreading) rather than plane, and if a second point is at a distance R2 = 400 km
from the source in the same direction, what is the total attenuation in decibels
due to both the spreading and the absorption? (Note: Because the medium is
assumed to be of infinite extent, the wave propagation will follow free-space
laws except for the absorption; Eq. 1–14 applies.)
5. A plane wave in free space is incident upon the plane surface of a dielectric
material at an incidence angle of 30 degrees. The dielectric constant of the mate-
rial e′r = 4. By what angle does the direction of propagation change as the wave
passes into the dielectric medium? (That is, what is the difference between the
angle of incidence and the angle of refraction?)
6. A boundary between two propagation media has a total power-reflection coeffi-

cient R = 0.64. (a) What is the power-transmission coefficient, T? (b) If the
reflection is partly specular and partly diffuse, and the specular coefficient is of
magnitude |G | = 0.7, what is the diffuse-power-reflection coefficient, Rd?
7. (a) A transmitting antenna is located on a mountain top at a height of 1,500

meters, overlooking the sea. A ship is steaming directly away from it. Assuming
that the only signal that can be effectively received is the one that propagates
via a direct (line-of-sight) path, at what maximum distance will the ship be able
to receive signals from the mountain-top antenna? (Assume that the transmitting
power is adequate for distances to the horizon. Neglect the height of the ship-
board antenna.) (b) An airplane at an altitude of 300 meters is also flying out to
sea, on the same course as the ship. At what maximum distance can it receive
the signals?
8. A radio station has a total radiated power Pt = 10,000 watts, and an antenna gain
Gt = 30. If free-space propagation is assumed, what is the rms electric intensity
of the radiated field at a distance R = 100 km (105 meters), expressed in volts
per meter?
9. A shipboard transmitting antenna radiates Pt = 1,000 watts and has a power gain
Gt = 100. Its height above the sea surface is such that at a certain elevation angle
the propagation factor is F = 2. An airplane at a range R = 50 km (5 × 104 meters)
is at an altitude corresponding to this elevation angle. What power density exists
at its receiving antenna due to the transmission from the ship?
10. An isotropic antenna is located at a height of 30.3 meters above a perfectly

reflecting plane surface. The wavelength of the radiated signal is 3 meters.
Assume the reflection coefficient G = −1. At a distant point whose elevation angle
is 30 degrees, what is the propagation factor F? (Note: to use trigonometric
functions in degrees rather than radians in Eq. 1–51, replace the factor 2p by the
number 360.)
CHAPTER 2
Transmission Lines
This chapter discusses transmission lines. Transmission lines must be addressed in the
study of antennas for three reasons: (1) A transmission line virtually always connects an
antenna to a transmitter or receiver, and is often regarded as a part of the antenna system;
(2) in some types of antennas transmission-line elements are integral parts of the antenna;
and (3) the principles of transmission lines are applicable to understanding some aspects
of antenna theory.
The material of this chapter, like that of chapter 1, is intended for review and refer-
ence. A complete treatise on transmission lines would provide material for an entire book.
Therefore only the basic aspects of the subject will be covered here, with emphasis on
antenna applications.
2.1. Basic Transmission-Line Concepts

A transmission line is a specific example of what is known in circuit theory as a linear
passive four-terminal network or, in more modern terminology, a linear passive two-port.
Its function is to convey electrical power or signals between two points appreciably
separated in distance.
The simplest form of transmission line, a pair of parallel wires insulated from each
other, is shown schematically in Fig. 2–1. It is called a two-wire balanced line. Basic
transmission-line principles will be discussed in terms of this type of line, but most of
them will be found to be applicable directly or with minor modifications to other forms
of lines, such as coaxial lines and waveguides.
As indicated in Fig. 2–1, the spacing of the wires, S, their diameter, d, and their length,
l, are important properties of the line. These are dimensional properties, but they are
important electrically. Also important, naturally, are the electrical properties—the con-
ductivity of the wires, and the permittivity, permeability, and loss characteristics of any
material within the field of the line.
In operation, a load of impedance ZL is connected to the load terminals of the line,
and a sinusoidal voltage Vi is applied to the input terminals. Given these quantities, the
47
48 Transmission Lines
transmission-line problem is to find the

l current and voltage at points along the
line and at the load. It is usually also of
Output particular interest to know what imped-
d s (receiving or
Input load end) ance the line presents at its input termi-
(Sending end) terminals nals to the source (generator) of the input
terminals
voltage. By definition this impedance, Zi,
FIGURE 2–1.
is the ratio of the input voltage Vi to the
Dimensional elements of a two-wire input current Ii, so that finding Zi is not a
transmission line. separate or additional problem.
Figure 2–2 indicates the nature of this
l electrical problem, which differs in an
Ii I(x) IL
essential way from that of the simple
Source Zi Vi V(x) VL ZL Load four-terminal network of circuit theory.
When the voltage Vi is applied at the input
Line terminals, it does not result in the imme-
x
diate appearance of a voltage VL at the
FIGURE 2–2. load terminals. A finite time is required
Electrical elements of the transmission-line for the voltage to travel the length of the
problem. line, in the same way that an electromag-
netic wave has a finite velocity in space.
In fact, voltage and current traveling along a transmission line are accompanied by an
electromagnetic wave in the space between the conductors, and the transmission-line
problem can be analyzed in terms of this field instead of in terms of the voltage and
current. The line is then regarded as a means of guiding the field so that it is confined to
a region near the line rather than spreading spherically in space. Because of this confine-
ment, the waves do not suffer the inverse-square-law decrease of power-density with
distance as do waves in free space.
Thus the transmission line may be considered either from the point of view of volt-
ages, currents, and electrical circuit theory or from the point of view of a guided elec-
tromagnetic wave. It is more conventional to adopt the former approach, at least initially;
and this will be done here. However, the wave concept cannot be avoided altogether;
reference will be made to the propagation of voltage and current waves along the line,
and the velocity of propagation will play an important part in the analysis. Eventually,
in the discussion of waveguides as transmission lines, the electromagnetic-wave approach
will necessarily be adopted.
2.1.1. Equivalent-Circuit Line Representation

A transmission line has inductance, capacitance, and resistance just as do “ordinary” four
terminal networks. The essential difference is that these properties are distributed uni-
formly and continuously along the line, whereas in ordinary circuits they are “lumped.”
For purposes of analysis, however, the transmission line can be represented by an
Basic Transmission-Line Concepts 49
L R1 L R1 L R1 L R1
Input C R2 C R2 C R2 C R2 Output
x=1
FIGURE 2–3.
Equivalent-circuit representation of a transmission line.
arrangement of many lumped-circuit elements (of short length) as shown in Fig. 2–3. In
other words, small lumped circuit elements are used to connote the infinitesimally small
amounts of inductance, capacitance and resistance that are distributed uniformly along a
transmission line.
Each identical circuit section in this representation corresponds to a unit length of the
line, that is, L is the inductance per unit length of the conductors, C is the capacitance
between the conductors per unit length, R1 is the series resistance of the conductors per
unit length, and R2 is the shunt or “leakage” resistance per unit length. For this repre-
sentation of a line to be valid, it is necessary to assume that the “unit of length” is a very
small fraction of the total line length, so that there will be very many of the identical
lumped unit-length sections. More specifically, it is necessary that the unit of length be
chosen small compared to a wavelength on the line, defined in the same way as for
electromagnetic waves in space:
v
λ= meters (2–1)
f
where v is the velocity of propagation of the voltage and current “waves” along the line,
meters per second, and f is the frequency in Hertz of the applied voltage.
These assumptions are necessary to justify certain steps in the analysis; but after the
analysis is made it is found that the results involve ratios of the unit-length circuit values.
For this reason, in application of the results, L, C, R1, and R2 can be expressed as values
for unit of length, for example, henrys per meter (H/m), farads per meter (F/m), ohms
per meter (W/m).
To describe the behavior of a line in these terms, it is helpful to consider two special
cases that are idealizations, that is, are never fully attainable in reality although they may
be closely approximated. The first of these is the line of infinite length (l → ∞ in Fig. 2–2)
and is discussed in sec. 2.1.2. The second is the lossless line discussed in sec. 2.1.3,
corresponding to R1 = 0 and R2 → ∞ in Fig. 2–3.
2.1.2. The Lossless Infinite Line

If a line is of infinite length, effects due to the presence of a load impedance ZL at the
load end do not appear, for a voltage Vi(t) applied at the instant t = 0 will never reach
the load end, traveling at the finite velocity v. If Vi(t) is sinusoidal with angular frequency
w = 2p f, it is described by
Vi ( t ) = V0 sin (ω t + φ ) (2–2)
in which V0 is the amplitude and f is the phase angle at t = 0. Since the line is lossless,
this voltage will travel down the line at velocity v with undiminished amplitude. If
distance along the line is denoted by x, the voltage at the point x will be given by
V ( x, t ) = V0 sin (ω t − β x + φ ) (2–3)
provided that enough time has elapsed for the voltage wave to reach the point x, that is,
at times t greater than t = x/v.
The quantity b, known as the phase constant, is defined as 2p/l. Equation (2–3) is
identical to the equation of a plane electromagnetic wave of equation (1–2), ch. 1, with
the substitution of a voltage V for the electric field intensity E.
The analysis of the transmission line of Fig. 2–3 leading to this result and to the further
results that will be presented is made by writing and then solving second-order differen-
tial equations, based on the assumptions that the circuit sections of Fig. 2–3 represent
infinitesimal lengths of the line. From the analysis, which will not be given here, it is
also found that the velocity of the voltage waves is
1
v= (2–4)
LC
for a lossless line, where L and C are the inductance and capacity per unit length of line.
In this case if L and C are in henrys and farads per meter, v is obtained in meters per
second.
Note the similarity of (2–4) to the equation for velocity of an electromagnetic wave
in a lossless medium, equation (1–18), ch. 1. Moreover, it turns out that if the transmis-
sion line is primarily immersed in air or a vacuum (except perhaps for solid insulating
supports of low loss at well-spaced intervals), the product of L and C will be a constant
such that v is equal to 3 × 108 meters per second—exactly the velocity of an electromag-
netic wave in space. Therefore the wavelength l, as given by (2–1), will be the same on
the line as for a wave in space.
These statements assume that the conductors are straight, smooth-surfaced, and are
made of nonferrous metal having magnetic permeability of the same value as that of air
or vacuum, that is, a relative permeability of unity. It may be noted in passing that copper-
plated steel wire does not violate this assumption. This is because, at radio frequencies,
the phenomenon called skin effect confines the current and fields to the surface of the
wire and the external region of space; therefore the steel core does not affect the
inductance.
But if a material having relative permittivity or relative permeability greater than one
occupies the region between and surrounding the conductors, L and C will have a larger
product. Then, v will be less than 3 × 108 meters per second, and is given by
c
v= (2–5)
µr ε r
where c = 3 × 108 meters per second, mr is the relative permeability, and er is the relative
permittivity. For most dielectric materials mr = 1. A lossless line is being discussed here,
where er is a real number. For a lossy dielectric, er is complex (see sec. 1.2.1), and then
er in (2–5) must be replaced by its real part (the dielectric constant).
When the voltage Vi(t) is applied to the input terminals of the line, a current flows
into the line. It is related to the input voltage by a quantity called the characteristic
impedance, Z0. That is,
Vi( t )
Ii(t ) = (2–6)
Z0
and Z0 is found to be given by
L
Z0 = ohms (2–7)
C
Equation 2–6 is of course Ohm’s law, and (2–7) is analogous to the equation for the wave
impedance of a propagation medium, equation (1–8), ch. 1. Therefore the current as a
function of distance x along the line is given by
V 
I ( x, t ) =  0  sin (ω t − β x + φ ) (2–8)
 Z0 
Or in other terms, as deduced from (2–2) and (2–6),
V0
I0 = (2–9)
Z0
which is analogous to equation (1–10), ch. 1, for electromagnetic waves in space.

Also, (2–8) is analogous to equation (1–3), ch. 1. In short, the voltage, current, and
characteristic impedance of a lossless infinite transmission line correspond to the electric
intensity, the magnetic intensity, and the impedance for a plane electromagnetic wave in
free space.
2.1.3. Reflection and Standing Waves

When the line is of finite length l, as in Fig. 2–2, the voltage wave applied to the input
terminals will reach the load end of the line in a time t = l/v, where the velocity v is given
by (2–4). If it is assumed that the load impedance ZL is at least partly resistive, the result-
ing current in ZL will cause power to be delivered to the load. The power flowing down
the line, since the voltage and current are in phase (as indicated by (2–3) and (2–8), is
given by
P = Vrms Irms (2–10)
where the subscript rms stands for “root mean square” or effective value; it is related to
the amplitudes by the well-known relations
V0
Vrms = = 0.707V0 (2–11)
2
and
I0
Irms = = 0.707I 0 (2–12)
2
From (2–9) through (2–12), it may be deduced that
2
Vrms
P= = I rms
2
Z0 (2–13)
Z0
It is to be emphasized that this result applies to the power flow in terms of in-phase
voltage and current in an infinite line, or in a finite line before the voltage wave reaches
the load.
The load impedance ZL may be purely resistive, purely reactive, or some combination
of resistance and reactance. Suppose that it is purely resistive and of value RL. Then the
power delivered to the load will be
VL2
PL = = I L2 RL (2–14)
RL
where VL and IL are the voltage and current at the load. Comparison of (2–13) and (2–14)
shows that if RL = Z0, all the power flowing down the line will be delivered to the load.
But if RL does not have this value, the two powers cannot be equal if the voltage and
current of (2–14) are assumed to be the same as the voltage and current of (2–13).
What actually happens is that some of the voltage and current are reflected. The
reflected voltage and current then travel back toward the input end of the line, at the
velocity v. Consequently, at every point x along the line, the voltage is that resulting from
linear super-position of the original (incident) voltage wave, of amplitude V0(i), and a
reflected wave, of amplitude V0(r). A similar statement applies to the current.
This situation is entirely analogous to the problem of interference between two
electromagnetic waves at a point in space. The voltage and currents at various points
on the line may either add or subtract from each other, depending on the difference
in their phase angles, which in turn varies from point to point. The equation of the
reflected voltage wave corresponding to (2–3) for the incident wave is obtained by sub-
stituting V0(r) in place of V0, and (2l − x) in place of x. The phase angle f also will in
general be changed in the process of reflection, depending on the phase angle of the
impedance ZL.
The result is that the voltage amplitude will vary along the line. There will be maxima
at half-wavelength intervals, with minima at positions halfway between the maxima. The
positions of these maxima and minima do not change with time, that is, they do not move
along the line. The resulting pattern (actually, an interference pattern) is therefore a
standing wave. A standing wave of current is also set up in the same way. The current
maxima coincide with the voltage minima, and vice versa (see Fig. 2.5 of sec. 2.2.6).
The arrival of the reflected voltage and current waves at the line input terminals results
in a changed voltage-current relationship at this point. Consequently, the effective input
impedance of the line is no longer equal to the characteristic impedance Z0, in general.
It is by definition the resulting ratio of voltage to current.
Because the effective input impedance undergoes a sudden change when the reflected
wave first reaches the input terminals, the applied voltage Vi may also undergo a change
at that instant (since the output voltage of the usual source, or generator, will vary when
the impedance presented to it varies). The voltage wave traveling down the line therefore
takes on a new value at this instant, and so the reflected wave coming back 2l/v seconds
later will also have a suddenly different value at that instant, resulting in a further read-
justment of the input voltage, and still further adjustments at future intervals of 2l/v
seconds. However, these readjustments become progressively smaller and gradually “die
out.” They are called transients. When they have died out, a steady state is said to exist.
(Parenthetically, if the source impedance were Z0, there would not be reflections at each
end of the transmission line).
For some purposes it may be important to analyze the transient behavior of the line,
but ordinarily only the steady-state behavior is important. This may be seen by consider-
ing a typical value of the time 2l/v required for a wave to travel down the line and be
reflected back to the input. For a line 100 meters long (328 feet) this time is
2l 2 × 100
t= = = 6.7 × 10 −7 second
v 3 × 10 8
or 23 µ sec. The steady-state condition is practically achieved after a few of these

reflection periods.
Nevertheless, in some very-short-pulse applications the transient effects would be
important with this length of line, if an appreciable reflected wave occurred at the load
end. Transient effects cannot be ignored if the time 2l/v is comparable to the reciprocal
of the highest modulation frequency fm of the rf applied voltage. For example, in televi-
sion transmission the video modulation may contain frequencies up to several megahertz,
so that 1/fm may be less than a microsecond, and transient effects in a 100-meter line
length would be of consequence. (The picture quality would be degraded. The remedy
would be to eliminate the reflections at the load by methods that will be described in
sec. 2–3.)
2.2. Transmission-Line Equations

2.2.1. Steady-State Lossless Line Equations
The equations of the steady-state case describe the standing waves of voltage and current
along the line. The values of V and I in these equations may be interpreted as either the
amplitudes or the rms values. Since the standing waves do not change with time, the
time variable t does not appear in them; the radio frequency time variation sin(wt + f)
has in effect been “factored out” of both sides of the equations. Moreover, the
steady-state equations are customarily written in terms of phasors, as discussed in
Appendix C.
The voltage equation is
 Z cos β ( l − x ) + jZ 0 sin β (l − x ) 
V ( x ) = Vi  L  (2–15)
 Z L cos βl + jZ 0 sin βl 
and the current equation is
Vi  Z 0 cos β ( l − x ) + jZ L sin β ( l − x ) 
I(x) = (2–16)
Z 0  Z L cos βl + jZ 0 sin βl 

The input impedance, Zi, may be obtained dividing V(0) by I(0), as given by (2–15) and
(2–16) for x = 0. The resulting equation is
 Z cos βl + jZ 0 sin βl 
Zi = Z 0  L  (2–17)
 Z 0 cos βl + jZ L sin βl 
This equation is often written in a modified form, obtained by dividing through both the
numerator and denominator of (2–17) by Z0 cos bl, which gives
 ( Z Z ) + j tan βl 
Zi = Z 0  L 0  (2–18)
 1 + j ( Z L Z 0 ) tan βl 
The definitions of the quantities in these equations are listed below for ready
reference:
Transmission-Line Equations 55
Zi —impedance presented to a source (generator) at the transmission-line input

terminals, ohms
ZL—impedance connected at load end of line, ohms
Z0—characteristic impedance of the (lossless) line, = L C ohms (Eq. 2–7)
V(x)—voltage on line at distance x from the input terminals, volts
Vi —voltage applied to the input terminals, volts
I(x)—current in line at the point x, amperes
l—total length of line (Fig. 2–2)
b —2p/l, where l is the wavelength in the same units as l and x, e.g., meters
It is apparent that the quantities V(x), I(x), and Zi are in general represented by complex
numbers,* since the quantity j = −1 appears in the equations. Even if j did not appear
explicitly, it is implicit in the quantity ZL, which is in general a complex impedance
having a real (resistive) part RL and an imaginary (reactive) part XL.† That is,
Z L = RL + jX L (2–19)
2.2.2. Some Important Special Line Conditions

Lengths of transmission lines have certain properties that are of special interest for par-
ticular values of the length l or of the load impedance ZL. As equation (2–18) indicates,
a transmission line is in general an impedance transformer. That is, if an impedance ZL
is connected at the load end, a different impedance Zi may be presented to a source at
the input end. As will be shown, sections of transmission line may also exhibit the prop-
erties of resonant circuits. Therefore transmission-line sections are often used in lieu of
more conventional circuits as transformers and as resonant circuits, especially at the
higher frequencies where the equivalent coils and condensers become so small that their
power-handling capabilities are limited, and losses may be high even at low powers.
These properties of lines lead to results of special interest for particular cases.
(1) ZL = Z0. This condition, when applied to equation (2–18), gives
 1 + j tan βl 
Zi = Z 0   = Z0 (2–20)
 1 + j tan βl 
That is, when the load impedance is equal to the characteristic impedance, the input
impedance is also equal to this value. The line then acts as a one-to-one transformer.
This equality of the load and characteristic impedances is called the matched load condi-
tion. The load impedance is said to be matched to the characteristic impedance of the line.
The one-to-one transformation of impedance results regardless of the length of the line, l.
* The reader is assumed to have a basic familiarity with the concept of complex numbers and
complex-number algebra. For reference and review, Appendix C covers the elements of this subject.
†
The subscript L here stands for “load.” The symbol XL often stands for “inductive reactance,” as opposed
to capacitive reactance XC. But here XL denotes the load reactance, which may be either inductive or
capacitive.
Moreover, if the condition ZL = Z0 is also substituted into (2–15) and (2–16), it is found
that the amplitudes of V(x) and I(x) remain constant for all values of x, that is, everywhere
on the line. Only the phases change, as would be expected from (2–3) and (2–8); in fact,
these equations of the infinite line now apply to the finite line (if and only if ZL = Z0). This
property is of great importance, for it provides a method of eliminating reflected waves
and transient effects. Since there is no standing wave, with maxima and minima, such a
line is called a flat line. It is important to note that since Z0 for a lossless line is an entirely
real (resistive) impedance, the matched-load condition requires that the load impedance
be nonreactive (purely resistive) as well as have the magnitude of Z0.
(2) l = l/2. (Line length equal to one-half wavelength.) For this case, the angle
bl that occurs in (2–18) becomes equal to p radians or 180 degrees. Since
the tangent of this angle is zero, the equation becomes
ZL
Zi = Z 0 = ZL (2–21)
Z0
But there are two respects in which the half-wavelength one-to-one transformer is infe-
rior to the matched line. First, there is a standing wave on the half-wavelength (or integral-
half-wavelength) line, which means there are reflections and transient effects. Second,
since the one-to-one transformation depends on the fact that l/l has a particular value, the
line will have this property only at one frequency and its integral multiples. If it is desired
to operate over a wide range of frequency without readjustment of the line length, the
matched-load line should be used. Nevertheless, the half-wavelength line is useful as a
one-to-one impedance and voltage transformer when its limitations are not of concern.
(3) l = l /4. (Line length equal to one-quarter wavelength.) In this case the angle
bl in (2–17) becomes equal to p/2, or 90 degrees. Note that (2–17) rather
than (2–18) is used because the tangent of 90 degrees is infinite, and conse-
quently (2–18) leads to difficulties that require careful handling.) Since
cos 90° = 0 and sin 90° = 1, (2–17) becomes
 jZ  Z 2
Zi = Z 0 0  = 0 (2–22)
 jZ L  Z L
The quarter-wavelength line transforms a load impedance ZL that is smaller than Z0 into
a value Zi that is larger than Z0; and vice versa. It is sometimes called an impedance
inverter for this reason. If ZL is resistive, Zi will also be resistive (real). Thus the quarter-
wavelength transformer is useful when it is desired to transform a resistive impedance
into a different resistive value, either larger or smaller. The desired transformation is
accomplished by choosing the appropriate value of Z0 in accordance with (2–22). The
procedure for obtaining a desired value of Z0 is described in sec. 2–4; it depends on the
ratio of the spacing of the line conductors, s in Fig. 2–1, to their diameter d. Since these
physical dimensions cannot practically have an infinite range of values, the transforma-
tion ratio of the quarter-wave transformer is subject to practical limitation. Nevertheless,
it is a very useful device because of its simplicity and the ease with which its behavior
is calculated. Lengths l equal to odd-number multiples of l/4 will have the same trans-
formation property, but as the length is made longer the sensitivity to a small change of
frequency becomes greater.
The voltage transformation of the quarter-wave transformer is found from (2–15) by
taking x = l, and as before bl = 90°. The result is
ZL
VL = − jVi (2–23)
Z0
which shows that when ZL is real the amplitude is changed by the factor ZL /Z0 and the
phase is changed by 90 degrees. (It should be noted here that (2–23) describes the trans-
formation of input voltage to output voltage, whereas (2–22) describes the transformation
of output impedance to input impedance, since these are the usual directions of interest.)
A transformation of the current also takes place, but in the inverse ratio, so that the
product of voltage and current is unchanged.
(4) Open-circuited and short-circuited lines. If a transmission line has no load

connected at the load end (open-circuited), the load impedance ZL is infinite
(or nearly so). If ZL in (2–17) becomes infinitely large, the terms jZ0 sin bl
in the numerator and Z0 cos bl in the denominator are so much smaller that
they can be ignored (considered equal to zero), and the equation becomes
Zi = − jZ 0 cot βl ( for Z L → ∞ ) (2–24)
Similarly if the load end of the line is short-circuited, the result is
Zi = jZ 0 tan βl ( for Z L = 0 ) (2–25)
In both cases the input impedance Zi is purely reactive (imaginary). This means that the
input terminals of such lines “look like” the terminals of either an inductance or a capaci-
tance, depending on the length of the line. If the length is between zero and a quarter
wavelength, so that bl is an angle between zero and 90 degrees, the open-circuited line
will be capacitive, a negative reactance in accordance with (2–24). In contrast, the short-
circuited line of the same length will be inductive, following (2–25). For lengths between
a quarter and a half wavelength, bl lies between 90 and 180 degrees, and both the tangent
and cotangent are negative; therefore the open-circuited line becomes inductive, and the
shorted line becomes capacitive. For each successive quarter wave increase of length,
their behaviors interchange in this way. Such line sections may therefore be utilized as
elements of inductance-capacitance circuits; and they are so used, especially at the higher
frequencies (shorter wavelengths) where the required line lengths are not great.
Open-circuited and short-circuited lines can also behave like resonant circuits. This
behavior occurs when the length of the line is an integral multiple of a quarter wave-
length; then tan bl → ∞, and cot bl = 0. The quarter-wave short-circuited line presents
an infinite impedance at its input terminals, like a parallel-resonant LC circuit; and the
quarter-wave open-end line presents a zero impedance at its input terminals, like a series-
resonant LC circuit. These behaviors are interchanged for each successive quarter-wave
increase of the line lengths. Transmission-line resonant circuits have many applications
in high-frequency radio circuitry and in antenna design.
2.2.3. Impedance-Admittance Relationships

It is often convenient, in transmission-line calculations, to work with admittance (Y),
conductance (G), and susceptance (B), which are respectively the reciprocals, in a some-
what restricted sense, of impedance (Z), resistance (R), and reactance (X). The restriction
applies only to the conductance-resistance and susceptance-reactance relationships. The
relationship of impedance and admittance is, without any restriction, given by
1
Z= (2–26)
Y
That is, they are reciprocals of each other. This equation holds when both Z and Y are
complex, as well as for purely resistive or reactive cases. That is, in general,
Z = R + jX (2–27)
and
Y = G + jB (2–28)
with the understanding that R and X refer to resistance and reactance in series, and G
and B refer to conductance and susceptance in parallel.
The restricted nature of the reciprocity between conductance and resistance, and
between susceptance and reactance, may be understood in terms of the diagrams of Fig.
2–4. In Fig. 2–4a, it is evident that Z = R + j0. Substituting this value for Z in (2–26),
and also applying (2–28), immediately gives Y = 1/R = G + j0. Thus in this simple case,
G = 1/R. In other words, the reciprocal relationship between G and R holds for a purely
resistive (conductive) circuit or branch of a circuit.
In Fig. 2–4b, the impedance is Z = jX (purely reactive). (The reactance in this diagram
is shown as a box, since for the purposes
of this discussion it may be either an
R1
R X G2 B2 inductance or a capacitance, or any com-
(G) (B) (R2) (X2)
X1 bination of the two.) If this value of Z is
substituted in (2–26), the result is Y = 1/jX
(a) (b) (c) (d) = −j(1/X). If this result is compared with
FIGURE 2–4. (2–28), it is evident that B = −1/X. Thus
Various two-terminal arrangements of for a purely reactive circuit, or branch of
resistance, reactance, conductance, and a circuit, the susceptance is the negative
susceptance. reciprocal of the reactance.
The restricted nature of the reciprocity relationships becomes evident when

Fig. 2–4c is considered. The impedance Z1 is of course simply R1 + jX1. Therefore the
admittance is
1 1 1  R1 − jX1   R1   X 
Y1 = = =   =  2 2
− j 2 1 2  (2–29)
Z1 R1 + jX1 R1 + jX1 R1 − jX1 R1 + X1   R1 + X1 
Again comparison with (2–28) makes it evident that for this circuit:
R1
G1 = (2–30)
R12 + X12
and
− X1
B1 = (2–31)
R12+ X12
On the other hand, for the parallel circuit of Fig. 2–4d, the corresponding relations are
simply G2 = 1/R2 and B2 = −1/X2. It is therefore apparent that the circuits of Fig. 2–4c
and 2–4d are equivalent if
R12 + X12 1
R2 = = (2–32)
R1 G2
and
R12 + X12 1
X2 = = (2–33)
X1 B2
“Equivalent” means that if the same voltage is applied to the terminals of either circuit,
the current that flows at the terminals will be exactly the same for both circuits. (Of
course, this will in general hold—for fixed values of inductance and capacitance—at only
one frequency.) This equivalence is of great importance and utility in impedance-
matching calculations. It means that the impedance at the input terminals of a transmis-
sion line, for example, can be regarded as either a series combination or a parallel
combination of resistance and reactance, whichever happens to be most convenient.
Equation (2–18) for the input impedance of a line can be converted, by applying (2–26)
to it, into an equation for the input admittance, Yi, of the line, where Yi = 1/Zi. The
equation is
 1 + j (Y0 YL ) tan βl 
Yi = Y0   (2–34)
 (Y0 YL ) + j tan βl 
The symbol Y0 denotes the characteristic admittance of the line, which is simply the
reciprocal of the characteristic impedance; that is, Y0 = 1/Z0. The symbol YL denotes the
load admittance, which is equal to 1/ZL.
The equations for open-circuited and short-circuited lines, (2–24) and (2–25), can also
be expressed in terms of the input susceptances. For the open-circuit termination, the
equation corresponding to (2–24) is
Bi = Y0 tan βl (2–35)
and for the short-circuit termination, the equation corresponding to (2–25) is
Bi = −Y0 cot βl (2–36)
Equations (2–35) and (2–36) can be obtained either by making the substitution Xi = −1/Bi
in (2–24) and (2–25), or by letting YL be zero for the open-circuit case and infinite for
the short-circuit case in (2–34.)
2.2.4. Reflection Coefficient and VSWR

Just as in reflection of electromagnetic waves from a surface (sec. 1–2), the fraction of
the incident wave voltage (or current) reflected from the load end of a line and the phase
change that occurs in the reflection are described by a reflection coefficient, r. It is a
complex number (see Appendix C), since it describes both a magnitude and a phase angle,
and can therefore be expressed in the form
r = r e jφ = r {cos φ + j sin φ } (2–37)
where |r| is the magnitude (modulus) of the complex quantity and f is the phase angle.
The magnitude of the reflection coefficient is the ratio of the amplitudes (or rms values)
of the reflected and incident voltage (or current):
V0(r )
r = (2–38)
V0(i )
The phase angle is the phase difference between the phases of the incident and reflected
waves relative to an arbitrary reference phase:
φ = φi − φr (2–39)
(Since the phase of the incident wave is the usual reference, ordinarily fi = 0, and
f = −fr.)
The reflection coefficient is determined entirely by the relationship of the load
impedance, ZL, to the characteristic impedance of the line, Z0. Analysis shows that the
relationship is
Z L − Z0 ( Z L Z0 ) − 1
r= = (2–40)
Z L + Z0 ( Z L Z0 ) + 1
This is a complex-variable equation, in general, since ZL may be complex, or reactive.

If ZL is real (resistive) and equal to or larger than Z0 (assumed also to be real, as it
will be for a lossless line), then r will be real and positive, which means that r = |r|, and
f = 0. If ZL is real and smaller than Z0, r will of course again be real, but negative. It
can be deduced from (2–37) that this means that f is 180° (since cos 180° = −1 and
sin 180° = 0). That is, in this case r = −|r|. But when ZL is reactive, or complex, f has
values other than 0 or 180°, and r is complex.
The phase relations referred to in the foregoing discussion are those of the incident
and reflected voltages at the load (i.e., immediately before and immediately after
reflection). The phases of both of these voltages at other points along the line will have
other values, since the phase of a wave (either in space or on a line or in a wave
guide) changes with distance from a reference point in accordance with (2–2) and (2–3).
In (2–3), the phase angle at an instant of time t1 at a distance x from the reference point
is wt1 − bx + f1 (where f1 is the phase angle when t = 0 and x = 0). (As previously given,
w = 2p f, where f is the frequency, and b = 2p/l, where l is the wavelength.)
In computing the phase angle the distance x is always measured in the direction of
travel of the wave, so that x increases in one direction for the incident wave and in the
other direction for the reflected wave. The two phase angles therefore do not have a
constant difference along the line (as they would if the two phases progressed in the same
direction along the line). The result is that at some points on the line the incident and
reflected voltages will be in phase and will add, resulting in a voltage maximum Vmax
given by
Vmax = V0(i ) + V0(r ) (2–41)
At other points the two voltage waves will be exactly out of phase and will therefore
subtract, resulting in a voltage minimum Vmin given by
Vmin = V0(i ) − V0(r ) (2–42)
the ratio of the maximum to the minimum voltage is called the voltage standing wave
ratio, VSWR, that is
Vmax V0(i ) + V0(r )

VSWR = = (2–43)
Vmin V0(i ) − V0(r )
Obviously the VSWR is a number equal to or greater than one. (It is equal to one when
there is no reflected wave, or V0(r) = 0. From (2–38) and (2–40) it is apparent that
VSWR = 1 when |r| = 0 and therefore when ZL = Z0.)
2.2.5. Standing-Wave Patterns

The nature of the standing-wave voltage pattern may be found by plotting V(x) as given
by equation (2–15), for given values of ZL and Z0. However, for this purpose it is more
convenient to obtain an equation for the voltage V relative to the voltage VL existing at
the load terminals, in terms of distance d measured back along the line from these ter-
minals. In manipulating (2–15) to obtain such an expression in turns out that both Vi and
l drop out, and a relatively simple formula is obtained:
 Z 
V ( d ) = VL  cos β d + j 0 sin β d  (2–44)
 ZL 
The magnitude (modulus) of V(d ), in terms of ZL /Z0 = RL /Z0 + jXL /Z0, with RL /Z0
abbreviated to R and XL /Z0 abbreviated to X, is
2 2
  X    R  
V ( d ) = VL cos β d +  2  sin β d  +  2  sin β d  (2–45)
 R +X 
2
  R + X 
2

Typical plots of |V(d)| are shown in Fig. 2–5 for a number of values of RL /Z0 and
XL /Z0.
Noteworthy features of these plots are
Voltage Current
the following. (1) The maxima are sepa-
RL = Z0 /2 rated by l/2 and the minima are separated
XL = 0
by l /2; the distance from a minimum to
Vmin Vmax
a maximum is l /4. (2) The minima are
more sharply defined than the maxima;
RL = 2Z0
XL = 0
consequently, in making standing wave
position measurements it is better to
dm measure the position of a minimum than
RL ≠ 0, ∞
the position of a maximum (greater accu-
XL ≠ 0, ∞ racy is possible). (3) When a resistive
(nonreactive) load has a value less than
dm
Z0, a voltage minimum is located at the
|V (d )|
ZL = 0 load terminals. (4) When a purely resis-
tive load has a value greater than Z0, a
dm voltage maximum is located at the load
terminals. (5) When the load is complex
ZL → ∞
(partially or wholly reactive), the voltage
9l/4 2l 7l/4 3l/2 5l/4 l 3l/4 l/2 l/4 0
at the load is neither a minimum nor a
d maximum. (6) When the load is either
FIGURE 2–5. zero or infinite, the standing-wave ratio is
Standing-wave voltage patterns for several infinite (since Vmin = 0) and the minima
ZL/Z0 conditions. have the form of cusps.
2.2.6. Determination of Load Impedance by

Standing-Wave Measurement
Equation 2–40 indicates that if the reflection coefficient is known, the load impedance
ZL can be determined. This is shown more explicitly by rearranging the equation as
follows:
1+ r 
Z L = Z 0
 1 − r 
(2–46)
The relationship of the magnitude of the reflection coefficient, |r|, to the standing-wave
ratio VSWR has been shown indirectly through (2–38) and (2–43). From these equations
it may be deduced that
VSWR − 1
r = (2–47)
VSWR + 1
Thus a measurement of the VSWR provides one ingredient necessary for determination
of ZL, in accordance with (2–46). The additional ingredient required is of course the phase
angle f of (2–37).
It can be shown that f is related to the distance dm, which as shown in Fig. 2–5 is the
distance from the load terminals to the first voltage minimum. The relation is
φ = π − 2β dm = π 1 − m  radians
4d
 λ  (2–48)
= 180 1 − m  degrees
4d
 λ 
Therefore, if the two quantities VSWR and dm are measured, ZL can be calculated,
through (2–46), (2–37), (2–47), and (2–48). Moreover, Zi can also be calculated through
(2–18), either by first calculating ZL or directly by substituting (2–46) into (2–18), which
gives
  1 + r  + j tan βl 
 1− r  
Zi = Z 0   (2–49)
1 + j  1 + r  tan βl 
 1− r  
In practical work (2–46) and (2–49) are solved, including the intermediate steps of (2–47)
and (2–48), by using a Smith Chart, as described in sec. 9.12.
2.2.7. Attenuation
The results presented thus far have all been based on the assumption that the line is loss-
less, which means that R1 in Fig. 2–3 is zero and R2 is infinite. In real transmission lines
these assumptions are of course not perfectly justified; there are some losses. When the
loss of a line is considerable, even the basic form of the transmission line equations is
considerably modified. However, when the losses are fairly small, as is usually the case,
the equations that have been given can be used for all purposes except computing the
power ultimately delivered to the load. This power will be slightly less than the power
delivered to the line at its input terminals. The power loss in decibels is given by multi-
plying the length of the line by a decibel attenuation constant. For most manufactured
lines, the value of this constant is published in the manufacturer’s sales literature.
Additionally, a useful collection of data on transmission lines and waveguides is
contained in Lowman and Simons (2007).
The attenuation constant (sec. 1.1.9) can also be expressed in terms of the conductor
resistance R per unit length and the leakage conductance G per unit length. Ordinarily,
transmission lines used with antennas are of the low-loss type. For these lines, the
attenuation constant, when expressed in terms of decibels per meter (dB/m), is
 R G 
α dB = 8.686  +
 2 Z 0 2Y0 
dB m (2–50)
Here Z0 is, as usual, the characteristic impedance of the line and Y0 is the characteristic
admittance (= 1/Z0). In terms of Fig. 2–3, R = R1 and G = 1/R2. When the line attenuation
is appreciable, a more complicated formula must be used. “Appreciable attenuation,” for
practical purposes, may be taken to mean more than about one decibel per wavelength
of the line, insofar as the validity of (2–50) is concerned.
The equations for transmission-line input impedance, voltage, and current, (2–15)
through (2–18), and subsequent equations derived from them, are accurate for most
antenna applications. However, this is not so if the total attenuation* of the line (adBl)
is greater than about a decibel. Attenuation reduces the amplitude of the reflected
wave as it travels back toward the line input, and the standing-wave ratio consequently
diminishes toward the input. Thus, with large attenuation, the standing-wave ratio as
calculated from equation 2–47 is valid only in the vicinity of the load end of the line.
Additionally, if the attenuation is great enough the input impedance of the line will be
essentially equal to the characteristic impedance regardless of the reflection coefficient
of the load. More detailed treatments of transmission lines give equations for high attenu-
ation, analogous to (2–15) through (2–18) (see, e.g., Lowman and Simons 2007,
p. 51–5).
* Total attenuation from ohmic loss is adBl without a standing wave. However, e.g., with VSWR = 2, the
attenuation is 1.25 times adBl, because of increased mean-squared voltage across the line (Ragan 1948,
pp. 30–31).
Impedance Matching and Power Division 65
2.3. Impedance Matching and Power Division

In their role as the connecting link between a transmitter and an antenna, or between an
antenna and a receiver, transmission lines affect the efficiency of power transfer. In some
cases they may be arranged to apportion a total amount of power among a number of
loads. Methods of adjusting the line impedance relations to achieve optimum results in
these functions will now be discussed.
2.3.1. The Matching Principle

A radio transmitter is in effect a generator possessing an internal electromotive force
(emf) and an internal impedance, which may be complex, that is, Zg = Rg + jXg. A
well-known theorem of a-c circuit theory states that such a generator will deliver the
maximum possible power to a load whose impedance is the complex conjugate of
Zg, that is, equal to Rg − jXg.
The ultimate load for the transmitter is the antenna, but the immediate load may be
the input impedance of the transmission line, Zi = Ri + jXi. Consequently, in order for
the transmitter to deliver its maximum power output it would seem necessary that
Ri = Rg and Xi = −Xg.
The value of Zi, as shown by (2–18), depends on the values of ZL, Z0, and l. It would
therefore seem possible to choose or adjust the values of these quantities to achieve the
desired value of Zi. But in practice the desired result is not usually accomplished in this
way. The quantity l, for example, may be determined primarily by the necessary separa-
tion of the transmitter and antenna and will not be independently adjustable. The possible
range of values of Z0 is practically somewhat limited. Moreover, it is usually desired to
eliminate or minimize standing waves on the line, and this means that ZL must be at least
approximately equal to Z0.
Therefore the impedance usually presented at the input terminals of the line will be
Zi = Z0. An impedance transformer is then inserted between the line input and the transmit-
ter output to transform the impedance Z0 into the value required for maximum transmitter
power output, or optimum performance where best operation from some standpoint other
than maximum power is required. The actual load impedance presented by the antenna to
the transmission line output terminals may not be equal to the desired value, Z0. Another
impedance transformer may be required between the load end of the line and the antenna
terminals, and this one must transform the antenna impedance, Za, into the value Z0.
Sometimes these transformers may consist of coupled coils of wire or of coil-
and-condenser circuits, especially likely at low frequencies. At frequencies in the multi-
megaHertz region, however, these transformers often consist of sections of transmission
lines in various arrangements. Some of the possible arrangements and their principles of
operation will be described.
The quarter-wave impedance-inverting transformer has been described in sec. 2.2.2.
Its principle of operation is given by (2–22). It is most useful when the desired transfor-
mation ratio is not great and can be predicted or determined to be stable, since this type
of transformer is not easily adjustable.
Short circuit 2.3.2. Stub Transformers

s A type of transmission-line transformer
frequently employed is diagrammed in
Zs
Fig. 2–6. The principal element of this
Line ZP Zy ZL Load transformer is a short-circuited section of
input line whose open end is connected across
P Q
y the main line at a certain distance,
y, from the load. This short-circuited
FIGURE 2–6.
branching section is called a stub, or
Single-stub impedance transformer. sometimes a tuning stub. At its point of
connection to the main line, P, it is
equivalent to a pure reactance, in accordance with (2–25).
The operation of this transformer is analyzed by dissecting the line at the point P
(figuratively speaking) and considering separately the dissected elements, then consider-
ing the result of putting them back together. As viewed from the point P, the section of
the main line to the right (toward the load) will present an impedance given by (2–18)
with l equal to y. The stub, as viewed from the point P, presents a reactance given by
(2–25) with l equal to the stub length s. For simplicity it will be assumed that the
characteristic impedances Z0 are the same for all the line sections.
When these dissected elements are reconnected, they are in parallel, and they repre-
sent, in their combined effect, the load impedance Zp of the main line at the point P. But
because they are in parallel, it is advantageous to calculate their admittances. The sum
of these admittances will then constitute a load admittance Yp for the main line at point
P, which can be inserted as YL into (2–34). If the objective is to present a matched load
to the main line at point P, however, the requirement is simply to make the combined
admittances of the stub and the load section PQ equal to Y0 = 1/Z0.
It can be shown that for some length y the admittance of the line section PQ will
consist of a conductance Gy and a susceptance By such that Gy = Y0. The necessary length
to achieve this result may be calculated from (2–34), or it may be determined more
speedily using a Smith Chart (sec. 9.12). The problem then is simply to eliminate the
susceptance By without changing the conductance Gy = Y0.
If the susceptance of the stub is Bs, the total admittance of the parallel combination
at P is
Yp = Gy + jBy + jBs (2–51)
If now it can be arranged that Bs = −By, these two susceptances in (2–51) will cancel
each other, leaving
Yp = Gy = Y0 (2–52)
This admittance presents a matched load to the main line at point P, and so there will be
no standing wave to the left of this point, that is, between point P and the line input. This
is the desired result.
Impedance Matching and Power Division 67
The susceptance of the stub, Bs, is computed from (2–36). By making s a suitable
length, it is demonstrable that any needed value of Bs can be achieved. Therefore, if the
two lengths y and s can be varied at will, the desired transformation of impedance
can be achieved. In practice the position and length of the stub are roughly determined
by Smith Chart calculations based on standing-wave measurements. Then the lengths
are experimentally adjusted to minimize the standing wave ratio (VSWR) on the
main line.
When it is not practical or advisable to provide a movable stub (one for which the
distance y can be varied), the same result can be accomplished by means of two fixed-
position stubs separated a suitable distance, usually between one-fourth and three-eighths
of a wavelength. The analysis of the impedance transformation proceeds on the same
principles, though it is of course somewhat more involved.
2.3.3. Power Dividers

For many purposes, particularly in array antennas, it is necessary to divide the power
from a single transmitter among two or more loads. The simplest case of two loads will
be considered first.
If the total transmitter power is first fed into a single transmission line, this line may
then be “branched” into two lines, as shown in Fig. 2–7. Again, since the branches are
in parallel, it is advantageous to work with the admittances. The net load admittance
presented to the main line at the branch point will be the sum of the two branch-line
admittances, YA and YB. It will usually be desired to make this sum equal to the charac-
teristic admittance Y0(M) of the main line. The requirement is simply
Y0( M ) = YA + YB (2–53)
If each branch line is terminated in its own characteristic admittance, Y0(A) and Y0(B), as
will usually be true, then of course YA and YB will be equal to the characteristic values.
Since the branch lines are in parallel, the voltage applied to each of them at the junc-
tion point is the same and is equal to the voltage output of the main line, VM. Therefore,
the power delivered to each branch is
PA = VM2 G A (2–54)
hA
nc ad
PB = VM2 GB (2–55) Bra Lo
Main line YA
Line YM
and the total power is input Br
YB an
Branch point ch
Ptotal = PA + PB = VM2 ( G A + GB ) (2–56) Lo
ad
B
and GA and GB are the conductance com-

ponents of the admittances YA and YB. FIGURE 2–7.
These equations establish the power- Parallel branching transmission lines.
division property of the junction. That is, the fraction of the total power delivered to each
branch will be
PA GA
FA = = (2–57)
Ptotal G A + GB
PB GB
FB = = (2–58)
Ptotal G A + GB
If each branch line is assumed to be terminated in its characteristic admittance, since

these admittances are real for lossless lines, the conductances GA and GB will simply be
equal to Y0(A) and Y0(B) respectively. If, in addition, it is desired that the main line be
terminated in its characteristic admittance at the junction, the requirement is
Y0( M ) = Y0( A) + Y0( B ) = G A + GB (2–59)
For this case (2–57) and (2–58) become
Y0( A)
FA = (2–60)
Y0( M )
Y0( B )
FB = (2–61)
Y0( M )
It would be possible to divide the power at a junction among more than two branching
lines. The analysis of the power division would be a simple extension of the foregoing
analysis; the fraction of power delivered to each branch would be equal to the conduc-
tance component of the input admittance of that branch, divided by the sum of the con-
ductances of all the branches. The conductances in the usual cases would be equal to the
characteristic admittances of the branches. It is not customary to divide power in this
way, however, partly because the required characteristic impedances of the branch
lines become impractically large and partly because there are mechanical construction
difficulties, especially with lines of the coaxial type.
When multiple power division is required, it is more common to employ one of the
successively branching line structures shown schematically in Fig. 2–8. In these diagrams
the two conductors of the lines are shown as single lines, for the purpose is solely to
show the branching arrangements.
In each of these arrangements the ultimate power division is accomplished by a
number of successive two-branch divisions. The required division of the total power to
the ultimate loads results in a system of simple simultaneous equations for the division
ratio required at each branch; this ratio is then achieved by the method that has been
described. Although eight loads are shown in Fig. 2–8, the methods are not restricted to
this number.
Forms of Transmission Lines 69
In addition to the power-division 1

requirement, it may also be required that 2
Eight loads
the voltages and currents have the same 3
4
phase at each load. This is readily achieved Input
5
in the arrangement of Fig. 2–8a, by
6
keeping the lengths of the parallel branches Branch point
7
equal. In Fig. 2–8b it would be necessary
8
to make the lengths denoted by d equal to (a)
a wavelength or an integral number of
Branch point
wavelengths; or d may be a half wave- d
length if the line is of the two-wire bal- Input
anced type and if the line is “turned over”
in each half-wavelength interval to reverse 1 2 3 4 5 6 7 8
Eight loads
the polarity of the voltage. (This cannot (b)
be done, of course, with a coaxial or other FIGURE 2–8.
unbalanced line type).
Alternative arrangements for division of power
It may also be necessary to have among multiple loads.
stepdown impedance transformers in the
branch lines to keep the characteristic
impedances required in the successive branches from building up to too high a value; in
short, there are numerous practical problems. Here the purpose has been primarily to
present basic principles.
It is possible also to have two lines branching with a series connection instead of par-
allel, although such an arrangement is less common. The analysis would be made in a
similar way, except that it would be more convenient to work with the impedances than
with the admittances.
2.4. Forms of Transmission Lines

Two-conductor transmission lines of the type thus far discussed may take a number of
physical forms. The principal forms are: (1) two-wire balanced lines; (2) coaxial (unbal-
anced) lines; (3) parallel-plate or “strip” lines (balanced or unbalanced). Examples of
these forms are shown in Fig. 2–9.
Each of these forms has its advantages and disadvantages. The two-wire line, which
is usually cheaper, may be fabricated from readily available components (e.g., ordinary
wire). Also it is balanced with respect to the ground and may therefore be used to connect
directly to balanced antennas. A disadvantage is that some radiation may occur, in the
manner discussed in chapter 1 in connection with Fig. 1–11. As long as the spacing of
the wires, s, is small compared to a wavelength, the radiation will be negligible; but at
extremely high frequencies (short wavelengths) this condition may not be fulfilled, and
appreciable radiation may result. This is of course undesirable, as radiation from the
transmission line does not usually have the desired directional and polarization charac-
teristics. It represents wasted power, at best, and at worst it may do harm by its presence
in undesired locations.
Insulating
spacers
Two-wire line
Insulating
“wafers”
Coaxial line
a b
Strip line
FIGURE 2–9.
Perspective sketches and transverse sections of two-wire, coaxial, and strip transmission lines.
Coaxial lines confine the fields entirely to the space between the inner and outer con-
centric conductors; therefore there is no radiation problem. Being unbalanced, these lines
require special balanced-to-unbalanced transformers (called “baluns”) to connect them
to any balanced source or load. However, sources and loads are in some cases unbalanced,
so that the unbalanced character of the line is as often an advantage as a disadvantage.
Coaxial lines have the advantage of permitting the interior to be completely sealed against
weather and contamination. The internal space may be pressurized to increase voltage
breakdown and to resist intrusion of moisture. Some coaxial lines are made with a flexible
braid outer conductor and flexible solid-dielectric material between the conductors. Such
lines are flexible and compact and find many uses at low and intermediate power levels.
For very high-power transmitting applications, coaxial lines employ solid copper pipe,
with outer-conductor diameters up to 9 or more inches.
Forms of Transmission Lines 71
The strip line is a modification of the two-wire line in which the flat faces of the
adjacent conductor surfaces are wider than the spacing between the conductors. Accord-
ingly the field is virtually confined to the space between the conductors, and there is very
little radiation. The two strips may be alike (balanced), or one of them may be a large
flat (grounded) plate. Often a solid dielectric material is used as a spacing material. It is
especially convenient for some purposes, usually at low power levels.
2.4.1. Characteristic Impedances

The characteristic impedances of these line types may be calculated from the transverse-
section geometry. For the two-wire line the formula is
log   ohms
276 2s
Z0 = (2–62)
εr  d
where s and d are the center-to-center conductor spacing and the conductor diameter
respectively, as shown in Fig. 2–9. The relative permittivity er is of the medium between
and around the conductors (er = 1 for air or vacuum). Typical values of Z0 for two-wire
lines range from about 200 to 800 ohms.
For a coaxial line the characteristic impedance is given by
log   ohms
138 b
Z0 = (2–63)
εr  a
with b the inner diameter of the outer conductor and a the outer diameter of the inner
conductor. The relative permittivity er is for the material between the conductors. Typical
values of Z0 range from about 20 to 100 ohms.
For the strip line the following formula is good as long as the dimension b (Fig. 2–9)
is much smaller than the dimension a:
377  b 
Z0 =   ohms (2–64)
εr  a 
The typical range of characteristic impedances is about the same as for coaxial lines.
Numerous variations of these basic forms are sometimes used; however, there is a
practical upper-frequency limit for any given type of line. Usually the limit is set either
by the losses in dielectric material or by the spacing of the conductors in relation to the
wavelength. When this spacing is an appreciable fraction of a wavelength, the open-wire
lines will radiate excessively. Coaxial and strip lines will act like waveguides, as will be
discussed in sec. 2–5, if the conductor spacing becomes comparable to or greater than a
half wavelength.
A basic property assumed for all the lines discussed is uniformity, that is, constant
spacings and conductor diameters. At joints, elbow turns, couplings, and so forth, there
may be discontinuities that cause reflections that can be troublesome if care is not taken
to minimize them. For broad-band impedance-transforming applications deliberately
nonuniform tapered lines are used. These lines have a characteristic impedance that
changes as the conductor diameters and spacings change, so that a gradual impedance
transformation takes place.
2.5. Waveguides
At frequencies too high for successful operation of conventional transmission lines, rf
power may be transmitted by means of waveguides. A waveguide in its commonest form
is a hollow pipe, sometimes circular but more often rectangular in cross section, and
occasionally of some other form. The dimensions of the cross section are such that an
electromagnetic wave can propagate in the interior of the guide. Being confined by the
walls of the guide, the field does not spread spherically as it would in free space, so there
is no inverse-square-law decrease of the power density. There is some attenuation due
to currents in the walls, but generally this loss is less than in a coaxial line of comparable
size. Moreover, since no insulating supports are required, there is virtually no “leakage”
loss and less tendency to flashover at high voltages.
Obviously the analysis of waveguide behavior must be made in terms of electromag-
netic field concepts rather than in terms of an equivalent-circuit representation. The
starting point is Maxwell’s equations, and their solution requires advanced mathematical
methods. Yet the results are that waveguides behave essentially like the conventional
transmission lines that have been described, although there are some important
differences.
2.5.1. Phase and Group Velocities

In transmission lines the wave velocity is independent of the frequency, and for air or
vacuum dielectrics the velocity is equal to the velocity of free-space waves, 3 × 108
meters per second. But in wave guides the velocity varies with the frequency. Moreover,
it is necessary to distinguish between two concepts of wave velocity, the phase velocity
and the group velocity.
The phase velocity is the apparent velocity of a particular phase of the wave, for
example, a crest, or electric-intensity maximum or minimum. The basic method of deter-
mining this velocity is to make a measurement of the wavelength, l. This may be done,
for example, by setting up a standing wave and measuring the separation of two minima
(see Fig. 2–5). The frequency f being known, the phase velocity may then be calculated
by an inversion of (2–1).
v ph = f λ (2–65)
The other concept of velocity may be understood by considering a pulse of radio

waves, of the type used in radar. If such a pulse of waves is propagated in a waveguide,
its velocity can be measured by standard radar timing techniques. For example, the pulse
Waveguides 73
might be transmitted through a long guide of length l, then reflected back to the sending
end, and the elapsed time t might be measured. The velocity of the pulse (group velocity)
is given by the formula
2l
vgr = (2–66)
t
(The factor 2 occurs because the total distance traveled is twice the length of the
guide.)
If these two measurements are made at the same frequency in a waveguide, it will be
found that in general the two velocities are not the same. At some frequencies they will
be nearly the same; at others they will be considerably different. The distinction between
these two velocities was not necessary for transmission lines because in conventional
lines the group and the phase velocities are equal.
To explain why these velocities are different in waveguides requires a detailed analy-
sis, which cannot be given here. It must suffice to define them and to state, at appropriate
points, which velocity is meant. Unless otherwise stated, the phase velocity will always
be meant; it is the one of importance whenever considerations of wavelength are involved.
Yet obviously the group velocity is also of great importance in other applications—radar
signal transmission, for example. In general it is the velocity at which signals of any
kind—that is, intelligence—are propagated; it is also the velocity at which energy is
propagated.
The phase velocity is always either equal to or greater than the group velocity. The
following relationship holds between them in evacuated or gas-filled waveguides:
( v ph )( vgr ) = c 2 = 9 × 1016 (meters per second)2 (2–67)
That is, the product of the two velocities is equal to the square of the free-space propaga-
tion velocity, 3 × 108 meters per second. This means that Vph is always greater than the
free-space velocity and Vgr is always less, unless both are equal to c.
It is a principle of modern physics that no form of matter or energy can travel with a
velocity greater than the free-space velocity of light (electromagnetic waves). However,
this principle is not violated by phase velocities, because it is the group velocity that
represents the velocity of propagation of energy.
Since the phase velocity in waveguide is greater than the velocity in free space, the
wavelength at a given frequency will be greater in the guide than in free space. Using
the conventional symbol c for the velocity in free space, the relationship of the
wavelength in the guide, lg, to the wavelength in free space, l0, is:
 v ph 
λ g = λ0 
 c 
(2–68)
The distinction between the guide wavelength and the free-space wavelength is an
important principle of waveguide theory.
2.5.2. Cutoff Frequency

Whereas transmission lines of the conventional type cannot be operated practically above
some maximum frequency, waveguides are characterized by a minimum frequency, the
cutoff frequency fc. This is an absolutely limiting frequency; frequencies below this value
will not be propagated in the guide. It is determined by the cross-sectional dimensions
of the guide in relation to the wavelength.
A relationship exists between the guide-wavelength lg at a particular frequency f and
the cutoff frequency fc:
c
λg = (2–69)
f − fc2
2
where c is the free-space propagation velocity. This can also be written in another
form:
λ0
λg = (2–70)
1 − ( fc f )
2
Comparison of this equation with (2–68) shows that the phase velocity is given by
 λg  c
v ph = c   = (2–71)
 λ0  1 − ( fc f )
2
It is evident from this expression that if f becomes less than fc, the phase velocity becomes
imaginary (in the complex-number sense); physically this means that the wave is not
propagated. Equations (2–69) and (2–71) also show that as the frequency f approaches
the cutoff frequency fc, the phase velocity and the guide wavelength become infinite.
Consequently, the group velocity tends to zero, according to (2–67).
The foregoing results have all presupposed that the waveguide is filled with air (or a
vacuum) or any gas having virtual unity relative permittivity. Waveguides are seldom
filled with solid material, but, if such a low-loss material were used, the foregoing
equations still hold if c is replaced by c ε r , where er is the relative permittivity of the
material in the guide.
2.5.3. Rectangular Waveguides

The commonest form of waveguide has a rectangular cross section, as illustrated in
Fig. 2–10. The greater of the two transverse dimensions is customarily denoted by a, and
the lesser dimension by b. The dimension a determines the cutoff frequency fc, according
to the relation:
c
fc = (2–72)
2a
Waveguides 75
or
λc = 2 a (2–73)
where lc is the cutoff wavelength. (It is

to be noted that the cutoff wavelength is
a free-space wavelength.) This relation b
states that cutoff occurs at the frequency
for which the largest transverse dimen-
sion of the guide is exactly a half wave- a
length in free space.
FIGURE 2–10.
Rectangular waveguide.
2.5.4. Modes of Propagation
At frequencies slightly above the cutoff
frequency the configuration of the electro- Direction of
E propagation
magnetic field is schematically represented
of wave
in Fig. 2–11. The electric field (E) vectors, b
as shown, are parallel and are perpendicu- H
lar to the wide face of the guide. Their a
amplitude is greatest midway between the
narrow walls and decreases to zero at these FIGURE 2–11.
walls, in a cosinusoidal fashion. The mag- Electric and magnetic field vectors in a
netic field (H) vectors, shown dashed, are transverse plane, TE10 propagation mode.
also parallel to each other and perpendicu-
lar to the electric vectors. The magnetic
intensity is constant in the vertical direc-
tion across the guide section. The wave is
propagating in the longitudinal direction
of the guide, perpendicular to the E and
H vectors. This particular arrangement FIGURE 2–12.
field, the only possible one just above Magnetic field configuration in a longitudinal
the cutoff frequency, is called the TE10 section parallel to wide face, TE10 mode.
mode of propagation. The letters stand for
transverse-electric, meaning that the electric field lines are everywhere transverse
(perpendicular to the guide walls). Transverse-magnetic (TM) modes and transverse-
electric-magnetic (TEM) modes may also be propagated under certain conditions.
In this particular view of the TE10, field, Fig. 2–11, the magnetic lines are also trans-
verse; but this is not true for all views. As viewed in a longitudinal plane of the guide,
the magnetic field is seen to have longitudinal components, but the electric field does
not. Figure 2–12 shows a longitudinal section parallel to the wide guide wall, in which
the magnetic field lines only are shown to illustrate this point.
The subscript numerals 10 refer to the number of half-sinusoid cycles in the transverse
electric field pattern parallel to the a and b cross-section dimensions, respectively.
At frequencies well above the cutoff frequency, higher-order TE modes of propagation,

with more complicated field configurations, are possible. It is undesirable to operate a
guide at a frequency at which these higher modes can propagate. The next higher mode
can be sustained by the guide when the free-space wavelength is equal to a, that is, at
just twice the cutoff frequency. Therefore a rectangular guide is usually operated within
the frequency range between fc and 2fc. Allowing higher modes to propagate is undesir-
able because they will not be properly coupled to the load, and therefore reflections and
standing waves will be set up, causing losses. The equations and discussion, therefore,
are based on the assumption of TE10-mode operation in rectangular guide unless
otherwise stated, at frequencies between fc and 2fc.
2.5.5. Impedance in Waveguides

For waveguides a characteristic wave impedance is defined, analogous to the character-
istic impedance of a conventional transmission line. It is also closely related to the
concept of wave impedance in free space, which was shown in chapter 1, Eq. 1–8, to
be 377 (or more accurately, 120p) ohms. At frequencies well above cutoff, the guide
characteristic impedance will also have this value, but more generally it is given by the
expression
377  λg 
Zc = = 377   ohms (2–74)
1 − ( fc f)
2  λ0 
In general Zc is greater than 377 ohms. At the cutoff frequency fc it becomes infinite, and
at f = 2fc it has the value 435 ohms, as found from (2–74).
The quantity Zc has the same general significance for waveguide as does the charac-
teristic impedance Z0 for ordinary transmission lines, but there are certain exceptions to
this statement. As shown (see Fig. 2–10) by (2–72), fc is determined by only the dimen-
sion a. Therefore two waveguides with the same a dimension and different b dimensions
will have the same value of fc and also the same value of Zc. But if these two waveguides
are joined together end to end, so that a wave propagating in one of them passes through
the junction into the other, a reflection will be set up at this point due to the disparity of
dimensions, in spite of the impedances being “matched.”
But for constant guide dimensions and a single mode of propagation, the characteristic
wave impedance of the guide, Zc, has the same significance as the characteristic impedance
of a two-conductor transmission line, with respect to reflection and standing waves. The
same principles of impedance matching and transformation apply, and the transmission-
line equations, (2–15) through (2–18) and beyond, can be used provided that the guide
wavelength, lg, is used in calculating b when using equation (1–32), ch. 1.
2.5.6. Impedance Matching in Waveguides

Just as reactive stubs are used for impedance transforming and matching in transmission
lines, various reactive devices are used in waveguides. Short-circuited waveguide stubs
Waveguides 77
Inducitive Capacitive
iris iris
b b
a a
FIGURE 2–13.
Inductive and capacitive irises (as they would appear if the waveguide walls were transparent).
may be employed in the manner described for transmission-line stubs. Reactance may
also be introduced into waveguides by means of irises, posts, and tuning screws. Their
roles in impedance transformation and matching are the same as were described for the
transmission-line tuning stub of Fig. 2–6; that is, the correct position of the reactance
with respect to the load and the amount of reactance (susceptance) needed are determined
in the same way.
Figure 2–13 illustrates capacitive and inductive irises in a rectangular waveguide. They
behave like susceptances, that is, shunt reactances. As shown, they consist of thin metallic
plates perpendicular to the guide walls and joined to them at the edges, with an opening
between them. When the opening is parallel to the narrow walls of the guide, the suscep-
tance is inductive; when it is parallel to the wide walls, it is capacitive. The amount of
the susceptance is determined by the size of the opening. Many variations of these
configurations may be used, for example, an unsymmetrical iris or a unilateral one.
A post placed across the narrow dimension of the guide acts as an inductive shunt
susceptance, of a value depending on its diameter and its position in the transverse
plane. A tuning screw projecting part way across the narrow guide dimension acts like
a capacitive susceptance and may be made adjustable. These devices are illustrated in
Fig. 2–14.
2.5.7. Transmission-Line-to-Waveguide Coupling

It is frequently necessary to transfer power from a coaxial transmission line into a wave-
guide, and vice versa. This is accomplished by means of any one of several coupling
devices such as those illustrated in Fig. 2–15.
These couplers may be used as wave launchers at the input end of a waveguide, or as
wave receptors at the load end, the operation in the one case being the reverse of the
Tuning
screw
Post
b b
a a
FIGURE 2–14.
Post and tuning screw in waveguide.
other. The various dimensions labeled l/4 are approximate; in practice, they are experi-
mentally adjusted for best operation. Note, however, that sometimes the approximate
dimension is l0/4 and in others lg/4. The couplers illustrated are representative of the
methods that may be employed; there are many variations of them in use.
2.5.8. Waveguide Junctions

Waveguides can be joined in various ways to result in power division among two or more
branches. The two basic forms of junction are illustrated in Fig. 2–16. The analysis of
the impedance relations for these junctions is complicated. In practice it is usual to
employ manufactured junctions designed for a particular frequency range.
Numerous special forms of waveguide multiple-branching couplers for special pur-
poses have been developed, too numerous to describe here. The magic-T is perhaps the
best known of them.
2.5.9. Other Forms of Waveguide

Rectangular guide, as mentioned in this section (Fig. 2–10), is the most commonly used,
but other forms are sometimes employed for special purposes. Circular guide, as the
name implies, is simply a round pipe. As might be surmised, the field configurations are
appreciably different from those in rectangular guide.
Waveguides 79
Quarter-wave Straight-through
probe coupler coupler
Coaxial line Coaxial line
Waveguide Waveguide
l0 /4
b b
l0 /4
lg /4
lg /4
Cross-bar coupler
Coaxial line Coaxial line
Waveguide
Waveguide
a lg /4
End view Side view
FIGURE 2–15.
Coaxial-line-to-waveguide couplers.
Transmission line analogs
B B
A A
C C
B
B
a A C b
A C
a
b
Series Waveguide Junction Shunt Waveguide Junction

FIGURE 2–16.
Series and shunt waveguide junctions and their transmission-line analogs.
A coaxial transmission line can be operated as a waveguide at frequencies above the

range in which ordinary transmission-line operation is possible. The field configuration
is appreciably different from configurations that exist in ordinary transmission-line opera-
tion. Other forms used for special purposes are ridge waveguide and trough waveguide.
Dielectric rods may also function as waveguides. The possibilities are almost endless,
and the subject of waveguide theory and practice is a vast one. Here only a brief intro-
duction to the basic ideas has been given.
2.6. Hybrid and Directional Couplers

In the transmission-line systems associated with antennas it is frequently necessary to
employ coupling devices that have special power division and directional properties.
These properties may be obtained with properly combined sections of ordinary balanced-
two-wire or coaxial transmission line and with waveguide. The resulting assemblies of
line sections are called directional couplers, of which hybrid couplers comprise an
important special class.
A section of ordinary transmission line is an example of what is called a two-port in
circuit theory. It has an input port and an output port, although the two are ordinarily
interchangeable. In a sense each of the two ports may simultaneously perform input and
output roles (as when reflections at the “output” end cause a reflected wave to travel back
toward the input end).
A port in circuit theory is, speaking somewhat loosely, a physical place of access to
a system or device. In ordinary low-frequency circuits it corresponds to a pair of termi-
nals, as it does also in balanced-two-wire and coaxial transmission lines. In closed-wall
waveguides and waveguide devices a port is often simply an opening through which
electromagnetic fields may be introduced into or exit from the internal space of the guide.
Sometimes a given port may serve as either an input port or an output port, at the will
Hybrid and Directional Couplers 81
of the user, or it may serve simultaneously

as both. (Additional qualifications on the Port 1 Port 3
definition of a port are necessary when
it applies to multiple-frequency-response Coupler
devices, such as heterodyne mixers, and
to multimode devices, such as wave- Port 2 Port 4
guides, but they are not of concern in the
present discussion.) FIGURE 2–17.
Hybrid and directional couplers are
A four-port coupler.
four-port devices. Figure 2–17 is a sche-
matic representation of a four-port
coupler, with the ports shown as terminal pairs, though restriction to this type of port is
not implied. The essential feature of any four-port is that if power is fed into any one of
the ports, some fraction of it may exit via each of the remaining three ports. If the device
is passive and lossless, these fractions must of course add up to unity. (“Passive” means
that the device does not contain any internal sources of power, such as oscillators or
amplifiers.) The couplers to be discussed are passive, and for the purposes of discussion
will be assumed lossless. (They are ordinarily nearly so.)
The feature of hybrid and directional couplers that distinguishes them from other four-
ports is that one of the three “output” ports is decoupled from the input port, so that the
fraction of input power available from it is virtually zero; moreover, the four ports are
thus decoupled in pairs. For example, if port 1 of Fig. 2–17 is taken as the input port,
all the output may appear at ports 3 and 4, and none at port 2. If then port 2 is taken as
the input port, it will be found that again all the power appears at ports 3 and 4, with
none going to port 1. Ports 1 and 2 thus comprise a pair of decoupled ports; ports 3 and
4 are similarly decoupled from each other.
In terms of this particular pairing, if port 1 is the input, a hybrid coupler divides the
input power equally between ports 3 and 4, with none going to port 2. A directional
coupler, on the other hand, divides the input power unequally, with (say) most of it going
to port 3, a small fraction to port 4, and none to port 2. If, however, port 2 is taken as
the input, the larger fraction will go to port 4, a small fraction to port 3, and none to
port 1.
In the preceding example, if 1% (1/100) of the input power at port 1 goes to port 4,
with 99% to port 3, the coupler is said to have a coupling factor of 20 dB (since
10 log 1/100 = −20). If the fractions are 0.1 and 99.9%, the coupling factor is 30 dB. A
hybrid coupler is therefore actually a 3-dB directional coupler, but the term directional
coupler is ordinarily reserved for couplers that divide the power unequally.
Hybrid couplers are used when power from a source is to be equally divided between
two output ports, when these ports are required to be decoupled from each other. (The
decoupling feature is not provided by the ordinary power dividers discussed in sec. 2–3.)
They are also used when power from two sources is to be fed simultaneously to a load
or loads and the sources must be decoupled from each other (so that there will not be an
exchange of power between them). In this case the two sources are connected to a
decoupled pair of ports, and the division of power between the remaining ports depends
Port 3 l /4 on the relative phasing of the source volt-

Port 2
ages as well as on the design of the
coupler. With proper coupler design the
l /4 l /4 output voltage at one of the output ports
will be proportional to the sum (in the
Port 1 Port 4 phasor sense) of the two input voltages,
and the output from the other will be pro-
portional to their difference. This fact has
an important application in simultaneous
lobing antenna systems, which are
discussed in sec. 7–6.
3l /4 Figure 2–18 is a diagram of a ring type
FIGURE 2–18. of hybrid coupler that exhibits the proper-
Hybrid ring coupler. ties just described. The ports are num-
bered to conform to the description of
behavior previously given, that is, ports
1 and 2 are a decoupled pair, as are also ports 3 and 4. If all the ports are properly
impedance-matched, power fed into port 1 will provide equally divided outputs at ports
3 and 4, and none at port 2. This result is due to the lengths of the ring sections between
the ports. The input at port 1 divides and goes both ways around the ring. The path dif-
ferences of the two routes determine whether or not a voltage will exist across the ter-
minal pair of a particular port. Since the path differences to ports 3 and 4 are respectively
a full wavelength and zero, the voltage waves at these points are in phase and therefore
add. At port 2, however, the path difference is a half wavelength; hence the oppositely
traveling voltage waves are out of phase and cancel. This behavior occurs of course only
at the frequency for which the ring sections have the lengths indicated, or odd-integer
multiples of them. The ring couplers, however, will perform reasonably well over a small
range of frequencies near the correct frequency. The ring may be two-wire balanced line,
coaxial line, or waveguide. (The line sections need not actually form a ring as long as
they are of the correct length.) Other methods of achieving this type of coupler action
also exist, and some of them perform adequately over a broader frequency range than
the ring coupler, at the expense of more complicated structure.
With input at port 1, the equal outputs at ports 3 and 4 will be out of phase with each
other. If they are required to be in phase, as they may be in particular antenna applica-
tions, port 3 can be used as the input port; then equal in-phase outputs will be obtained
at ports 1 and 2, with no output at port 4. Alternatively, port 2 can be used as the input
port.
In a directional coupler there are in a sense two parallel channels, with “strong” bidi-
rectional coupling between the ports of a given channel and weak unidirectional coupling
between the channels. The description of directional-coupling behavior that has been
given for Fig. 2–17 considers ports 1 and 3 to comprise one channel and ports 2 and 4
the other. The coupler is directional in the sense that if a particular port is considered an
output port with respect to both ports of the other channel (main channel), it is apparent
that output will occur when power flows one way in this main channel, but practically
References 83
Channel 2–4
Coaxial line
Port 2 Port 4
Coupling loop
Port 1 Port 3
Coaxial line Main channel (1–3)

(inner conductors
shown as dashed lines)
FIGURE 2–19.
Coaxial-line directional coupler.
no output will occur when it flows the other way. The ratio of these outputs for equal
power flow in the two directions of the main channel is called the directivity factor. A
typical value is 40 dB.
Figure 2–19 shows one form of directional coupler using coaxial line sections. The
coupling between the two channels is by means by a short loop of the inner conductor
of the 2–4 channel projecting into the interconductor space of the 1–3 channel. The plane
of the loop is parallel to the inner conductor of the line into which it projects. It couples
to both the electric and the magnetic field components in such a way that the currents
that each type of field induces reinforce in the coupler branch leading to port 2 and cancel
in the branch leading to port 4 (when the flow in the 1–3 channel is from port 1 to port
3). Many other coupling mechanisms exist for achieving a similar result.
In practice, directional couplers are often constructed with a matched-load termination
on one port, such as port 4 in the above example, and only ports 1, 2, and 3 are made
accessible. Then when there is a reflected wave in the 1–3 channel, which results in an
output to port 4, the power thus received is absorbed in the load. (Otherwise it could be
reflected and reach port 2, which would be harmful to the intended operation of the
coupler.) Thus it might be considered that the coupler is not virtually lossless, as assumed.
However, this load, though “built in,” is not truly a part of the coupler itself, so that it
does not invalidate the assumption of losslessness.
The principal uses of directional couplers in antenna applications are for measurement
and monitoring of the power flow and the VSWR, as discussed in secs. 9.8.3 and
9.11.1.
References
Lowman, R. V., and R. N. Simons, “Transmission Lines and Waveguides,” ch. 51 in
Antenna Engineering Handbook, 4th ed., J. L. Volakis (ed.), McGraw-Hill, 2007.
Ragan, G. L. (ed.), Microwave Transmission Circuits, McGraw-Hill, 1948.

1. In Eq. 2–3, the indicated parameters have the following values:
V0 = 1 volt
w = 2p f = 6.28 × 108 radians per second (f = 108 Hz)
t = 2 × 10−9 second
b = 2p/l = 2.09 radians per meter (l = 3 meters)
f = 5.02 radians
Make a plot of the resulting values of V as a function of x, using rectangular
coordinates, with V as ordinate and x as abscissa. This plot shows how the
instantaneous rf voltage varies with distance along the line at a particular instant
of time.
2. A transmission line has an inductance L = 9.99 × 10−7 henry per meter of length
and a capacitance between the conductors of C = 1.11 × 10−11 farad per
meter. (a) What is the characteristic impedance, Z0? (b) What is the velocity of
propagation, v?
3. An air-dielectric transmission line has a characteristic impedance Z0 = 50 ohms,
and length l = 25 meters. It is being operated at a frequency f = 7.5 MHz
(7.5 × 106 Hz). The wavelength is therefore l = 40 meters, from equation (2–1).
An impedance consisting of 25 ohms resistance in series with 30 ohms inductive
reactance is connected at the load end of the line (ZL = 25 + j30 ohms). A voltage
Vi = 100 volts rms is applied at the input end of the line. (a) What is the rms
current at a distance x = 12 meters from the line input terminals? (b) From equa-
tion (2–16), what is its phase angle relative to the phase of the input voltage?
4. An antenna is to be fed from a transmitter by a line of characteristic impedance
Z0 = 500 ohms. The line is several wavelengths long, and it is desired to have
no standing waves (VSWR = 1). The impedance presented by the antenna at its
input terminals is 80 ohms, purely resistive. Therefore a quarter-wavelength
section of different characteristic impedance is inserted between the end of the
500-ohm line and the antenna terminals. What must the characteristic impedance
of this quarter-wave-transformer section be in order to have no standing wave
on the 500-ohm line?
5. A section of transmission line of length l = 0.2 l and of characteristic impedance
Z0 = 70 ohms is short-circuited at the “load” end. (a) At its input end, what is
the impedance? (b) If the frequency is 107 Hz, what inductance must a coil have
in order to have this reactance value?
6. By measuring the rf voltage amplitude along a line with voltmeter, it is found
that the voltage standing wave ratio (VSWR) is 4. (a) What is the magnitude of
the reflection coefficient of the load, |r|? (b) If the characteristic impedance of
the line is 600 ohms, and the load is known to be purely resistive and to have a
value greater than 600 ohms, what is its resistance?
7. If in Problem 6 the load is not purely resistive, suppose that it is found (by sliding
the voltmeter along the line) that a voltage minimum exists at a distance
dm = (5/24)l from the load terminals. (The VSWR is still 4.) (a) What is the
(complex) reflection coefficient, r? (b) From equation (2–46), what is the value
of the load impedance, ZL?
8. The conductors of a transmission line have a series resistance of 0.01 ohm per
meter of line and a leakage conductance of 10−5 mho per meter. (a) What is the
attenuation, in decibels, of a 1,000-meter length of this line, assuming that
Z0 = 50 ohms and that the wavelength is less than 10 meters? (b) If a transmitter
delivers 1,000 watts of rf power at the input terminals of this line, and if an
antenna connected at the load terminals has a nonreactive impedance equal to
the characteristic impedance of the line, how much power is delivered to the
antenna?
9. A low-loss transmission line system connects a single source of power to eight
loads in the manner of Fig. 2–8a. The loads are resistive (nonreactive) and are
each 400 ohms in value. If the power divides equally at each branch point of the
system, and if all line sections are completely free of standing waves, what must
the characteristic impedance of the input line be? (Assume that no impedance
transformers are used, so that the impedance [admittance] presented at each
branch point depends solely on the characteristic impedances [admittances] of
the branching lines.)
10. A rectangular air-filled waveguide has lateral dimensions a = 7.2 cm, b = 3.4 cm.
(a) What is the cutoff frequency, fc? (b) At an operating frequency of
f = 3 × 109 Hz, for which the free-space wavelength is l0 = 10 cm, what is the
guide wavelength, lg?
CHAPTER 3
Antenna Parameters
An antenna is a device used to radiate or receive electromagnetic waves. To the layman,

a rod or wire is probably the most familiar form of antenna. Antennas of this type are
seen virtually everywhere—on rooftops, automobiles, boats and ships, aircraft, even on
spacecraft. Another familiar type of antenna is the loop, made of wire formed into a large
planar coil. Loop antennas are used on ships and aircraft for direction finding and are
also commonly used as “built-in” receiving antennas for home radio and television. The
layman also sees other forms of antennas, such as arrays, parabolic and corner reflectors,
horns, slots, helixes, lenses, and log periodics that are discussed in chapters that follow.
The increasing use of these more sophisticated forms of antennas reflects advances that
have been made in antenna theory, the more demanding requirements of modern electron-
ics systems, and the exploitation of higher frequencies where more exotic forms of
antennas are feasible.
The ideal antenna is, in most applications, one that will radiate all the power delivered
to it by a transmitter (usually via a transmission line) in the desired direction or directions
and with the desired polarization. Practical antennas can never fully achieve this ideal
performance, but their merit is conveniently described in terms of the degree to which
they do so. For this purpose, certain parameters of antenna performance are defined. This
chapter is primarily concerned with these definitions and related terminology, which will
be needed in the later chapters for discussing the performance of specific forms of
antennas.
The principal parameters of antennas are associated with the radiation pattern, the
radiation efficiency, the input impedance, and the bandwidth. Parameters are defined
under each of these categories such as the gain, beamwidth, beam polarization, minor
lobe level, radiation efficiency, aperture efficiency, effective area, radiation resistance,
various “bandwidths,” and others that have specialized applications. Some of these
parameters are interrelated or correlated. For example, if the beamwidth is given, the
gain can thereby be estimated, though it cannot be calculated exactly from the beamwidth
without additional information.
This book is primarily concerned with the electromagnetic design and performance of
antennas. Often of almost equal practical importance are the mechanical design factors,
87
88 Antenna Parameters
some of which will be mentioned throughout the book in connection with related discus-
sions of electromagnetic behavior. It is desirable, however, for the reader to have some
general “feeling” for the structural factors in the design of antennas, as an aid in relating
the somewhat abstruse electromagnetic concepts to real physical structures. A thumbnail
sketch of some of these aspects of antenna design will be given here, before the defini-
tion of electromagnetic parameters is taken up.
3.1. Antenna Structures

3.1.1. Size
The sizes of antennas range from microminiature to gigantic. There is a general propor-
tionality between antenna size and the wavelength at the frequency of operation, but
this relationship is not hard and fast. Large antennas are sometimes used at short
wavelengths (high frequencies) to obtain a highly directional radiation pattern (beam)
and high gain in a preferred direction. However, there is a practical extreme beyond
which further increase of size produces little or no additional gain because the required
precision of construction or maintenance of phase relationships is not practically attain-
able. This limitation is in the size region of perhaps a few thousand wavelengths, but an
antenna of as much as 100 wavelengths in any dimension is considered “electrically
large.”
Moreover, at long wavelengths (low frequencies) very small antennas may be used
for reception when efficiency is not important. An antenna appreciably less than a half
wavelength is termed “electrically small.” Even a physically large antenna may be elec-
trically small at very low frequencies.
In general, however, the largest antennas are those used at the very lowest frequencies
(especially for transmitting, where radiation efficiency is important). An example of an
extremely large VLF antenna is the navy’s installation (actually two antennas) at Cutler,
Maine, where wires supported by towers 1,000 feet high extend over an area of 2 square
miles. In contrast, a half-wave dipole at the microwave frequencies may be considerably
less than an inch long.
3.1.2. Supports
As mentioned in sec. 1–3, an antenna must be located “in the clear” for good results—
away from large conducting or absorbing objects. Accordingly there must often be some
supporting structure to place the radiating element or elements in a clear location (which
often is synonymous with a high location). Antennas are supported by such devices as
towers, masts, and pedestals. Towers are used when great height is required. Masts may
be quite high, but they are often as short as a few feet. Pedestals are the base structures
of antennas such as reflectors and lenses, for which height is not so important as strength.
Sometimes an antenna may be mounted directly on a vehicle, such as an automobile,
ship, aircraft, or spacecraft; no intermediate “support” is required. Moreover, towers and
masts are sometimes themselves used as antennas rather than as supports. In the standard
Antenna Structures 89
broadcast band (535–1,705 kHz), for example, vertical towers of heights up to several
hundred feet are used as transmitting antennas.
3.1.3. Feed Lines

An important part of any antenna installation is the transmission line used to connect the
transmitter or receiver to the antenna. In fact the design of a feed line and any necessary
impedance-matching or power-dividing devices associated with it are usually considered
part of antenna design. The line connects to the antenna at its input terminals or input
port. At the very lowest frequencies, the earth (ground) is a part of the antenna electrical
system, and one terminal of the antenna input is therefore a rod driven into the ground
or a wire leading to a system of buried conductors (especially if the earth is dry in the
vicinity of the antenna). The other terminal is then usually the base of a tower or other
vertically rising conductor. Towers used in this way are usually supported at the base by
a heavy insulator or insulators (series feed), but occasionally they are directly grounded
and fed by connecting the feed wire a short distance up from the ground (shunt feed).
At somewhat higher frequencies, up to perhaps 30 MHz, the antenna may be a hori-
zontal wire strung between towers or other supports (from which it is insulated). The
feed line is then often a two-wire balanced line connected at the center of the antenna,
either to the two terminals provided by a gap in the antenna wire (series feed) or to two
points somewhat separated on the unbroken antenna wire (shunt feed). Sometimes the
feed line is connected at the end of the horizontal span, or elsewhere off center, but center
feed is preferred because it results in better balance of the currents in the feed wires.
Two-wire-line spacings range from less than an inch to 12 inches or more.
At still higher frequencies, up to perhaps 1 GHz and occasionally somewhat higher,
coaxial feed lines are commonly used. They are favored because the two-wire-line
spacing becomes too great a fraction of the wavelength to prevent appreciable radiation
and because waveguides below 1 GHz are quite large and expensive (though sometimes
used). (Two-wire lines of small spacing may also be used in low-power applications.)
Coaxial line diameters range from a fraction of an inch up to 9 inches or more. Above
1 GHz, waveguides are commonly used, with some use of small-diameter coaxial lines
in low-power noncritical applications. In addition, microstrip constructed using printed
circuit boards are used to reduce cost and bulk. When the antenna rotates on a pedestal,
or has other motion with respect to its support, the feed line must contain flexing sections
or rotating joints.
3.1.4. Conductors
Metals are the usual conducting materials of antennas. Metals of high conductivity, such
as copper and aluminum (and its alloys), are naturally preferred. Brass may be used for
machined parts. Magnesium is sometimes used where ultralight weight is important,
usually in an alloy and with a protective coating or treatment. Where strength is
of primary importance, steel may be used, either with or without a coating or plating of
copper. The conductivity of unplated steel is adequate when it is used in the form of
sheets or other large-surface-area forms (as for the surface of a paraboloidal reflector).
Antenna wire is sometimes made with a steel core for strength and to minimize stretch-
ing, and with a copper coating to increase the conductivity. Such wire is virtually as good
a conductor as solid copper, since rf currents are concentrated near the surfaces of con-
ductors (skin effect). For this reason brass and other metals are sometimes silver plated
when exceptionally high conductivity is required. For the same reason large-diameter
conductors may be hollow tubes without loss of conductivity. At low radio frequencies
the conductivity of large-diameter conductors may be increased, compared to a solid
conductor, by interweaving strands of small-diameter insulated wires; the resulting con-
ductor is called Litz wire. This technique is most effective below about 500 kHz.
At higher frequencies it is not effective because the currents tend to flow only in the
outer strands.
Conductor size in antenna design is determined by many factors, principally the per-
missible ohmic losses and resultant heating effects in some cases, mechanical strength
requirements, permissible weight, electrical inductance and capacitance effects, and
corona considerations in high-voltage portions of transmitting antennas. Corona is mini-
mized by large-diameter conductors, by avoidance of sharp or highly curved edges, and
by using insulators with metal end caps bonded to the insulating material, so that small
air gaps between wires and insulators do not exist. Corona can occur on metal supports
of the antenna as well as on the antenna conductor itself, as a result of induced
voltages.
3.1.5. Insulators
The conducting portions of an antenna not only carry rf currents but also have rf voltages
between their different parts and between the conductors and ground. To avoid “short
circuiting” these voltage, insulators must sometimes be used between the antenna and its
supports, or between different parts of the antenna. Insulators are also used as spacer
supports for two-wire and coaxial lines and to “break up” guy wires used with masts and
towers so that resonant or near-resonant lengths will not occur. The maximum permissible
uninterrupted length of guy wire sections is about 1/8 wavelength. (A half-wavelength
section would be resonant.) Insulators used to support long heavy spans of wire must be
of high strength. Typical insulating materials for such insulators are glass and ceramics.
Other “low-loss” materials such as polystyrene and other plastics are used where less
strength is required. Very large and heavy insulators are necessary in high-power trans-
mitting applications to prevent flashover. Coaxial lines and waveguides in high power
applications may be filled with an inert gas, or dry air, at a pressure of several atmo-
spheres, to increase the voltage-breakdown level. Printed circuit boards rely on low loss
dielectric insulators.
3.1.6. Weather Protection

Since antennas are ordinarily “outdoors,” they must withstand wind, rain, ice and snow,
lightning, and sometimes corrosive gases or salt-laden air. Protection against wind and
Radiation Pattern 91
ice loads is primarily a matter of mechanical strength and bracing. Guy wires are used
with tall structures or towers to prevent their overturning in high winds. At very low rf
frequencies, where long spans of wire may be used, ice is sometimes melted from the
wires by passing heavy 60-Hertz currents through them to produce heating.
Sometimes an antenna (such as a rotating paraboloidal reflector or lens) is totally
enclosed in a protective housing of low-loss insulating material that is practically trans-
parent to the electromagnetic radiation. Such a housing is called a radome. Radomes are
commonly used on some types of aircraft antennas for aerodynamic reasons. They are
also sometimes used on large microwave ground based antennas, permitting the antenna
itself to be a lighter structure, at the cost of the expensive radome and some loss of
rf power in the radome material. Electrical properties of radomes are addressed in
Appendix E.
Protection against lightning-induced currents and static-charge buildup is necessary
for some types of antennas, such as AM, FM and television broadcasting towers, or any
structure that stands high above its surroundings if the conducting path to ground is not
heavy and direct. Insulators may be protected by horn or ball gaps, and static may be
drained by connecting high-ohmage resistors across insulators.
3.2. Radiation Pattern

The radiation pattern of an antenna is one of its most fundamental properties, and many
of its performance parameters pertain to various aspects of the pattern. As mentioned in
sec. 1.3, there is a reciprocal relationship between the processes of radiation and recep-
tion by an antenna. Although it is customary to speak of the antenna pattern as a radiation
pattern, it is a “reception pattern” as well because it also describes the receiving proper-
ties of the antenna. However, it is perhaps somewhat easier to discuss the radiation
pattern, which describes the relative strength of the radiated field in various directions
from the antenna, at a far-field distance (sec. 3.2.5) that is fixed or constant.
3.2.1. Coordinate Systems

An antenna pattern is three dimensional, and therefore its description requires a three-
dimensional coordinate system. Two possibilities are: (1) cartesian (rectangular) coordi-
nates (x, y, z) and (2) spherical coordinates (r, q, f). The spherical coordinate system is
more appropriate because the radiation pattern may be expressed in terms of the electric
field intensity, for example, at some fixed distance r from the antenna, at all points on
the spherical surface at that distance. Specific points on the surface are then defined by
the direction angles q and f. The pattern then becomes a function of only two independent
variables, since r is a constant, and this fact greatly simplifies matters.
It is sometimes convenient also to refer to the corresponding cartesian system. Figure
3–1 shows the relationships of these two coordinate systems, for some arbitrary point in
space designated P. That is, the dashed lines parallel to the x-, y-, and z-axes define the
cartesian coordinates of P, whereas the spherical coordinates are shown in terms of the
line (of length r) going directly from the origin to the point P, and the projection of this
(q = 0°) line onto the xy-plane. This projection is

z xz-plane
designated r, and, as the reader may
recognize, r and f comprise a two-
yz-plane P
q1 xy-plane dimensional polar coordinate system in
–x
the xy-plane.
Origin r1
(x, y, z, z1 The following analytical relationships
–y r, p = 0 y
exist among the various coordinates:
f1 p x1
1 r = x 2 + y2 + z2 (3–1)
y1 P¢
x
(f = 0°)
z z
cos θ = = (3–2)
x 2 + y2 + z2 r
–z
y
tan φ = (3–3)
FIGURE 3–1. x
Three-dimensional cartesian coordinate system.
Coordinates of an arbitrary point P are shown x = r sin θ cos φ (3–4)
as x1, y1, z1. Corresponding spherical
coordinates are also shown as r1, q1, f1. Also, y = r sin θ sin φ (3–5)
polar coordinates of projection of P (designated
P′) onto xy-plane are shown as r1, f1.
z = r cosθ (3–6)
ρ = x 2 + y 2 = r sin θ (3–7)
x = ρ cos φ (3–8)
y = ρ sin φ (3–9)
The variables x, y, and z range from minus infinity to plus infinity, whereas r and r
range from zero to infinity (always positive). The angle q covers the range of 0 to 180
degrees (0 to p radians), and f goes from 0 to 360 degrees (0 to 2p radians). The angle
q is called the angle of colatitude, or sometimes the polar angle, and f is the angle of
longitude. The system of angular coordinates used to describe positions on the surface
of the earth (latitude and longitude) is essentially a spherical coordinate system with the
radial coordinate fixed, except that the geographic latitude has the value zero at the
equator where the colatitude is 90 degrees, and is +90 degrees at the north pole and −90
degrees at the south pole, where the corresponding values of colatitude are 0 and 180
degrees respectively.
It is in fact sometimes convenient to use the geographic system of coordinates in dis-
cussing antenna patterns that consist of a single main lobe of radiation. The axis of the
lobe is assumed to lie in the xy-plane, so that its longitude angle is f = 0°. Then the angle
q is measured in the xz-plane and corresponds to the latitude angle of geographic coor-
dinates. However, in this context q is usually referred to as the elevation angle, and f is
called the azimuth angle. The value of q corresponding to the beam axis is called the
elevation angle of the beam. A positive elevation angle is equal to the complement of
the colatitude angle in the corresponding system of true spherical coordinates.
3.2.2. Definition of Pattern

If an antenna is imagined to be located at the center of a spherical coordinate system, its
radiation pattern is determined by measuring the electric field intensity over the surface
of a sphere at some fixed distance, r. Since the field E is then a function of the two vari-
able q and f, it is written E(q, f) in functional notation.
A measurement of the electric field intensity E(q, f) of an electromagnetic field in free
space is equivalent to a measurement of the magnetic field intensity H(q, f), since the
magnitudes of the two quantities are directly related by E = 377H in equation (1–10),
ch. 1, for the far field. (Vectorially, of course, they are at right angles to each other; also,
their phase angles are equal.) Therefore the pattern could equally well be given in terms
of E or H. It is customary, however, to discuss patterns in terms of the electric field
intensity, E.
The power density of the field, p(q, f), can also be computed when E(q, f) is known,
the relation being: p = E2/377 (equation [1–9]). Therefore a plot of the antenna pattern
in terms of p(q, f) conveys the same information as a plot of the magnitude of E(q, f).
In some circumstances, the phase of the field is of interest, and a plot may be made of
the phase angle of E(q, f) as well as its magnitude; this plot is called the phase pattern
of the antenna. But ordinarily the term antenna pattern implies only the magnitude of E
or p. Sometimes the polarization properties of E may also be plotted, thus forming a
polarization pattern.
3.2.3. Patterns in a Plane

Although the total pattern of an antenna is a function of q and f, the pattern in a particular
plane that passes through the peak of the main beam is often of interest. In fact, there is
no satisfactory way of making a single plot of the entire pattern on a plane piece of paper.
The q versus f pattern is usually represented in terms of pattern cuts in two planes that
form 90 degree angles with each other, with the origin of a spherical coordinate system
on their intersection line. The f = 0° direction is taken to lie along the intersection line,
and the q = 0° direction is of course then perpendicular to this line and lies in one of the
planes. This plane then becomes equivalent to the xz-plane of Fig. 3–1, the other being
equivalent to the xy-plane. These are called principal planes of the coordinate system,
and the patterns in them are the principal-plane patterns of the antenna.
The principal-plane pattern cuts do not, of course, convey the full q and f pattern
information; this would require an infinity of plane patterns. For example, planes formed
by rotating the xy-plane about the x-axis in small increments of the angle q are appropri-
ate for plotting additional plane patterns, which, in the aggregate, present a good picture
of the total pattern. Alternatively, the planes may be formed by rotating the xy-plane
about the y-axis or the xz-plane about the z-axis. Figures 6–32, 6–33, and 6–34 are
examples of calculated patterns of amplitude versus q and f. Still another method of

depicting three-dimensional pattern information is to plot contours of constant signal
strength on the surface of a sphere containing the antenna at its center. But ordinarily
only the principal-plane patterns are given, as they convey an adequate picture of the q
versus f pattern for many purposes.
Patterns in a plane involve only one angle, and hence are polar-coordinate plots, except
that the radial coordinate is the field strength or power density, rather than distance.
However, because fortuitously in a given direction the field strength E is inversely pro-
portional to the distance, assuming far field free-space propagation, equation (1–11),
ch. 1, a pattern that represents field strength as a function of angular direction at a fixed
distance from the antenna is identical to a plot of distance for a constant field strength.
Therefore a field-strength pattern can be interpreted in either of these ways, by simply
changing the labeling of the radial coordinate scale.
Although the pattern in a plane is thus appropriately represented by polar coordinates,
it would be possible to employ cartesian coordinates, through the equations (3–7) to
(3–9). If this were done, the shape of the pattern would be unchanged; but because inter-
pretation of the meaning of the pattern in terms of the cartesian coordinates would be
relatively difficult, this is never done. It is fairly common, however, to plot the pattern
on rectangular-coordinate graph paper but in terms of the direction angle as the abscissa
and field strength or power density as the ordinate. This type of plot distorts the appear-
ance of the pattern geometrically but pre-
f serves the interpretability of an angle
0° representation and makes the plotting
and reading of the low-amplitude por-
tions of the pattern easier. Figure 3–2
compares these two representations. Note
270° 90° that it is easier to locate the angular posi-
E = 0.5 tions of nulls (zeros) of the pattern on the
rectangular plot.
E = 1.0 When the radiation of an antenna is
180° polarized so that the E-vector lies in a
(a)
plane (usually one of the principal
1.0 planes), the pattern in this plane is some-
times referred to as the E-plane pattern;
0.5
and the pattern in the plane perpendicular
E
0
to it, in which the H-vector lies, is called
180° 270° 0° 90° 180° the H-plane pattern.
f
(b)
FIGURE 3–2. 3.2.4. Absolute and Relative
Comparison of plane pattern plotted in polar Patterns
and rectangular form. The same pattern is
represented in both cases and the coordinates If the radiation pattern is plotted in terms
are the same. Only the form of the plot is of the field strength in electrical units,
different: (a) polar plot; (b) rectangular plot. such as volts per meter or the power
density in watts per square meter, it is called an absolute pattern. An absolute pattern
actually describes not only the characteristics of an antenna but also those of the associ-
ated transmitter, since the absolute field strength at a given point in space depends on
the total amount of power radiated as well as on the directional properties of the antenna.
An absolute pattern in terms of constant-field-strength contours plotted on a map is a
most useful representation for a radio broadcast station, since it defines the geographic
regions within which various received-signal levels are available to listeners.
Often, however, the pattern is plotted in relative terms, that is, the field strength or
power density is represented in terms of its ratio to some reference value. The reference
usually chosen is the field level in the maximum-field-strength direction. The field
strength or power density is given the value unity in this direction and fractional values
in other directions. This type of pattern provides as much information about the antenna
as does an absolute pattern, and therefore relative patterns are usually plotted when it is
desired to describe only the properties of the antenna, without reference to an associated
transmitter (or receiver).
It is also fairly common to express the relative field strength E or power density p in
decibels. This coordinate of the pattern is given as 20 log(E/Emax) or 10 log(p/pmax). The
value at the maximum of the pattern is therefore zero decibels, and at other angles the
decibel values are negative (since the logarithm of a fractional number is negative).
3.2.5. Near-Field and Far-Field Patterns

Antenna radiation patterns are graphical representations of the radiation, usually as a
function of directional coordinates and in the far-field of the antenna. The far-field region
is where the angular field distribution is essentially independent of the distance from the
antenna (IEEE 1992, p. 482). For an out-of-doors pattern measurement, it is therefore
important to choose a distance sufficiently large to be definitely in the far-field region.
The minimum permissible distance, for a far-field measurement, depends on the
dimensions of the antenna in relation to the wavelength. In free space, the general rule-
of-thumb (see Appendix F) for the minimum separation distance rmin is
D2
rmin = 2 (3–10)
λ
where D is the diameter of the smallest stationary sphere enclosing the radiating portions
of the antenna as it is rotated for measurement and l is the wavelength. The factor of 2
in (3–10) is sufficient for sidelobe levels higher than −20 dB, but a factor of 12 may be
required for the measurement of −50 dB sidelobe levels (see sec. 7.10).
Sometimes, however, additional criteria are necessary for assurance that a measured
pattern replicates a far-field measurement. While reviewing the manuscript, E. B. Joy
added two additional criteria: (1) rmin = 20l, that pertains to electrically small antennas
and assures that the reactive near field is negligible at rmin and (2) rmin = 20D, that limits
the relative down range far field decrease across the antenna under test to an acceptable
value. As with the factor 2 in (3–10), choice of the factors 20 depends on the accuracy
of measurement required. According to Hollis, Lyon, and Clayton (1970, sec. 14.2), the
relevant effects are usually considered negligible if rmin ≥ 10l and rmin ≥ 10D.
Fields and patterns of an antenna are generally dependent on the distance from the
antenna, broadly separated into near-field and far-field regions. The near-field region is
further subdivided between the reactive near-field and radiating near-field regions.
However, the total field present is actually the vector sum of the reactive (i.e., induction
by time-varying electric and magnetic fields) and radiating fields. At distances from an
antenna that are small compared to a wavelength (or small compared with the antenna
dimensions if the antenna is large), field strengths within reactive near-field regions may
greatly exceed the radiated fields. This is because the nonradiating electric and magnetic
fields, caused by charges and currents on and within the antenna, decrease with range r
as 1/r2 and 1/r3. As a consequence, these reactive fields decrease more rapidly with
increased range than the radiation fields, which vary as 1/r. At some range the reactive
and radiating power densities are equal. This range may serve as a demarcation between
the reactive and radiating near-fields. For a dipole that is electrically very short, or an
equivalent radiator, the outer reactive field boundary is commonly taken to be at a dis-
tance l/2p from the antenna surface (IEEE 1992, p. 1080). Thus, for the VLF band and
lower frequencies, having wavelengths of 104 meters and longer, antennas are electrically
very short. Consequently, the reactive field extends to distances of 1.6 kilometer (about
one mile) and farther. Thus, some of the original “wireless” communications were within
the reactive near zone. Today, short-range communications between equipments is a
developing application of reactive fields. Radio frequency identification (RFID) used for
product inventory is another relatively new application of reactive fields. From near field
measurements with antennas large compared to l, it is known that the strengths of reac-
tive fields are negligible compared to the radiating near fields at distances of 4 or 5l
(sec. 9.1.2). Like all varying fields, reactive fields can, at intense levels, be hazardous to
people. For example, applications of reactive fields include RF heating of metals through
induction and of insulating materials through dielectric loss. Unlike radiating fields, the
magnitudes of electric and magnetic reactive fields are not generally predictable, one
from the other. Thus, an unfamiliar reactive field environment may require monitoring,
separately, the electric and magnetic fields.
The radiating near-field region is that portion of the near-field region between the
reactive near-field and the far-field regions. As already noted, within the far-field the
radiation pattern is essentially independent of the distance from the antenna. However,
within radiating near-field regions, radiation patterns are dependent upon distance from
the antenna. This is because the path length differences between various parts of the
antenna and an observation point, in terms of phase differences, depend on distance from
the antenna center. Consequently, for an antenna small compared to a wavelength, the
radiating near-field region may not exist. Following optical terminology, the radiating
near-field and far-field regions are sometimes called the Fresnel and Fraunhofer regions,
respectively.
Pattern, gain, directivity, and polarization of an antenna are basically far field terms.
Their definitions rely on the simplicity of the far field and do not include the complexities
of reactive near fields. Accordingly, cal- 0

culated radiating near and far fields of an –5
antenna are meaningful for determining
E-field (dB)
–10
antenna patterns at different distances,
–15
only if there is assurance that the strength
of the reactive near field is negligible. –20
The general methods for calculating –25
radiation patterns versus distance are –30
described in Appendix G, and radiating –6 0 6
Observation angle (degrees)
near-field Mathcad calculations for array R=64m
R=320m
and reflector antennas are included in the R=1600m
accompanying website. Figure 3–3 shows FIGURE 3–3.
results of calculating radiation within the Calculated patterns at different distances from
near- and far-field regions for a linear an antenna, for which 2D2/l is 320 m.
antenna, with constant amplitude and
phase over its aperture. The antenna width
D and wavelength l are 4 m and 0.1 m, respectively, giving the calculated 2D2/l
distance of 320 m to the boundary between the radiating near-field and far-field
regions.
Figure 3–3 includes three patterns, corresponding to distances of 10D2/l (1,600 m),
2D2/l (320 m), and 0.4D2/l (60 m). Notice that the 2D2/l and 10D2/l patterns are nearly
identical, except that the first nulls are much deeper for the larger distance. For the
0.4D2/l pattern, the first nulls are not apparent, the maximum field strength is reduced,
and sidelobe levels are increased. Thus, it is seen that the 2D2/l rule-of-thumb provides
useful approximate patterns if very accurate low-sidelobe measurements are not
required.
3.2.6. Free-Space and Earth-Reflection Patterns

The antenna pattern calculated for an antenna in free space will obviously not be observed
in the presence of the earth, which both absorbs and reflects, as described in sec. 1–4.
Vertical-plane free-space and earth-reflection patterns of an isotropic radiator under
certain conditions are compared in Fig. 1–15, which indicates how drastically the two
patterns may differ.
Antenna patterns are usually given for the free-space condition, it being assumed that
the user of the antenna will calculate the effect of ground reflection on this pattern for
the particular antenna height and ground conditions that apply in the particular case.
Some types of antennas, however, are basically dependent on the presence of the ground
for their operation, for example, certain types of vertical antennas at low frequencies.
The ground is in fact an integral part of these antenna systems. In these cases, the pattern
must include the effect of the earth. Patterns of antennas intended for a specific environ-
ment or vehicle, such as a ship or aircraft, are usually measured so as to include
the effects of reflections from the metallic surfaces of the deck, airplane wing, and so
forth. It is always advisable to indicate clearly the conditions under which a pattern
measurement was made, or the conditions assumed in calculating it, if any possible ques-
tion could exist.
An antenna mounted so that earth-reflection interference effects occur will have its
vertical-plane pattern drastically affected, as Fig. 1–15 indicates. Horizontal-plane pat-
terns may be affected as to absolute values, but relative values will not be affected if the
earth in the vicinity is “smooth.” That is, the shape of the horizontal pattern will not be
affected, unless the earth is irregular.
3.3. Directivity and Gain

In discussions of directivity and gain, the concept of an isotropic radiator, or isotrope, is
fundamental. This concept was introduced briefly in sec. 1.1.4. Essentially an isotrope
is an imaginary, lossless antenna that radiates uniformly in all directions. Its pattern is a
perfect spherical surface in space; that is, if the electric intensity of the field radiated by
an isotrope is measured at all points on an imaginary spherical surface with the isotrope
at the center (in free space), the same value will be measured everywhere. Therefore, if
an isotrope radiates a total power Pt in watts and is located at the center of a transparent
(or imaginary) far field sphere of radius R meters, the power density over the spherical
surface is
Pt
pisotrope = watts per square meter (3–11)
4π R 2
Equation (3–11) is true because Pt is distributed uniformly over the surface area of the
sphere, which is 4pR2 square meters.
Actually an isotropic radiator is not physically realizable; all actual antennas have some
degree of nonuniformity in their radiation patterns. A nonisotropic antenna will radiate
more power in some directions than in others and therefore has a directional pattern. A
directional antenna will radiate more power in its direction of maximum radiation than an
isotrope would, with both radiating the same total power. Thus, since the directional
antenna sends less power in some directions than an isotrope does, it follows that it must
send more power in other directions, if the total powers radiated are the same.
Directivity D is a quantitative measure of an antenna’s ability to concentrate radiated
power per unit solid angle in a certain direction, and thus D is highly dependent on the
three-dimensional pattern of an antenna. D will be explicitly defined in materials that
follow. On the other hand, gain G is the ratio of the power radiated per unit solid angle
to the power per unit solid angle radiated by a lossless isotrope, each having the same
input power. Unless otherwise specified, the direction applicable to D and G is that for
which maximum radiation occurs.
Over the years there have been several gain-related terms used. In fact, gain once was
not defined in terms of signal strength relative to any specific antenna type (Terman
1943). Gain is sometimes specified relative to the gain of a one-half wavelength dipole,
whose gain exceeds the isotrope by the factor 1.64 (2.15 dB). Then, gain is specified in
Directivity and Gain 99
dBd (gain above the gain of a lossless one-half wavelength dipole). However, almost
universally, gain is expressed relative to the lossless isotrope and when specificity is
desired, it is referred to as absolute gain or expressed in terms of dBi.
3.3.1. Definitions of Directivity and Gain

The IEEE definitions of directivity and gain are expressed in terms of radiation intensity
U(q, f), which is a range-independent quantity that is the product of power density
p(q, f, r) and range squared r2, that is, U(q, f) = p(q, f, r)r2. Specifically,
• Directivity is the ratio of the radiation intensity in a given direction from the
antenna to the radiation intensity averaged over all directions (IEEE 1993,
p. 362). Notes: (1) The average radiation intensity is equal to the total power
radiated by the antenna divided by 4p (area of sphere in steradians). (2) If the
direction is not specified, the direction of maximum radiation intensity is
implied.
• Gain is the ratio of the radiation intensity, in a given direction, to the radiation
intensity that would be obtained if the power accepted by the antenna were
radiated isotropically (IEEE 1993, p. 547). Notes: (1) Gain does not include
losses arising from impedance and polarization mismatches. (2) If the antenna
is without dissipative (I2R) losses; then, in any given direction its gain is equal
to its directivity. (3) If the direction is not specified, the direction of maximum
radiation intensity is implied.
Sections 3.3.2 through 3.3.5 that follow discuss the mathematics of directivity and
gain, and relationships between them. To accomplish this, the concepts of solid angle
and radiation intensity are first introduced.
3.3.2. Solid Angle

Solid angle is the angle that, seen from the center of a sphere, includes a given area on
the surface of the sphere. The value of the solid angle is numerically equal to the size of
that area divided by the square the radius of the sphere (Jurgenson and Brown 2000).
Mathematically the solid angle is unitless, but for practical reasons the steradian (s.r.) is
assigned so that 1 steradian = 1 radian2. Since radians are dimensionless, steradians are
also dimensionless.
Figure 3–4 shows a cross-hatched area dA in spherical coordinates that is bisected by
the solid angle dW. Since dA = r2 sin q dq df = r2dW, the value of the cross-hatched solid
angle in Fig. 3–4 is dW = sin q dq df. In general, however, solid angles can have other
shapes. The solid angle of area A/r2 that subtends all of a sphere may be determined as
follows:
A 2π π  2π
Ω= = ∫ ∫
r2 0 0
sin θ dθ  d φ = ∫ 2dφ = 4π (3–12)
 0
Z 3.3.3. Radiation Intensity

q=0 The power P radiated by an antenna is
q r sin q df equal to p(q, f, r)dA integrated over a
surface enclosing the antenna, where
r sin q r dq
dA = r2 sin q dq df
= r2 dW r = distance from origin to surface of
q dq sphere
p(q, f, r) = power density at q, f, and r
Y
f df dA = r2 sin q dq df = incremental area at
r q, f, and r normal to the propagation
r df
direction
f=0
f p(q, f, r)dA = incremental power in area
X dA at q, f, and r
FIGURE 3–4.
The solid angle dW subtends area dA, and Now, using the coordinates of Fig. 3–4,
consequently the solid angle dW = sin q dq df.
let an antenna be at the center of the
sphere. Then, the total radiated power
P can be determined by summing the incremental power p(q, f, r)dA over the surface,
as follows:
2π π
P= ∫ p (θ , φ, r ) dA = ∫ ∫ p (θ , φ, r ) r 2 sin θ dθ dφ (3–13)
surface 0 0
The power density varies as 1/r2, and thus the product p(q, f, r)r2 is independent of dis-
tance r from antenna. This range-independent product is defined as the radiation intensity
U. Therefore
U (θ , φ ) = p (θ , φ, r ) r 2 (3–14)
and the incremental power that crosses surface area dA is
dP = p (θ , φ, r ) dA = p (θ , φ, r ) r 2 sin θ dθ dφ = U (θ , φ ) d W
where dW = sin q dq df. Thus, the radiation intensity U(q, f) is dP/dW, that is, the incre-
mental power per unit solid angle in direction (q, f). U(q, f) is usually expressed in units
of watts per steradian (i.e., watts per square radian). Then, from (3–13) and (3–14), total
radiated power P is
2π π 2π π
P= ∫ ∫ p (θ , φ, r ) r 2 sin θ dθ dφ = ∫ ∫ U (θ , φ ) dΩ (3–15)
0 0 0 0
Directivity and Gain 101
In other words, total radiated power equals the sum of all radiation intensity that encloses
the antenna.
3.3.4. Directivity
Directivity D is a quantitative measure of an antenna’s ability to concentrate energy
in a certain direction. Specifically, D is the ratio of the maximum radiation intensity
Umax to the average radiation intensity Uav. Then
Directivity = U max /U av (dimensionless) (3–16)
If the radiation is isotropic, the radiation intensity in every direction is Uav. Thus, from
(3–12) and (3–15), total radiated power P is a sphere’s solid angle 4p times Uav and is
given by (3–17)
P = 4π U av (3–17)
Now, by using (3–15), (3–16) and (3–17), directivity can be expressed as
U max 4πU max 4πU max

= = 2π π
(3–18)
Uav P
∫ ∫ U (θ , φ ) sin θ dθ dφ
0 0
Now, by shifting the constant Umax to the denominator, it is seen from (3–19) that D is a
function the relative value on U(q, f) (the bracketed term) that has a maximum of unity:
4π
D= 2π π
(3–19)
 U (θ , φ ) 
∫ ∫  U max  sin θ dθ dφ
0 0
The bracketed term of (3–19) can be replaced by functions, at a fixed distance, of

power density p(q, f, r) or electric field strength E(q, f, r). This is because, with r
constant,
U (θ , φ )/U max = p(θ , φ, r )/p(θ , φ, r )max = |E (θ , φ, r )/E (θ , φ, r )max |2 (3–20)
Therefore, with r fixed, D can be determined with relative values of either U(q, f),
p(q, f, r), or |E(q, f, r)|2.
It is also to be noted that directivity D of (3–19) can be expressed as
4π
D= (3–21)
ΩA
where WA is known as the beam solid angle. For the special case of an isotrope, since an
isotrope radiates equally in all directions, WA = 4p and D = 1. In general, however,
2π π
 U (θ , φ ) 
ΩA = ∫ ∫  U max 
sin θ dθ dφ (3–22)
0 0
and with r constant, U(q, f)/Umax can be replaced with either p(q, f, r)/p(q, f, r)max or
|E(q, f, r)/E(q, f, r)max|2. Use of the term WA in (3–21) helps to emphasize that directivity
is a measure of an antenna’s ability to concentrate energy in a certain direction. Finally,
it is important to again underscore that directivity is calculated by integrating relative
values of an antenna’s radiation pattern, and this does not require knowledge of an abso-
lute value.
3.3.5. Gain
Gain, previously defined in sec. 3.3.1, is the ratio of the radiation intensity in a given
direction to the radiation intensity that would be obtained if the power were radiated
isotropically. If the direction is not specified, the direction of maximum radiation intensity
Umax is implied. In addition, for isotropically radiated power, the radiation intensity is
Uav. Therefore, for a lossless antenna, gain G equals directivity as defined previously by
(3–16).
Gain is determined, in principle, by comparing the radiation intensity if both the actual
test antenna and an isotrope have the same input power. The isotrope is assumed to radiate
all of its input power, but some of the power delivered to the actual antenna may be dis-
sipated in ohmic resistance (i.e., converted to heat). Thus, gain takes into account the
antenna efficiency as well as its directional properties. The efficiency factor k is the ratio
of the power radiated by the antenna to the total input power. Thus, the relationship
between gain G and directivity D is
G = kD (3–23)
where 0 ≤ k ≤ 1 to account for dissipative (I2R) losses.

It is apparent that the quantity of greatest significance to a system designer is gain.
The antenna theorist, on the other hand, finds the concept of directivity convenient, for
it depends only upon the pattern of the antenna. The theorist can compute the directive
gain of a half-wave dipole, for example, by application of Maxwell’s equations and the
assumption that the dipole has no ohmic loss. The directivity of an isotrope, incidentally,
is by definition unity, and the gain of an isotrope with efficiency factor k is (also by defi-
nition) equal to k.
Antenna gain is a power ratio. Values of gain of practical antennas may range from
near zero (total loss) to as much as 106 or more. As with any power ratio, antenna gain
may be expressed in decibels. The antenna gain G expressed in decibels is GdB = 10 log G,
Effective Area and Friis Transmission Equation 103
where the word log denotes the common logarithm (base 10). Directivity in decibels is
calculated using the same formula, with D substituted for G. Gain is expressed in decibels
more commonly than as a power ratio.
3.4. Effective Area and Friis Transmission Equation

Although there is a reciprocal relationship between the transmitting and receiving proper-
ties of reciprocal antennas, it is sometimes more convenient to describe the receiving
properties in a somewhat different way. Whereas the gain is the natural parameter to use
for describing the increased power density of the transmitted signal due to the directional
properties of the antenna, a related quantity called the effective area, is a more natural
parameter for describing the reception properties of the antenna.
The effective area of an antenna Ae is defined as follows. Suppose that a distant trans-
mitter radiates a signal that, at the receiving-antenna location, has a power density pr.
This is the amount of power incident on, or passing through, each unit area of any
imaginary surface perpendicular to the direction of propagation of the waves.
A receiving antenna exposed to this field will have radio-frequency current and voltage
induced in it, and at the antenna terminals this voltage and current represent radio-
frequency power that can be delivered to a load (e.g., the input circuit of a receiver). In
principle the power available at these terminals can be measured (although in practice it
may be so small a power that amplifying equipment will be needed to measure it). This
power, however small, is the received signal power, Pr. It can be deduced that in order
for this amount of power to appear at the antenna output terminals, the antenna had to
“capture” power from the field over a surface in space (oriented perpendicularly to the
direction of the wave) of area Ae such that
Pr = pr Ae (3–24)
The area Ae, which bears this relationship to Pr and pr, is the effective area of the antenna.
(The conception of the capture process thus conveyed is a much oversimplified one, but
it has validity for the present purposes.) Since received power depends on the antenna
polarization, effective area is a function of the polarization of the incident field.
As might be supposed, there is a connection between the effective area of an antenna
and its physical area as viewed from the direction of the incoming signal. The two areas
are not equal, however, although for certain types of high-gain antennas they may be
nearly equal. But some antennas of physically small cross section may have considerably
larger effective areas. It is as though such an antenna has the ability to “reach out” and
capture power from an area larger than its physical size, as in the case of a dipole
antenna.
As is discussed in sec. 5.7.6, there is also a relationship between the gain of a lossless
antenna and its physical size. This relationship suggests that there may also be a connec-
tion between the gain and the effective area, and this indeed turns out to be true. The
equation relating the two quantities is
Gλ 2
Ae = (3–25)
4π
where l is the wavelength corresponding to the frequency of the signal. (This relation-
ship may be proved theoretically and verified experimentally.)
Because of this connection between the effective area and the gain, (3–24) may be
rewritten, with the right-hand side of (3–25) substituted for Ae. Thus:
pr Gλ 2
Pr = (3–26)
4π
Therefore the concept of the effective area of an antenna is not a necessary one. As (3–26)
shows, it is possible to calculate the received-signal power without knowing Ae. As seen,
the effective area definition has a conceptual value, however, and is a convenient quantity
to employ in some types of problems. We now consider the relationship between transmit
and receive powers with two antennas separated by far-field distance R. This relationship
is highly useful, and it is known as the Friis Transmission Equation (Friis 1946). The
following abbreviations are used:
Pr = received power
Pt = input power of transmit antenna
Aet = effective aperture of transmit antenna
Aer = effective aperture of receive antenna
Gt = gain of transmit antenna
Gr = gain of receive antenna
Now, from equation (3–11) for a lossless isotropic radiator and the definition of gain, the
power density pr at distance R from a transmit antenna may be expressed as (Pt/4pR2)Gt.
Recall that another antenna in the presence of pr will receive power Pr that equals pr Aer.
Thus Pr becomes (Pt/4pR2)AerGt. Finally, we use relationships, based on (3–25), between
Gr and Aer and between Gt and Aet to express Pr as functions of Aet Aer and of GrGt. The
result is the Friis Transmission Equation which is given in (3–27) that follows.
G G λ2
Pr =  t 2  Aer Gt = Pt et2 er2 = Pt t 2r 2
P A A
(3–27)
 4π R  R λ (4π ) R
When reciprocity applies, as is usually so, the effective transmit aperture area Aet of
an antenna equals its effective receive aperture area Aer. Similarly, transmit gain Gt and
receive gain Gr of an antenna are equal. Then, because of reciprocity, the transmitter and
receiver locations can be interchanged. Thus, either antenna can be designated as the
transmit antenna and the other as the receive antenna.
Equation (3–27) underscores the practical significance of gain. For example, a transmit
power of 1,000 watts and a transmit antenna gain of 10 (10 dB) will provide the same
received power as will a transmit power of 500 watts and a transmit antenna gain of 20
Beamwidth 105
(13 dB). Obviously, this relationship has great economic significance. Sometimes it may
be much less expensive to double antenna gain (add 3 dB) than it would be to double
transmit power (though in other cases the converse may be true). Generally, it is desirable
to use as much antenna gain as feasible, because it increases received signals.
3.5. Beamwidth
When the radiated power of an antenna is concentrated into a single major “lobe,” as
exemplified by the pattern of Fig. 3–2, the angular width of this lobe is the beamwidth.
The term is applicable only to antennas whose patterns are of this general type. Some
antennas have a pattern consisting of many lobes, for example, all of them more or less
comparable in their maximum power density, or gain, and not necessarily all of the same
angular width. One would not speak of the beamwidth of such an antenna, but a large
class of antennas do have patterns to which the beamwidth parameter may be appropri-
ately applied.
3.5.1. Practical Significance of Beamwidth

If an antenna has a narrow beam and is used for reception, it can be used to determine
the direction from which the received signal is arriving, and consequently it provides
information on the direction of the transmitter. To be useful for this purpose, the antenna
beam must be “steerable,” that is, capable of being pointed in various directions. It is
intuitively apparent that for this direction-finding application, a narrow beam is desirable
and the accuracy of direction determination will be inversely proportional to the beam-
width, assuming no errors in other parts of the system (although in practice this is not
always a good assumption). Its relation to direction-finding accuracy is one significant
aspect of the beamwidth parameter. This, however, is by no means its only practical
significance.
In some applications a receiver may be unable to discriminate completely against an
unwanted signal that is either at the same frequency as the desired signal or on nearly
the same frequency. In such a case, pointing a narrow receiving-antenna beam in the
direction of the desired signal is helpful; the resulting greater gain of the antenna for the
desired signal, and reduced gain for the undesired one, may provide the necessary dis-
crimination. If the directions of the desired and undesired signals are widely separated,
even a relatively wide beam will suffice. But the closer the two signals are in direction,
the narrower the beam must be to provide effective discrimination. (In radar systems the
analogous problem is the ability to distinguish two targets that are close together in
bearing, that is, in direction angle; this ability is called angular resolving power or
resolution.)
Finally, as the foregoing discussion of gain and directivity (sec. 3.3) indicated, there
is a relationship between the solid-angle width of an antenna beam and its directivity
(also see sec. 3.11). Thus, with some exceptions, generally a narrower beam implies a
greater gain. Since gain is usually a desirable property, this relationship constitutes an
additional virtue of narrow beamwidth.
On the other hand, there are situations that call for a wide beam. For example, a
broadcasting station must radiate a signal simultaneously to listeners in many different
directions—typically, over a 360 degree azimuth sector. Any narrowing of the beam to
obtain gain must therefore be done in the vertical plane. At the low frequencies of the
AM broadcast band (535–1,705 kHz), and at lower frequencies, such vertical-plane beam
narrowing is not feasible because it would require an impractically large (high) antenna.
In television and FM broadcasting in the VHF and UHF bands, however, this means of
obtaining gain while preserving 360 degree azimuthal coverage is much used. A situation
in which an antenna beam must be moderately broad in the vertical plane is that of a
search radar antenna on a ship that rolls and pitches. It is usually required that the beam
of such antenna be directed at the horizon. But if the beam has, say, a 2-degree vertical
beamwidth, and the ship rolls 20 degrees (which is not unusual), it is evident that at times
no part of the beam will remain directed at the horizon. The vertical beamwidth in this
case must be of size comparable to the maximum roll angle of the ship, unless some
method of stabilizing the beam is employed (as is sometimes done).
3.5.2. Beamwidth Definition

As illustrated in Fig. 3–2, an antenna beam is typically round-nosed. Defining beamwidth,
therefore, is a problem. As seen clearly in the rectangular plot, Fig. 3–2b, it would be
possible in this case to cite the width of the beam between the two nulls (zero values)
on either side of the maximum, since these are two definitely measurable points. Not all
beams have such nulls, however, although they are present ordinarily. Moreover, it is
logical to define the width of a beam in such a way that it indicates the angular range
within which radiation of useful strength is obtained, or over which good reception may
be expected. From this point of view the convention has been adopted of measuring
beamwidth between the points on the beam pattern at which the power density is half
the value at the maximum. (On a plot of the electric-intensity pattern, the corresponding
points are those at which the intensity is equal to 1/ 2 or 0.707 of the maximum
value.)
The angular width of the beam between
0 these points is called the half-power
1 beamwidth. When a beam pattern is
2 plotted with the ordinate scale in decibels,
3 as is frequently done, the half-power
Decibels
4 Beam pattern
5 points correspond to the minus 3 dB
6 points. For this reason the half-power
10° beamwidth is often referred to as the 3 dB
beamwidth. Figure 3–5 illustrates the pro-
–10° –5° 0° +5° +10° cedure of determining the 3 dB beam-
Angle width on a rectangular pattern plot. Only
FIGURE 3–5. the nose region of the beam pattern is
Determination of half-power (3 dB down) shown. As indicated, this beamwidth is
beamwidth. approximately 10 degrees.
Minor Lobes 107
This criterion of beamwidth, although adequate and convenient in many situations,

does not always provide a sufficient description of the beam characteristics. Beams
have different shapes. An additional description may be given by measuring the width
of the beam at several points, for example, at −3 dB, −10 dB, and at the nulls (if they
are present). Some beams may have an asymmetric shape, for some special reason, and
the specific nature of the asymmetry may be important. Special methods of describing
such beams can be employed. In the final analysis the best description of a beam is a
plot of its pattern.
3.5.3. Principal-Plane Beamwidths

An antenna beam occupies a solid angle, and a single beamwidth figure refers only to
the pattern in a particular plane. It is apparent that the beam may have different widths
in different planes through the beam axis. (The axis is the direction of maximum radia-
tion, a line from the antenna passing through the nose of the beam.) Therefore it is cus-
tomary to give the widths of the beam in two planes at right angles, usually the principal
planes of the coordinate system. The beamwidths in planes at intermediate angles will
generally have intermediate values, so that giving just these two beamwidth figures
conveys considerable information concerning the solid-angular shape of the beam. When
the beam is linearly polarized, the beamwidth in the plane containing the E-vector (plane
of polarization) is sometimes called the E-plane beamwidth, and that in the perpendicular
plane the H-plane beamwidth.
3.6. Minor Lobes

A directional antenna usually has, in addition to a main beam or major lobe of radiation,
several smaller lobes in other directions; they are minor lobes of the pattern. Those
adjacent to the main lobe are sidelobes, and those that occupy the hemisphere in the
direction opposite to the main-beam direction are back lobes. Minor lobes ordinarily
represent radiation (or reception) in undesired directions, and the antenna designer there-
fore attempts to minimize their level relative to that of the main beam. This level is
expressed in terms of the ratio of the power densities in the mainbeam maximum and in
the strongest minor lobe. This ratio is often expressed in decibels.
Since the sidelobes are usually the largest of the minor lobes, this ratio is often called
the sidelobe ratio or sidelobe level. A typical sidelobe level, for an antenna in which
some attempt has been made to reduce the sidelobe level, is 20 dB, which means that
the power density in the strongest sidelobe is one percent of the power density in the
main beam. Sidelobe levels of practical well-designed directional antennas typically
range from about 13 dB (power-density ratio 20) to about 40 dB (power-density ratio
104). Attainment of a sidelobe level better than 30 dB requires very careful design and
construction, but better than 50 dB (105) has sometimes been accomplished.
Figure 3–6 shows a typical antenna pattern with a main beam and minor lobes, plotted
on a decibel scale to facilitate determination of the sidelobe level, which is here seen to
be 25 dB.
0 In some applications sidelobes are not

–5 especially harmful unless their level
Main beam
–10 becomes comparable to the main-beam
–15
Decibels
Highest sidelobes level; in other applications it may be

–20 important to hold the sidelobe level to an
–25 absolute minimum. In most radar systems,
–30 for example, a low sidelobe level is impor-
–35 tant. Since signals received in a sidelobe
–60° –30° 0° 30° 60° are ordinarily indistinguishable from
Angle measured from nose of beam
signals received from the main beam, a
FIGURE 3–6. large target located in the direction of one
Decibel pattern plot, indicating sidelobe level. of the antenna sidelobes (or even a back
lobe) may appear to the radar as though it
were a target in the main beam. Such sid-
elobe echoes create “clutter” on the radar output that may mask main-beam echoes from
smaller targets and overload the data-processing personnel or equipment. Therefore radar
antennas are often designed for sidelobe levels of −25 dB or better in the horizontal-plane
patterns. Sidelobes in the vertical-plane patterns may also be harmful in height finding
radars but do no harm in search radars that provide angle information only in the azimuth
plane, except to the extent that they represent wasted power.
3.7. Radiation Resistance and Efficiency

In a large class of antennas the radiation is associated with a flow of rf current in a con-
ductor or conductors. Thus, when a current I flows in a resistance R, an amount of power
P = RI2 will be dissipated and converted into heat. In an antenna, even if there were no
resistance in the conductors, the electrical energy supplied by the transmitter is radiated
and it is in a sense “lost.” It is customary to associate this “loss” of power through radia-
tion with a fictitious “radiation resistance,” that bears the same relationship to the current
and the radiated power as an actual resistance bears to the current and dissipated power.
If the power radiated by the antenna is P and the antenna rms current is I, the radiation
resistance is
P
Rr = (3–28)
I2
The concept of radiation resistance is applicable only to antennas in which the radia-
tion is associated with a definite current in a single linear conductor. Even then, the defi-
nition of radiation resistance is ambiguous. This is because of standing waves, the current
is not the same everywhere along a linear conductor. It is therefore necessary to specify
the point along the conductor at which the current will be measured. Two points some-
times specified are where the current has its maximum value and the feed point (input
terminals). These two points are sometimes at the same place, as in a center-fed dipole,
Radiation Resistance and Efficiency 109
but they are not always the same. And in principle, any point may be specified. The value
then obtained for the radiation resistance of the antenna depends on what point is speci-
fied; this value is the radiation resistance referred to that point.
The above words “maximum current” refer to the rms current in that part of the
antenna where the current is maximum. It does not mean the peak value of the current
located where the current is greatest. The reader will recall that, in accordance with equa-
tion (1–3), by definition the amplitude I0 of a current sine wave is actually the peak value
of current during its cycle. Furthermore, the reader is cautioned that, in some texts, for-
mulas for radiation resistance are written as a function of current “amplitude” I0, instead
of rms current. Thus, use of the wrong combination of formula and current level will
cause the radiation resistance to be incorrectly calculated.
The radiation resistance of some types of antennas can be calculated, and yet not for
others. Sometimes radiation resistance can be obtained by measurement (see section 9.10,
chapter 9). Typical values of the input radiation resistance of actual antennas range from
a fraction of an ohm to several hundred ohms. The very low values are undesirable,
because they imply large antenna currents (i.e., P = I 2R). Therefore, there exists the
possibility of considerable ohmic loss of power, that is, dissipation of power as heat
rather than as radiation. An excessively high value of radiation resistance is also undesir-
able, because this requires that a very high voltage (i.e., P = E 2/R) be applied to the
antenna.
Antennas always have ohmic resistance, although sometimes it may be so small as
to be negligible. The ohmic resistance is usually distributed over the antenna; and since
the antenna current varies, the resulting loss may be quite complicated to calculate. In
general, however, the actual loss can be considered to be equivalent to the loss in a ficti-
tious lumped resistance placed in series with the radiation resistance. If this equivalent
ohmic loss resistance is denoted by R0, the full power (dissipated plus radiated) is
I 2(R0 + Rr), whereas the radiated power is I 2Rr. Hence the antenna radiation efficiency
k of (3–23) is given by
Rr
kr = (3–29)
R0 + Rr
It must be acknowledged, however, that this definition of the efficiency is not really very
useful even though it may occasionally be convenient. The fact is that both R0 and Rr
are fictitious quantities, derived from measurements of current and power; Rr is given in
these terms by (3–28), and R0 is correspondingly equal to P0/I 2. Making these substitu-
tions into (3–29) gives the more basic definition of the efficiency:
Pr
kr = (3–30)
P0 + Pr
where Pr is the power radiated, and P0 is the power dissipated.

3.8. Input Impedance

Rf current usually reaches to the antenna through a transmission line. The antenna input
is at the point where the transmission line is connected to the antenna. At this location
the antenna presents a load impedance to the transmission line, for example, ZL of Fig.
2–2. This impedance is the input impedance of the antenna, and an impedance mismatch
occurs unless the input impedance equals the characteristic impedance of the line Z0. A
mismatch leads to loss of power transfer and creates a standing wave on the line. Aside
from the loss of power transfer, standing waves are the source of increased voltage along
the transmission line. Consequently, because input impedance of the antenna in relation
to Z0 controls the standing wave ratio, it is a very important parameter. Measurement of
the antenna input impedance is discussed in sec. 9.11.
In addition to its importance in determining whether or not there will be a standing
wave on the transmission line, the input impedance determines how large a voltage must
be applied at the antenna input terminals to obtain the desired current flow and hence
the desired amount of radiated power. The impedance is in fact equal to the ratio of the
input voltage Ei to the input current Ii, by definition:
Ei
Zi = (3–31)
Ii
This impedance is in general complex. If the antenna input is at a current maximum,

and if there is no reactive component to the input impedance, Zi will be equal to the sum
of the radiation resistance and the loss resistance; that is:
Z i = Ri = Rr + R0 (3–32)
The input impedance will be nonreactive for this feed point if the antenna is resonant (as
in the case of a wire approximately an integral number of half wavelengths long, dis-
cussed in secs. 4.2–4.4) or if the antenna is terminated by a resistance of proper value
at its far end so that there is no standing wave of current on it (as described in sec. 4.4).
In other cases, however, the input impedance may consist effectively of the radiation and
loss resistances in series with a very high reactance.
If this reactance has a large value, the antenna input voltage must be very large to
produce an appreciable input current. If in addition the radiation resistance is very small,
the input current must be very large to produce appreciable radiated power. Obviously
this combination of circumstances, which occurs with the “short dipole” antennas that
must be used at very low frequencies, results in a very difficult “feed” impedance-match-
ing problem. Methods of solving it are described in sec. 4.1 and 4.11.
The concepts of radiation resistance and feed point or input impedance have been
discussed here in terms of antennas that have currents flowing in linear conductors as
the basis of the radiation. As will be discussed in secs. 4.7, 4.8, and 4.10, other types of
antennas exist with radiating properties difficult to analyze in these terms. They are, for
example, fed by waveguides rather than by transmission lines. The equivalent of an input
Bandwidth 111
impedance can be defined at the point of connection of the waveguide to the antenna,
just as waveguides have a characteristic wave impedance analogous to the characteristic
impedance of a transmission line. It is difficult to define a “radiation resistance” for
such antennas.
Even for some types of antennas consisting of current-carrying conductors this is dif-
ficult, and it may even be difficult to define an input impedance. This is true, for example,
of an array of dipoles, when each dipole is fed separately; sometimes each dipole, or
groups of dipoles, will be connected to separate transmitting amplifiers and receiving
amplifiers. The input impedance of each dipole or group may then be defined, but the
concept becomes meaningless for the antenna as a whole, as does also the radiation
resistance. Both these terms have meaning primarily for simple linear-current radiating
elements; but they comprise a very large class of antennas.
3.9. Bandwidth
All antennas are limited in the range of frequency over which they will operate satisfac-
torily. This frequency range, whatever it may be, is called the bandwidth of the antenna.
If an antenna were capable of operating satisfactorily from a minimum frequency of
195 MHz to a maximum frequency of 205 MHz, its bandwidth would be 10 MHz. It
would also be said to have a 5 percent bandwidth (the actual bandwidth divided by the
center frequency of the band, times 100). This would be considered a “moderate” band-
width. Some antennas are required to operate only at a fixed frequency with a signal that
is narrow in its bandwidth; consequently there is no bandwidth problem in designing
such an antenna. But in other applications much greater bandwidths may be required;
in such cases special techniques are needed. Such techniques are known and will be
described in secs. 7–1 and 7–2. In fact, some recent developments in broad-band anten-
nas permit bandwidths so great that they are described by giving the numerical ratio of
the highest to the lowest operating frequency, rather than as a percentage of the center
frequency. In these terms, bandwidths of 20 to 1 are readily achieved with these antennas,
and ratios as great as 100 to 1 are possible.
The bandwidth of an antenna is not as precise a concept as some of the other parame-
ters that have been described, because many factors are involved in what is meant by
“operating satisfactorily.” The principal ones are input impedance, radiation efficiency,
gain, beamwidth, beam direction, polarization, and sidelobe level.
These may not all be involved in every case, because one or another of them may be
so much more critical than the others, in a particular case, that it alone determines the
bandwidth. The two basic factors involved, between which a distinct separation may be
made, are the antenna pattern and the input impedance. Accordingly, the terms pattern
bandwidth and impedance bandwidth are sometimes used to emphasize this distinction.
The beamwidth, gain, sidelobe level, beam direction, and polarization are parameters
associated with the pattern bandwidth, whereas input impedance, radiation resistance,
and efficiency are associated with the impedance bandwidth.
The definition of the bandwidth of an antenna is less precise than the definitions of
other parameters for still another reason: there is no established criterion of “satisfactory”
operation. In some applications, for example, an impedance variation of a factor of 2

over the operating frequency band may be acceptable; in others only a 10 percent varia-
tion may be tolerated. Similar statements apply to variations of beamwidth, gain, and so
forth. When an antenna bandwidth figure is given, therefore, it should always be accom-
panied by a statement of “satisfactory operation” on which it is based, unless these criteria
are fairly well established for the particular application involved.
Two categories of bandwidth requirement exist, and sometimes they allow different
approaches. In some cases (category 1) the antenna is required to handle frequencies over
a wide band simultaneously. This is called instantaneous bandwidth. An example is an
antenna that must function properly when passing short pulses (wide bandwidths). There
are also cases (category 2) in which the range of frequencies is covered over a period of
time, but at any one time the bandwidth requirement is only a fraction of the long-term
requirement. This is sometimes called tunable bandwidth. An example of this is a com-
munication antenna that may operate with its carrier frequency anywhere within a band
of 15 MHz, for instance, in the region of 300 MHz, a bandwidth requirement of 5 percent.
But when it is operating at a particular time, its transmitted signal may cover a bandwidth
of, for example, 50 kHz—a bandwidth of less than 0.02 percent. It is possible, in such
a situation, for an operator to readjust the antenna impedance, if necessary, when the
frequency of operation is changed.
These examples illustrate fundamental differences between instantaneous and tunable
bandwidths. The first category presents the most difficult problem of course, because the
antenna must meet the requirement without the possibility of any adjustments being made
in going from one frequency to another. However, even when the bandwidth requirement
is of the second type, it may be desirable to meet it without requiring adjustments to be
made; the decision is primarily an economic one.
3.10. Polarization
The radiation of an antenna may be linearly, elliptically, or circularly polarized, as defined
in sec. 1.1.4 (also see Appendix B). Polarization in one part of the total pattern may be dif-
ferent from polarization in another. For example, the polarization may be different in the
minor lobes and in the main lobe, or may even vary in different parts of the main lobe.
The simplest antennas radiate (and receive) linearly polarized waves. They are usually
oriented so that the polarization (direction of the electric vector) is either horizontal or
vertical. Sometimes the choice is dictated by necessity, at other times by preference based
on technical advantages, and sometimes one polarization is as good and as easily achieved
as the other.
For example, at the very low frequencies it is practically impossible to radiate a hori-
zontally polarized wave successfully (because it will be virtually canceled by the radia-
tion from the image of the antenna in the earth); also, vertically polarized waves propagate
much more successfully at these frequencies (e.g., below 1,000 kHz). Therefore vertical
polarization is practically required at these frequencies.
At the frequencies of television broadcasting (between 54 and 698 MHz), horizontal
polarization has been adopted as standard. This choice was made to maximize signal-to-
Interdependencies of Gain, Beamwidths, and Aperture Dimensions 113
noise ratios; it was found that the majority of man-made noise signals are predominantly
vertically polarized. (Interestingly, in Great Britain the opposite choice was made, because
from the pure wave-propagation point of view, vertical polarization provides maximum
signal strength.)
At the microwave frequencies (above 1 GHz) there is little basis for a choice of hori-
zontal or vertical polarization, although in specific applications there may be some pos-
sible advantage in one or the other. Of course in communication circuits it is essential
that the transmitting and receiving antennas be polarization matched.
Circular polarization has advantages in some VHF, UHF, and microwave applications.
For example, in transmission of VHF and low-UHF signals through the ionosphere, rota-
tion of the polarization vector occurs, the amount of rotation being generally unpredict-
able. Therefore if a linear polarization is transmitted it is advantageous to have a circularly
polarized receiving antenna (which can receive either polarization), or vice versa.
Maximum power transfer is realized when both antennas are either left- or right-circularly
polarized. (This applies, for example, to transmission and reception between the earth
and a space vehicle.) Circular polarization has also been found to be of advantage in
some microwave radar applications to minimize the “clutter” echoes received from rain-
drops, in relation to the echoes from larger targets such as aircraft.
It is apparent that the polarization properties of an antenna are an important part of
its technical description—a parameter of its performance. Sometimes it may be desirable
to provide a “polarization pattern” of the antenna, that is, a description of the polarization
radiated as a function of angles within a spherical coordinate system, although such a
complete picture of the polarization is not ordinarily required.
3.11. Interdependencies of Gain, Beamwidths, and

Aperture Dimensions
Aperture area controls directivity and gain, linear dimensions control beamwidths and
area, and thus directivity and gain depend on beamwidths. These interdependencies of
directivity and gain, beamwidths, linear dimensions, and area on one another are impor-
tant to the antenna design process, and are outlined below.
Equation (3–25) relates effective receive area Ae to gain G and wavelength l. Because
of reciprocity, it is applicable to both effective transmit and receive areas. The general
relationship from antenna theory (Silver 1949, sec. 6.4) for directivity D is related to
maximum effective area Aem and wavelength l as follows:
D = (4π Aem )/λ 2 (3–33)
Since gain G = kD from (3–23), we can express G in terms of an effective area Ae and
l as
G = (4π Ae )/λ 2 (3–34)
where k = Ae/Aem.
The definitions of Aem and Ae are defined by (3–33) and (3–34). Physical aperture Ap
is another frequently used aperture term, but it is less clearly defined. Obviously, Ap is a
measure of the physical size of an antenna and it is an area over which radiation is trans-
mitted or received. Physical area for a horn or reflector antenna can be readily visualized.
Presumably Ap for a stub antenna is its cross-sectional area, but what about one attached
to an automobile which affects its radiation pattern? Another definition sometimes used
is aperture efficiency eap, where
ε ap = Ae /Ap (3–35)
Depending on how Ap is defined, eap can exceed unity. For a uniformly continuously
illuminated aperture, that is, one for which the amplitude and phase are both constant
over the physical area, Aem = Ap. This can be shown to be true from results of integrating
the total radiation pattern for a uniformly illuminated aperture and calculating D from
(3–33), as done by Silver (1949, p. 183). Thus, for a lossless, uniformly continuously-
illuminated aperture eap = Ae/Ap = 1 and
D = (4π Ap )/λ 2 (lossless, uniform illumination) (3–36)
The actual gain of an antenna is of course less than D of (3–36), because of losses that
may occur from large and extraneous sidelobes, impedance mismatches, and dissipative
(I 2R) losses.
Theoretically, the half-power beamwidth qHP for a uniformly illuminated line
source is
θ HP = 50.8 λ d (θ HP in degs., λ d dimensionless)
or (3–37)
θ HP = 0.887 λ d (θ HP in rads., λ d dimensionless)
For the equations of (3–37), l and d are in the same units. The equations of (3–37) can
be derived from the pattern for a uniformly illuminated aperture, which is discussed in
sec. 6.10.
Now consider a rectangular aperture. Its gain can be determined with (3–36) by
expressing physical aperture Ap in terms of principal plane beamwidths, qHP and fHP,
using (3–37). Accordingly, let dq and df denote the linear dimensions that control the
beamwidths in q and f directions, respectively. Then, theoretically, the gain G for a loss-
less, uniformly continuously-illuminated aperture is
G = 4π (dθ dφ )/λ 2 = 4π [(50.8λ )2 /(θ HP ⋅ φ HP )]/ λ 2 = 32,429/(θ HP ⋅ φ HP ) (3–38)
with G dimensionless and qHP and fHP in degrees.

Stutzman (1998) investigated equations for a number of theoretical patterns, including

those for this case of a rectangular aperture. He notes that, ordinarily, patterns of real
antennas do not have perfectly deep nulls. Also, because of the effects of installation
environments, they do not have sidelobes that decrease uniformly at wide-angles, as do
theoretical patterns. Further, the gain of an antenna is also reduced by lossy materials
(I 2R losses), and the radiated power may be reduced by impedance mismatches. Stutzman
suggests, without detailed measurement data being available, that a reasonable approxi-
mation, as a general rule, for an antenna having a beam normal to the aperture, is
G = 26,000/(θ HP φ HP ) (G dimensionless, θ HP and φ HP in degs.) (3–39)
The approximate, “typical” gain of (3–39) is nearly 1.0 dB less than that of (3–38), which
is for a lossless, uniformly illuminated rectangular aperture. Furthermore, the use of
(3–39) is recommended only if the known antenna efficiency (“k” in G = kD) is nearly
1. The point being that directivity and thus gain are reduced by spurious radiation not
readily detected and thus not ordinarily accounted for when determining directivity.
For example, if only the usual principal-plane measurements are made, there may be
unsuspected out-of-plane spurious radiation.
References
Friis, H. T., “A Note on a Simple Transmission Formula,” Proceedings of the IRE, vol.
34, May, 1946, pp. 254–56.
Hollis, J. S., T. J. Lyon, and L. Clayton, Microwave Antenna Measurements,
Scientific-Atlanta, Inc., 1970. This publication may be available through MI
Technologies, Inc. (www. mi-technologies.com).
Jurgenson, R., and R. G. Brown, Geometry, Houghton Mifflin Co., 2000.
IEEE Standard 100-1992, The New IEEE Standard Dictionary of Electrical and Electronic
Terms, 1993.
Silver, S., Microwave Antenna Theory and Design, McGraw-Hill, 1949.
Stutzman, W. L., “Estimating Directivity and Gain of Antennas”, IEEE Antennas and
Propagation Magazine, August, 1998, pp. 7–11.
Terman, F. E., Radio Engineers’ Handbook, McGraw-Hill, 1943, p. 785.

1. Indicate, by placing a letter T or F in each of the boxes, whether you think that
the following statements are true (T) or false (F):
(a) The largest antennas are those used to obtain high gain at high
frequencies.
(b) A steel tower can be used to support an antenna, or sometimes it can itself
be the antenna.
(c) The place at which the feed line connects to an antenna is called its input
terminals or input port.
(d) Guy wires of antenna towers should be interrupted by insulators spaced a
half-wavelength apart.
(e) Solid copper is always the preferred form of conductor for an antenna.
(f ) A radome is a dome-shaped radar antenna.
2. A directional antenna has a maximum electrical dimension of 50 meters, and its
operating frequency is 100 MHz (108 Hz). A field strength measurement is made
at a distance of 1 km from this antenna, in the main beam. (a) Is this a near-field
or a far-field measurement? (b) What is the approximate distance at which the
near field ends and the far field begins?
3. The solid angle subtended by the sun as viewed from the earth is W = 6 × 10−5
steradian. A microwave antenna, designed to be used for studying the microwave
radiation from the sun, has a very narrow beam whose “equivalent” solid angle,in
the sense of the denominators of equations (3–19) and (3–21), is approximately
equal to that subtended by the sun. Assume that this antenna has no minor lobes.
What is its approximate directivity D? Express this gain as both a power ratio
and a decibel value.
4. An antenna radiates a total power of 100 watts. In the direction of maximum
radiation, the field strength at a distance of 10 km (104 meters) was measured
and found to be E = 12 millivolts (0.012 volt) per meter. (a) What is the directiv-
ity D of this antenna, assuming free-space propagation to the measuring point?
(b) If its efficiency factor is kR = 0.92, what is its gain G? (c) If the wavelength
is l = 3 meters, what is the effective area Ae in square meters?
5. If the operating wavelength of an antenna, whose gain G = 30,000, is l = 0.1
meter, (a) what is its effective area, Ae, and (b) If the actual cross-sectional area
of the antenna is the same as its effective area, and if the outline of this area is
circular, what is the antenna diameter? (Recall from elementary geometry that
the area of a circle of diameter d is p(d/2)2.)
6. Derive the two forms of the Friis Transmission Equation given in equation
(3–27). In doing so, start with equation (3–11) and the definition of gain. (These
two forms of the Friis equation are in fact of greater practical value than
equations that include power density.)
7. A dipole slightly shorter than a half wavelength, fed at the center, has an input
impedance that is purely resistive and of value 75 ohms. The conductor has some
ohmic resistance, and the resulting ohmic loss of power is equivalent to what
would result if the dipole were a perfect conductor but had a “lumped” resistance
of value R0 = 8 ohms connected in series with it at the feed point. (a) What is
the radiation resistance, Rr, of the dipole? (b) What is the radiation efficiency
factor, kr? (c) The directivity of a dipole approximately a half-wavelength long
is D = 1.64. What is the gain, G, of this particular dipole?
8. The two VHF television bands are 54 to 88 MHz (low band) and 174 to 216 MHz
(high band). (a) What are the percentage bandwidths of each of these two bands?
(b) If a single receiving antenna is designed to perform satisfactorily at all fre-
quencies between the bottom end of the low band and the top end of the high
band, what is its bandwidth, expressed in the conventional way?
9. Assume N isotropes are placed along a line, and are energized equally in ampli-
tude and phase by the same power source. By using equations (1–9) and (3–11),
what is the theoretical gain perpendicular to the line of isotropes?
10. List three quantities pertaining to the electric field that should be plotted, as
functions of the direction angles q, f, to present all possible information about
the antenna pattern in the far field. (Note: Do not include the magnetic intensity
or the power density, as these are both derivable from a knowledge of the electric
field.)
CHAPTER 4
Basic Radiators and

Feed Methods
The preceding chapters have been largely groundwork, covering principles, concepts,
and terminology, as well as some basic antenna theory. It is now possible to discuss
specific antennas.
Initially the basic forms of radiating structures will be discussed. They are sometimes
used by themselves as simple antennas; they may be very effective in some applications.
They are also used in combination with each other and with other components to form
more complicated antennas having special properties such as high directivity, large band-
width, omnidirectionality, steerable beams, low noise, and so on. These more complicated
antennas are described in chapters 5 through 8. The properties of the basic radiators will
be discussed in terms of their behavior in free space, except for specific types that require
the presence of the earth for their operation, as will be indicated in the discussion. It will
also be assumed, except where specifically otherwise stated, that current-carrying por-
tions of the antennas are perfect conductors. The conductors of actual antennas are
usually very good so that this assumption is reasonably well approximated in practice.
When some conductor losses do occur, the principal effect is to reduce the efficiency, as
defined by equation (3–30), chapter 3. Ordinarily the pattern (and hence the directivity
and beamwidth) are not seriously affected, although appreciable conductor resistance
would affect these parameters also.
4.1. Short Dipoles

Perhaps the simplest and most important radiating structure, from the viewpoint of
antenna theory, is the electrically short dipole with uniform current along its length.
“Electrically short” means “short compared to a half wavelength.”
Throughout this book, the term short dipole is understood to mean simply a dipole
that is much shorter than a half wavelength, but not necessarily one with a uniform
current. The term is generally applied to any dipole that is no longer than about a tenth
of a wavelength (0.1l). Isolated short dipoles do not ordinarily have uniform current
throughout their length, but approximate uniformity of the current may be achieved by
capacitive end loading, as will be described later in this section.
119
120 Basic Radiators and Feed Methods
A short dipole that does have a uniform current will be called an elemental dipole.
Such a dipole will usually be considerably shorter than the tenth-wavelength maximum
specified for a short dipole. The term infinitesimal dipole may be used to imply extreme
shortness, as required in certain mathematical analyses. Other terms sometimes used for
an elemental dipole are elementary dipole, elementary doublet, and Hertzian dipole. Part
of the importance of this type of dipole is that many more complicated antennas can be
analyzed by considering them to be assemblages of many elemental dipoles. For example,
a long-wire antenna may be regarded as composed of many elemental dipoles connected
end-to-end. Although the current is considered constant (at any instant of time) along the
length of each elemental dipole, the currents in different dipoles may be different, both
in magnitude and phase; therefore, a nonuniform current in the long wire can be approxi-
mated by this representation.
If a center-fed short dipole is initially in a neutral condition and then a current starts
to flow in one direction, one half of the dipole will acquire an excess of charge and the
other a deficit (since a current is a flow of electrical charge). There will then be a voltage
between the two halves of the dipole. If the current then reverses its direction, this charge
unbalance will be first neutralized and then reversed. Therefore an oscillating current will
result in an oscillating voltage as well (or vice versa). If the current oscillation is sinu-
soidal, the voltage oscillation will also be sinusoidal and approximately 90 degrees
lagging the current in phase angle; that is, the short dipole is capacitive in nature, from
the viewpoint of its current-voltage relationship.
Since such a dipole can be regarded as one in which electric charge oscillates, it is an
oscillating electric dipole. It is distinguished from an oscillating magnetic dipole, which
is equivalent to a bar magnet whose magnetic strength and polarity oscillate. An example
of a magnetic dipole is discussed in sec. 4.5.1.
The current, though uniform throughout the length of the elemental dipole at any
instant, is assumed to vary sinusoidally in time, according to (4–1) that follows
I (t ) = I 0 sin (2π ft + α ) (4–1)
where I(t) is the current at any time t, I0 is the amplitude of the current (the rf peak value),
f is the frequency in Hertz, t is the time in seconds, and a is the phase angle (which
simply means that when t = 0, I(t) = I0 sin a).
4.1.1. Dipole Radiation

It is a well-known fact of elementary electrical theory that current in a wire is accompa-
nied by a magnetic field surrounding the wire, and also that a voltage existing between
two conductors, or different parts of a conductor, is associated with an electric field in
the intervening and adjoining space. A dipole antenna, therefore, will be surrounded by
an electric field and a magnetic field. The nature of these fields at an instant of time in
the vicinity of the dipole is indicated in Fig. 4–1.
The current and voltage of an rf dipole change direction (polarity) at a rate determined
by the frequency f and undergo a sinusoidal variation of their magnitudes as shown (for
Short Dipoles 121
the current) by (4–1). At a given instant,

+
however, they have a particular direction,
and the field lines have a corresponding
I
direction, as shown in Fig. 4–1. When the
current direction reverses, the direction of
–
the magnetic field lines reverses also; and
(b)
when the voltage polarity reverses, the
(a)
electric field reverses direction.
The field lines are shown only in the FIGURE 4–1.
immediate vicinity of the dipole, and only Configuration of electric and magnetic fields
a few lines are shown. In principle, (approximate) in the immediate vicinity of an
however, they fill the entire space around elemental-dipole radiator. (a) Electric field
lines. (b) Magnetic field lines.
the dipole and would extend to an infinite
distance from it if the dipole were located
in empty space and if the current had been flowing in it for an infinite time. But the fields
do not exist at an infinite distance as soon as the current starts flowing. They travel
outward from the dipole at the finite speed c = 3 × 108 meters per second (in empty space
and at practically the same speed in air).
Therefore, by the time the field lines corresponding to a given polarity of the dipole
have reached a distance from the dipole equal to 12 (c / f ) , which is a half wavelength as
defined by equation (1–1) of chapter 1, the dipole polarity has reversed, and the new field
lines being set up in the region immediately adjoining the dipole have directions opposite
to those at the distance 12 (c / f ) = λ /2 . As the oscillation of current and voltage in the
dipole continues, the outward-traveling field lines will evidently have opposite directions
at half-wavelength intervals along the direction of travel. This oscillating outward-
traveling field is an electromagnetic wave, which has been radiated by the dipole. The
approximate configuration of the electric-field component of this wave, in a region
extending several half wavelengths from the dipole, is shown in Fig. 4–2a, and the mag-
netic-field component is shown in Fig. 4–2b.
It is noteworthy that some of the electric field lines, rather than terminating on charges
in the conductors, form closed loops in space—something that cannot occur in the elec-
trostatic case (where the field is due to a steady or d-c voltage). This field configuration,
resulting in radiation, is an electrodynamic phenomenon; it happens only with varying
currents and voltages, and with moving fields. At much greater distances than those
shown in Fig. 4–2, however, the “closed ends” of the loops virtually disappear, leaving
only field lines that are transverse to the direction of propagation of the waves, as in Fig.
1–1, chapter 1.
Not all the fields in the vicinity of the dipole represent radiation. When the dipole polar-
ity reverses, some of these field components “collapse” upon the conductor, causing
induced voltage and current associated with the inductance and capacitance of the dipole
conductor. These are the field components that decrease rapidly with distance from the
dipole, as discussed in sec. 3.2.5. One component, the static field, decreases as the inverse
cube (1/r 3) of the distance r. A second component, the time varying reactive field,
decreases as the inverse square (1/r 2) of the distance. These two components comprise the
Dipole axis near field of the dipole. The third compo-

nent, which decreases as the inverse first
power of the distance (1/r), is the radia-
tion field. Because of their more rapid
decrease with distance, the static and reac-
tive fields become very much smaller than
the radiation field at great distances from
the dipole; in fact, at a distance of a few
wavelengths, the radiation field is so very
much the strongest component that the
near-field components are virtually negli-
gible. Close to the dipole, however, the
near-field components are much stronger
than the radiation components.
The existence of a radiation field was
(a) unnoticed in the earliest experiments with
electricity, because of the dominating
reactive fields at short distances. Its exis-
tence was first demonstrated about 1887
by the classical experiments of the German
physicist Heinrich Hertz, though it had
been predicted theoretically about 25
years earlier by the English mathematical
physicist James Clerk Maxwell (see
Appendix A).
Electric and magnetic fields represent
energy. The static and reactive fields repre-
sent stored energy that is returned to the
conductor, just as the fields associated with
coils and condensers of an electric circuit
return energy to the circuit when the current
or voltage maintaining the field is removed.
(b)
The radiated fields, on the other hand,
FIGURE 4–2. represent electromagnetic energy flowing
Configuration of electric and magnetic field outward into space. From the viewpoint of
lines in radiation field of an elemental-dipole the dipole conductor, radiated energy is
radiator. (a) Electric field. (b) Magnetic field. “lost” in the same sense that electrical
Electric field lines are shown in a plane energy is lost when it is converted into heat
containing the dipole axis, and magnetic field
lines are shown in the plane through the center energy in a resistance. As mentioned in
of the dipole perpendicular to its axis. (Dipole sec. 3.7, this energy loss is ascribed to a
is perpendicular to plane of paper in Figure hypothetical “radiation resistance,” which
4–2b.) is a helpful concept for theoretical con-
siderations involving the antenna input im-
pedance, efficiency, and radiated power.
Short Dipoles 123
In the radiation field there is a continual exchange of energy between the electric and
magnetic field components. As part of his theory, Maxwell postulated that an electric
field acting in empty space causes a “displacement current” just as it causes an electron
current in a conductor. He further hypothesized that this displacement current has the
same ability to set up a magnetic field as does an electron current. Many experiments
confirm this theory, which is now unquestioned. A varying electric field therefore causes
a varying displacement current, which in turn results in a varying magnetic field. The
magnetic field, in its turn, creates a varying electric field in accordance with Faraday’s
law. The cycle is thus complete; each field component sustains the other, and in regions
remote from conductors or charges (empty space), one cannot exist without the other, in
the fixed ratio given by equation (1–10), chapter 1.
That electric and magnetic fields represent energy may be demonstrated by placing a
charged object in an electromagnetic field. The field will exert a force upon such an object
and cause it to move if there are no restraining forces. This effect occurs, for example,
when radio waves encounter the earth’s ionosphere, in which there are free electrons.
These electrons are accelerated by the field, which means that kinetic energy is
imparted to them. The energy is supplied by the electromagnetic field. Similarly, an
electromagnetic field impinging on a receiving antenna transfers energy (power) to it,
which may be amplified by the receiver and eventually converted into sound variations
in a loudspeaker, light variations on a television screen, or some other form of
intelligence.
4.1.2. Pattern of an Elemental Dipole

If an elemental dipole (length < 0.1l) is placed at the origin of a spherical-coordinate
system (Fig. 3–1) with its axis (direction of its length) parallel to the q = 0 direction (the
z-axis of the corresponding cartesian-coordinate system), the radiated field may be con-
sidered for any point in space whose position is given by the coordinates r, q, f; and the
electric field intensity will then be designated E(r, q, f). The geometry of this situation
is shown in Fig. 4–3.
Analysis based on Maxwell’s equations shows that the radiation field (far field) of this
radiator is given by
q = 0°
(z) f = 90°
( y)
60π Il sin θ
E (r , θ , φ ) = (4–2)
λr Dipole
l (x) f = 0°
where l is the length of the elemental
dipole and I is the dipole current, in
amperes. If l, l, and r are in meters, and
I is in rms amperes, E is obtained from
this formula in rms volts per meter. FIGURE 4–3.
The fact that the angle f does not Relationship of elemental dipole radiator to
appear in this expression means that for spherical coordinate system.
q fixed values of r and q, E does not vary

330°
0° as f varies, that is, E is independent of f.
30°
Pattern Stated otherwise, the pattern of E in any
plane parallel to the xy-plane (i.e., a plane
300° 60°
E = 0.5 in which f varies but q does not) is a
E = 1.0 perfect circle. It also means that the
270° 90° pattern of E in all planes containing the
z-axis (q = 0-axis) will be exactly alike;
for example, the pattern in the xz-plane
240° 120° will be the same as the pattern in the
yz-plane. Finally, it means that all signifi-
210° 150° cant information concerning the three-
180° dimensional pattern is contained in just
FIGURE 4–4. one of these plane patterns in which q is
Relative electric-intensity pattern of elemental the angle variable. The relative pattern in
(or short) dipole in a plane perpendicular to the such a plane, as given by (4–2), is shown
dipole axis. (E = |sin q|). (Note: The range of in Fig. 4–4.
the angle q in this diagram is 0–360°, to
indicate that the pattern is shown in a whole
Figure 4–4 shows only the variation of
plane. Strictly, however, since the spherical the rms value of E (or the amplitude,
colatitude coordinate q has a maximum value depending on whether I in (4–2) is taken
of 180°, the angles on the left side of the to be the rms value or the amplitude of
diagram should go from 0 to 180° like those on the current), as functions of r and q. The
the right side.) polarization is also of interest. It is linear,
and in the q = 90° plane (the plane through
the dipole center perpendicular to the dipole axis) the polarization is parallel to the dipole
axis. At any general field point the polarization is by definition the direction of the electric
field lines. These lines, in the radiation field, are always perpendicular to a line from
the dipole center to the field point, and they lie in a plane containing this line and the
dipole axis.
The magnitude of E, given by (4–2), and its direction in space (polarization) are two
properties of a radiation field that must be specified for a description of the field. A third
quantity also to be specified is the relative phase of E as a function of q and f for a fixed
value of r. For the elemental dipole field the phase is constant for a fixed value of r; that
is, it does not vary as q and f are varied. The absolute phase of E may be expressed as
a function of the distance r and the phase of the current I. Relative to the phase of I, the
phase angle of the electric field E at distance r is
π 2π r π ⎛ 4r ⎞
γ = − = ⎜1 − ⎟ radians (4–3)
2 λ 2⎝ λ⎠
Note that −2pr/l is the delay of phase, in radians, created by the separation distance r.
Equation (4–3), as well as (4–2), applies only in the far field, which means at values of
r that are of the order of l or greater.
Short Dipoles 125
The relative-power-density pattern may be obtained from (4–2) by making use of the
relation p = E2/377 from (1–9). (It is useful here to note that the number 377 actually
represents the quantity 120p.) The result is
30π I 2 l 2 sin 2 θ
p (r , θ , φ ) = (4–4)
λ 2r 2
This formula gives p in watts per square meter when I is in rms amperes and all lengths
are in meters.
4.1.3. Radiation Resistance

The total power radiated Ptotal by an elemental dipole is found by integrating p(r, q, f)
over the surface of an imaginary sphere at a fixed value of r. Then, in accordance with
equation (3–15), chapter 3
2π π 30π I 2 l 2 2π π  sin 2 θ  2
Ptotal = ∫
0 ∫0 [ p (r, θ , φ )] r 2 sin θ dθ dφ = λ2
∫0 ∫0  2 
r sin θ dθ dφ
 r  (4–5)
60π 2 I 2 l 2 π 790 I 2 l 2
=
λ2
∫0 sin θ dθ =
3
λ2
watts
The radiation resistance may now be calculated, from (3–28), by dividing Ptotal by I2.
The result is
l2
Rr = 790 ohms (4–6)
λ2
For example, if l/l = 0.1, Rr = 7.9 ohms. (It will be recalled that this value of l/l was
stipulated at the beginning of this section to be approximately the maximum permissible
length for an antenna to qualify as an elemental dipole.) The dipole current required for
one kilowatt of power with this value of radiation resistance is 11.3 amp (a rather large
current). This result indicates a major disadvantage of short dipoles, namely, the very
large currents required for radiation of appreciable power. Therefore, radiators with larger
values of radiation resistance are preferred when they can be used; but this is not always
possible, and short dipoles are very useful radiators under these circumstances.
4.1.4. Directivity
The directivity of an elemental dipole can be computed by using foregoing results and
the fundamental definition of directivity, equation (3–16), that follows:
D = U max U av = p (r , θ , φ )max p (r , θ , φ )av

providing p(r, q, f)max and p(r, q, f)av are evaluated at the same distance r. The numerator
of (3–16), the equation above, is available from (4–4) with q = 90°, so that sin2 q = 1
(the maximum value). The denominator is Ptotal/4pr2, where 4pr2 is the area of a sphere
of radius r and Ptotal is given by (4–5). Then, complete expression for the directivity (with
790 written as 80p2) is
 30π I 2 l 2   4π r 2 λ 2 
D= = 1.5 (4–7)
 λ 2 r 2   80π 2 I 2 l 2 
Thus the directive is 1.5 regardless of the exact length l in relation to the wavelength l,
as long as the radiator qualifies as an elemental dipole. In fact, this result applies to any
short dipole.
4.1.5. Beamwidth
The pattern of Fig. 4–4 (the two circles formed by heavy lines) is created by a “slice”
through the three-dimensional pattern. However, the total pattern in space is shaped like a
doughnut (with a pin-sized hole). The two lobes of Fig. 4–4 are really cross sections of the
same lobe, and the angular width of this doughnut-shaped lobe will now be discussed.
The half-power beamwidth may be determined either by measuring the angular width
of the pattern, as plotted in Fig. 4–4, between the points of electric intensity equal to
0.707Emax, or from analysis of (4–4). The only angle-dependent term in this expression
is sin2q, and it determines the pattern and the beamwidth. The quantity sin2q has its
maximum value of unity at q = 90°. Therefore, the half-power beamwidth is determined
by finding the values of q where sin2q equals 12 . The values are q = 45° and q = 135°
and thus the half-power beamwidth is 135° − 45° = 90°.
4.1.6. Input Impedance

The input impedance of a short dipole, when it is fed at a small gap near its center, con-
sists of the radiation resistance in series with a large value of capacitive reactance
(equivalent to a small series capacitance). There is also, in series, an equivalent loss
resistance R0, included previously in (3–26), that may be large enough to be significant
in relation to Rr. An equivalent circuit of the dipole, as it looks from the viewpoint of
the transmission line, is shown in Fig. 4–5. Of course there is not actually a “series con-
denser” in the dipole; this circuit merely represents the impedance in a lumped-circuit
equivalent form. Because the value of the equivalent series capacitance is very small, it
represents a high value of capacitive reactance, which must be “tuned out” by including
an inductive reactance of equal value in the feed circuit. Then the voltage supplied by
the transmission line need not be extremely high. The current must be high, however, in
order to radiate appreciable power, as indicated previously by (3–25). (In effect this
equation shows that the radiated power is P = I2Rr; so if Rr is small, I must be large to
make P large.)
Short Dipoles 127
Transmission Transmission
line C R0 line
l
Rr
Input terminals
Dipole input
terminals
(a) Dipole
(b)
FIGURE 4–5.
Equivalent and actual circuits of center-fed short dipole. (a) Equivalent circuit. (b) Center-fed
dipole and feed line.
This high current, flowing through the inductive and capacitive reactances, produces
very high voltages across the feed-circuit inductor and across the antenna input terminals,
even though the transmitter itself does not have to supply a high rf voltage (because the
inductive and capacitive reactances cancel, leaving only Rr and R0 as the effective trans-
mitter load). If the input resistance Ri (= R0 + Rr) is very small, a more complicated
arrangement than that shown in Fig. 4–6 may be required to provide an impedance
transformation in addition to reactance cancellation. Therefore feeding a short dipole
antenna is somewhat difficult. In particular, the inductance in the feed circuit, when high
power is being radiated, must be very large both to withstand the high voltage and to
carry the heavy current without excessive loss. These are expensive requirements.
4.1.7. Short Dipole with Nonuniform Current Distribution

An isolated elemental dipole cannot actually be achieved as a physical reality; it is more
of a concept, like the isotrope, than a practical radiator. Some types of dipole radiators
do exist that approximate the properties of the elemental dipole. If an actual short dipole
is fed in the manner shown in Fig. 4–5b, current will flow in the dipole conductor, and
radiation will occur. The current in the conductor, however, will not be uniform as
assumed for an elemental dipole, and therefore equations (4–2) to (4–6) will not apply
to its radiation without modification. The necessary modification turns out to be very
simple.
Because the short dipole conductor is “open” at both ends, that is, not connected to
anything, the current at these points must be zero. (Physical theory exists to support this
statement, although here it will simply be assumed to be intuitively obvious that a current
Transmission Transmission
line L C R0 line
l
Rr
Tuning coils
Dipole input dipole
terminals
(a)
(b)
FIGURE 4–6.
Method of tuning out dipole capacitive input reactance with series inductance (L).
(a) Equivalent circuit. (b) Dipole with input tuning coils. Reactance cancellation results
when 2p fL = 1/2pfC.
cannot exist where it has “no place to go.”) At the same time, current can exist elsewhere
in the dipole. This statement would seem to violate the physical principle of continuity
for electric current, which implies that a current must have the same value everywhere
along a continuous conductor. It will be recalled that this is not true in a transmission
line on which a standing wave exists (sec. 2.1). The current in the dipole is a standing
wave. It may be thought of as two equal and opposite currents (the reverse current is due
to reflection from the open end) that, at the ends of the dipole, have exactly opposite
phases and thus cancel, resulting in zero current. The cancellation is incomplete (because
of the changing phase relationship) at points a short distance from the end and becomes
progressively less complete going from the ends to the center. The principle of continuity
is satisfied for the currents traveling in each direction, separately.
When the dipole is very short compared with a wavelength, as assumed, the current
will vary approximately linearly along the dipole from end to center. This means that a
graph representing the current as ordinate I, and distance from the end to the center of
the dipole as abscissa x, is a pair of sloping straight lines forming a triangle, as shown
in Fig. 4–7, with zero value at the left end (end of the dipole, x = −l/2), maximum value,
Imax, at the center, x = 0, and zero value again at the right end (x = +l/2, l being the total
length of the dipole). It can be shown that the short dipole with linear current distribution
is like an elemental dipole with an “effective length” equal to half its actual length, and
a uniform current equal to the current at the center of the actual dipole. Equations (4–2)
to (4–6) may be used to describe the behavior of the actual short dipole if l in
these equations is replaced by (l/2). With this modification, the entire discussion of the
Short Dipoles 129
elemental dipole, including the feedpoint imped-

ance considerations, applies to the actual short Imax
Graph of
dipole. The directivity and the beamwidth— dipole current
in fact, the total pattern—are the same for the I
open-ended short dipole and for the elemental
dipole. –l /2 0 +l /2
x
Feed-point gap
4.1.8. Short Vertical Antenna with l
Ground Image
Dipole
At very low frequencies, below 500 kHz, for
example, the earth in most localities is a nearly Transmission
perfect reflector of radio waves. Since a wave- line
length at these frequencies is physically quite
long (600 m or about 2,000 ft. at 500 kHz, for
example, and 30 km or about 19 miles at FIGURE 4–7.
10 kHz!) it is difficult to get a horizontal antenna Current distribution in a short open-
high enough above the earth to produce appre- ended center-fed dipole. (Gap at center
ciable low-angle radiation (based on the princi- is assumed to be of negligible length.)
ples illustrated in Figs. 1–14 and 1–15).
Moreover, antenna lengths that are an appre-
ciable fraction of a wavelength are quite long, and expensive.
If a vertical antenna is erected with its base at the ground, it will be imaged in the
earth in accordance with the principle of images (see sec. 1.2). The phase of the
equivalent current in the image conductor is such that the antenna-plus-image may be
considered a single antenna in free space. Since the height of the vertical antenna-plus-
image, 2h, will usually be a small fraction of a wavelength, the radiation is like that of
a short dipole in free space. This combination of a half dipole, of height h, and its image
in a reflecting surface is known as a monopole.
The pattern, however, is actually only half the free-space pattern, since the earth “cuts
off” the other half. For a given current at the base of the antenna, the total radiated power
is only half as great as it would be for the antenna-plus-image in (actual) free space with
the same maximum current, as is found by substituting p/2 for p in the q-integration of
(4–5). Therefore, the radiated power and radiation resistance are only half as great as the
values calculated on the free-space antenna-plus-image basis. Then, for a fixed current
I, because of a smaller radiation resistance, both the input power and the radiated power
are reduced.
As already discussed in connection with Fig. 4–7, the effective length of a short dipole
being one-half its overall length may be used. Thus, although the length of the antenna-
plus-image is 2h, the effective monopole length is half as much, that is, h.
The total radiated power Ptotal of a short monopole is determined as follows:
(1) replace the length l of the elemental dipole of (4–5) with h, the effective
length of the short monopole (i.e., the half-dipole plus image in ground
plane), and
(2) substitute p/2 for p in the q-integration of (4–5), because the earth’s block-
age; this causes the factor 790 to be replaced by 395.
Then, by using the Ptotal for the monopole, its radiation resistance becomes
Ptotal 395h 2
Rr = = (4–8)
I2 λ2
It is to be recalled that h is the actual height of the half-dipole of the monopole, which
is the same as the effective height of the half-dipole plus its image. Alternatively, Rr may
be expressed in terms of the effective height (i.e., length) of the half-dipole he, which
equals h/2. Then, Rr = 1580h2e/l2.
Recall that, for a monopole, the total power radiated is effectively concentrated into
half the solid angle of the free-space case. Then, by the reasoning given in chapter 3,
e.g., equation (3–21), the directivity will be twice that of a free-space short dipole. Thus
for the monopole, D = 3, as compared to D = 1.5 given in (4–7) for the elemental dipole.
The pattern in the horizontal plane is uniform (circular), that is, equal signal strength is
radiated in all horizontal directions. The radiation at the peak of the beam is vertically
polarized. The half-power vertical beamwidth is half that of the free-space short dipole
(i.e., 45 degrees) since the earth eliminates half of the pattern. However, calculations of
field strength at a distance in the actual earth environment cannot be made on the basis
of these “semi-free-space” results because the propagation of the vertically polarized
waves depends on the semi-guiding effect of the earth’s surface, and at very great dis-
tances the ionosphere plays a part, as discussed in sec. 1.4.
A vertical radiator of this type may take the form of a steel tower with its base insu-
lated from earth; then it is fed by connecting the source (transmitter) between the tower
base and the ground, with an inductance in the feed line to compensate for the capacitive
reactance of the antenna. The ground is a source of appreciable loss unless care is taken
to minimize the resistance of the ground connection. In high-power transmitting instal-
lations an elaborate network of buried wires is used to make a good connection.
4.1.9. Top-Loaded Antenna

As has been indicated in the discussion of input impedance, in this section, the low radia-
tion resistance of a short dipole together with the high capacitive reactance component
of the input impedance create a difficult feed problem. As (4–8) above shows, the radia-
tion resistance can be increased by increasing the effective height. This can be done either
by increasing the actual height, or by changing the current distribution in the antenna so
as to increase the effective height. As has been explained in the discussion of a short
dipole with nonuniform current distribution, with a linearly decreasing current that has
zero value at the end of the radiator the effective height is only half the actual height. If
the current can somehow be made uniform, the effective height becomes equal to the
actual height, and the radiation resistance is increased by a factor of 4 (because the effec-
tive height is squared in the radiation-resistance formula, i.e., the numerical constant 395
in (4–8) becomes 1580 if h is interpreted as the effective height).
Short Dipoles 131
Flat top
Insulator Insulator
Support Vertical Support

tower radiator tower
Insulator
Feed wire
Ground
FIGURE 4–8.
A top-loaded short-monopole antenna, semischematic.
Current uniformity can be either totally or partially accomplished by “top loading”*

the antenna. This procedure consists of running wires approximately horizontally from
the upper end of the dipole. These wires do not radiate, but their capacitance to ground
results in larger current at the upper end of the vertical conductor. If this “flat top” of
horizontal wires is sufficiently extensive, and the vertical section is of “short dipole”
length, the vertical current distribution may be made practically uniform (as assumed in
analysis of elemental dipoles) so that the effective height is equal to the actual height.
When the top loading does not produce approximately uniform current in the vertical
section, the result is an effective height less than the actual height but greater than the
effective height with no top loading.
The top loading also has the effect of reducing the magnitude of the capacitive-
reactance component of the input impedance. This means that much less inductance is
needed in series with the feed line, and also that for a given radiated power the voltage
across this inductance and at the antenna base will be greatly reduced. This effect is
perhaps the greatest benefit of top loading in high-power installations, since voltage
breakdown and corona losses on the antenna are serious problems.
Top-loaded short vertical radiators at very low frequencies take many physical forms.
A typical one is illustrated in Fig. 4–8. The flat top, as shown, is strung between two
vertical towers, and the vertical radiating portion is supported at the center of the flat-top
span. The flat top is insulated from the towers, and the antenna is fed at the bottom of
the down lead through an inductance; the other side of the feed line is grounded through
a network of buried conductors.
A dramatic form of this type of antenna is the United States Navy’s Jim Creek VLF
installation (in the state of Washington), which operates in the 10 to 30 kHz frequency
region. This flat top is a conductor supported between two mountains about 2 kilometers
* The equivalent procedure for a dipole would be called end loading. The term top loading is used simply
because the end of the monopole is at the top. On a dipole, both ends would be “loaded,” symmetrically.
apart, and the vertical radiating portion is close to 300 meters in height. Another United
States Navy installation, in Cutler, Maine, has towers nearly 300 meters high; but even
these heights are short compared to the wavelength at these frequencies (l = 10,000
meters at f = 30 kHz).
To illustrate the loss problem with electrically short vertical antennas using ground
images, typical radiation efficiency factors (kr, equation 3–30) range from 0.05 to 0.5
(often expressed as 5 to 50 percent). The majority of the loss occurs in the ground resis-
tance, though some occurs in the feed-line tuning coil and, in very high-power installa-
tions, in insulator leakage losses and corona. (Efficiencies of high-frequency antennas
are often close to 100 percent.)
4.2. Current and Voltage in Longer Antennas

Electrically short antennas have been discussed in some detail because they are of both
theoretical and practical importance. At higher frequencies, beginning in the MHz region
and especially in the HF region and above, longer electrical lengths become feasible, and
their use results in certain advantages. In particular, the difficulties encountered in feeding
short dipoles are eliminated, and higher directivities may be obtained. The input imped-
ance of longer antennas may be made nonreactive, and the radiation resistance is usually
much higher than that of a short dipole.
The variation of current and voltage along the length of longer antenna wires is more
complicated than for short or elemental dipoles. For an elemental dipole it was assumed
that the current is uniform (constant) along the conductor at any instant of time, although
varying in time according to (4–1). For a short dipole without capacitive end loading,
the current varies in a linear (straightline) fashion, as shown in Fig. 4–7. The voltage,
though not shown in Fig. 4–7, is of opposite polarity on either side of the feed-point gap
and has virtually constant amplitude along the conductor length.
The patterns of variation of the current and voltage along the conductor are called
the current and voltage distributions of the antenna. These distributions are important
in understanding the radiation properties of various antenna lengths and feed
arrangements.
The current and voltage distributions on open-ended long antenna wires are basically
similar to the standing waves of current and voltage on an open-ended two-wire trans-
mission line. This standing-wave pattern is shown in Fig. 4–9a. (The voltage distribution
is the same as shown in Fig. 2–5, chapter 2, for ZL → ∞.) The voltage maxima occur at
the end of the line, and at other points an integral number of half wavelengths from the
end. The voltage has zero values (nulls) at points an odd number of quarter wavelengths
from the end. The current has maxima at the voltage minima, and the current minima
are at the voltage maxima.
These patterns are plotted with both positive and negative “amplitudes” (I0 and V0) to
emphasize the phase reversals. The same phase information is conveyed by the phase
distribution patterns of Fig. 4–9b, which show that the phases of the voltage and current
are constant in the intervals between the nulls and that a sudden change of 180 degrees
occurs at the nulls. It is also shown that the current and voltage are everywhere 90 degrees
Current and Voltage in Longer Antennas 133
out of phase with each other. (A 270- l/4

degree phase difference is equivalent to
I0 (m) V0 (m)
90 degrees.)
0 I0,
If now the wires in the end quarter- V0
I0
wavelength section of the open-ended V0
two-wire line are bent outwards at right l/2
angles to their original directions in the x
plane of the line, as shown in Fig. 4–10a, (a)
the result is a half-wave center-fed dipole. Voltage + 180°
Figure 4–10b shows the current and Current + 90°
voltage distribution on the dipole. Essen- 0°
tially it is the same pattern that existed on – 90°
these wires before they were bent to form – 180°
the dipole, with zero current at the ends (b)
and maximum current at the center. The Transmission line
Input end Open end
voltage is maximum at the ends and zero
at the center. These features are the same FIGURE 4–9.
as those of the short open-ended dipole, Voltage and current standing wave patterns on
whose current distribution is shown in an open-ended uniform transmission line,
Fig. 4–7. The difference is that the varia- assuming no line losses. (a) Amplitude.
tion is no longer linear; it is sinusoidal. If (b) Phase angle.
the current and voltage are expressed as
functions of the distance x measured along the wire from one of the open ends of the
antenna, the phasor* amplitudes are given by the equations:
2π x 
I 0 ( x ) = jI 0(m ) sin 
 λ 
(4–9)
2π x 
V0 ( x ) = V0(m ) cos 
 λ 
(4–10)
where I0(m) and V0(m) are the maximum amplitudes. These equations do not show the time
variation, or instantaneous values; they are obtained by multiplying the amplitudes by
the factor sin (2p ft + a), as in (4–1).
The space–time relationships of current and voltage on a half-wave dipole are some-
what difficult to visualize at first. As an aid in this effort, Fig. 4–11 shows the instanta-
neous patterns of current and voltage on the dipole at several instants during a single
rf cycle. In these diagrams, the rf period T is the time for completion of a single cycle;
* In these expressions, the j factor indicates that the current is 90 degrees out of phase with the voltage,
and the positive and negative sign changes resulting from the sine and cosine functions of x indicate the
180-degree phase changes that occur at the standing-wave nulls. I0(m) is the value of I0(x) that occurs at
x = l /4, and V0(m) is the value of V0(x) that occurs at x = 0 and x = l /2.
Half-wave dipole
l /4
l /2 V0
Two-wire line
l /4
I0
(a) (b) – 90° 0° + 90°
(c)
FIGURE 4–10.
Current-voltage distribution on half-wave dipole. Comparison with Figure 4–9 indicates
correspondence with distribution on open-ended transmission line. (a) End-quarter-wave section
of two-wire line bent back to form half-wave dipole. (b) Amplitude distribution of voltage and
current on half-wave dipole. (c) Voltage phase distribution (solid) and current phase (dashed).
t=0 t = 3T/8 t = 3T/4
l /2
t = T/8 t = T/2 t = 7T/8
t = T/4 t = 5T/8 t=T
FIGURE 4–11.
Instantaneous distributions of half-wave-dipole current (dashed lines) and voltage (solid lines)
at various times (t) during one rf cycle (period T = 1/f ). (At t = 0, T/2, and T the voltage is
zero everywhere, and at t = T/4 and 3T/4 the current is zero everywhere.) The t = T diagram is
identical to the t = 0 diagram, indicating that the cycle is complete and starting over.
it is equal to 1/f, where f is the frequency in Hertz. The patterns are drawn assuming that
at zero time (t = 0) the current is at its instantaneous maximum value, corresponding to
a = p/2 in (4–1).
Linear antennas longer than a half wavelength are also used. They may or may not be
fed at their centers. On each side of the feed point, the current and voltage distributions
are determined by (4–9) and (4–10) above. For correct operation of such an antenna,
Current and Voltage in Longer Antennas 135
with a balanced two-wire line, the lengths +

of wire on either side of the feed point V0 I0
must be the same or must differ by an

0 I0, Amplitude
integral number of half wavelengths. V0
Linear antennas of this type, on which
standing waves of current and voltage –
exist owing to reflections from the end of l
an open-ended wire, are called resonant + 180°
antennas.
+ 90°
At the gap in the antenna wire across V0 I0
which the feed line is connected, the 0° Phase
antenna voltage distribution undergoes a
– 90°
180-degree phase reversal, but the current
phase is the same on either side of the – 180°
gap. Therefore, if the feed point is at a (a)
current maximum (voltage minimum), V0 V0
+
the distributions over the entire antenna I0 I0
length will be the same as they would be
in an unbroken wire of the same length I
0 0, Amplitude
V0
(case 1). This situation exists when the
wire lengths on each side of the feed point
–
are odd integral multiples of a quarter
wavelength. But if the feed point is at a l
voltage maximum, which will be the case
if the wire lengths on each side are inte- V0
+ 90°
gral multiples of a half wavelength, the I0
antenna phase pattern is not the same as 0° Phase
it would be on an unbroken wire (case 2). – 90°
The voltage-current-phase patterns for (b)
these two cases are shown in Fig. 4–12, FIGURE 4–12.
for a total antenna length of one wave-
Voltage-current distributions on a full-wave
length. Antennas of the first type, having antenna fed (a) at end or current maximum in
the unbroken-wire type of distribution 4(a) and (b) at center (voltage maximum) in
(obtained by feeding either at one end 4(b).
or at a current maximum), are properly
termed long-wire resonant antennas; those of the second type, in which the feed point
is a voltage maximum, are actually two-element collinear arrays. (If the total antenna
length is one wavelength, as in Fig. 4–12b, each element of this collinear array is a half-
wave dipole, and for longer total lengths each element may itself be a long-wire antenna.)
Because arrays are discussed in chapter 5, only the true long-wire types will be discussed
in this chapter, although some types of collinear and other long-wire arrays are sometimes
loosely referred to as long-wire antennas.
Actually these sinusoidal distributions of current and voltage are approximations
rather than exact descriptions. They are slightly modified, on an actual antenna, by the
radiation resistance of the antenna and by the fact that the antenna wires are not equiva-
lent to a uniform transmission line. The radiation resistance, as well as any actual resis-
tance in the antenna wire, results in a small component of current that is in phase with
the voltage, rather than 90 degrees out of phase. But the sinusoidal approximation is quite
good for linear antennas whose conductors are very thin compared to their length, and
of high conductivity. (It is also assumed that the antenna wire is not close to any large
irregular conducting bodies or dielectric material that would disturb the uniformity of
the electrical environment. In fact, a free-space environment is assumed, but the assumed
distributions apply reasonably well in practical situations.)
Antennas may also be designed to have uniform current and voltage amplitudes along
their lengths, that is, no standing waves. This result is achieved by terminating the end
of the antenna wire in a resistive load so that no reflection occurs. In one form of such
an antenna (Beverage or wave antenna), the wire runs approximately horizontally above
the earth, and the input terminals consist of one end of the wire and the ground. The ter-
minating resistor is connected between the other end of the wire and the ground. In
another form (rhombic antenna) long wires form an array in the shape of a diamond
(rhombus) in a horizontal plane. The two sides of the diamond are fed at one vertex, and
the terminating resistor is connected between them at the other vertex. The current and
voltage are approximately constant along the wires, but there is a gradual decrease of
both with increasing distance from the feed point, owing to the radiation losses and the
ohmic loss in the wire. The current and voltage are in phase with each other everywhere,
rather than approximately 90 degrees out of phase as with standing-wave distributions,
but their phases change linearly with distance along the wire in the amount of 2p radians
or 360 degrees for every wavelength. This description is characteristic of traveling waves,
as described by (1–2) and (1–3) of chapter 1 for waves in space, and by (2–3) of chapter
2 for waves on wires. Antennas having traveling-wave current and voltage distribution
are called nonresonant antennas or traveling wave antennas.
4.3. The Half-Wave Dipole

The radiation patterns of linear antennas that do not qualify as “short dipoles” may be
found by considering them to be composed of a number of elemental dipoles placed end
to end. For example, a dipole a half wavelength long might be approximated by five
tenth-wavelength elemental dipoles end to end. The current in each elemental dipole
would (by definition) be constant and equal to the average current in the corresponding
section of the half-wave dipole, as indicated in Fig. 4–13. The current distribution is a
half cycle of a sinusoid with the maximum at the dipole center, as in Fig. 4–10. The
current has a constant phase angle everywhere on a half-wave dipole so that all the
elemental dipoles are assumed to be in phase.
The radiated field intensity at a distant point (field point) due to each elemental dipole
was given previously by (4–2), and its phase angle was given by (4–3), with the distance
r taken to be the distance to the field point from the center of the elemental dipole; that
is, r will have a slightly different value in computing the field-point contribution of each
elemental dipole. These slight distance differences will not significantly affect the relative
The Half-Wave Dipole 137
intensities of the individual elemental Assumed

elemental-
dipole fields, but they will affect the rela- dipole Actual
current sinusoidal
tive phases significantly. The total field at values current
a distant point is the phasor sum of the distribution
I0(m)
contributions of the individual elemental
dipoles, in accordance with the principle
of interference in sec. 1.2.
This method of analysis, as described 0.1l 0.1l 0.1l 0.1l 0.1l
thus far, is obviously an approximation
0.5l
and a rather crude one, when the half-
wave dipole is dissected into only five FIGURE 4–13.
elemental dipoles. The accuracy of the Approximation of actual half-wave dipole by
approximation may be increased by dis- five tenth-wave-length elemental dipoles, with
secting it into more elemental dipoles of currents equal to average current in
corresponding segment of half-wave dipole.
shorter individual lengths. But the labor
of calculation is also thereby increased, if
the phasor summation process is employed as described.
The approximation may be made exact, however, and the tedious summation process
avoided, by applying the methods of calculus. The half-wave dipole is then considered
to be composed of an infinite number of infinitesimal dipoles, and the phasor summation
of their fields at a distant point is expressed as an integral. The resulting expression for
the magnitude of the electric field of the half-wave dipole, at a distance r in the direction
q, f, obtained by solving the integral expression, is
 cos  π cos θ  
60 I  2 
E (r , θ , φ ) =   (4–11)
r  sin θ 
 
where E is in rms volts per meter if r is in meters and I is the rms current in amperes at
the center of the dipole. This pattern is seen to be a slightly more complicated mathemati-
cal expression than that of the short or elemental dipole of (4–2), but the patterns are
only slightly different. They are compared in Fig. 4–14. The half-wave dipole has a
slightly narrower beamwidth—78 degrees compared to 90 degrees for the short dipole.
Consequently its directivity is slightly greater—1.64 compared to 1.5 for the short dipole.
The power-density ratio is 1.64/1.5 = 1.093, and the field-strength ratio 1.047.
The slightly greater directivity of the half-wave dipole is thus almost insignificant. Its
advantage lies primarily in its increased radiation resistance and reduced or nonexistent
feed-point reactance. The radiation resistance for an exactly half-wavelength dipole is
found, by the method illustrated in the case of the elemental dipole, to be 73.1 ohms,
referred to the maximum current point (dipole center). Therefore this is also the resistive
component of the input impedance when the dipole is fed at the center. There is also a
small reactive component of 42.5 ohms, inductive. This small inductive reactance may
q = 0° be eliminated by shortening the dipole to

Isotrope about 95 percent of a half-wavelength
Dipole (i.e., about 0.475l). The radiation resis-
axis Short dipole tance (and input impedance) is then about
Half-wave 65 ohms (Kraus and Marhefka 2002, p.
dipole
182). The pattern (beamwidth and gain)
q = 90°
is not significantly affected by this slight
shortening.
These properties make the half-wave
dipole especially attractive as a radiator
for many purposes, at frequencies for
q = 180°
which its length is not excessively large
or minutely small. The results given for
FIGURE 4–14.
the radiation resistance and input imped-
Half-wave-dipole, short-dipole, and isotrope ance are free-space values for a conductor
patterns compared for equal power radiated very thin compared to the length, with
(electric-intensity patterns), in plane containing
dipole axes. no ohmic resistance. Consequently they
become somewhat modified for conduc-
tors of appreciable diameter or when the
dipole is close enough to the ground or other conductors to result in “coupling.” Usually
ohmic loss is small enough to be disregarded. Dipoles of larger diameter, and of special
shapes and configurations, are discussed in sec. 7.1.
When a vertical quarter-wavelength radiator is erected with its base at or just above
the ground, it is imaged in the earth so that its radiation may be analyzed as if it were a
half-wave dipole in free space, subject to the same modifications as discussed for the
short vertical dipole imaged in the ground. When the quarter-wave vertical antenna
(monopole) is fed at its base with the other side of the feed line connected to ground, its
radiation resistance and input impedance are just half the values for the half-wave dipole
in free space, and the directivity is twice as great. The radiation is vertically polarized
at the peak of the beam. Vertical radiators of other lengths may be similarly analyzed,
that is, by use of the image principle. Vertical-tower radiators of heights up to about 5/8
wavelength are much used for broadcasting and other applications in the medium-
frequency range of about 500 to 3,000 kHz. (Even higher vertical antennas may be used
if they are “sectionalized” so that they become, in effect, collinear-array antennas,
described in sec. 5.2). A monopole too short to be a quarter-wavelength high, yet too
long to be classed as a “short monopole,” may be capacitively top loaded (in the same
general manner as described for short vertical monopoles) to result in virtually quarter-
wave performance.
4.4. Long-Wire Antennas

Antennas consisting of a single straight wire, either unbroken or with a feed-point gap
at a current maximum when the antenna has standing-wave current distribution, are
classed as long-wire antennas if their length is substantially greater than a half wave-
Long-Wire Antennas 139
length. Such antennas are not properly called dipoles. On the other hand, the half-wave
antenna is commonly called a half-wave dipole, even though a true electric dipole is
equivalent to two equal and opposite polarity point charges separated by a definite dis-
tance. The elemental dipole and the short dipole are, in essence, equivalent to oscillating
electric dipoles, but longer antennas are not. However, usage sanctions the term for the
half-wave dipole.
The radiation patterns of long-wire antennas may be determined by the method
described for the half-wave dipole by considering them to be composed of end-to-end
infinitesimal dipoles. The current amplitude and phase in each infinitesimal dipole are
taken to be the values indicated by the current distributions calculated from (4–9), and
as shown for a one-wavelength wire in Fig. 4–12 for a resonant antenna. For a nonreso-
nant antenna a constant current amplitude along the wire is assumed, with a linear phase
change corresponding to a traveling wave of current (2p radians or 360 degrees per
wavelength). These current-distribution assumptions are valid for a thin wire of perfect
conductivity, ignoring the effect on current distribution of the radiation losses. Therefore,
the results are approximate, but useful in that they indicate the general nature of the
radiation patterns.
4.4.1. Patterns of Resonant Antennas

As shown in Fig. 4–14, the pattern of a short or half-wave dipole consists of a single
doughnut-shaped lobe of radiation (appearing as two oppositely directed lobes in a “slice”
or plane pattern containing the dipole axis). Long-wire radiators have more than one
three-dimensional lobe, taking the form of cones of radiation. The axes of the cones
coincide with the axis of the wire, and the sides of the cones are inclined at various angles
with respect to the wire. As for the short dipole, the patterns are uniform (circles) in the
plane perpendicular to the axis of the wire.
There will be one cone-shaped lobe for each half wavelength of wire length, for both
the standing-wave and traveling-wave antennas. The lobes are symmetrically disposed
with respect to the plane that bisects the wire. Therefore, if there is an odd number of
half wave-lengths, one lobe will be perpendicular to the wire, like the short-dipole lobe
except that it is thinner (narrower beamwidth), more like a pancake than a doughnut
when the wire is many wavelengths long. When the wire length is an even number of
half-wavelengths, there is no perpendicular lobe.
For the standing-wave or resonant antenna, the amplitude of the electric field strength
(pattern) is given by
 cos  nπ cos θ  
60 I   2 
E (r , θ , φ ) =   (4–12a)
r  sin θ 
 
where n is the number of half wavelengths in the wire length, assumed to be an odd
number, and as usual E is in rms volts per meter if r is in meters, and I is the rms current
q = 0°
q = 0° Wire
axis
Wire
axis
q = 90° q = 90°
q = 180°
n=3
(a) q = 180°
n=4
(b)
FIGURE 4–15.
Patterns of resonant long-wire antennas.
at a current maximum, in amperes. For a wire an even number of half-wavelengths long,

the equation becomes
 sin  nπ cos θ  
60 I   2 
E (r , θ , φ ) =   (4–12b)
r  sin θ 
 
The nature of these patterns is shown in Fig. 4–15 for an odd and an even number of
half-wavelengths.
In these formulas, it is assumed as usual that the antenna is located at the origin of a
spherical coordinate system with the axis of the wire along the q = 0° axis (z-axis, Fig.
3–1) and that I is the rms current at a current-maximum point of the sinusoidal standing
wave. The patterns shown are for long wires of modest length (n = 3 and n = 4). As the
number of half wavelengths is made larger, the number of lobes increases proportionately
and the lobes of maximum radiation lie closer to the wire. Because of the factor “sin q ”
in the denominator of (4–11) and (4–12), an envelope of the lobe pattern is, in three
dimensions, a circular cylinder parallel to the axis of the wire. In a plane containing the
wire axis, the edges of the envelope are straight lines parallel to the wire. The effect is
shown in Fig. 4–16 for a many-lobed pattern.
These patterns, and equations (4–11) and (4–12), are for antennas an integral number
(n) of half wavelengths long. The total length, however, should be shortened by about 5
percent of one half wavelength to eliminate a reactive

component of input impedance. This will not affect
the pattern appreciably.
q = 0°
4.4.2. Radiation Resistance and Directivity
The radiation resistance of a long-wire resonant
antenna n half wavelengths long, in free space, is
given approximately by
q = 90°
Rr = 73 + 69 log10 n (4–13)
for values of n greater than 2 (Brainerd et al. 1942).

The equation is also approximately correct for
n = 1.
The angle of maximum radiation, that is, the angle
that the strongest lobe makes with the wire axis (this q = 180°
is also the lobe closest to the axis, as Figs. 4–15 and
4–16 show), is given approximately by (4–14) below.
This formula is quite accurate for small values of n
FIGURE 4–16.
and gives a result close enough for most purposes,
even for large values of n. Pattern of five-wavelength resonant
antenna (n = 10).
n −1
cos θ max = (4–14)
n
The maximum directivity may be attained through knowledge of the maximum field
strength Emax, the radiated (transmitted) power Pt, and other previously developed rela-
tionships. By substituting cosqmax into (4–12a) or (4–12b), according to whether n is odd
or even, the value 60I/(rsinqmax) for the maximum field strength Emax is obtained.* The
power Pt is I2Rr from equation (3–28). Furthermore, the maximum and average power
densities pmax and pav are (Emax)2/377 and Pt/4pr2, respectively. Then, the maximum
directivity Dmax becomes
U max pmax 120

Dmax = = = (4–15)
U av pav Rr sin 2 θ max
It is interesting to note that this formula gives the correct result for a half-wave dipole,
Dmax = 1.64, when Rr is taken as 73 ohms and qmax = 90°; also, (4–13) and (4–14) are
correct for n = 1.
* When (4–14) is substituted into (4–12a) and (4–12b), their numerators become, respectively,
60I|cos {(n − 1)p/2}| and 60I|sin{(n − 1)p/2}|. For n odd, |cos{(n − 1)p/2}| = 1, and for n even,
|sin{(n − 1)p/2}| = 1. The denominators become r sinqmax.
4.4.3. Patterns of Nonresonant Antennas

A long wire with a traveling-wave current of uniform amplitude I has an electric-intensity
pattern that is given by
 60 I sin θ   π L 
E (r , θ , φ ) =   sin  λ (1 − cos θ ) (4–16)
 r (1 − cos θ ) 
where L is the length of the wire. This pattern has the same number of lobes as a resonant
wire of the same length, and the maxima and minima occur at approximately the same
positions. Their magnitudes, however, are quite different, as shown by the pattern of a
3/2-wavelength nonresonant wire in Fig. 4–17. As seen there, the lobes directed toward
one end of the wire are much larger than those at the other end of the pattern. The lobe
nearest the axis of the wire and pointed in the direction of the traveling wave of current
on the wire is the largest. The smallest lobe is the one at the other end of the pattern, the
magnitudes increasing progressively toward the large-lobe end.
This type of pattern has an advantage when it is desired to radiate or receive in pre-
dominantly one direction, rather than two. Suppression of the pattern in one direction is
accomplished by eliminating the reflected current at the end of the wire by means of a
resistive termination. This usually takes the form of a resistor connected from the end of
the wire to ground. Such termination can be successful, however, only if the height of
the antenna above ground is a very small fraction of a wavelength (otherwise the con-
nection would have reactance as well as resistance and would not be a reflection-free
termination). The correct value of the resistor, being connected between the end of the
wire and ground, is half the value that matches the impedance of a transmission line
consisting of the antenna wire and its earth image. If the antenna height is h and the wire
diameter is d, the resistance is from equation (2–62), chapter 2:
R = 138 log10   ohms

4h
(4–17)
q = 0°  d
However, this formula should be used only as a rough

q = 90°
guide. The usual practice is to adjust the resistance until
no standing wave exists on the antenna wire.
q = 180°
4.4.4. Polarization
The radiation from a long-wire antenna is linearly polar-
FIGURE 4–17.
ized, but the polarization (electric field) direction is not the
Pattern of three-half- same in all parts of the pattern. (This is true even of short
wavelength nonresonant and half-wave dipoles, but for them the variation is not as
long-wire antenna (L = 1.5l).
(Compare with resonant wire great because there is only one lobe perpendicular to the
of same length, n = 3 pattern wire; in the perpendicular plane the polarization is simply
of Figure 4–15.) parallel to the wire.) The polarization in one of the oblique
lobes of a long-wire antenna, or in fact in any particular part of the pattern, may be
determined by the following procedure (assuming free-space propagation):
(i) Draw a line from the center of the antenna in the direction of interest.
(ii) Form the plane that contains both the antenna wire and this direction line.
(iii) At any point on this direction line the polarization (electric field) vector is
perpendicular to the direction line and lies in the plane thus formed.
(This procedure is in fact applicable to the radiation from any straight-wire radiator,
including short dipoles and half-wave dipoles.)
4.4.5. Effect of the Ground and Other Factors

The foregoing discussion of long-wire antennas has assumed perfect thin-wire conductors
and a free-space environment. Ground reflection affects the vertical-plane pattern in the
manner discussed in sec. 1.4.3 and illustrated for a special case by Fig. 1–15. The pres-
ence of the ground also affects the radiation resistance and the input impedance because
of mutual coupling between the antenna and its image. Resistance of the wire itself is
usually small, but in a very long wire the total resistance may be appreciable. In addition
to its direct effect on the input impedance, this resistance also changes the form of the
current distribution on the wire, in both the resonant and nonresonant antennas, and so
affects the radiation pattern. Therefore the behavior calculated for the free-space perfect-
conductor antenna serves primarily as a general guide to what will be observed with
practical antennas. When correction for ground effects is made, the theoretical patterns
agree quite well with those that actually occur.
4.4.6. Uses of Long-Wire Antennas

Both resonant and nonresonant long-wire antennas are used for transmitting and receiv-
ing in the MF and HF range, from perhaps 500 kHz to 30 MHz. They provide a simple
and effective method of obtaining a directional pattern and gain. As will be described in
Ch. 5, these properties can be further enhanced when long-wire antennas are used as
elements in an array.
The single terminated wire used as a nonresonant antenna will not be effective for
horizontal polarization, as already discussed, because of its small-fraction-of-a-wave-
length height. As the discussion of polarization indicates, however, a lobe that makes a
small angle with the axis of a horizontal wire in a vertical plane will radiate or receive
waves whose electric field vector has an appreciable vertical component. Such an antenna
is sometimes used as a rather highly unidirectional receiving antenna for vertically polar-
ized waves. In this use, the long-wire nonresonant antenna is known as a Beverage or
“wave” antenna (Beverage, Rice, and Kellog 1923).
The functioning of a Beverage antenna is now described. For a wave with propagation
direction slanted between horizontal and vertical, there will be both a vertically- and a
horizontally-polarized (V- and H-POL) component. Since the earth is an imperfect con-
ductor, the H-POL component is not totally diminished. Thus, the incoming H-polarized
fields, from the ground, induce voltages along the antenna that add in phase at the receiv-
ing end. Waves propagating along the wire from the opposite direction are, ideally,
absorbed by the terminating resistor. Therefore, the Beverage antenna provides a highly
directive pattern in the horizontal plane for vertical polarization. Beverage antennas
are not ordinarily used for transmitting, because the power absorbed in the terminating
resistor results in poor radiation efficiency.
4.5. Loop Antennas

Another basic form of radiator is the loop, which in its fundamental form is a single-turn
coil of wire. A current can be made to flow in the loop by breaking it at some point and
connecting the terminals of a transmission line (or other source) at the gap in the loop,
as indicated in Fig. 4–18.
4.5.1. The Small Loop

The radius of the small loop, a, is assumed to be very small compared to the wavelength
l, so that the current in all parts of the loop will be of the same amplitude and phase at
any instant. An analysis of the radiation from such a loop may be made by considering
it to be made up of many elemental dipoles connected together. Since dipoles are straight
rather than curved, the figure thus formed will be a polygon rather than a circle. However,
the approximation to a circle can be made as good as desired by taking the elemental
dipoles to be sufficiently short or, ideally, infinitesimal. The fields of the individual
dipoles are then superposed in the manner described for analyzing the half-wave dipole,
sec. 4.3. Here not all the electric vectors of the separate field components are parallel.
Although this complicates the mathematics, the principle is the same. The superposition
of nonparallel fields was discussed in sec. 1.2.
From such an analysis it is found that the field pattern of a loop has exactly the same
shape as that of a single elemental dipole oriented with its axis coincident with the loop
axis (i.e., with its axis perpendicular to the plane of the loop) (Fig. 4–4). However, the
vector directions of the electric (E ) and magnetic (H ) components of this field are inter-
changed, relative to the E and H direc-
y tions of the elemental-dipole field. The
2a polarization is linear but perpendicular to
that of the corresponding electric dipole.
I Therefore a loop with its axis horizontal
x (f = 0°) radiates maximum field intensity in the
I Loop, radius a plane of the loop, the pattern being dough-

nut-shaped; but with axis horizontal the
(Loop is in a
z Feed line xy-plane) polarization is vertical, rather than hori-
zontal. If the plane of the loop is horizon-
(q = 0°)
tal, the radiation pattern in the horizontal
FIGURE 4–18. plane is uniform (a circle), like that of a
Basic form of loop antenna. vertical dipole, but the polarization is
Loop Antennas 145
horizontal. The small loop is an oscillating magnetic dipole, which is equivalent to a bar
magnet whose magnetic strength and polarity oscillate. Because the pattern has the same
shape as the elemental electric dipole, the directivity is the same (D = 1.5) and so is the
half-power beamwidth (BW3dB = 90°).
Now consider Fig. 4–18 and assume an observer is on the z-axis, which extends
outward from the page, and where angle q = 0°. Recall that the current I is assumed
constant and of the same phase everywhere on the loop. Then, the radiation reaching the
observer from any two diametrically opposite located elemental dipoles are equal in
magnitude and opposite in phase, and thus cancel. In other words, at q = 0° the pattern
has zero amplitude, as indicated in Fig. 4–4 for the dipole. Now notice that as q is
increased by being removed from the z-axis, the path length difference (and thus the
phase difference) increases between the radiation from diametrically opposite elemental
dipoles. Therefore, the pattern’s amplitude increases with increases in q. Furthermore,
because of the symmetry of the loop about the z-axis, it is clear that the amplitude is
constant and independent of q. Therefore, analogous to the pattern of a short dipole, the
E-field pattern of a loop antenna is donut shaped.
The polarization can also be determined from considering Fig. 4–18. Let an observer
be anywhere on the x-y plane and far removed from the loop. Then, recalling that the
polarization from each elemental dipole within the loop is aligned with the direction of
its current, one can discern that the vectorial sum of the radiation at each angle q is per-
pendicular to the z-axis. But, because of loop symmetry, the electric field lines are neces-
sarily concentric about the z-axis. Thus, if the z-axis is vertical, the polarization is
horizontal.
The relationship between the loop current and the radiated power is quite different for
a loop, since the radiation resistance of the elemental dipole depends on the ratio of its
length to the wavelength, and the geometry of a loop is not comparable. The formula for
the radiation resistance of a small loop is
31, 200 A2
Rr ≅ ohms (4–18a)
λ4
where A is the area of the loop and l is the wavelength. Since A can be expressed in
terms of circumference C as C2/4p, (4–18a) can also be expressed as
4
Rr ≅ 197  
C
(4–18b)
λ
In these formulas, A, C, and l must be expressed in the same units of length.

The elemental-dipole pattern (with E and H directions interchanged) and these radia-
tion-resistance formulas apply only to loops that are small compared to a wavelength.
This criterion is considered to be met if the loop diameter is less than 0.1 wavelength.
For this maximum size of loop, the radiation resistance is about 2 ohms. A very large
current, therefore, is needed for radiation of appreciable power, as was also found to be
true of short dipoles (sec. 4.1).
It turns out that the results found for a circular loop apply equally well for a loop of
any shape as long as its dimensions are sufficiently small compared to a wavelength; the
radiation resistance depends only upon its area, following (4–18a). The loop may be
square, triangular, or even irregular in shape.
At very low frequencies it is common to make a loop with more than one turn of wire.
If the number of turns is N, the resulting radiation resistance is found by multiplying the
one-turn value of (4–18a) or (4–18b) by N2. It is necessary that the total length of wire
be small compared to the wavelength if the small-loop behavior is to apply, but if the
frequency is very low (wavelength very long) this requirement is not difficult to meet.
Because of the large wavelength, however, the radiation resistance may still be a very
small value in spite of the number of turns, and the loop may have a significant ohmic
resistance also. Therefore the radiation efficiency will be poor.
When such loops are used for receiving, the terminals of the loop may be connected
to a very high-impedance receiver input circuit, so that the quantity of primary interest
is the voltage induced in the loop rather than the power delivered. When the loop has its
plane in the direction of a properly polarized incoming signal, if the incident-wave field
intensity is E volts per meter, the induced voltage will be
2π NAE
V= volts (4–19)
λ
where N is the number of turns of the loop and A is its area.

Small loops are often used for receiving as direction finders when the received signals
are vertically polarized. The direction of the received signal is determined by orienting
the loop with its axis toward the signal direction. The “null” (minimum value) of the
pattern is in this direction and is very sharp (corresponding to the pattern in the direction
of the axis of a dipole, Fig. 4–2). Thus, when the orientation of the loop is adjusted for
minimum signal, the direction of the signal is accurately indicated. There is a twofold
ambiguity in the direction, however, because a null exists in the pattern on both sides of
the loop. This ambiguity may be resolved in various ways. A common way is to combine
(in the receiver input circuit) the loop output with the output of a small vertical dipole,
with a 90 degree phase difference (produced by the circuit arrangement). If the loop and
dipole signal amplitudes are equal, the resulting combined pattern has only one null and
is therefore unambiguous. However, loop direction finding is successful only when the
loop can be located in an environment free from nearby large reflecting objects that may
result in signals arriving at the loop position from more than one direction. This destroys
the null effect, or at the least it destroys the sharpness of the null.
4.5.2. Other Loop Antennas

This section describes some other types of loops. Smith (2007) discusses these and other
configurations, and includes an extensive annotated list of references.
Loops are fed by coaxial cables or two-wire lines. An interesting coax-fed loop con-
figuration has some features common with the short vertical monopoles on automobiles.
Loop Antennas 147
In both cases, the outer conductor of the coaxial cable is attached to the base plate. Also
in both cases, the center conductor passes through a hole in the conducting base plate,
from which an image is formed. In the monopole arrangement, an extension of the center
conductor serves as the half-dipole above the base plate. However, for the loop, the
extension is bent into a half circle, whereby its image in the base plate forms the other
half of the loop.
For light-weight receiver applications, a low-loss magnetic ceramic (ferrite) is com-
monly used as the core within a multiturn loop to improve efficiency. A reflecting back-
plane (parallel with the plane of the loop) can be used for providing a unidirectional
pattern and increased directivity. Resonant loops (discussed below) are also used as the
elements of phased arrays to increase directivity: an example being coaxially positioned
loops to function as the reflector, driven element, and directors of a Yagi-Uda array
(Fig. 5–11).
Small-loop analysis is applicable to loop wire-lengths of roughly 0.1l or less, so
that the current distribution around the loop is approximately uniform. Larger loops are
also used, especially at the higher frequencies. Ordinarily the patterns of larger loops
have multiple lobes, and the current distributions on these loops affect the patterns
considerably.
The Alford loop, shown in Fig. 4–19, is an example of a larger loop. It is more efficient
than, and has a pattern similar to, that of a small loop. It consists of a square one-turn
loop with quarter-wavelength sides, and it is fed with opposite phases at opposite corners.
The other two corners are capacitively connected. The out-of-phase feed is achieved by
transposing one branch of the feed line as shown. The capacitors are commonly open-end
sections of transmission line; with the reactance given in chapter 2 by (2–24). The radia-
tion resistance of the Alford loop is about 80 ohms. Its radiation pattern is similar (though
not identical) to that of a small loop, but it is much more efficient.
As the circumference C of a loop increases and approaches a wavelength, the peak of
the beam moves toward the loop axis. Near this resonant length, the far field pattern is
nearly the same as two parallel dipoles separated by approximately the loop diameter.
The input impedance varies significantly
with C, having large peaks in both the l/4 Capacitive
stub
resistance and reactance for circumfer-
ences near odd multiples of l /2. Near the Transposed
resonant loop length of l, the input resis- feed line
tance is about 100 ohms and the reactance l/4

component is relatively small. Thus, the
input impedance for a loop with a one l Capacitive
circumference can be readily matched to stub
a transmission line.
The pattern of the resonant C = l loop
can be made unidirectional and the direc- Feed line
tivity increased in axial direction by
placing the loop over a planar reflector. FIGURE 4–19.
With spacings between the reflector and The Alford loop antenna.
loop in the range of 0.05l to 0.2l, measurements have given directivities of approxi-
mately 10, and input impedances that can be readily matched (resistance R ≤ 135 W).
The reflector used was square reflector and had different side dimensions between 0.6l
and 1.9l.
As already noted, when the loop circumference is increased to value near one wave-
length, the maximum of the far field pattern is along the loop axis. Resonant loops of
this type are used as the elements of a Yagi-Uda array to form a unidirectional beam
along the axes of the loops and of the array. Details for choosing the dimensions of the
loops and their spacings for a Yagi-Uda array are given by Balanis (2005, p. 599).
4.6. Helical Antennas

Another basic form of radiator is the helix, which is a wire (conductor) wound in the
shape of a screw thread and used as an antenna in conjunction with a flat metal plate
called a ground plane. Helixes are mostly used at relatively high frequencies so that their
dimensions are appreciable compared to the wavelength. Theory and practice in the art
of helical antennas has been developed largely by John Kraus and his associates at the
Ohio State University (Kraus and Marhefka 2002, pp. 222–42).
As shown in Fig. 4–20, the helix is fed at one end, usually being connected to the
center conductor of a coaxial transmission line whose outer conductor is connected to
the ground plane. The basic geometry of the helix is described in terms of its diameter
D and its turn spacing S. For an N-turn helix the total length of the antenna is equal to
Feed wire a
d z (q = 0°)
Feed line y
(coaxial)
f
Ground plane
S
d–diameter of wire
D–diameter of helix
S–turn spacing
f–spacing of helix
from ground plane
a–pitch angle
FIGURE 4–20.
Helical antenna.
Helical Antennas 149
NS, and the circumference C = pD. The length of the wire per turn of the helix is
L = S 2 + C 2 = S 2 + π D 2 . The pitch angle a, an important parameter of the helix, is
the angle that a line tangent to the helix wire makes with the plane perpendicular to the
axis of the helix; and it can be found from this relation: sin a = S/L or tan a = S/pD = S/C.
The properties of helical antennas are described in terms of these geometric parameters.
Many different radiation characteristics may be obtained by varying their magnitudes in
relation to the wavelength l.
The feed wire (Fig. 4–20), which connects the terminus of the co-axial center conduc-
tor to the beginning of the actual helix, lies in a plane through the helix axis and is
inclined with respect to the ground plane at approximately the pitch angle of the helix.
Variation of its geometry affects the input impedance of the antenna.
When the dimensions of the helix are very small compared with the wavelength, the
maximum radiation is in the plane perpendicular to the helix axis, and the radiation
pattern is a combination of the equivalent radiation from a short dipole positioned on the
helix axis and a small loop also coaxial with the helix. This type of helix is known as a
normal mode helical antenna (NMHA), and it is widely used in wireless handsets. To
reduce cost of manufacture, NMHAs have been fabricated by printing a conducting path
into a groove in a small, cylindrical plastic post or mast (see Fig. 36–2, Vardaxoglou and
James 2007).
For a NMHA, the patterns of the two equivalent radiators are of course the same, but
the linearly polarized components are at right angles, and the phase angles at a given
point in space are 90 degrees apart. Therefore, as explained in sec. 1.1.3, the resultant
field is elliptically polarized or circularly polarized, depending on the field-strength ratio
of the two components. This ratio depends on the pitch angle a. When a is very small,
the loop type of radiation predominates; when it becomes very large, the helix becomes
essentially a short dipole. In these two limiting cases the radiated polarization is linear,
in one having loop polarization, and in the other, dipole polarization. For intermediate
values of a the polarization is elliptical, and at a particular value of a it will be circular.
Wheeler (1947) showed that this result is obtained when S = p 2D2/2l, which corresponds
to a value of a given by
πD
tan α = (4–20)
2λ
The analysis of the helix leading to these conclusions may be made by considering it to
be equivalent to a number of small loops having the same diameter as the turns of the
helix, with their planes parallel and their axes in line with the helix axis and spaced the
same as the helix turn spacing. Then these loops are considered to be connected by short
dipoles parallel to the helix axis and of length equal to the helix turn spacing. The radia-
tion field of the helix is equivalent to that obtained by superposition of the fields of these
elemental radiators.
When the diameter and spacing (D and S ) are appreciable fractions of a wavelength,
an entirely different radiation pattern is obtained. The maximum intensity is radiated in
the direction of the helix axis, in the form of a directional beam with minor lobes at
oblique angles. The radiation in the main lobe is circularly polarized. It is this feature of
the helix, in this mode of radiation (the axial mode), that probably accounts for most of
the practical applications of this type of antenna.
A typical helical antenna operating in this mode has a circumference C of approxi-
mately one wavelength and spacing S approximately a quarter wavelength. The antenna
will operate quite well over a range of frequency so that these dimensions are noncritical.
The ground plane (which may be either a solid sheet or a wire grid or mesh) should be
at least 34 of a wavelength in diameter. The pitch angle may range from about 12 to 18
degrees; approximately 14 degrees is optimum. The gain and beamwidth depend on the
helix length (equal to NS, Fig. 4–20, where N is the number of turns). The feed-point
impedance is resistive and of the order of 150 ohms at the frequency for which C = l.
At higher and lower frequencies the resistive component varies, and a reactive component
appears. Detailed design information are available (Kraus and Marhefka 2002, ch. 8).
In terms of a three-dimensional spherical coordinate system (Fig. 3–1) with the q =
0° axis coincident with the helix axis, the beam (pattern) has axial symmetry, that is, it
is the same in any plane containing the axis (does not depend on the longitude angle f).
The 3-dB beamwidth, obtained empirically, is given approximately by the formula
52 λ3
θ 3 db = degrees (4–21)
C NS
The formula assumes that the pitch angle is between 12 and 15 degrees, that N is equal
to or greater than 3, that NS (the helix length) is not greater than 10 wavelengths, and
that C is between 0.75l and 1.33l.
The directivity, subject to the same assumptions, is given by
12 NSC 2
Dmax = (4–22)
λ3
In some applications it is necessary to pay attention to whether the helix is wound

with a right-hand or a left-hand pitch (analogous to right-hand and left-hand screw
threads). This determines whether the wave will be right- or left-hand circularly polar-
ized. (A receiving antenna designed to receive right-hand circular polarization cannot
receive left-hand circular, and vice versa.)
Helical antennas have found considerable application in space telemetry applications
at the ground end of the telemetry link with ballistic missiles, satellites, and space probes,
at HF and VHF. The circular polarization is useful in this application because of the
polarization rotation of waves produced by the ionosphere (Faraday effect, sec. 1.4.6).
4.7. Horn Radiators

The radiators thus far discussed are based on the concept of fields set up by alternating
currents in wires (in the generalized sense of the term “wire,” which includes tubing,
Horn Radiators 151
pipes, and bars). The most basic of them is the elemental dipole, since in principle all
other current-carrying conductors can be regarded (mathematically and conceptually) as
an assemblage of elemental dipoles, and the radiation field is then deduced by applying
the principle of linear superposition to the fields of the individual dipoles.
Another class of radiators is based on the existence over a surface of a specific elec-
tromagnetic field configuration. The intensity, phase, and polarization of the field over
this surface, or aperture, are analogous to the current amplitude, phase, and direction in
antennas represented by an assemblage of dipoles. The description of the variation of
these field quantities over the aperture is called the aperture distribution. When this dis-
tribution is known, it is possible in principle to calculate the radiation pattern, just as it
is possible for a wire or arrangement of wires in which the current distribution is known.
As for radiation due to currents, the analysis of radiation due to the field distribution of
an aperture is based on Maxwell’s equations.
An example of an aperture over which the field distribution is known is the cross
section of a waveguide, in which a particular known mode is propagating, as described
in sec. 2.5. It is a well-known experimental fact that if such a guide is “sawed off ” in a
plane perpendicular to the axis of the guide, leaving an open end, radiation will occur
from this open mouth. The resulting field pattern can be calculated from the known
configuration of the field for the particular waveguide mode. This is the simplest case of
a waveguide horn radiator.
It should not be surprising that radiation occurs under these circumstances because
the fields inside a waveguide are propagating in essentially the same way that fields
propagate in free space; the only difference is that they are constrained from spreading
spherically by the walls of the guide. When this propagating field reaches the guide mouth
it continues to propagate in the same general direction except that, in accordance with
Huygen’s principle, it also spreads laterally, and the wavefront eventually becomes
spherical, although there is a “near field” region in the vicinity of the mouth of the guide
in which the wavefront is more complicated. It can be thought of as a transition region
in which the changeover from guided propagation to free-space propagation takes place.
This changeover involves a change of phase velocity and a change in the characteristic
wave impedance, from those of the guide to the free-space values, c = 3 × 108 meters
per second and Zc = 377 ohms.
Because the waveguide impedance is ordinarily different from this free-space value,
the radiating open end does not usually present a matched-impedance load to the guide,
resulting in an undesirable standing wave. This can be eliminated by some form of trans-
former matching device, such as those described in sec. 2.5. A better method, however,
is to flare the walls of the guide. If this is done properly, it results not only in a matched
impedance but also in a more concentrated radiation pattern, that is, narrower beamwidth
and higher directivity. This flared structure is what is ordinarily meant by the term horn
radiator.
Various possible flaring arrangements, resulting in different types of horns, are shown
in Fig. 4–21. As shown, a rectangular guide may be flared on the narrow walls, the wide
walls, or both. A sectoral horn is flared in only one dimension. If the flare is in the direc-
tion of the electric vector, as when the broad walls are flared with the TE10 mode in
fH dE
dE
f
dH
fE
Sectoral H-plane horn Pyramidal horn dH
L
L
dE
f d
f
Conical horn
dH
Sectoral E-plane horn

FIGURE 4–21.
Waveguide horn types.
rectangular guide, the result is an E-plane sectoral horn; when the narrow walls are flared,
the radiator is an H-plane sectoral horn. Flaring both walls results in a pyramidal horn.
A conical horn is formed by uniform flaring of the walls of a circular waveguide.
If the flare angle f is too great, the wavefront at the mouth of the horn will be curved
rather than plane. This means that the phase distribution over the aperture will be non-
uniform, resulting in decreased directivity and increased beamwidth. On the other hand,
too small a flare angle results in a small aperture area for a given length L of the horn.
The directivity is proportional to the aperture size for a given aperture distribution. Thus,
there is an optimum flare angle that provides the maximum gain for a given horn length,
and therefore the optimum flare angle is intermediate between slight and abrupt. In other
words, an optimum horn or optimum-gain horn is one that compromises in aperture phase
error to maximize its gain for a given length L, and therefore its flare angle is intermedi-
ate between slight and abrupt.
Graphs of radiation patterns are available for a variety of horn types and dimensions
(Balanis, ch. 13, 2005; Love 1993). These graphs, known as universal patterns, were
generated by aperture theory and have been validated, generally, by measurements. The
“universal” patterns provide details of the major lobe shapes for large as well as small
horns, with aperture widths as small as 1.5 or 2l. Even so, designs for center-fed reflec-
tors and lenses often require feed-horn patterns with dimensions even smaller than 1.5l.
(Section 6.3.3 provides guidance on design for horn aperture widths less than 2l).
Horn Radiators 153
It is apparent that the flare angle must be made smaller as the length is increased to
maintain a given maximum phase variation across the aperture. Therefore there is a
practical limit to the gain that can be obtained with a horn radiator; very high gain
requires an excessive horn length. For moderate gains, however, horn radiators are very
useful. They are of course especially appropriate when the feed line is a waveguide. Their
bandwidth is then essentially the bandwidth of the guide—2 : 1 typically, for the TE10
mode in rectangular guide and a sectoral or pyramidal horn.
The published literature on optimum-gain horns includes somewhat different results
for beamwidths versus horn dimensions, depending on how the horn dimensions are
specified and the details of analysis. Kraus and Marhefka (2002, p. 339) provide simple
equations for the beamwidths of optimum horns, that were obtained from analyses of
measured beamwidths versus horn flare angle, for various horn lengths. Those equations
for the 3 dB beamwidths of “optimally” tapered horns follow:
56
θE = degrees (4–23a)
dE
and
67
θH = degrees (4–23b)
dH
where E and H refer to the horn’s E and H plane patterns. The symbols dE and dH are
the aperture dimensions (widths), expressed in wavelengths along the E and H planes.
Use of (4–23a) and (4–23b) does not require detailed horn dimensions, yet they
provide good “first” beamwidth estimates for an “optimumly” tapered horn. As already
mentioned, the universal patterns are considered accurate when detailed horn dimensions
are available. It is to be noted that optimum gain horns are often called standard gain
horns, because of their wide laboratory usage. When detailed horn dimensions are avail-
able and more accurate calculated beamwidths are desired, the reader is referred to Bird
and Love (2007, pp. 14–17). These authors also include an equation for the gain of an
optimum-gain antenna, which follows
6.5dE d H 6.5 A
G= = 2 (4–24)
λ2 λ
where A = dEdH is the area of the horn-mouth opening (aperture). From (3–39) of chapter
3, a recommended formula for estimating gain is
26, 000
G= (4–25)
θ Eθ H
By using (4–23a) and (4–23b) with (4–24), one finds a difference of only 0.3 dB in gain
from that provided by (4–25).
Incidentally an unflared open-mouthed wave-

guide provides a slightly greater directivity than
s (4–24) indicates, because of its more nearly con-
stant phase distribution. For such a radiator the
f h
factor 6.5 in (4–24) becomes approximately 10.
s
For a typical rectangular guide A is approximately
equal to l2/4; hence G is about 2.5 Thus this simple
radiator has a somewhat greater gain than 1.64, the
gain of a half-wave dipole.
Coaxial
line feed Many special forms of horns are used. There are
too many to describe in detail, but their basic prin-
ciple is the same as those of the commoner types
FIGURE 4–22.
that have been discussed. One special horn that
Biconical horn. deserves a brief discussion, however, is the biconi-
cal horn, pictured in Fig. 4–22 . It consists of two
conical metal surfaces with their axes collinear (in line with each other) and their vertexes
opposed. As with sectoral horns, there is an optimum flare angle that depends on the
mode of excitation employed (Terman 1943). With the coaxial-line feed shown, the
polarization will be vertical, and the optimum-flare criterion is h = 2 sλ . Since sin(f/2)
= h/2s, the optimum flare angle f is the one that satisfies the relation
φ
sin   = λ 2 s (4–26)
 2
Horizontal polarization may be excited in this horn by means of a small loop antenna
with its plane perpendicular to the cone axes and lying between the vertexes, and the
loop axis collinear with the cone axes. The optimum flare angle is then given by
φ
sin   = 3λ 4 s (4–27)
 2
The radiation pattern is omnidirectional in the horizontal plane with the cone axes
vertical, the vertical beamwidth and directivity depending on the dimension h. For
optimum flare angles the directivity is given by
D = m  
2h
(4–28)
 λ
where for vertical polarization (assuming vertical cone axes) the factor m ⯝ 0.8, and
for horizontal polarization m ⯝ 0.6. The vertical beamwidths may be estimated using
(4–23a) for vertical polarization and (4–23b) for horizontal polarization, with dE and dH
replaced by h.
Slot Radiators 155
The omnidirectional horizontal pattern y

is the attractive feature of the biconical
horn, in the VHF and UHF bands, espe-
cially with horizontal polarization. Few x (f = 0°)
antennas provide a truly omnidirectional
horizontal-plane pattern and horizontal l/2
polarization.
z (q = 0°)
Slot
4.8. Slot Radiators
Infinite thin
If a narrow slot-like opening is cut in a Transmission conducting sheet
large flat sheet of metal, and properly line feed in yz plane
connected to a source of rf power, it will
radiate in a manner that bears a certain
resemblance to the radiation by a dipole
FIGURE 4–23.
of the same dimensions as the slot. In fact,
if the conducting sheet is a plane of infi- Radiating slot in an infinite plane conducting
nite extent, the radiation pattern will have sheet.
exactly the same shape as that of the cor-
responding dipole except that the electric and magnetic vectors are interchanged. (The
same relationship exists, incidentally, between the patterns of a short dipole and a small
loop.) Also, the impedance properties of the slot are somewhat different. These comple-
mentary properties of a slot in an infinite plane conductor, and a thin flat dipole of exactly
the same dimensions as the slot, are predicted by an important result of electromagnetic
theory known as Babinet’s principle. It also relates the impedance properties of the two
kinds of radiators.
Figure 4–23 shows such a slot and a two-wire transmission-line feed connected to it.
For the half-wavelength-long slot shown, the radiation pattern will have the same angle
dependence as that of a half-wave dipole, as given by (4–11) and plotted in Fig. 4–14.
Note that only the term in square brackets of (4–11) applies; i.e., the relative patterns are
the same. Also, since the E and H vectors are interchanged, the polarization is opposite
to that of the corresponding (complementary) dipole. It is this fact that allows the three-
dimensional “doughnut” pattern to exist unaffected by the presence of the conducting
plane, and on both sides of it, except for the infinitesimally thin slice eliminated by the
metal sheet. That is, the required electric field intensity can exist in the space at the
surface of the conducting plane because the electric field lines are perpendicular to
the plane. (It is a well-known principle of electromagnetic theory that they could not so
exist if they were parallel to the surface.)
Thus beamwidth and directivity are the same as for the half-wave dipole; q3dB = 78°
and D = 1.64. The input impedance Zs for the connection shown in Fig. 4–23 is related
to the input impedance of a center-fed half-wave dipole, Zd, by the formula
(120π )2 35, 530

Zs = = (4–29)
4Zd Zd
Since Zd = 73.1 + j42.5 ohms for a thin half-wave dipole, a thin (narrow) half-wave slot
is found, from the above formula, to have impedance Zs = 363 − j211 ohms. A thin slot
of length 0.475l, the complement of a dipole having nonreactive input impedance of 67
ohms, will have Zs = 530 ohms, nonreactive.
There is no such thing in the real world as an infinite plane conducting sheet, but if a
slot is cut into a sheet that is very large compared to the slot, then the behavior predicted
above will be realized to a high degree of approximation. Slots cut into sheets of even
moderate size will radiate effectively, but their exact behavior is not as readily
predictable.
The slot as described will radiate on both sides of the sheet. If radiation on one side
only is desired, the “back” side of the slot may be enclosed by a box, or cavity. The
field distribution along the slot is then affected by the dimensions of the cavity. The
problem of theoretical design is quite complicated, and design is often determined by
experiment.
A unidirectionally radiating slot may also be obtained by cutting it in a proper position
and orientation in the wall of a waveguide. Figure 4–24 shows the appearance of several
slots in a TE10-mode rectangular waveguide that will radiate, and two that will not radiate.
A waveguide slot, if it is to radiate, must be positioned so that it interrupts currents that
would otherwise flow across its length in the inner walls of the guide. This is equivalent
to saying that there must be a component of magnetic field (H ) parallel to the slot at the
C C
Radiating Radiating
slot slot
C
C
C
Nonradiating
Radiating slots
slot
FIGURE 4–24.
Radiating and nonradiating waveguide slots in rectangular guide, TE10 mode.
Patch or Microstrip Antennas 157
inner surface of the guide. The field configurations in waveguide for various modes are
given in advanced engineering textbooks and handbooks. The examples of radiating slots
shown in Fig. 4–24 by no means exhaust the possibilities. To be most effective the
waveguide slot should be resonant, which it will be if the length is approximately half a
wavelength. However, the exact length for resonance depends on the position of the slot
in the guide.
A waveguide slot does not radiate the total power flowing in the guide; it “extracts”
or “couples out” some fraction of it. The remaining power continues on down the guide
to whatever additional load there may be further on. Or if it encounters a “short circuit”
(end cap), it is reflected. By impedance matching devices it may then be possible to
couple all the power to the slot. Waveguide slots, however, are seldom used singly;
usually an “array” of them is cut along the length of the guide, as will be described in
sec. 5.9.
Slot radiators need not be complements of half-wave dipoles. Many other radiating
shapes are possible, but all may be analyzed in terms of their complementary radiators.
For example, an annular-ring slot in an infinite plane sheet will have a radiation pattern
exactly like that of a loop of the same size, with E-fields and H-fields interchanged. Slot
radiators are practically restricted to the higher frequencies by the requirement of a con-
ducting sheet considerably larger than the slot. They are very useful when a radiator must
be devised that will not project from a surface, for example, in an airplane wing or fuse-
lage. (Here the opening may be covered by a protective cover of low-loss dielectric
material.)
4.9. Patch or Microstrip Antennas

The geometry of a microstrip is well adapted to construction by printed-circuit tech-
niques, in which the strip conductor and adjoining ground planes can be deposited on
low-loss substrate material by mass-production methods. An unbalanced microstrip
geometry is illustrated in chapter 2, Fig. 2–9 (lower right). The balanced form has the
strip enclosed both below and above by a metallic ground plane. This transmission-line
technology can be combined with ordinary printed-circuit techniques to permit construc-
tion of monolithic phased-array modules containing not only the transmission line
components, but also phase shifters, couplers, control circuitry, radiating elements, and
solid-state transmitting and receiving amplifiers. This fabrication technique greatly
reduces the cost as well as the bulk of large phased arrays. Moreover, the “sheet” con-
struction need not be completely planar; it can be made to conform to the surface of a
vehicle such as the fuselage of an airplane. An array thus constructed in nonplanar form
is known as a conformal array.
A patch or microstrip antenna is commonly a metal rectangular or circular metal
surface (the patch) on a dielectric ground plane, such as a printed circuit board. The patch
or microstrip antenna is useful as a single element antenna and as the element-type within
multielement arrays. Their features include lightweight and ruggedness with low profile,
and bandwidths typically less than a few percent. Patch antennas can be made that radiate
either a linear or a circular polarization, and the use of two operating frequencies is
possible. Because of their ruggedness and

RADIATING lightweight, patches are especially useful
EDGES for airborne or space applications (Howell
a TOP
VIEW 1975; Munson 1974). Practical patch
FEED antennas have been made for a wide range
POINT
of wavelengths, covering at least from
PATCH
UHF into the millimeter wavelengths.
b The top and side views of a rectangular
SLOT 1 PATCH SLOT 2 SUBSTRATE patch or microstrip antenna are illustrated
SIDE in Fig. 4–25. The patch is energized by a
t
VIEW microstrip transmission line at the patch’s
GROUND PLANE left side, where the dipole launches radia-
FIGURE 4–25. tion. Instead, the patch can be energized
Microstrip antenna, a thin conducting patch by a coaxial line in a configuration called
on a dielectric substrate above a conducting a pin fed patch. Then, the outer coaxial
ground plane (from Carver and Mink 1981; conductor is connected to the ground
© 1981, IEEE).
plane of the printed circuit board and the
center conductor is routed through a hole
in the printed circuit board and connected to the patch. Sometimes proximity coupling
is used, where the connection is electromagnetic instead of direct.
Figure 4–25 depicts a patch antenna, where a wave propagates in a dielectric substrate
within parallel plates between the patch and the ground plane. The parallel plate medium
is a resonant structure. With a relative dielectric constant er, the between-plates wave-
length is λ / ε r . The length, electrically, from the left to the right patch sides is approxi-
mately one-half l, and thus the physical length is ≈ λ /2 ε r . This produces a phase
difference between the left and right edges (designated as slots 1 and 2 in Fig. 4–25) of
approximately 180°, making the vertical components of the edge E-fields of opposite
polarity. However, the edge horizontal components are in time phase. The terminology
underscores the fact that the fringing of the electric fields at the patch edges is similar
to the electric fields that emanate from narrow slits. Therefore, the pattern has maximum
amplitude in the direction normal to the patch (i.e., upward) and it is horizontally polar-
ized (pointed to the right).
Patch antenna design and development became an active field in the early 1970s. By
January 1981, a special issue of the IEEE Transactions on Antennas and Propagation
was published that contained a number of especially significant papers (see, e.g., Carver
and Mink 1981; Mailloux, McIlvenna, and Kernweis 1981). At that time, the perceived
needs included, for example, increases in impedance bandwidth beyond the then avail-
able few percent, improved techniques to allow multifrequency operation, and small
microstrip phase shifters for use in electronic scanning patch arrays.
Robust developments continued in the patch antenna field, resulting in reliable fabrica-
tion techniques, dramatically increased bandwidths, and reduced sizes. Novel designs
now include patches that are stacked, stacked patches that are probe fed, and multiple
shorting posts on stacked patches. For example, the literature describes stacked-patch
antennas having bandwidths up to octave widths (Targonski, Waterhouse, and Pozar
Basic Feed Methods 159
1998; Waterhouse 1999) and electrical sizes small enough for hand-held mobile com-
munications at frequencies less than 2 GHz (Waterhouse, Targonski, and Kokotoff
1998).
4.10. Surface-Wave and Leaky-Wave Antennas

In the years during and since World War II some rather exotic forms of antennas have
been devised, primarily in the category of VHF, UHF, and microwave antennas. Two
such types of antennas that may be in the class of basic radiators are the surface-wave
and leaky-wave antennas. Their theory and applications are too complex and too exten-
sive for a full discussion here (see, e.g., Kraus and Marhefka 2002, pp. 734–41).
These antenna types are analyzed in terms of guided electromagnetic waves propa-
gated along a surface or other guiding structure that does not fully confine them.
Radiation may take place at discontinuities of the structure, or gradually along an aper-
ture. Prominent examples of such antennas are the polyrod antenna, the cigar antenna,
the zigzag, the holey plate, and the mushroom antenna. Some of these antennas find use
in applications requiring a high-gain radiator with a low silhouette, as on a streamlined
aircraft. They are primarily microwave or near-microwave devices, but they are by no
means limited to the microwave region. In fact, the Yagi-Uda antenna, usually discussed
as an array antenna (sec. 5.3), can also be regarded as a surface-wave antenna. This
antenna is often used at frequencies as low as the 20 meter (14 MHz) amateur radio
band.
4.11. Basic Feed Methods

The usual arrangement for feeding power to an antenna from a transmitter, or to a receiver
from the antenna, is indicated in block-diagram form in Fig. 4–26. The transmission line
can be any of the types discussed in secs. 2–4 and 2–5. The function of transformer A is
to match the antenna feed-point impedance to the characteristic impedance of the line
and also to make a transition, if necessary, from one form of line to another (e.g., from
balanced to unbalanced, or from coaxial line to waveguide). Transformer B serves the
same functions between the transmission line and the internal impedance of the transmit-
ter or receiver. The two transformers insure not only that the transmitter will be correctly
“loaded” by the antenna and that the antenna will be correctly loaded by the receiver,
but also that there will be no standing waves on the transmission line. Elimination of
standing waves is desirable, but not essential. If standing waves are to be permitted, one
transformer can be omitted, and the other transformer adjusted for correct loading of the
Transformer Transmission line Transformer Transmitter

Antenna A (no standing waves) B or receiver
FIGURE 4–26.
Block diagram of basic antenna feeding arrangement.
transmitter (or, in reception, optimum transfer of received signal power from the antenna
to the receiver).
Sometimes the transmission line can be eliminated altogether by making a direct con-
nection from the antenna to the transmitter output terminals or the receiver input termi-
nals. Here also only one transformer is required. This situation can exist at very low
frequencies, where a “line” of even a few hundred feet in length may be in effect a direct
connection. Any conductor of length less than about a hundredth of a wavelength may
be so regarded, since no appreciable standing wave pattern can exist on a conductor so
electrically short. The direct connection can also exist at higher frequencies when the
antenna is “built into” a receiver or transceiver.
The transformers at low frequencies are inductor-and-capacitor devices. In addition to
impedance step-up or step-down, they also can provide reactance cancellation; both
effects may occur in the same circuit elements or they may be separated. At the higher
frequencies the transformers may be composed of transmission-line or waveguide ele-
ments, as described in secs. 2–3 and 2–5.
At frequencies up to about 30 MHz, two-wire balanced transmission lines can be used,
since at these frequencies radiation due to the line spacing being a significant fraction of
a wavelength is not a serious problem. Line impedances range from slightly less than
100 ohms to perhaps 800 ohms. The lower impedances are achieved with close-spaced
wires embedded in low-loss polyethylene plastic. The higher impedances result with
air-insulated wires or tubing. The wire lines have spacing bars of porcelain or other
insulating material at intervals of a few feet or more, depending on the spacing and the
wire stiffness.
When the higher-impedance lines are used to feed an ordinary half-wave dipole at its
center, or a long-wire antenna at a current maximum, a transformer must be used if the
impedances are to be matched. One possible type of transformer is shown in Fig. 4–27.
Antenna (half-wave dipole)
Transmission line Insulators

Guy wire feed Insulators
Matching stub
Shorting bar
Transmitter-receiver
house
FIGURE 4–27.
An antenna installation for the HF range (3–30 MHz), showing method of center feeding a
half-wave dipole with two-wire balanced line and matching stub. Line section above stub has
standing wave, but with proper adjustment there is no standing wave on the feed line to left of
the stub.
It is the shorted-stub arrangement whose l /2

principle is explained in chapter 2. (See s
Fig. 2–6.) An alternative method, which Half-ware

a dipole
can be used when the antenna feed-point Fanned-out
impedance is purely resistive, is the (delta)
section of line
quarter-wave transmission-line trans-
former, whose transformation ratio is Two-wire balanced
given by equation (2–22). (These methods transmission line
are of course not limited to dipole and
long-wire antennas; they are of general
applicability.)
Another method, which can be used for
feeding a wire antenna at a current FIGURE 4–28.
maximum, is the “delta match,” illustrated Delta-match feed method applied to a half-
in Fig. 4–28. In this method there is no wave dipole.
gap in the antenna at the feed point; the
line spacing is gradually increased until it
spans a section of the antenna of length denoted as s in Fig. 4–28. The effective imped-
ance presented to the transmission line increases as s is increased. This spacing is adjusted
until an impedance match results, as indicated by absence of standing waves on the line.
Methods of measuring the voltage standing-wave ratio (VSWR), and of thereby deter-
mining the antenna input impedance, are described in chapter 9.
Another method of obtaining an impedance match to a dipole with a two-wire line of
moderate impedance is to use a folded dipole, as described in sec. 7.1. In its simplest
form the folded dipole has an input impedance that approximately matches a 300-ohm
line.
Although it is not absolutely essential to operate the transmission line without standing
waves, it may be important to do so in high-power applications to minimize line losses
due to the high currents at the current maxima and to avoid excessive voltages at the
voltage maxima. Radiation losses are also greater when there are standing waves, and
the loading adjustment is more critical. A line operating with an appreciable standing
wave is called a resonant line; one with little or no standing wave is called
nonresonant.
A half-wave dipole or a long-wire antenna can be fed at one end (maximum-voltage
point) with a two-wire resonant line. The antenna is connected to one side of the line,
and the other side is simply left open, as indicated in Fig. 4–29. A line used in this way
is called a “Zepp” feeder. An advantage of this arrangement is that the antenna can be
operated at any integral multiple of the frequency for which the antenna length is a half
wavelength. The antenna pattern will be different for each such frequency, in accordance
with (4–12a) and (4–12b). The value of n in these equations is of course 1 for half-
wavelength operation, 2 for full-wave operation, and so on. The frequency for which
n = 1 is called the fundamental frequency of the antenna, and those for successively
higher values of n are called harmonic frequencies. (For example, the frequency for
n = 3 is called the third-harmonic frequency.)
Half-wave or
resonant-long-wire
antenna
Insulators Insulator
Zepp feeder
FIGURE 4–29.
Method of feeding a half-wave or resonant-long-wire antenna at one end by means of a
two-wire line with standing waves (Zepp feeder).
The pure Zepp feeder arrangement, shown in Fig. 4–29, omits transformer A in Fig.
4–26. The line is operated with a standing wave. When the antenna is operated at more
than one of the possible frequencies (values of n) transformer B must in general be
adjusted differently for the different frequencies. In the HF band, where such antennas
are most commonly used, this transformer typically consists of a variable inductance-
capacitance circuit (ARRL Antenna Book 2007). This reference and its earlier editions
contain much practical information on antennas and feed systems of the type described
in this chapter and on some of those described in chapter 5.
A matching circuit (i.e., transformer A) can also be incorporated at the antenna end
of this type of feeder, to eliminate standing waves on the line. The stub arrangement of
Fig. 4–27 (and Fig. 2–6, chapter 2) is used for this purpose. When a half-wave vertical
dipole is fed in this way, the resulting arrangement is called a “J” antenna. This matching
will in general be effective at only one frequency.
As the Zepp feeder indicates, “balanced” two-wire lines may be used to feed an unbal-
anced antenna, though at the sacrifice of perfect balance of the line currents. (Therefore
such lines will radiate somewhat more than would a perfectly balanced line.) In general,
however, balanced two-wire lines are preferred when the antenna is fed at a point of
symmetry—where the structure is electrically balanced with respect to the ground on
both sides of a center feed point. But when one side of the feed point is the ground, or
a metallic ground plane, the favored transmission line for feeding the antenna is coaxial
line. Examples of such antennas are monopoles, antennas whose axes are perpendicular
to a ground plane in which they are imaged. All the feed methods described for two-wire
lines are applicable in these cases, except that the center conductor of the coaxial line
connects in the manner indicated for one of the conductors of a two-wire line, and the
coaxial outer conductor connects to the base of the antenna (ground).
Transmission-line feed is appropriate to such radiators as dipoles, long wires, loops,
and helixes, whose radiation is based on currents flowing in wires. Waveguide feed is
more appropriate for horns and waveguide-slot antennas. However, these “rules” have
exceptions; it may at times be convenient Probe

l /4 coupler
to use a coaxial line to feed a horn—for
example, when the frequency is low Closed
enough so that a waveguide would be end Horn
very large and expensive, but yet high Coaxial line
enough to make a horn radiator feasible.
(This might be the case at a frequency in
FIGURE 4–30.
the region of 500 MHz.) Then a line-to-
waveguide transition (described in sec. Arrangement for feeding a horn radiator with a
coaxial transmission line.
2.5) is used, as indicated in Fig. 4–30.
Coaxial lines are also often used when
the ultimate load is balanced because of
their nonradiating properties and the pro-
tection that the outer conductor affords
against weather and physical damage, and Balanced Unbalanced
(a)
because they have lower losses and higher
Dipole or other
voltage breakdown rating for the same balanced load
conductor spacing. A wide variety of
methods for accomplishing the connec- Larger-diameter
outer-conductor
tion of an unbalanced (coaxial) line to a section l/4
balanced (two parallel-conductor) load (Connected
are possible. The device used to make to coaxial-line
outer-conductor
an unbalanced-to-balance connection is at bottom; no Coaxial-line
called a balun, an abbreviation of the connection at top)
words balance and unbalance. Examples
of baluns are shown in Fig. 4–31.
At low frequencies inductor-and- (b)
capacitor arrangements are used, as sug- l/2

gested by Fig. 4–31(a). At frequencies
above about 100 MHz, transmission-line
transformers are customary. Typically, as Balanced line
now briefly addressed, the basic configu-
rations include either a l/4- or and a l /2- Coaxial-line
length transmission line.
Figure 4–31(b) shows “Bazooka” balun (c)
that improves the coupling between a FIGURE 4–31.

coaxial (coax) line and a dipole (or the Examples of balanced-to-unbalanced line
balanced two-conductor load). Note that, coupling devices (baluns). (a) Coupled coils.
in feeding a dipole in Fig. 4–31(b), the (b) “Bazooka” balun. (c) Half-wave-line balun.
inner and outer coax conductors are con-
nected to the right- and left-hand arms of the dipole. Ordinarily, current within a coax at
high frequency travels along the outer surface of the inner conductor and the inner surface
of the outer conductor. However, because of radiation from the dipole elements and from
the coax-dipole junction discontinuity, and without the bazooka balun present, additional
current flows vertically along the outer surface of the outer coax conductor. This vertical
current creates vertically polarized radiation, yet the radiation of the dipole, per se, is hori-
zontal. Therefore, without the l /4-length “bazooka” sleeve, the effective antenna gain is
reduced because the antenna will simultaneously radiate horizontally and vertically polar-
ized waves. The essence of this balun is the l /4 length conducting coaxial section, formed
by the larger diameter conducting sleeve and the outer conductor of the coax line, with the
l /4 section open at its top and connected its bottom to the outer conductor of the coaxial
line. In other words, looking from the top and downwards, the impedance of the l/4 coax
(formed by the sleeve and the outer conductor of the coax line) appears as an open circuit.
Thus, the vertically moving outer coaxial-line surface current, and hence the vertically
polarized radiation, is appreciably reduced.
Figure 4–31c shows a half-wave-line balun for coupling between the bottom coaxial
(coax) line and the balanced two-conductor line. To recognize how this balun functions,
it may be helpful to recall that the voltages between points on a transmission line are
equal if separated by a distance l /2, but opposite in phase. Then, one will recognize that,
at the ends of the l/2-length coaxial line, the voltages are of equal amplitude but of
opposite phase. Therefore, the voltage across the balanced line is twice that at the input
coaxial line. This balun may provide a 4-to-1 impedance transformation. First, note that
the two coax lines formed by the l /2 length line are in parallel across the input coax.
Thus, an impedance-matched condition may exist if the l/2 length coax characteristic
impedance is 2Z0, where Z0 is the characteristic impedance of the input coax line. Note
that the double voltage across the two output conductors is consistent with the ends of
the l /2 length line being in series with one another. Thus, an impedance matched condi-
tion exists if the two-conductor output line is terminated with an impedance of 4Z0, where
Z0 is the characteristic impedance of the input coaxial line.
Figures 4–31(b) and (c) show only simple baluns that provide connections between
coaxial and two-conductor balanced lines. Munk (2002) discusses numerous other balun
configurations; including those that use multiple transmission lines for increased band-
width, as well as baluns where connections are made between coax and balanced printed
lines, and between unbalanced and balanced printed lines.
References
ARRL Antenna Book, 21st ed., American Radio Relay League, 2007.
Balanis, C. A., Antenna Theory: Analysis and Design, 3rd Edition, John Wiley & Sons,
2005.
Beverage, H. H., C. W. Rice, and E. W. Kellog, “The Wave Antenna: A New Type of
Highly Directive Antenna,” Transactions of the AIEE, vol. 42, 1923, p. 215.
Bird, T. S., and A. W. Love, “Horn Antennas,” ch. 14 in Antenna Engineering Handbook,
4th ed., J. L. Volakis (ed.), McGraw-Hill, 2007.
Brainerd, J. G., ed., Ultra-High-Frequency Techniques, Van Nostrand, New York, 1942,
p. 415. (The formula is there given as Rr = 72.5 + 30logen.).
Carver, K. R., and J. W. Mink, “Microstrip Antenna Technology,” IEEE Transactions on
Antennas and Propagation, January 1981, pp. 2–24.
Howell, J. Q., “Microstrip Antennas,” IEEE Transactions on Antennas and Propagation,

January 1975, pp. 90–93.
Kraus, J. D., and R. J. Marhefka, Antennas, 3rd ed., McGraw-Hill, 2002.
Love, A. W., “Horn Antennas,” ch. 15 in R. C. Johnson, Antenna Engineering Handbook,
McGraw-Hill, 1993, pp. 15-6 through 15-9 and 15-35.
Munk, B. A., “Baluns, etc.,” ch. 23 of Kraus and Marhefka, Antennas, 3rd ed.,
McGraw-Hill, 2002.
Mailloux, R. J., J. F. McIlvenna, and N. P. Kernweis, “Microstrip Array Technology,”
IEEE Transactions on Antennas and Propagation, January 1981, pp. 25–37.
Munson, R. E., “Conformal Microstrip Antennas and Microstrip Phased Arrays,” IEEE
Transactions on Antennas and Propagation, January 1974, pp. 74–78.
Smith, G. S., “Loop Antennas,” ch. 5 in Antenna Engineering Handbook, 4th ed., J. L.
Volakis (ed.), McGraw-Hill, 2007.
Targonski, S. D., R. B. Waterhouse, and D. M. Pozar, “ Design of Wide-band Aperture
Stacked Patch Microstrip Antennas,” IEEE Transactions on Antennas and Propagation,
September 1998, pp. 1245–51.
Terman, F. E., Radio Engineers’ Handbook, McGraw-Hill, 1943, p. 836.
Vardaxoglou, Y. C., and J. R. James, “Mobile Handset Antennas,” ch. 36 in Antenna
Engineering Handbook, 4th ed., J. L. Volakis (ed.), McGraw-Hill, 2007.
Waterhouse, R. B., “Design of Probe-Fed Stacked Patches,” IEEE Transactions on
Antennas and Propagation, December 1999, pp. 1780–1784.
Waterhouse, R. B., S. D. Targonski, and D. M. Kokotoff, “Design and Performance of
Small Printed Antennas,” IEEE Transactions on Antennas and Propagation, November
1998, pp. 1629–33.
Wheeler, H. A., “A Helical Antenna for Circular Polarization,” Proceedings of the IRE,
vol. 35, December 1947, pp. 1484–88.

1. At a distance of r = 1 meter from a particular short-dipole antenna, the reactive
component of the total electric field is equal in strength to the radiation compo-
nent. At a distance r = 10 meters in the same direction, which of these two
components is stronger? How much stronger is it? That is, what is the ratio of
their strengths?
2. A thin, perfectly conducting wire is fed as a short dipole at its center; its total
end-to-end length is 0.04 wavelength. The wire has no capacitive end loading
so that its effective length (in terms of an equivalent elemental dipole) is less
than its actual length. The rf current at the feed point (center) is 10 amp, rms.
(a) What is the radiation resistance? (b) How much total power is radiated?
3. The dipole described in Problem 2 is connected to another transmitter of the
same frequency (and wavelength), but of higher power output, so that the feed-
point current is now 25 amp. (a) At a point half way between the center and one
end of the dipole, what is the current? (b) What is it at a point at either end,
which is at a distance from the center equally 90 percent of the distance from
the center to the end? (c) What is the current at the end of the dipole?
4. A half-wave dipole in free space is center-fed with a current of 10 amp (rms).
At a distance of 1,000 meters from the dipole, which is in the far field, and in a
direction that is 60 degrees from the dipole axis, what is the field strength in
volts per meter?
5. A medium-frequency (MF) radio station (for which a long-wire horizontal
antenna is practical) is required to communicate with only four other stations.
These stations are in four different directions; one is due north, one due south,
one east, and one west. It is desired to utilize a single horizontal long-wire reso-
nant antenna that will have four major lobes, one directed at each of the four
other stations. (The major lobes are those nearest the wire axis.) (a) How many
half wavelengths long should this antenna be? (b) What are the two possible
directions of the antenna wire? Suggestions: Assume that equation (4–14) is valid
for the value of n involved. Draw a diagram showing the relative positions and
directions of the stations, and consider possible orientations of the antenna wire.
Determine from this the required value of qmax in equation (4–14). Then, using
equation (4–14), solve for n. The value obtained for n will not be an exact integer;
the correct value is the integer nearest to the value found.)
6. A helical antenna with a ground plane (as in Fig. 4–20) has a turn diameter and
spacing that are appreciable fractions of a wavelength, so that it radiates in the
axial mode. The circumference of a turn, C, is equal to the wavelength, l, and
the turn spacing S is equal to 0.25l. The number of turns is N = 16. (a) What is
the 3-dB beamwidth of this antenna? (b) What is its directivity?
7. A rectangular waveguide pyramidal horn has aperture dimensions, dE and dH of
Fig. 4–21, of one wavelength. (a) Determine approximate E- and H-plane half
power beamwidths. (b) What is the approximate directivity?
8. A slot in an infinite plane sheet of metal is in the form of an annular ring so that
it is the complement of a small loop antenna. (The central portion of the sheet,
inside the annulus, is supported by low-loss insulating material.) (a) Is the
maximum radiation of this annular slot in the direction perpendicular to the plane
of the sheet, or is it parallel to it? (b) What is the polarization of the field near
the metal sheet at a point distant from the slot?
9. A thin vertical monopole antenna is 95 percent of a quarter wavelength in height
above a ground plane of infinite extent and perfect conductivity. It is fed at a
gap at its base by a coaxial line. No transformer is used at the feed point, yet
there are no apparent standing waves on this line. What is the approximate
characteristic impedance of the line?
10. Several means of delivering power to an antenna from a transmitter are listed
below, and identified by capital letters:
(A) Coaxial line.
(B) Balanced two-wire line followed by an impedance transformer to provide
nonresonant operation (no standing waves).
(C) Waveguide.
(D) Direct connection.
(E) Resonant two-wire line.
After each of the antenna types below, write in one of the above capital letters
to indicate which form of line or connection you consider most appropriate. (Use
each letter once and only once.)
(i) Automobile radio antenna for broadcast-band reception ⵧ (535–1,605 kHz)
(ii) Helical antenna with ground plane ⵧ
(iii) Long-wire antenna fed at one end ⵧ
(iv) Horn radiator ⵧ
(v) Half-wave dipole having 20-meter length, fed at gap in center ⵧ
CHAPTER 5
Arrays
This chapter discusses array antennas that are designed so that the major lobe (the main
beam) is pointed in one fixed direction or more. Commonly, array antennas are also
designed to electronically, and thus almost instantly, change beam pointing direction.
Such a beam pointing antenna is called a phased array, electronic scanning array, or
electronically steered array (ESA). The general principles of arrays are discussed in the
present chapter, and chapter 8, titled Electronically Steered Arrays, focuses on concepts
and techniques more directly applicable to electronic beam steering.
The term array, as applied to antennas, means an assembly of radiating elements in
an electrical and geometric arrangement of such a nature that the radiation from the ele-
ments “adds up” to give a maximum field intensity in a particular direction or directions
and cancels or very nearly cancels in others. Obviously this principle can be used for the
design of a directional antenna with increased gain.
The long-wire antennas discussed in sec. 4.4 may in a sense be regarded as arrays,
since they are analyzable as an assembly of elemental dipoles in a geometric configura-
tion that provides directionality and gain. (The resonant long-wire antennas also might
be regarded as an array of half-wave dipoles.) The term array, however, is usually
reserved for arrangements in which the individual radiators are separate rather than part
of a continuous radiator.
5.1. Basic Array Theory

All the elements of an array (the individual radiators) are usually alike, or at least very
similar. Perhaps the commonest array elements are the half-wave dipole, the half-wave
microstrip patch, the half-wave slot, and the open ended waveguide, but practically all
the basic radiators discussed in chapter 4 are employed as elements of arrays.
The radiation pattern of an array (in free space) depends on four factors: (1) the rela-
tive positions of the individual radiators with respect to each other; (2) the relative phases
of the currents or fields in them; (3) the relative magnitudes of the individual-radiator
currents or fields; and (4) the patterns of the individual radiators. The basic theory of
arrays is developed in terms of the first three factors, on the assumption that the individual
radiators are isotropic point sources. An isotropic point source is one that radiates with
169
170 Arrays
uniform intensity in all directions and has no physical size, and here it is assumed that
it does not block or otherwise affect the radiation of the other elements of the array. An
array radiation pattern can be calculated on the basis of these assumptions, and then a
correction to it can be made to take into account the fact that the individual radiators in
practical cases do affect each other and do not radiate isotropically.
5.1.1. The Two-Isotropic-Element Array

The simplest array is one that consists of just two isotropic point-source radiators. It will
be discussed in some detail because it illustrates most of the principles of arrays of any
number of individual radiators. The two-element array will be further specialized by
assuming that the individual radiators are of “equal strength.” If they are thought of as
radiating a field intensity proportional to a current that flows in them (in some unspeci-
fied fashion), the currents in the two are assumed to be equal. (An isotropic point-source
radiator is a fiction, and the idea of a current is introduced solely to illustrate the meaning
of “equal strength.”) The only remaining variables are the spacing of the two radiators
and their phases.
The meaning of the term phase as applied to the radiating elements of an array is also
most easily illustrated by considering that their radiation is related to currents that flow
in them. Suppose that at some distant point the fields of the two radiating elements are
examined and are found to be in phase with each other. If then the phase of the current
in Radiator 2 is changed by an amount a radians whereas the phase of the current in
Radiator 1 is not changed, it will be observed that the two fields at the distant point are
now out-of-phase by the amount a radians. A phase can also be ascribed to radiators of
the aperture type; the phase may be taken as that of the field at some reference point in
the aperture.
The phase of a radiating element in an array is always discussed in relation to the
phases of the other elements. In other words, if the phases of the two radiators in a two-
element array are changed by the same amount, the array radiation pattern is unaffected;
but changing the phase of one without changing the other will affect the pattern because
it affects the way that the individual fields of the two radiators add up at a specified
distant point.
The significant geometry of a two-element isotropic-point-source array is shown in
Fig. 5–1. The two sources are positioned a distance d apart on the y-axis of an xy coor-
dinate system, each distant d/2 from the origin. The point P is a distant point at which
it is desired to calculate the field strength that results from the radiation of the array. The
coordinates of P denoted r and f are two coordinates of the spherical-coordinate system
shown in Fig. 3–1, chapter 3. The third spherical coordinate, q, is not shown because
only a plane in which q is constant is being considered here, namely, the xy-plane in
which q = p/2 radians (90°).
In the parlance of electromagnetic theory (and antenna theory which is really an
application of electromagnetic theory), P is the field point, that is, the point at which the
field is to be calculated. Since the position of P will be specified only in general terms,
it can represent any point in space; hence, an expression for the field strength at P applies
Basic Array Theory 171
r1
r r2
f1
Radiator 1
f
d x
f2
Radiator 2
FIGURE 5–1.
Array of two isotropic point sources.
to all points and thus defines the radiation pattern of the array. Calculation of the field
due to the array at an arbitrary point P is basic to the understanding of array theory. This
calculation is here made for the two-element isotropic-point-source array, with the two
sources of equal strength (intensity).
As shown, the distance from Radiator 1 at point R1 to P is r1, and from Radiator 2 at
point R2 to P the distance is r2 (R1 and R2 are shown in Fig. 5–2). The two sources are
assumed to lie on the y-axis of a cartesian coordinate system, and P is assumed to be in
the xy-plane. Therefore, in terms of the corresponding spherical coordinate system
(Fig. 3–1), the direction of P from R1 is f1, and from R2 it is f2.
The distances r1 and r2 are assumed to be very large compared to the distance d, the
separation of the two radiators. If the difference of these two distances to the field point
is d = r2 − r1, the maximum possible value that d can have, whatever the location of P,
is equal to d, the radiator separation. (This equality will occur when P lies on the y-axis;
i.e., when f1 = f2 = 90° or 270°.) This means, since both r1 and r2 are very much larger
than d, that d will always be very much smaller than either r1 or r2. The very important
conclusion that may then be drawn is that if the two radiators are of equal strength, the
amplitudes of their separate fields at P will be very nearly the same; that is, both will
be reduced in strength, because of the distance traveled, by virtually the same amount.
172 Arrays
(This reduction in strength due to the spherical spreading of waves radiated from point
sources in free space was discussed in sec. 1.1.6).
On the other hand, the relative phases of the two fields at P, from R1 and R2, will be
very importantly affected by d. The resulting phase difference of the fields due to d is in
fact equal to −2pd/l radians or −360d/l degrees. To this difference must be added the
initial phase difference, that is, the phase difference of the two radiators themselves, a.
The total phase difference y of the two fields at P will be*
2πδ
ψ =α − radians (5–1)
λ
The resultant field at P is determined by the superposition of two fields of equal ampli-
tude, which will be denoted E0, and of phase-difference y. This involves phasor addition,
and it is mathematically the same as the addition of vectors separated in direction by the
angle y. The magnitude of this phasor addition was given in sec. 1.2.4, and for the case
of equal amplitudes the result was found from (1–28) to be
ψ
E = 2 E0 cos   (5–2)
 2
where, at observation point P, E0 is the amplitude of each field and y is the phase dif-
ference of the two fields.
This is a first step toward finding the radiation pattern of the array, which is an expres-
sion of E as a function of f, the angle of the direction of P from the center of the array.
To obtain such an expression, it is evidently necessary to express y in terms of f. Equa-
tion (5–1), an expression for y in terms of d and a, is already available. Since a is a
(presumably) known quantity (the phase difference of the array elements), the only vari-
able here is d, which can be expressed in terms of f. To see how this can be done, it is
helpful to consider an enlarged diagram, Fig. 5–2, showing the region in the immediate
vicinity of the array. As this diagram shows, because P is so distant in relation to the
element separation d, the lines labeled r1, r2, and r can be considered almost parallel to
each other, so that also f1 = f2 = f to a very close approximation. If a construction line
R1Q is drawn as shown dashed, so that it is perpendicular to the line R2P (also designated
r2), the distances R1P and QP can be considered equal. Therefore, the distance R2Q is
the difference, d, between R1P and R2P (i.e., between r1 and r2). The angle R2R1Q is
equal to f, since R1R2 is perpendicular to the x-axis and R1Q is perpendicular to R2P.
The triangle R1QR2 is a right triangle, with base d, hypotenuse d, and angle f opposite
the base. Therefore,
δ
sin φ = (5–3)
d
* In (5–1) y and a denote phases of Radiator 2 field and current with respect to those of Radiator 1, that is,
the Radiator 1 phase is the reference phase.
r1
f1
P
R1
f
f
d x
r2
f2
R2 d
Q
FIGURE 5–2.
Enlarged portion of two-element-array geometry.
or
δ = d sin φ
Substituting this result into (5–1), and then substituting the resultant expression for y
into (5–2), gives the following equation for the magnitude of the field at P:
α π d sin φ 
E (φ ) = 2 E0 cos  − (5–4)
2 λ 
This is the desired expression for the magnitude of the field at P as a function of the
angle f that the direction of P makes with the line perpendicular to the line of the array.
This equation gives the shape of the pattern in the xy-plane (which may also be called
the q = p/2 plane). Recall from (5–1) that y is expressed in radians. Thus, for calcula-
tions using degrees, p in (5–1) and (5–4) is replaced by 180°, because p radians = 180°.
The absolute-value brackets are used to indicate that the field intensity being calculated
is proportional to the amplitude or to the rms value and is therefore a positive number,
174 Arrays
although the quantity inside the brackets may be sometimes positive and sometimes
negative.
In order to obtain the relative pattern, for which the field strength in the maximum-
intensity direction has the value unity, (5–4) must be divided by the maximum value of
E. In this example E = 2E0, because the separate fields of each radiator are assigned the
intensity E0.
It is evident that this pattern depends on the spacing of the elements d and their phasing
a, as stated at the outset. It is worth noting that the result just presented for a two element
array is very similar to that of the field due to an antenna and its image in a conducting
earth, which was considered in sec. 1–4. The differences are that in the antenna-and-
image configuration, the phase difference of the two has the fixed value p radians or 180
degrees, and the distance designated 2h in Fig. 1–14 may be much larger than d is usually
permitted to be in an array (although sometimes 2h may be comparable to d ). Otherwise
the two situations are the same, mathematically. Physically there is still another differ-
ence, in that the interference pattern of the antenna-plus-image occupies only half of
three-dimensional space, the other half being occupied by the reflecting earth.
5.1.2. Pattern Versus q and f

Equation (5–4) describes the array pattern in the xy-plane, in which the angle q of a
three-dimensional coordinate system is constant (q = p/2 radians, or 90°). Therefore the
pattern in this plane is not a function of q, and q does not appear in the equation.
The complete pattern is actually a “figure of revolution,” obtained by rotating the xy-
plane pattern about the y-axis (line of the array). In other words, all patterns in planes
through the y-axis are identical. For example, the pattern in the yz-plane is identical in
shape and size to the pattern in the xy-plane. In this plane, however, the pattern equation
is expressed as a function of the angle q instead of the angle f; in other planes, both
angles are involved. Equation (5–4a) gives the pattern of amplitude versus q and f
α π d ⋅ sin θ ⋅ sin φ 
E (θ , φ ) = 2 E0 cos  −  (5–4a)
2 λ
The quantity sin f in (5–4) is merely replaced by the product sin q ⋅ sin f. In the xy-plane,
when q = 90°, sin q = 1; hence (5–4a) becomes (5–4) in this plane, as it should.
5.1.3. Parallel-Dipole Two-Element Array

The pattern defined by (5–4) can be plotted as a function of the angle f for various values
of the parameters d and a. Although the various resulting patterns pertain basically to an
array of isotropic elements, they also represent the patterns of an array of two dipoles in
the plane perpendicular to their axes, if the dipoles are parallel to each other and perpen-
dicular to the line joining their centers (line of the array). In terms of Fig. 5–1 the dipoles
have their centers at the positions of the isotropic point sources on the y-axis, but their
axes are parallel to the z-axis. In the xy-plane each dipole considered individually will
have an omnidirectional pattern, that is, uniform radiation in all directions in that plane.
Consequently the dipole-array pattern will be represented by (5–4) in the xy-plane only.
These two-element dipole-array patterns for a number of values of d and a within the
ranges of greatest practical interest are shown in Fig. 5–3, as originally published by
G. H. Brown (1937). They may be thought of as the horizontal-plane patterns of a pair
of vertical dipoles or monopoles separated by a distance d and with currents having phase
difference a.
As shown in Fig. 5–3, the effect of the spacing and phasing of the dipoles is rather
remarkable, and the variations in the patterns are extreme. The light-line circles represent
the E-field pattern that would be obtained from a single dipole located at the midpoint
of the array (with the array dipoles removed) radiating the same total power as the array.
The heavy-line plots are the electric-intensity radiation patterns. A great deal can be
learned about arrays in general by studying these two-element patterns.
It should be noted that when the two elements are in phase (a = 0°), the radiation has
a maximum in the direction perpendicular to the line joining the elements. That is,
because the distances from the elements to the field point (P, Fig. 5–1) are equal in that
direction (f = 0°), the phase difference due to path difference is zero also; hence the total
phase difference of the superimposed fields is y = 0°. Consequently, the fields add
directly, and the maximum possible resultant field is obtained. When a = 0° and a pattern
maximum is in the direction perpendicular to the array line, the antenna is called a
broadside array.
It should also be noted that for certain conditions the resultant field in some directions
is zero—when the sum of the radiator phase difference (a) and the phase difference due
to path difference (2pd/l) is an odd integral multiple of p radians. The fields of the indi-
vidual radiators are in this case of equal amplitude and opposite phase, so they cancel.
This occurs, for example, when a = 0° and d = l/2, in the f = 90° and 270° directions,
since in these directions the field phase-difference due to path difference is 180 degrees,
whereas the phase difference due to radiator phase difference is zero. The same result
occurs when the 180-degree net phase difference is due to a combination of path differ-
ence and radiator phase difference. These directions of zero intensity in a pattern are
called nulls.
Finally, it is to be noted that certain combinations of d and a result in maximum radia-
tion in the direction of the line joining the array elements. The array is then said to be
operating as an endfire array. The radiation of an endfire array may be either bidirectional
(radiation lobes in both directions along the line of the array) or unidirectional (a lobe
in one direction and a null in the opposite direction). Examples of unidirectional endfire
arrays are the case of d = l /8 with a = 135°, and the case of d = l/4 with a = 90°. When
d is greater than l /2, the array will have lobes in more than two directions, and may
even be simultaneously broadside and endfire.
Although the two-element array is the simplest of all arrays, the variety of patterns
obtainable makes it useful in many applications. In radio broadcasting in the MF band,
for example, two vertical towers with appropriate spacing and phasing can be used to
produce a horizontal-plane pattern (possibly one depicted in Fig. 5–3) to favor certain
176 Arrays
7/8
3/4
5/8
1/2
d/l
3/8
1/4
1/8
0° 45° 90° 135° 180°

a
FIGURE 5–3.
Horizontal-plane patterns of two vertical dipoles (or monopoles). (Proceedings IRE, 25,
78–145, January 1937.)
directions, or to minimize radiation in certain directions, as may be required by the

Federal Communications Commission to avoid interfering with the coverage area of
another station on the same frequency.
Equations (5–2), (5–4), and (5–4a) and most equations in chapter 5 are based on
assumptions that the radiating elements and input networks are lossless, and that each
element radiates a fixed amount of power independent of element spacing or relative
phasing. Thus, these assumptions include the input and radiated powers being equal and
that there is no mutual coupling between elements. These oversimplifications are com-
monly used, so that the most basic of array concepts can be more easily addressed. In
fact, although the coupling effects may sometimes be negligible, mutual coupling in
principle causes changes in an element’s radiation resistance and its input impedance,
and it may also cause a change in an element’s radiation pattern. Consequently, with a
change in either element spacing or phasing, a fixed element current does not imply that
there is a fixed radiated power from an element or conversely.
As already mentioned, in Fig. 5–3 the resulting field strength from the two elements
is compared with the field strength from only one of the elements, and with the inputs
for the two cases (two elements versus one element) always having the same radiated
power (losses assumed zero) for each pattern. However, with constant power radiated,
because of mutual-coupling effects on dipole radiation resistance and input impedance,
the dipole currents may be different for the different spacings and phasings. Kraus and
Marhefka (2002, pp. 529–52) include clear explanations on how to calculate the Fig. 5–3
patterns as well as patterns for other constant power, two-dipole arrays. Additionally, on
page 535, Kraus and Marhefka note that Brown (1937) considered it important to provide
analyses based on constant antenna input power, because “most transmitters are essen-
tially constant power devices which can be coupled to a wide range of antenna imped-
ance.” The subject of mutual impedance is further addressed in secs. 5.1.4 through 5.1.7
of this chapter, and in chapter 8 in connection with electronically steered arrays.
5.1.4. Mutual Coupling

As indicated in the last two paragraphs, effects of a radiating element’s input (driving-
point) impedance and radiation resistance caused by mutual coupling between array ele-
ments need to be taken into account when calculating array performance. Therefore, the
feed system of an array cannot be designed based on the input impedance of an isolated
dipole, nor can the radiated power be calculated based on the radiation resistance of an
isolated-dipole. In other words, mutual impedances should be included in the calculations
of input impedance and radiation resistance. Furthermore, although usually of a lesser
effect, the actual patterns of an element may be significantly affected by the mutual cou-
pling. An element pattern, when operating within an array environment, is known as a
scan element pattern (SEP). An SEP, previously known as an active element pattern, is
found experimentally by connecting a receiver to only one element of the array, with all
the other elements connected to terminating impedances of the value that would be effec-
tive if the array were operated normally. The pattern measurement of this element is
accomplished in the usual fashion (see chapter 9). The currents induced in the single
178 Arrays
receiving element (as well as in the terminated elements) will be the same as in actual
array operation, including mutual-coupling effects. The pattern measured under these
conditions is in general different from the pattern of an isolated element. Although much
of this discussion has been in terms of dipole elements, mutual coupling occurs with all
types of array elements.
In the practical design of arrays, the effects of mutual coupling are often evaluated
experimentally, because theoretical calculations are usually very difficult, although the
theory is helpful in understanding the general nature of the effect. The electromagnetic
analysis of element coupling involves complex mathematics and detailed use of
Maxwell’s equations. Such analysis will of course depend on the shape and size of the
elements and on their electrical properties, element spacing, locations of the elements
within the array aperture, wavelength, polarization, and the amplitudes and phases of the
element input currents. The total of the electric and magnetic fields near an element
includes the fields generated by its currents and the coupling from reactive near-fields
from the other elements and possibly from nearby objects. For antennas with dimensions
of many wavelengths, the fields of an element may also include significant radiating
near-field components from other elements. Computer codes are used to compute the
effects of mutual coupling among the elements of an array (Hansen 2007, pp. 20–24).
5.1.5. Input and Mutual Impedances, and Radiation Resistance

Because of mutual coupling among elements, the amplitude and phase of the input
current of a specific array element cannot be adjusted independently of other elements.
In other words, the driving-point (i.e., input) impedance of an individual element might
differ considerably from its self-impedance, because of its mutual coupling with other
array elements. In a multielement array, terminal voltages and element currents behave,
conceptually, like a conventional mesh circuit network. Thus, the sum of the voltages
around a current loop equals zero, as follows:
V1 = I1Z11 + I 2 Z12 + . . . + I n Z1n

V2 = I1Z12 + I 2 Z 22 + . . . + I n Z 2 n
. (5–5)
.
.
Vn = I1Z1n + I 2 Z 2 n + . . . + I n Znnn
where
V1, V2, etc. = voltages applied to elements 1, 2, etc.
I1, I2, etc. = currents flowing in elements 1, 2, etc.
Z11, Z22, etc. = self-impedances of elements 1, 2, etc.
Z12, Z2n, etc. = mutual impedances between elements denoted by the subscripts
The driving-point (input) impedance offered to the voltage applied to the kth element
is the phasor ratio (in general complex) Vk/Ik, obtained by dividing the right-hand side
of the above equation by Ik. Thus for k = 1, the input (i.e., driving-point) impedance
follows:
V1 I I
Z1input = = Z11 + 2 Z12 + 3 Z13 + . . . (5–6)
I1 I1 I1
From above, it is seen that the input or driving-point impedance of a particular element
is not only a function of its own self-impedance, but also a function of the relative cur-
rents flowing to the other elements and possibly in nearby objects. Thus, it is important
in an array to design the feed system to match the input impedances instead of simply
the self-impedances.
Radiation resistance is the resistance Rr that, inserted at the point where a current I is
flowing, dissipates the same energy that radiates from the antenna. Thus Rr equals radi-
ated power divided by I 2 (sec. 4.1.6). In an array composed of several radiating elements,
the total radiation resistance of an element may be expressed in terms of the radiation
resistance of the individual element and its mutual impedances. Specifically, the effective
radiation resistance encountered by a voltage V1 applied to element 1 of an antenna array
is the resistance component of the driving-point (input) impedance Z1input offered to V1,
as calculated by (5.6). The power radiated by element 1 delivered by the voltage V1 is
the square of the current multiplied by this effective radiation resistance. Likewise, the
total power delivered to an element is the square of the current at the input multiplied
by the resistive component of Z1input. This resistive component consists of the radiation
resistance Rr, plus whatever additional resistance is needed to account for the dissipated
(heat) losses that may arise from the presence of earth, and from the conductors and
dielectrics within the antenna. The radiation resistance to the voltages applied to other
elements in the array and the power delivered by these other voltages is determined in
the same way. The total radiated power is then the sum of the powers supplied by the
various applied voltages. Likewise, total power input to the elements is the sum of the
powers delivered to all elements.
5.1.6. Principle of Pattern Multiplication

Equation (5–4) describes the pattern in one plane of either a two-element array of iso-
tropic radiators or a two-element dipole array if the dipoles have their axes perpendicular
to the line of the array and parallel to each other. Equation (5–4a) describes the pattern
versus q and f for the array of isotropic elements, but it does not describe pattern versus
q and f for the dipole array. In planes other than the one perpendicular to the dipole axes,
the pattern is affected by the nonisotropic patterns of the individual dipoles. The question
to be answered is, what is the complete far-field pattern of any array?
The answer is provided by the principle of pattern multiplication, which is described
below by use of (5–7). If an array consists of elements whose electric field patterns are
all alike, and described by an expression that will be denoted Ee(q, f), and if the array
pattern for isotropic point source elements has a pattern that will be denoted by Ea(q, f),
then the pattern of the array of actual elements, EA(q, f), is given by the formula
180 Arrays
E A(θ , φ ) = kn ⋅ Ee(θ , φ ) ⋅ Ea(θ , φ ) (5–7)
That is, the pattern of the array of actual elements is obtained by multiplying together
the pattern of a single element and the pattern of the array as calculated for isotropic
point-source elements. The function EA(q, f) is sometimes referred to as the antenna
pattern factor, which is the product of the element factor Ee(q, f) and the array factor
Ea(q, f). Squaring both sides of this equation converts it into a relationship of power-
density patterns, in which the squares of the electric-intensity factors may be replaced
by corresponding power-density pattern factors, pA(q, f), pe(q, f), and pa(q, f), in accor-
dance with equation (1–9), chapter 1.
The factor kn is included in this equation as an arbitrary numerical constant to be
adjusted to whatever value is required to make EA(q, f) a true relative pattern. That is,
it is desired that EA = 1 for the particular values of q and f in the direction of maximum
field intensity, in accordance with the definition of a relative pattern. If both Ee and Ea
are relative patterns, it may happen that EA will automatically be a relative pattern also,
that is, with kn = 1. This will happen only when a maximum direction of Ee coincides
with a maximum direction of Ea, which it will not necessarily always do, although it
often does. The factor kn is called a normalizing constant.
As an example of the application of (5–7), the pattern versus q and f will be deter-
mined of a two-element array of dipole radiators, spaced a distance d apart on the y-axis,
with phasing a. For the case of isotropic radiators instead of dipoles, the pattern is given
by (5–4a). The pattern of a half-wave dipole with its axis parallel to the z-axis is given
by equation (4–11), sec. 4–3. (This is the complete dipole pattern, even though it involves
only the angle q rather than both q and f, because the pattern for this particular dipole
orientation is independent of the angle f. That is, the pattern in any plane corresponding
to a particular value of f is exactly like the pattern in a plane corresponding to any other
value of f.) The complete pattern follows in (5–8). It is a function of q and f, of a two-
element dipole array with this orientation of the dipole axes and is attained by multiplying
the patterns of (4–11), sec. 4–3 and (5–4a) together:
⎡ cos ⎛ π cos θ ⎞ ⎤
α
E A(θ , φ ) = kn cos ⎛⎜ −
π d ⋅ sin θ ⋅ sin φ ⎞ ⋅⎢ ⎝2 ⎠⎥
⎝2 ⎟
⎠ ⎢ ⎥ (5–8)
λ ⎢ sin θ ⎥
⎣ ⎦
In using the principle of pattern multiplication, certain rules apply:

1. The pattern to be used for the array elements is the pattern that a single element
would have in the array environment, i.e., the “active pattern”—not the pattern
in isolation other elements or objects.
2. The pattern of the array factor to be used is the one calculated for isotropic
point-source elements spaced and phased the same as the actual elements.
The phasing cannot be calculated on the basis of a transmission-line feed
system designed for the input impedances of the elements operating as isolated
elements. Their actual input impedances will be affected by mutual coupling

with other elements.
3. All the array elements must be alike; they must have similar patterns and
polarizations, similarly oriented. For example, if the elements are dipoles, they
must have their axes parallel to one another. If the elements are horns, they
must be of the same form and size, and all “aimed” in the same direction.
The principle of pattern multiplication and the calculation of array patterns in general
have been discussed from the relative-pattern point of view. To attain the absolute value
of the E-field pattern, one must return to the definitions of directivity D and gain G. As
discussed in sec. 3.3.4, the directivity of an antenna, D, can in principle be calculated if
the relative pattern is known, although the actual process may be very difficult. Addition-
ally, D can be measured, as discussed in chapter 9. If both D and the total radiated power,
Pt, are known, the absolute pattern may be calculated. To do this, one multiplies the rela-
tive E-field pattern by 30Pt D R from equation (1–45), chapter 1, where R is the dis-
tance from the antenna to a far-field point.
5.1.7. Effects of Element Directivity and Mutual Coupling

An interesting conclusion may be drawn from the principle of pattern multiplication
concerning the use of directive elements in arrays. The conclusion is that if the pattern
of the array factor itself, that is, (5–7) has a much narrower beamwidth than the element-
pattern beamwidth, the overall pattern beamwidth and gain are determined primarily by
the array factor. Element beamwidth and gain then have little effect for a given array
configuration (fixed number of elements). However, the use of directive elements may
permit achieving a desired antenna beamwidth and gain with fewer elements spaced
farther apart than would be permissible with isotropic or wide-beamwidth elements,
resulting in economy of elements and a simpler feed system, but not in a reduction of
overall antenna size.
Experience shows that an active element pattern (i.e., SEP) may be narrower than the
pattern of the same element when isolated from surrounding objects (see, e.g., Mailloux
1994, p. 324). Then, for large arrays having many elements, the dominating array factor
tends to mitigate the element pattern narrowing effect of mutual coupling. However, the
array factor does limit the angle over which an array beam can be electronically scanned
without appreciable pattern degradation.
The previous discussions on mutual coupling are mostly about the inadequacy of using
element impedance, radiation resistance, and radiation patterns applicable to an isolated
array element environment. Consequently, one might logically ask why, in the remaining
parts of chapter 5, is there so much emphasis on the use of (a) array element character-
istics that are in isolation of other elements and (b) array factors that, by definition,
assume no mutual coupling? Firstly, these subjects are basic to the principles of arrays
and to understanding array performance. In addition, although likely different in an array
environment, knowledge of such gross features as self-impedance, polarization, general
element pattern shape, and resonant frequency are useful, generally, toward a preliminary
design. As discussed later throughout much of chapter 5, the amplitudes and phases of
182 Arrays
element excitation currents are critical to the control of the overall array patterns. These
quantities are needed when calculating array factors, yet they do not include, explicitly,
effects of mutual coupling. Furthermore, these are the currents needed, when effects of
mutual impedance are included, to determine the required element input voltages and
radiated power. One must also recognize that element beam shapes change, because of
being within an array environment. However, because of the dominating effect of the
array factor on overall array patterns (sec. 5.1.6), useful pattern shape approximations
are often possible without the use of measured in-array “active” element patterns.
5.2. Multielement Uniform Linear Arrays

When more than two elements are used in an array, the principle of calculating the pattern
(from which in turn the beamwidth and directivity can be computed) is the same as for
a two-element array except that the fields of all the elements must be superposed at the
field point. The simplest type of multielement array is one in which all the radiators are
in a line, with equal spacing between adjacent pairs, as shown in Fig. 5–4. The method
f = 90°
y –z P
Array
elements r1
r2
d
d/2
d x f = 0°
r3
d
r4
Array
elements
FIGURE 5–4.
Linear array of four radiating elements.
Multielement Uniform Linear Arrays 183
of analysis is suggested by showing a field-point P joined by ray lines to each element.

Such an array is called a linear array. When all the elements are radiating with equal
intensity, and the phase difference between adjacent elements is constant, the array is
called uniform. This section deals with the properties of uniform linear arrays.
5.2.1. Radiation Pattern

If the array contains N isotropic point-source elements, with equal spacing d and phase
difference a between adjacent elements, the pattern in a plane containing the line of the
array can be shown to be
π d sin φ α  
sin  N  −
  λ 2  
Erel = (5–9)
π d sin φ α 
N sin  − 
 λ 2
where f, as in (5–4), is the angle between the direction of the field point and a perpen-
dicular to the line of the array. For N = 2 this expression can be shown to be equivalent
to (5–4) (despite the seeming dissimilarity). As in the two-element case, the pattern versus
q and f is obtained simply by replacing sin f by the product sin f ⋅ sin q. The factor N
in the denominator is a normalizing factor, as discussed for (5–7); that is, in this case
kn = 1/N.
The pattern has its maximum values at angles f such that the quantity (pd sin f)/
l − a /2 = 0; that is, when (pd sin f)/l = a /2. The pattern will have additional maxima
at angles for which this quantity is equal to an integral multiple of p radians. At the
maxima, (5–9) is equivalent to
Erel = lim 
sin Nx 
lim 
sin Nx 
 or
 N sin x 
(5–10)
x → 0  N sin x  x → mπ
where m is any integer. For either limit, this expression is an “indeterminate form” (zero
divided by zero), but application of a method of differential calculus shows that the limit
is unity. The pattern has secondary maxima, or minor lobes, when the numerator of (5–9)
attains values expressed by sin[(2m + 1)(p/2)], where again m is any integer (i.e., m = 1,
2, 3, . . .).
The variety of possible patterns for an array with a given number of elements, obtained
with different values of d and a, is very great, but many of the possibilities are primarily
curiosities, of no practical value. Two basic types of pattern are of special interest.
5.2.2. Broadside Linear Arrays

When a = 0°, all the elements are “in phase,” and pattern maxima occur at f = 0° and
f = 180°, that is, in the directions perpendicular to the line of the array, which is then
184 Arrays
Dipoles called a broadside array. The pattern will

be maximum in these directions regard-
less of the element spacing, d. These will
l/2
be the only primary maxima if d is less
than l. If d = l, additional maxima occur
at f = 90° and 270°; as d is increased still
d
further, additional maxima occur as cones
of radiation about the axis of the array.
FIGURE 5–5. They are known as grating lobes (so
Linear horizontal array of six vertical dipoles. called because they are analogous to
the lobes observed in the optical study
of a diffraction or reflection grating).
d Dipoles Ordinarily, therefore, the spacing of ele-
ments in a broadside array is kept less
l/2 than a wavelength.
However, it has been shown that,
FIGURE 5–6.
subject to this restriction, there is an
Collinear broadside array. advantage in spacing the elements of an
array by somewhat more than half a
wavelength. For a two-element array the optimum spacing is about 0.7l, for four ele-
ments it is about 0.8l, and for a large number of elements the optimum is about 0.95l
(Tai 1964). The directivity increases gradually as the spacing is increased until the
optimum is reached, then drops rather sharply with further increase. It is to be empha-
sized, however, that the advantage of the wider spacings is in the directivity obtainable
with a given number of elements. In terms of the ratio of the directivity to the total length
of the array, there is no advantage in the wider spacing.
The foregoing observations are based on a uniform linear broadside array of point-
source isotropic radiators. They apply also to a similar array of dipoles with their centers
on the array line and their axes perpendicular to the array line and parallel to each other,
as in Fig. 5–5, with respect to the pattern in the plane perpendicular to the dipole axes,
that is, the xy-plane in Fig. 5–4. If the array line is horizontal, the dipole axes are verti-
cal, and the radiation is vertically polarized. (It is also possible to make the array line
vertical and have the dipole axes horizontal, and the radiation will then be horizontally
polarized.) The spacing effects for an array of isotropes also apply to the dipoles in this
case because their individual patterns are omnidirectional in the plane perpendicular to
their axes.
The dipoles can also be arranged with their axes parallel to the line of the array, which
is then called a collinear broadside array. This arrangement, pictured in Fig. 5–6, allows
a center-to-center dipole spacing somewhat greater than is permissible with parallel
dipoles, because the patterns of the individual dipoles have nulls in the f = 90° and
f = 270° directions (i.e., in line with their axes and the line of the array). Therefore the
wide-angle grating lobes of radiation that appear with greater spacings are suppressed,
for element spacings somewhat greater than a wavelength, in accordance with the prin-
ciple of pattern multiplication.
l /4 l /2
Quarter-wave
stubs
Feed line
FIGURE 5–7.
A six-half-wave Franklin antenna (six-element collinear dipole array).
The in-phase currents in the individual dipoles, required for a broadside pattern, may
be obtained by properly connecting a branched transmission line to the feed point of each
dipole. In other words, if the total line length from the transmitter to each dipole is the
same, as in Fig. 2–8a, chapter 2, the dipoles will be fed in phase. (Care must be taken
to see that the same side of the line is connected to the same side of each dipole; revers-
ing this connection reverses the phase.)
In an alternative feed arrangement, shown in Fig. 5–7, the feed line is connected
between the ends of one pair of dipoles, which causes them to be in-phase. (This is a
high-impedance feed point, since the ends of the dipoles are voltage-maximum points.)
Then quarter-wave stubs are connected between the ends of the additional adjacent pairs
of dipoles, providing simultaneously a feed connection and proper phasing. The l /4 stubs
provide open circuit impedances, and their l/2 round-trip lengths cause a phase reversal
at each of their input terminals. This arrangement is known as a Franklin antenna. (Recall
that if the dipoles are directly connected end to end, without the quarter-wave stub, the
current in alternate half-wave sections will be out of phase; the antenna will be of the
long-wire type, and the pattern will be entirely different.)
The collinear array can also be fed by transmission lines connected between the ends
of pairs of adjacent dipoles. This has an advantage over center-feeding each dipole indi-
vidually, in that only one transmission-line branch is required for each pair of dipoles.
Also, the relatively high feed-point impedance is usually more easily matched to a two-
wire line. The dipole spacing is then restricted to one-half wavelength (as is also true for
the Franklin antenna), but this is a fairly satisfactory spacing.
Broadside arrays may also be formed from other types of elements, such as horns,
slots, microstrip patches, helixes, and polyrods. If the individual elements are themselves
directional, their maximum radiation should be in the broadside direction. If they are
unidirectional radiators, such as sectoral or pyramidal horns, waveguide slots, microstrip
patches, axial-mode helixes, and polyrods, a unidirectional broadside array results. When
186 Arrays
the elements themselves are fairly direc- 90°

tional, they can be spaced considerably
farther apart than isotropes or parallel
dipoles, because their directionality sup-
presses the unwanted grating lobes that
would occur with isotropic elements.
Figure 5–8 shows a typical principal 180° f = 0°
plane pattern of a uniform broadside
linear array of isotropic elements, in the
plane containing the line of the array.
Section 5.5.3 includes patterns and equa-
tions for calculating patterns for broad-
side linear arrays, including cases where
270°
array elements are not identical.
FIGURE 5–8.
Approximate pattern of an eight-element
uniform broadside linear array of point-source
5.2.3. Endfire Arrays isotropic elements.
If in (5–9) the phase-difference a between
adjacent elements is equal to 2pd/l radians, the condition for a maximum of radiation is
satisfied when f = 90°, and for this value only, provided that d is less than l/2. The
maximum field intensity is radiated in a direction along the line of the array, “off the
end” rather than broadside; hence the name endfire array. The maximum is toward only
one of the ends of the array, rather than in both the endfire directions. In terms of Fig.
5–4, if the progressive phase change is a retardation going in the direction of the positive
y-axis by the amount a per element, the beam will be in the f = 90° direction. If the sign
of a is changed, or its amount is increased by 180 degrees without changing the element
spacing, the beam will be in the direction f = 270°.
Although the condition stated above for the value of a results in an endfire array, it
does not result in an endfire pattern with the maximum possible directivity and narrowest
possible beam. It has been shown by Hansen and Woodyard (1938) that an endfire beam
with somewhat greater gain results if the phase change per element satisfies the following
condition:
2π d π 
α =  +  (5–11)
 λ N
where N is the number of elements in the array. This relationship is referred to in the
antenna literature as the Hansen-Woodyard condition. (This condition does not neces-
sarily result in a unidirectional pattern, however, as does the basic endfire condition.)
The correct phase of current or field to each radiating element of an endfire array
requires a feed system in which the current or field must travel through a longer path in
the transmission line or waveguide to each successive element along the array, the
increased length of line or guide per element being equal to al/2p, where a is the
Dipoles
Direction
of beam
Transmission
line d
Input
terminals
FIGURE 5–9.
One form of feed arrangement suitable for an endfire array.
required phase change per element expressed in radians and l

is the wavelength. If a waveguide line is used, l in this formula
is the guide wavelength, lg given as (2–68) in chapter 2. Figure
5–9 illustrates this principle for a four-element endfire dipole
array. Incidentally the basic endfire analysis assumes isotropic
radiators, but it applies also to other radiators that have radiation
along the line of the array.
The feed system must also be a matched-impedance system,
that is, one with hardly any standing wave on it, in order to
produce the required phasing of the elements in accordance with
this formula. There must be an impedance match at each branch
point of the line, in accordance with principles described in
chapter 2; see, for example, Fig. 2–7. The impedance relations FIGURE 5–10.
must also be adjusted so that equal currents are fed to all the Typical endfire array
dipoles in order for equation (5–9) to apply. A typical endfire pattern in a plane
containing the line of
array pattern is shown in Fig. 5–10, for isotropic point-source the array.
radiators, in a plane containing the array line. The beam has axial
symmetry; for example, if the beam axis is horizontal, the hori-
zontal and vertical beamwidths are the same.
Some of the patterns shown in Fig. 5–3 are those of two-element endfire arrays. A few
of them have not only a maximum in one endfire direction (the “front” direction) but
also a complete null, or zero field intensity, in the opposite (back) direction. The cases
d = l /8 with a =135° and d = l/4 with a = 90° are examples. These particular patterns
are advantageous in applications where a high front-to-back ratio is desired. The condi-
tion for a null in one endfire direction is
188 Arrays
α = π 1 ±  radians
2d
(5–12)
 λ
(Choosing the plus or minus sign merely corresponds to changing the particular endfire
direction in which the null occurs.) Satisfying this null condition does not insure a
maximum in the other direction. In fact, if d is an odd multiple of a half wavelength,
(5–12) corresponds to an in-phase condition of the elements (a = 0° or an integral mul-
tiple of 360°). This produces a broadside array, with nulls in both endfire directions.
However, with some spacings, of which examples are given above, a maximum does
occur in the direction opposite the null, producing a unidirectional pattern. In fact, the
two-element array with d = l /8 and a = 135° simultaneously satisfies the condition for
a null in the back direction and the Hansen-Woodyard condition for maximum forward
directivity for that particular spacing.
Incidentally, the null condition of (5–12), applies not only to the two-element array
but also to any uniform linear endfire array with an even number of elements, since the
elements cancel in pairs in the null direction. It may also be noted that the null is never
absolute because of unavoidable imperfections in the spacing and phasing of the ele-
ments, and many-element arrays are more susceptible to such imperfections than those
of few elements.
Section 5.5.6 includes further discussion on endfire arrays and Fig. 5–18 that compares
the beamwidths of an ordinary endfire array and an endfire array that satisfies the Hansen-
Woodyard condition.
5.3. Parasitically Excited Endfire Arrays

It is not necessary to feed each element of an endfire array by direct connection to a
transmission line. If only one dipole of such an array is directly fed, or driven, the field
that it sets up will cause currents to flow in adjoining elements. This process is called
parasitic excitation, and the elements thus excited are parasitic elements. Endfire arrays
employing this principle are known as Yagi-Uda antennas (often called simply “Yagis”).
According to Ma (1993, pp. 3–13), this type of array was first described in Japanese by
S. Uda in 1926, and subsequently in English by H. Yagi in 1928. As noted in sec. 4.10,
the Yagi-Uda antenna can also be regarded as a surface-wave antenna.
Parasitic excitation cannot be employed in broadside arrays. To produce the in-phase
currents required for a broadside pattern, a full-wavelength element spacing would be
required. There would then be two endfire lobes of radiation as well as the two broadside
lobes, and this type of pattern is not a true broadside pattern. Therefore broadside arrays
are always driven rather than parasitic. (There are some antennas that might be considered
exceptions to this statement. In them parasitically excited elements produce a broadside
beam, but the exciting source is not itself part of the array.)
The phases of the currents in parasitic dipole elements are determined by their spacing
from the adjacent element, and also by their lengths. A parasitic dipole electrically exactly
a half wavelength or slightly longer will be inductive, and the phase of its current will
Parasitically Excited Endfire Arrays 189
Driven element
C
Beam direction
Reflector
Directors
Feed line
FIGURE 5–11.
A Yagi-Uda parasitic endfire array with one reflector and five directors.
lag the induced emf. A dipole shorter electrically than a half wavelength will be capaci-
tive, and the current in it will lead the induced emf. Comparatively close spacing of
elements is used in parasitic arrays to obtain good excitation, and the induction fields of
the elements play a major role, so that an exact analysis is very complicated. It is known,
however, that properly spaced dipole elements that are electrically slightly shorter than
a half wavelength act as directors, reinforcing the field of the driven element in the direc-
tion away from the driven element. Thus a line of directors may be used, and each one
will excite the next one. On the other hand, an element that is electrically one-half-
wavelength long or slightly longer will act as a reflector, if correctly spaced, reinforcing
the field of the driven element in a direction toward the driven element from the reflector.
Therefore if a reflector element is placed adjacent to a driven element, another element
placed beyond the reflector will not be appreciably excited, as very little field exists
beyond the first reflector. For these reasons a Yagi-Uda endfire array usually consists of
one driven element, one reflector on one side of it, and a number of directors on the other
side of it.
A typical Yagi-Uda configuration is shown in Fig. 5–11. Antennas of this type offer
the advantages of a unidirectional beam of moderate directivity with light weight, sim-
plicity of feed-system design, and low cost. The design becomes critical, however, if
high directivity is attempted through the use of many elements. Up to five or six may be
used without difficulty, and arrays of thirty or forty elements are possible. The input
impedance of a Yagi-Uda array tends to be low, and the bandwidth is limited to around
2%, typically. Directivity of around 10 dB is readily achieved with a moderate number
of elements (five or six). Higher gains may be achieved by making a broadside array of
which the elements are Yagi-Uda arrays.
190 Arrays
5.4. Planar and Volume Arrays with Uniform Aperture Distribution

5.4.1. Planar Arrays
As implied in the discussion of Yagi-Uda endfire arrays, there is a practical limit to the
directivity that can be obtained from a linear array by increasing the number of elements.
The limit is reached sooner in a parasitic array than in a driven one, and sooner in a
driven endfire array than in a broadside one, but there is a practical limit in all forms of
linear arrays. The emphasis is on the word “practical”; in theory there is no limit. When
the number of elements is made extremely large the correct phasing of the currents
becomes extremely critical. Very slight errors of phasing will nullify the expected advan-
tage of additional elements, and the practical problem of maintaining the necessary exact
phasing is virtually impossible to solve. A linear array that results in about a 0.1° beam-
width represents approximately a practical limit. (Such an array would have about a
thousand elements.)
The directivity of an array can be increased, however, by combining broadside linear
arrays to form a broadside planar array. A broadside linear array of isotropic elements
has a pattern that may have a narrow beamwidth in a plane containing the line of the
array, but it radiates uniformly in all directions of a plane perpendicular to the line of the
array. Thus a many-element linear broadside array has a pattern that was earlier described
as like a “pancake” (as compared with the relatively fat “doughnut” pattern of a dipole).
But when broadside linear arrays are arranged in parallel rows with correct phasing to
form a broadside planar array, the pattern becomes two separate lobes, each having a
maximum in one of the two opposite directions perpendicular to the plane of the array.
The two lobes are identical (mirror images of each other) and are characterized in general
by two beamwidths measured in planes perpendicular to each other through the beam
axis. (The two planes are usually chosen parallel to and perpendicular to the line of the
basic linear arrays of which the plane array is formed.)
If the number of linear arrays thus “stacked” in rows is equal to the number of ele-
ments in one of the linear arrays, and the spacing of the rows is equal to the spacing of
elements in each linear array, the total plane array is a square. In fact, if the plane of the
array is vertical, it is impossible to say whether it consists of horizontal linear arrays
stacked vertically in rows, or vertical linear arrays arranged horizontally in columns; the
array is in fact a matrix of rows and columns. It is often convenient for purposes of
analysis however, to consider the planar array as a plane assembly of parallel linear
arrays. When the array is rectangular rather than square the linear arrays are usually
considered to run parallel to the longer dimension.
A common array element is the dipole, and the dipole axes are ordinarily parallel to
one side of the array. In a vertical planar array, the dipoles are placed vertically if vertical
polarization is desired and horizontally to obtain horizontal polarization. Then linear
arrays considered in one direction are collinear arrays, and in the other direction they are
parallel-dipole arrays. Figure 5–12 illustrates a vertical planar dipole array that is two
(vertical) dipoles high and four dipoles wide (a “two-by-four”). Such arrays are some-
times called curtains.
Planar and Volume Arrays with Uniform Aperture Distribution 191
z 5.4.2. Pattern Versus q and f

Dipoles
y The theory of planar arrays is best
x approached assuming isotropic point-
source radiators. Consider Fig. 5–13 that
consists of equal amplitude, in-phase
(a = 0°) isotropes with element spacings
dh and dv in the horizontal (y-axis) and
dv vertical (z-axis) directions. The equation
of the horizontal-plane broadside pattern
l/2
for a horizontal row of elements is given
Dipoles
nh = 4
nv = 2 by (5–9), with a = 0°. This linear array
dh pattern, when expressed in terms of q and
f is, in accordance with (5–4a), attained
FIGURE 5–12. by replacing sin f with sin f sin q. Thus,
A two-by-four, dipole broadside planar array in the pattern for a horizontal broadside
y-z plane. (Beam is in the x direction.) linear array, parallel to the y-axis, is
π d sin φ sin θ ⎞ ⎤
sin ⎡⎢ N h⎛ h
z
⎣ ⎝ λ ⎠ ⎥⎦
Eh(θ , φ ) =
π d sin φ sin θ ⎤
N h sin ⎡⎢ h
q ⎣ λ ⎦⎥
dh (5–13)
A vertical column of elements in the

dv planar array is a line array along the z-axis
of Fig. 5–13. Obviously the pattern of
y
such an array will have the same shape
(for a given number of elements N ) as a
horizontal linear array, but rotated in the
coordinate system to agree with its verti-
cal axis. This results in a different mathe-
matical expression for the pattern, which
f
x is
FIGURE 5–13.
π d cos θ ⎞ ⎤
A three-by-seven broadside planar array of sin ⎡⎢ N v ⎛ v
equal-amplitude, equal-phase isotropes in y-z ⎣ ⎝ λ ⎠ ⎥⎦
Ev(θ , φ ) =
π d cos θ ⎤
N v sin ⎡⎢ v
plane. (Beam is in the x direction.)
⎣ λ ⎥⎦
(5–14)
Note that the angle f does not enter into (5–14), even though it describes the complete
pattern. This is because of the orientation of the vertical array with respect to the coor-
dinate system.)
192 Arrays
Equations (5–13) and (5–14) express separately the patterns of horizontal and vertical
linear arrays into which the plane array may be decomposed. Thus, the question is, what
is the pattern of the complete array? Note that the array can be viewed as consisting of
vertical line arrays that are equally spaced along the y-axis (the horizontal), where each
vertical array is a triplet, containing three isotropic elements. Thus, the array may be
viewed as a horizontal linear array of the triplets, and (5–14) expresses the pattern of
each triplet. Therefore, the planar array pattern may be attained by application of (5–7),
the principle of pattern multiplication. This principle can be applied in the present case
by considering the planar array to be a horizontal uniform linear array, whose individual
elements are vertical arrays (the triplets). Therefore the complete planar array pattern is
attained by simply multiplying Eh and Ev of (5–13) and (5–14); namely
π d sin θ sin φ     π dv cos θ  

sin  N h h sin N
  λ    v  λ  
E (θ , φ ) = Eh(θ , φ ) × Ev(θ , φ ) = (5–15)
π d sin θ sin φ   π dv cos θ 
N h N v sin  h  sin  
 λ λ
Equation (5–13) gives the pattern versus q and f of a planar broadside array of equal
amplitude, equal phase, and uniformly spaced isotropic point sources. The subscripts h
and v are used with N and d to denote the number of elements and their spacing in both
the horizontal and vertical dimensions. Here the factor kn of (5–7) is equal to 1/NhNv. It
is to be noted that use of the principle of multiplication for (5–15) is possible because
of the uniformity of the array, that is, the array description in the x direction is indepen-
dent of position along the y-axis; and conversely, the array description in the y direction
is independent of position along the x-axis. This independence between the x and y direc-
tions causes the planar array mathematics to be separable in x and y. However, in general,
complete patterns of planar arrays cannot be attained from the product of principal plane
patterns (see sec. 5.7.3).
There is a further step required, because the usual planar array employs half-wave
dipoles or elements of another type, rather than isotropic elements. Then, the complete
pattern is obtained by multiplying (5–15) by the relative pattern of the element. If dipole
elements are used and are placed with their axes vertical (i.e., lying along the z-axis),
their patterns are given by equation (4–11) (only the part in the square brackets is needed
for the relative pattern). If instead, dipole elements are horizontal (parallel to the y-axis),
the relative pattern (corresponding to (4–11) for a vertical half-wave dipole) is
π
cos  sin θ sin φ 
2 
E (θ , φ ) = (5–16)
1 − sin 2 θ sin 2 φ
(This additional multiplication will not be performed explicitly here because the principle
has already been adequately illustrated.)
Planar and Volume Arrays with Uniform Aperture Distribution 193
5.4.3. Volume Arrays and Plane Reflectors
Just as a planar array was formed by stacking linear arrays in parallel rows, a volume
array can be formed by combining planar arrays. As noted, a single broadside planar
array has a bidirectional pattern. If two such arrays are placed parallel to each other a
distance d apart and phased so that the corresponding array of the two isotropic elements
would be a unidirectional endfire array in accordance with (5–12), the resulting complete
pattern will be unidirectional. Ordinarily, two planes of elements will suffice. The
complete pattern is then obtained by further application of the principle of pattern
multiplication.
Still another method of obtaining a unidirectional beam is to place a plane reflecting
surface parallel to a single planar array, with a suitable spacing between the array and
the reflector. The resulting pattern may be calculated by considering the actual array in
combination with its image in the reflector, according to the image principle described
in sec. 1.2. The spacing between the planar array and its image will be twice the spacing
of the array and the reflector, and the phases will be opposite, that is, 180 degrees apart.
Thus the array plus its image are in effect a volume array, two elements deep.
The reflector method has advantages over the actual volume-array method, in that
since there are actually only half as many dipoles the feed system is less complicated.
Also the reflector effectively prevents back radiation except for a small amount due to
diffraction around the edges of the reflector, provided that the reflector is either a solid
sheet of metal or a fine-mesh screen. However, a fairly coarse mesh or even a parallel-
wire grating (wires parallel to the antenna polarization) gives satisfactory results, with
much-reduced weight and wind resistance, if the spacing between reflector wires in the
direction perpendicular to the polarization is a small fraction of a wavelength. The reflect-
ing surface should extend some distance beyond the edge elements of the array—usually
a distance comparable to the spacing between elements.
The dipole-to-reflector spacing, within certain limits, is not critical. The useful range
of spacings is from about 1/16 to 1/4 wavelength. Close spacing gives somewhat greater
gain with an array of few elements, but at the expense of low dipole-feed-point imped-
ance and reduced bandwidth. With large arrays the gain does not depend appreciably on
the spacing, within the permitted range. Too large a spacing, however (greater than half
a wavelength), will “split” the pattern and result in more than one beam. A good com-
promise among the competing factors—directivity, ohmic losses, and bandwidth—is a
spacing of about 1/8 wavelength. (Ohmic losses increase with the closer spacings because
of lowered radiation resistance and higher currents.)
The two-element-deep volume array, consisting of two large-area vertical planar
arrays, may be preferable at the lower frequencies because a reflector of sufficient size
would be much more expensive and would have excessive wind resistance. Such anten-
nas, known as “double curtain” arrays, can be used in the frequency range from about 3
to 30 MHz or higher. Above 30 MHz the use of a reflector instead of a true volume array
becomes increasingly advantageous. A compromise method is the use of parasitic dipoles
as the reflecting device. Planar arrays with plane reflectors, sometimes called “mattress”
or “bedspring” antennas, are much used in the VHF and UHF bands. As radar antennas
194 Arrays
q P (xp, yp, zp) ri P (xp, yp, zp)
R
q
R
y z
f
f y
x
(a) (b)
FIGURE 5–14.
Spherical coordinates with array elements in x-y plane. (5–14a) shows point P at distance R
from origin; (5–14b) shows point P and distance R, and distance r1 from a position xi,yi on the
x-y plane.
they are often mounted on a rotating platform so that the beam “scans” the horizon, typi-
cally at rotation speeds of 1 to 30 rpm, depending on the antenna size. Such reflecting
screens are commonly as large as 30 feet or more with hundreds of elements, and much
larger ones can be used on occasion for special purposes.
5.5. Linear Array Pattern Calculations

Section 5.5 includes general equations for calculating the patterns of array factors, includ-
ing cases where array elements are not excited with equal currents. Patterns for broadside
and endfire arrays are determined, comparisons are made between the patterns for linear
arrays and continuous linear aperture antennas, and the dependence of grating lobes on
element spacing is discussed.
5.5.1. General Equations for Array Patterns

An array factor is the radiation pattern of an array antenna when each of its array ele-
ments is considered to radiate isotropically (IEEE 1993, p. 55). Now assume isotropic
array elements of arbitrary amplitude and phase are located on the x-y plane of Fig. 5–14.
Then, the array factor F(q, f) is the complex sum (phase and amplitude) of the electric
fields in the far zone from each element as follows:
F (θ , φ ) = A0 exp ( − jkδ 0 ) + A1 exp ( − jkδ1 ) + + A( N −1) exp ( − jkδ ( N −1) )
N −1
(5–17)
2π
= ∑ Ai exp  − j δ i 
i=0
 λ 
Linear Array Pattern Calculations 195
where
Ai = ai exp( jai), where ai is the amplitude and ai is the phase of the radiated E-field
of the ith element
k = 2p/l
d i = ri − R
ri = distance from ith element to an observation point in the far zone
R = distance from the reference point on the array to the observation point in the far
zone
N = total number of isotropic elements
An array factor is a complex function of angle. Therefore, a graph of the amplitude of
an array factor F(q, f) is obtained by plotting |F(q, f)|. To obtain a normalized amplitude
pattern of an array factor, one must of course divide |F(q, f)| by the peak amplitude of
|F(q, f)|.
Note that di is the distance between the ith element and an observation point that
exceeds the distance between the origin (x = y = z = 0), the array reference point, and
the observation point, said point being in the far zone. Thus, when including the effects
of path length differences, the relative far-zone phase contribution of the ith element yi
follows:
2π
ψ i = α i −   δ i (5–18)
 λ 
Then, the array factor of (5–17) may be expressed as

N −1 N −1
2π   
F (θ , φ ) = ∑ ai exp jψ i = ∑ ai exp  j  α i −   δi 
λ   
(5–19)
i=0 i=0
From Appendix H, for observations in the far zone, di may be expressed as
δ i = − sin θ ( xi cos φ + yi sin φ ) (5–20)
where
xi , yi is the i th element position within the x -y plane
and
θ , φ are the angles from the origin to the observation point
Recall that F(q, f), as defined, is the pattern calculated under the assumption that each
array element radiates isotropically. Then, the far-field pattern E(q, f) of an array having
identical elements can be calculated by using the principle of multiplication. In reality,
however, element patterns of physically identical elements are never the same when the
196 Arrays
elements are actually placed in an array. However, in principle, if each element pattern
is described by the function f(q, f), the far-field pattern E(q, f) is
E (θ , φ ) = f (θ , φ ) ⋅ F (θ , φ ) (5–21)
It is to be noted the F(q, f) and E(q, f) are generally complex, and therefore their ampli-
tude patterns are created by plotting magnitudes only, that is, |F(q, f)| and |E(q, f)|.
In summary, (5–17) and (5–19) are equal and are general expressions for the array factor
if each array element is an isotrope; and (5–21) is the far-zone E-field array pattern if the
pattern of each element can be expressed as f(q, f). A note of caution: effects of mutual
coupling between elements may cause the “in-place” element patterns (called scan element
pattern, SEP) to differ from their “free space” patterns. Because of this, even for an array
of physically identical elements, the actual array pattern may not equal F(q, f) times the
“free-space” element pattern f(q, f), as indicated by (5–21). Conventionally, however,
(5–21) is used as a “best” estimate until more detailed information is available.
5.5.2. Linear Array Patterns

In this section, the radiation pattern in the x-y plane is investigated for a linear array of
elements located along the y-axis. With these constraints, from Fig. 5–14a it can be seen
that q = 90° and each xi = 0, and from (5–20) one attains di = −yi sin f. Thus, the array
factor of (5–19) becomes
N −1
2π  
F ( 90°, φ ) = ∑ ai exp  j  α i +  λ
 yi sin φ  

(5–22)
i=0
In (5–22) the element amplitudes, phases, and separation distances are arbitrary. Equation
(5–9), previously introduced in sec. 5.2 is also for arrays along the y-axis, but it is appli-
cable only to arrays for which the element amplitudes, phase differences, and physical
separations are all equal.
Equation (5–22) can be further simplified if, between adjacent elements, the separation
distance d and the phase difference a are constant, as is common in practice. Furthermore,
frequently the input current amplitude and phase of an array are symmetrical about the
array center. Then, F(90°,f) with its phase reference at the array center can be expressed
as follows:
( N −1) 2 2π
F ( 90°, φ ) = ∑ an exp  jn  α +   d sin φ   (5–23)
n =−( N −1) 2    λ 
where
an = amplitude of the element located on the y-axis at yn = n · d
a = phase difference between adjacent elements
d = separation distance between elements along the y-axis

N = total number of isotropic elements
As previously mentioned, (5–9) in sec. 5.2 pertains only to arrays having n equal
amplitude isotropic elements, with equal amplitude, equal spacing d, and equal phase
difference a between adjacent elements. For this special case, (5–23) can be further
simplified and becomes (5–9), except for the fact in (5–23) the total number of elements
is designated as N, not n.
5.5.3. Broadside Linear Array Patterns

A broadside array is one in which the main beam direction is normal to the line of the
array. Then, for a linear array located along the y-axis of Fig. 5–14a, f is either 0° or
180°. Then, with equal phase excitation (i.e., a = 0), the pattern from (5–23) has equal-
amplitude maxima at 0° and 180°. Consequently, each maximum equals the sum of the
element amplitude terms an, and the broadside pattern of an array of isotropic radiators
is bidirectional. Broadside arrays are made unidirectional through use a reflecting back
plane or elements with unidirectional element patterns. In further discussions on broad-
side arrays in this book, it is assumed that means have been taken to assure unidirectional
performance.
Figure 5–15 shows patterns for two arrays with the same array factor, each array having
element excitations with equal amplitudes (an = 1) and equal phases (a = 0). In each case,
the element spacing d is l /2 and the number of elements N is 10. Each array pattern is
normalized so that its peak is unity (0 dB) and graphed versus f over ±90°. For one array
the elements are isotropic radiators, and for the other the pattern f(q, f) of each element
pattern is |cos f| for f between ±90°. The |cos f| dependence approximates the pattern of
a simple element. The array factors are obtained by calculating the absolute value of
(5–23), and (5–21) is used to apply pattern multiplication for obtaining the effects of the
|cosf| element patterns. By comparing the two patterns of Figure 5–15, it may be seen that
the |cosf| element patterns do not reduce
Relative amplitude in dB
the magnitude of the major side lobes,

0
but the minor lobes are reduced at angles –10
near 90° from the main beam. Thus, –20
although the major side-lobe to main-lobe –30
–40
ratio is the same for the two arrays, the –50
array with the cosf elements has a slightly –60
–90 –45 0 45 90
higher directivity (greater beam concen- Angle in degrees from beam center
tration). It is also to be noted that (a) side
FIGURE 5–15.
lobes are large (approximately −13 dB)
when all elements are excited with equal Radiation patterns of two ten-element broadside
arrays, with element excitations of equal
amplitude and (b) wide-angle element amplitudes and phases. Heavier pattern for
patterns can affect a reduction in the wide- isotropic element patterns; thinner pattern for
angle sidelobe levels. cos f amplitude element patterns.
198
Arrays
0 5.5.4. Grating Lobes

–10
–20
As mentioned for broadside arrays in sec.
–30
5.2.2, periodically occurring major lobes
–40
exist when the spacing between uniformly
–50 spaced elements is l or greater. These
–60 periodically occurring lobes are called
–1 –0.5 0 0.5 1 grating lobes. As a cautionary note, if a
Sine of angle from beam center
beam is focused to point in a direction
FIGURE 5–16. other than broadside, a grating lobe may
Radiation pattern of a ten-element linear array occur at smaller element spacings. For
of isotropic radiators spaced l /2 apart (heaviest arrays designed for changing the beam-
curve), and a five-element linear of isotropic
radiators spaced l apart (thinner curve).
pointing direction (chapter 8), element
spacing is ordinarily limited to l /2 or
somewhat less, to avoid grating lobes.
Figure 5–16 includes two patterns, one for a ten-element linear array of isotropic radia-
tors spaced l /2 apart (heaviest curve) and the other for a five-element linear array of
isotropic radiators spaced l apart (thinner curve). The curves are graphed as a function
of sin f, Notice that, graphed in this manner, the grating lobes at sin f = ±1 (i.e.,
f = ±90°) are the same magnitude and shape as the major lobe at f = 0 (i.e., sin f = 0).
5.5.5. Comparisons: Broadside Array and Continuous Aperture Antenna

Reflector and lens antennas are discussed in chapter 6. Unlike the discrete excitation of
array apertures at each element location, the illumination of reflector and lens apertures
is continuous. In the limit of very close spacing between elements, array apertures
become continuous. A comparison of the patterns from discrete and continuous apertures
is included here, because knowledge of the patterns from continuous apertures provides
useful insights about patterns from arrays. As will be shown, the patterns of the two
aperture types are similar if: (1) the array element spacing is l /2 or less and (2) the array
amplitudes and phases are matched point-to-point with the continuous aperture distribu-
tions. There are well-known closed-form mathematical relations between continuous
aperture distributions and their resulting patterns. Therefore, because of the similarities
of discrete and continuous aperture patterns, much is known about array patterns based
on previously acquired continuous aperture patterns.
From sec. 6.10, chapter 6, the far-zone amplitude pattern in a principal plane, normal-
ized to unity, from a uniformly illuminated (constant amplitude and phase) continuous
aperture is
E ( u ) = sin u u (5–24)
where
u = (π L λ ) sin φ
and f denotes the angle measured from normal to the aperture of width L. In comparison,
from (5–9), for a broadside linear array of isotropes having equal amplitude and phase
with spacing d, the normalized amplitude pattern is
π d sin φ ⎞ ⎤
sin ⎡⎢ N ⎛
⎣ ⎝ λ ⎠ ⎥⎦ (5–25)
π d sin φ ⎞ ⎤
N sin ⎡⎢⎛
⎣ ⎝ λ ⎠ ⎥⎦
where N is number of elements. With the approximation n times d used as the array
aperture width L, the numerator of (5–25) becomes sin u as in (5–24). Of course, the
array width is actually N − 1, but the difference between N and N − 1 diminishes for
large values of N. Therefore, for a continuous aperture and an array with a large number
of elements, each of the same width, the zeros of E(u) and E(f) occur at practically the
same angle f.
There are also other close similarities between the broadside patterns for closely
spaced arrays and continuous aperture antennas. For example, because sin x ≈ x, if x is
small and in radians, the numerator of (5–25) becomes approximately u at near-broadside
where f is small. Figure 5–17 includes patterns of a uniformly excited broadside array
with spacing d of l /2 and of a continuous aperture, each of the same length. The patterns
are virtually equal at angles near the direction of the main beam. As also apparent from
Fig. 5–17, the peaks in the minor lobes differ increasingly as the angle f becomes closer
to ±90°. For aperture widths of about 5l and larger, the side-lobe peaks at wide angles
(f near ±90°) may differ by several decibels. Even so, at these wide angles the field
strength, and thus the energy, in either pattern is small compared to the energy in the
major lobes. Therefore, since the two major lobe patterns are almost identical, the direc-
tivity of the two antennas of Fig. 5–17 are virtually the same.
For most practical purposes, the approximate equality of broadside antenna
patterns, whether from closely spaced (l /2 or less) elements or continuous apertures, is
retained provided the array amplitudes and phases are matched point-to-point with the
continuous aperture distributions. This
near-equality of patterns exists provided
also that the apertures are of equal length 0

–10
and exceed 5l. –20
–30
–40
–50
5.5.6. Endfire Array Patterns –60
–90 –45 0 45 90
Endfire arrays are discussed in sec. 5.2.3. Angle in degrees from beam center
The present section includes pattern cal- FIGURE 5–17.
culations of endfire arrays having ele-
Radiation pattern of a ten-element linear array
ments aligned along the y-axis (Fig. 5–14). of isotropic radiators spaced l/2 apart (heaviest
Endfire array patterns may have two prin- curve), and a continuous linear aperture of 5l
cipal lobes, in opposing directions along width (thinner curve).
200 Arrays
the array axis. On the other hand, there will be only one major lobe (unidirectional
pattern) if the array element spacing d is less than l /2. An endfire array can provide
higher directivity for a given number of elements than a broadside array. However,
endfire arrays require larger dimensions in the along-beam direction. For an ordinary
(conventional) endfire array, the phase difference between elements is opposite to the
phase delay due to element spacing, thereby causing the radiation of all elements to add
constructively at the center of the beam. Thus a = −2pd/l for an ordinary endfire array,
where a is the phase difference between elements.
The directivity is said to be extraordinary if it satisfies the Hansen and Woodyard
(H-W) condition, and then its directivity is larger than that of the ordinary endfire array.
For an H-W array, there are N equal amplitude isotropic elements and the phase a
between elements is
2π d π 
α =  +  (5–26)
 λ N
The patterns of ordinary and extraordinary endfire arrays are now compared.
Assume the elements are equally spaced isotropic elements of equal amplitude and
located on the y-axis of Fig. 5–14a. Now consider the pattern of an endfire linear array
in the x-y plane, for which q = 90°. Then, with N = 10, each amplitude an = 1, and the
element spacing d of l /4, (5–23) becomes
4.5
π
F ( 90°, φ ) = ∑ exp  jn  α +   sin φ   (5–27)
n =−4.5    2 
Now assume that the beams of the ordinary endfire array and of the one satisfying the
Hansen and Woodyard condition are pointed in direction f = 90°. The phase a between
elements for an ordinary endfire array equals the phase delay due to propagation between
elements; thus, a = −p/2 because d = l /4. From (5–26), for the extraordinary (H-W)
condition, with d = l /4 and n = 10 ele-
Normalized array factor
1 ments, a = −0.6p.
0.8 The normalized patterns for the two
0.6 unidirectional endfire arrays having 10
0.4 elements and l /4 element spacing, using
0.2 (5–27), are included in Fig. 5–18. Notice
0 that the Hansen-Woodyard array has a
–90 –45 0 45 90 135 180 225 270
Angle in degrees from array axis substantially narrower beamwidth, and
FIGURE 5–18. larger side lobes. On the other hand, the
shape of the major lobe of the ordinary
Normalized patterns for ten-element endfire
arrays with d = l/4 element spacing: ordinary endfire array is flatter at its peak.
endfire (bold line) and endfire satisfying From (5–27), the values of the array
Hansen-Woodyard condition (fine line). factors along the y-axis are
Array Tapering for Side-Lobe Reduction 201
4.5
π π
F ( 90°, 90°) = ∑ exp  jn  − +   = 10 ordinary enddfire array (5–28)
n =−4.5   2 2
and
4.5
π
F ( 90°, 90°) = ∑ exp  jn  −0.6π +   = 6.392 extraorrdinary endire array (5–29)
n =−4.5   2
Recall that electric field strength is proportional to array factor F. Although (5–29) is
smaller than (5–28), the directivity of the Hansen-Woodyard array is larger. This is
because the Hansen-Woodyard pattern (and thus energy) is more closely confined to the
y-axis. Additionally, F(90°, 90°) is largest in (5–28), because the radiation from all ele-
ments adds in-phase but not so for (5–29). Kraus and Marhefka (2002, pp. 115) report
the directivities obtained by integrating the patterns, including the minor lobes, for the
two ten-isotrope arrays. They give the approximate directivities for the ordinary and
Hansen-Woodyard arrays to be approximately 11 and 19, respectively. For these directivi-
ties, the Hansen-Woodyard directivity is the largest by the factor 1.73.
A focused broadside array of ten isotropes has in theory a directivity of 10. Thus, from
the directivities given above and the same number of elements, lossless endfire arrays
can provide greater gain, as well as larger directivity, than a broadside array.
5.6. Array Tapering for Side-Lobe Reduction

5.6.1. Introduction
A principal advantage of a directional antenna is the gain that it provides in the desired
direction, but sometimes it is equally important to minimize radiation toward or reception
from an undesired direction. The matter is especially serious in congested areas where
interference from other systems is a problem or for radar when signals from objects
located in minor-lobe directions may obscure weak target signals of interest.
The minor lobes immediately adjacent to the main lobe are side lobes. Since these are
usually the largest of the minor lobes, and hence of greatest importance, the words “side
lobes” are often used when the more inclusive term “minor lobes” would be more appro-
priate. (Note: confusion is occasionally also caused by the fact that a “low side-lobe
level” means a high main-beam to largest minor lobe ratio. That is, the smaller the side
lobes are, the larger will be the number, whether expressed as a power-density ratio or
as a decibel value to describe their level relative to the main lobe.)
A technique called “tapering” is used to reduce side lobes, and it refers to providing
a variation in the element excitations (e.g., element currents) in a linear array. Namely,
minor lobes are reduced if the center elements of the array radiate more strongly than
the end elements, the strengths being gradually tapered from center to end according to
some prescription.
When array elements are excited with equal amplitude and phase to maximize
antenna gain, the ratio of main-beam power density to the first-side-lobe power density
202 Arrays
is approximately 20, or 13 dB. A side-lobe level −13 dB below the major lobe peak is
often unacceptably high. It is common to design high-gain narrow-beam antennas for
side-lobe levels down at least 30 dB, especially for radar use. A −30 dB level is consid-
ered good, and −40 dB excellent. A level of 50 dB below the main lobe is very difficult
to achieve, but even somewhat better side-lobe levels have been obtained. Planar arrays
are typically tapered by different amounts in the horizontal and vertical dimensions,
because side-lobe requirements versus principal plane differ depending on an antenna’s
application.
5.6.2. Amplitude Tapering of Broadside Arrays: General

This section compares the patterns for five-element arrays having uniform, binomial,
optimum, and edge distributions; and its content follows Kraus (1988, pp. 159–62).
Broadside arrays having uniform distributions (all element amplitudes and phases equal)
ordinarily provide maximum directivity, but their minor lobes are relatively large. A five-
element uniform amplitude distribution provides a 23° half-power beamwidth and peak
side lobes of −12.4 dB relative to the peak of the main beam. The binomial distribution,
although generally impractical, has no side lobes, relatively large beamwidth and small
directivity. Moreover, the theoretical absence of side lobes is not likely to be achieved
in practice, because of unavoidable imperfections in the antenna geometry, phasing, and
feed system. A five-element binomial distribution provides a 3 dB beamwidth of 31°.
The relative amplitudes for the binomial distribution are in accordance with Pascal’s tri-
angle, as shown in Table 5-1 for N elements.
The distribution having the narrowest 3 dB beamwidth and the strongest side lobes is
the so-called edge distribution. It contains only the two outer-edge elements, and no ele-
ments in between. Thus the five-element, l /2 spacing edge distribution array is identical
to a two-element array having 2l spacing. Such an array has many side lobes, and each
has the same amplitude as the center lobe. The 3-dB half-power beamwidth for this array
is 15°.
The names Tschebyscheff, Tchebyscheff, Tchebyschev, Chebyshev, and Chebishev
appear in the technical literature. All of these names refer to the one person who devel-
oped the well-known polynomials, for whom they are named. Chebyshev is used here,
following the IEEE standard (IEEE 1993).
TABLE 5-1 Relative Amplitudes of Binomial Distributions

for Three- to Eight-Element Arrays
N Relative amplitudes
3 1 2 1
4 1 3 3 1
5 1 4
1 6 4
6 1 5
10 10 5 1
7 1 6 15 20 15 6 1
8 1 7 21 35 35 21 7 1
The Dolph-Chebyschev (D-C) distribution is considered the optimum aperture distri-

bution, but its use for more than about eight elements is impractical. In theory, the Dolph-
Chebyschev distribution optimizes the relationship between beamwidth and side-lobe
level. Specifically, if the maximum side-lobe level is specified, the beamwidth between
the first nulls is minimum. A five-element Dolph-Chebyschev distribution designed for
−20 dB side lobes has a 3 dB beamwidth of 27°.
Figure 5–19 includes the patterns for the distributions discussed above. By considering
this figure, certain general conclusions can be made, namely: (a) low side lobes can be
obtained if the amplitude tapers down to small values out at the aperture edges, and (b)
an abrupt change in amplitude versus aperture position creates large side lobes. In addi-
tion to the Dolph-Chebyschev amplitude distributions for beam shaping and side-lobe
reduction, there is a wide variety of distributions that can be used; and these depend on
methods available for implementation, desired beamwidth, and side-lobe levels. As will
be discussed later in sec. 5.6, a method devised by T. T. Taylor probably achieves the
best balance among all three factors: side lobes, beamwidth, and directivity.
5.6.3. Dolph-Chebyshev Synthesis

Array design by Dolph-Chebyshev synthesis produces the narrowest beamwidth subject
to a given peak side-lobe level. In theory, all side lobes of a D-C array are of the same
level. Dolph-Chebyshev synthesis is a mathematical procedure that equates the currents
in an array to a Chebyshev polynomial. The required element current distributions as a
function of number (three up to eight) of elements that provide calculated side lobes (in
10 dB steps) as low as −40 dB are available (see, e.g., Ma 1993). A sample of tabulated
values appears below, as an indication of the general nature of D-C distributions. This
table is for an eight-element linear array designed for a −30 dB side-lobe level. The array
elements are numbered 1 through 8 progressively from one end of the array to the other,
numbers 4 and 5 being the two center elements.
Element Numbers Relative Currents

1 and 8 0.26
2 and 7 0.52
3 and 6 0.81
4 and 5 1.00
The current distribution is symmetrical about the center of the array; hence the ele-
ments can be grouped in pairs as shown. The current values are normalized to the value
unity at the center of the antenna. As the table shows, for this −30 dB side-lobe level,
the end-element currents are only 26 percent of the center-element current. For half-
wavelength element spacing, the 3-dB beam width of this array is 16.4 degrees and
directivity is 6.7 (8.3 dB).
Unfortunately, antennas having a large number of elements and designed for low side
lobes using D-C aperture distributions are impracticable. For example, since the side-lobe
levels are constant (independent of direction), directivity and thus gain decrease with
204 Arrays
23° 31° 27°
Uniform Binomial Optimum
1 1 1 1 1 1 4 6 4 1 1 1.9 1
(a) (b) 1.6 1.6
(c)
15°
Edge
1 0 0 0 1
(d)
FIGURE 5–19.
Normalized far-field patterns of five-element broadside arrays. Only the upper half of each
pattern is shown. All elements have same phase, and the relative amplitudes are included below
each figure. From John D. Kraus, Antennas, 2nd ed., ©1988, p. 160, with permission of The
McGraw-Hill Companies.
increase in array size. This is because, as the principal beamwidth is reduced, a larger
fraction of the radiated power is within the side lobes. Also, for increasingly larger arrays,
the required array element excitation becomes nonmonotonic, requiring high peak cur-
rents at the end elements that causes large I 2R losses. This feature reduces available gain.
Another problem with low side-lobe D-C designs is that the element currents must be
controlled to an impracticably high accuracy, which contributes to them being unrealiz-
able. Therefore, aside from the application to arrays with a small number of elements,
D-C patterns are important for identifying the narrowest beamwidth theoretically avail-
able for a given peak side-lobe level. A simplified method is described in the next section
for calculating theoretical, yet unattainable, D-C patterns for arrays having a large
number of elements.
5.6.4. Simplified Calculations for Dolph-Chebyshev Patterns

Van der Mass (1954) discovered a relatively easy way to obtain the theoretical D-C pat-
terns for arrays with a large number of elements. This was done by replacing the array
with a continuous line source formed by letting the number of array elements approach
infinity. Then, line-source D-C patterns can be obtained without the tedious calculations
of D-C synthesis, and the beamwidths are almost as narrow as had the calculations been
with discrete elements.
The “optimum” line-source patterns of Van der Mass, like other D-C patterns, has side
lobes of constant amplitude (Mailloux 1994, pp. 127–28), and are therefore physically
unrealizable. These far-field E-field patterns are expressed as
E ( u ) = cos u 2 − (π A)
2
(5–30)
where u = (pL/l)sin q
A = (1/p)arccos h(R)
R = ratio of electric field strength of the peak side lobes to main beam, and
where arcos h is the inverse hyperbolic cosine, l is wavelength, and q is the observation
angle measured from the normal to the aperture.
The 3 dB beamwidth for a D-C pattern expressed in radians is
BWHPCheb = (λ L ) β 0
where (5–31)
(
0.5
β0 = (2 π ) ⎡⎣(arccos h( R ))2 − arccos h(0.70 R )2 ⎤⎦
As an example of the optimum yet unrealizable 3 dB beamwidth for a D-C pattern in

accordance with equation (5–31), we now calculate b0 for R = 100 (i.e., 40 dB) side
lobes. Then
206 Arrays
0.5
β0 = ( 2 π ) ⎡⎣( 5.298)2 − ( 4.952 )2 ⎤⎦ = 1.20
Therefore, the “optimum” 3 dB beamwidth for constant 40 dB peak side lobes, although
unrealizable, expressed in radians, is
BWHPCheb = 1.20 (λ L ) (5–32)
The broadside beamwidths for continuous line apertures are discussed in sec. 6.10.
The narrowest broadside continuous line-source beamwidth is one that has an aperture
distribution of constant amplitude and equal phase (uniform distribution). The 3 dB
beamwidth for this basic aperture, expressed in degrees is
BWHPu = 50.8 (λ L ) (5–33)
with l /L dimensionless. When expressed in radians, (5–33) becomes
BWHPu = 0.887 (λ L ) (5–34)
Therefore (5–31) and (5–34) can be used to provide a comparison between the most
efficient distribution (which gives −13 dB peak side lobes) and the D-C pattern (which
gives the narrowest major lobe for a specified constant side-lobe level). However, as
already stated, D-C patterns are not always practical or realizable.
Figure 5–20 is included to compare the patterns for a uniform aperture distribution
and a D-C distribution with 26 dB side lobes. Each aperture is 2 m wide and the fre-
quency is 3 GHz. Although the beamwidth for the uniform aperture is slightly narrower
than provided by the D-C distribution, the much larger side lobes created by the uniform
distribution are apparent.
0
–10
5.6.5. The Taylor Distribution
Decibels
–20
The Taylor aperture distribution is often
–30 used where narrow broadside beams and
low side lobes are needed. Taylor (1955)
–40
–45 –30 –15 0 15 30 45 analyzed the deficiencies of Dolph-
Angle from Broadside in Degrees Chebyshev arrays and formulated a com-
bined radiation pattern and aperture
Constant Amplitude and Phase
Dolph-Chebychev Distribution distribution that provides narrow beam-
widths, and yet has good efficiency for
FIGURE 5–20.
large arrays. His analysis began with the
Two patterns, one for a uniform amplitude line-source D-C patterns given by Fig.
distribution and the other for a Dolph-
Chebychev distribution with 26 dB side lobes. 5–20 above. In so doing Taylor developed
Each aperture is 2m wide and the frequency is a method for avoiding the D-C directivity
3 GHz. (and gain) loss problem caused by no
decay in the side lobes. This was accomplished by approximating, arbitrarily closely, the
D-C pattern with a physically realizing pattern. He retained an approximate constant
side-lobe pattern close to the main beam, but let the wide-angle side lobes decay in
amplitude like the side lobes from a uniform continuous line source.
Taylor’s method is basically designed to apply to continuous aperture distributions
rather than to an array of discrete elements, but it can be applied to linear arrays of a
large enough number of elements (about twenty or more). This is the case where the
Dolph-Chebyshev distribution results in poor directivity. Taylor’s method gives improved
directivity for a given peak minor-lobe level, because it allows the far-out side lobes to
diminish in amplitude.
A parameter n, called nbar, defines the number of nondecaying side lobes on each
side of the main lobe. Thus as n becomes large, the Taylor pattern (and aperture distribu-
tion) approaches the Dolph-Chebyshev pattern (and distribution). Then, ordinarily, one
uses the largest n that still provides a monotonically declining amplitude distribution out
to each end of the aperture. In doing so, the beamwidths obtained are almost as narrow
as with the D-C pattern—and the resulting aperture efficiency is excellent.
The Taylor aperture distribution provides a beamwidth wider than that of a D-C pattern
by an approximate factor s. From Mailloux (1994 p. 129), the 3 dB beamwidth for a
Taylor distributed aperture can be expressed as
BWHPTaylor = σ ( BWHPCheb ) = σ (λ L ) β0
where (5–35)
−0.5
σ = n ⎡⎣ A2 + ( n − 1 2 )2 ⎤⎦
with
n = number of equal-amplitude side lobes on one side of the main beam
L = total length of aperture
A = (1/p)arccos h(R)
R = design side lobe voltage ratio
Now to compare the beamwidths of the unrealizable D-C pattern of (5–30) and a
realizable Taylor pattern, consider an array with n = 4 and 40 dB peak side lobes. Then,
from the definitions given above
A = (1 π ) arccos h(100 ) = 1.686 (5–36)
and
−0.5
σ = 4 ⎡⎣(1.686 )2 + ( 4 − 0.5)2 ⎤⎦ = 1.03 (5–37)
Thus for n = 4 and 40 dB peak side lobes, the 3-dB width of the main beam of the
Taylor pattern is only 3 percent wider than a D-C pattern having 40 dB side lobes.
208 Arrays
Recall from (5–32) that the beamwidth for a D-C pattern with 40 dB side lobes is
1.20 (l /L).
As already mentioned, in addition to the Taylor 3-dB beamwidth being broader than
that of the Chebyshev pattern, in principle the n = 4 Taylor pattern contains four equal-
amplitude side lobes on either side of the major lobe. Also, in principle, the remaining
side lobes have widths and amplitudes like those from a continuous uniform aperture
distribution.
Figure 5–20 is useful for roughly illustrating the formation of a Taylor pattern by a
merger of two patterns: one from a uniform aperture distribution and the other from the
D-C distribution with 26 dB side lobes. Each aperture is 2m wide and operates at 3 GHz.
Ideally, the Taylor side lobes predominate in close to the major lobe and out to angles
where the side lobes of the uniform amplitude pattern are the smallest. Then, the uniform
amplitude pattern dominates a Taylor pattern. From Fig. 5–20, a reasonable guess is that
a resulting Taylor pattern might have four or five side lobes (n = 4 or 5) at the −26 dB
level on either side of the major lobe. However, care is required in a Taylor design when
selecting the nbar and side-lobe levels, because calculated Taylor patterns are seldom as
precise as theory predicts. In addition, the choice of n is not arbitrary, because increasing
it retains more of the side lobes at the design side-lobe level and this makes the calculated
Taylor closer to a D-C pattern. In other words, increasing n eventually leads to aperture
illuminations that do not decrease monotonically down to the aperture edges (Mailloux
1994, p. 133). Thus, the act of increasing n too much can destroy the usefulness of the
aperture distribution.
In summary, the Taylor distribution is widely used, because it is a realistic way of
providing the narrowest beamwidth from a broadside array, given a specific peak side-
lobe level and an aperture width. Neither the theory nor design for Taylor distributions
is simple. However, the necessary equations for calculating aperture distributions and
patterns are available (see, e.g., Mailloux 1994, pp. 128–29; Balanis 2005, pp. 408–10).
Additionally, tables are available that give required array element amplitudes, for a
variety of n values, for Taylor arrays having side-lobe levels for 20 to 40 dB, inclusive,
in 5 dB steps (Reference Data for Radio Engineers 1979, pp. 27–29 through 27–31).
5.6.6. Other Low-Side-Lobe Array Distributions

A tapering effect is also achieved by “thinning,” or space-tapering, that is, by removing
some of the elements. Suppose that initially an array has elements that are uniformly
spaced, and fed with equal amplitudes. Then, if some elements are removed, with more
elements removed near the edges than near the center of the array, the average density
of elements will exhibit a taper from center to edge. This taper will have a side-
lobe-suppressing effect, somewhat like that of an amplitude taper in an unthinned (filled)
array. The pattern of element removal may be random, rather than regular, to avoid the
creation of the grating lobes that occur when the spacing of array elements is large.
Clearly, the gain is less than that of the filled array of the same size and taper. The primary
application of the thinned space-tapered technique is with arrays that have many
elements.
Planar Arrays: Patterns, Directivity, and Gain 209
As discussed previously, the patterns of arrays and continuous apertures are similar if
(1) the array element spacing is l/2 or less and (2) the array amplitudes and phases are
matched point-to-point with the continuous aperture distributions. Consequently, from
previously developed knowledge about continuous apertures there are simple amplitude
distributions that are known to provide low side-lobe levels. For example, a cosine-
shaped amplitude distribution with its peak at the aperture center will provide a reason-
ably efficient, low side-lobe pattern. Digital filter algorithms (Harris 1978) also provide
useful aperture amplitude distributions, the Hamming function being an example. Bodnar
(2007; pp. 55–16 through 55–19) provides graphs that show beamwidth versus side-lobe
level and aperture efficiency versus side-lobe level, for the better-known uniform aperture
distributions. These graphs are useful for estimating performance available from array
antennas.
5.7. Planar Arrays: Patterns, Directivity, and Gain

The subject of broadside planar array patterns is introduced in sec. 5.4. There it is
assumed the array elements were located in the y-z plane, and the pattern equations
considered are only for the simplest case of elements having equal amplitude and equal
phase. Sections 5.7.1 through 5.7.5 that follow discuss general equations for calculating
patterns of planar arrays, as well as simpler special cases. Finally, sec. 5.7.6 addresses
the beamwidths and gain of practical planar arrays.
5.7.1. Introduction
A planar array pattern can be calculated by summing the vector contributions of each
array element at every direction and at a distance where far zone approximations are
valid. Then, for an array located in the x-y plane, the array factor is
F (θ , φ ) = ∑ ai( x, y ) exp jψ i (5–38)

i
where
q, f = angles to a far-zone point from the aperture origin
ai = amplitude of element i
2π
yi = α i ( x , y ) − δ i ( x, y ), the phase contribution of element i at the observation point
λ
ai = phase of element i
di = −sin q(xi cos f + yi sin f), which is the difference in distances to an observation
point in the far zone (see Appendix H), from (a) a point xi, yi on the aperture
and from (b) the aperture origin
The reader will recognize that (5–38) is a summation that includes each ai, ai, and di
over all x and y values. Consequently, the calculation of F(q, f) is, in general, complicated
because di (contained in yi) is a function of both q and f. Furthermore, for calculations
210 Arrays
z of patterns in the radiating near field, the

di above must be replaced by equation
(H-4), Appendix H.
P
q
5.7.2. Patterns Versus q and f:
Equal Element Spacings
dx
y Pattern calculations versus q and f are
made in this section using Fig. 5–21,
which provides a systematic method for
dy
f describing the x and y coordinates of a
planar array. Since the element spacings
x
dx and dy are constants, the x and y coor-
dinates can be expressed as xm = mdx and
FIGURE 5–21. yn = ndy. Similarly, one attains
A Planar array in the x-y plane. Element
spacings are dx and dy, respectively, along the 2π
x and y axes. ψ mn = α mn − δ mn (5–39)
λ
and
δ mn = − sin θ ( md x cos φ + nd y sin φ ) (5–40)
Then, by letting Amn = amn exp jamn, where amn is an element amplitude and amn is its
phase, a general expression for the planar array factor is as follows:
F (θ , φ ) = ∑ ∑ Amn exp [ j ( ms x + ns y )] (5–41)

m n
with
2π
sx = d x sin θ cos φ (5–42)
λ
and
2π
sy = d y sin θ sin φ (5–43)
λ
where
dx and dy are the element spacings along the x and y axes
m is an integer such that mdx denotes a coordinate of an array element on x-axis
n is an integer such that ndy denotes a coordinate of an array element on y-axis
5.7.3. Separable Aperture Distributions

Planar arrays that have a large number of elements are usually constructed so that
Amn = AmAn = am exp( jmax) × an exp( jnay). Here am and an are element amplitudes, where
am is a function of x only and an a function of y only. In addition, ax and ay are fixed phase
differences between adjacent elements in the x and y dimensions, respectively. Expressing
Amn in this manner makes it mathematically separable between the x and y coordinates.
Methods by which the element excitations are made separable are as follows:
(a) by stacking above one another identical x-axis line arrays (each with element
excitation Am), and exciting the separate x-axis arrays in accordance with
An, or
(b) by placing alongside one another identical y-axis line arrays (each with
element excitation An), and exciting the separate y-axis arrays in accordance
with Am
Then, with Amn expressed as the product of Am (a function of x only) and An
(a function of y only), equation (5–41) can be expressed as
F (θ , φ ) = ∑ am exp jm (α x + sx ) × ∑ an exp jn (α y + sy ) (5–44)

m n
Thus, for separable array distributions, we see from (5–44) that a planar array factor is
the product of two terms. On careful examination, it may be seen that the first and second
terms are the array factors for line arrays along the x and y axes, respectively. In other
words,
F (θ , φ ) = Fx (θ , φ ) × Fy(θ , φ ) (5–45)
Thus, Fx(q, f) and Fy (q, f) are array factors for line sources along the x and y axes, as
follow
2π
Fx (θ , φ ) = ∑ am exp jm (α x + sx ) = ∑ am exp ⎡⎢ jm ⎛⎜ α x + d x sin θ cos φ ⎞⎟ ⎥⎤ (5–46)
m m ⎣ ⎝ λ ⎠⎦
2π
Fy(θ , φ ) = ∑ an exp jn (α y + s y ) = ∑ an exp ⎡⎢ jn ⎛⎜ α y + d y sin θ sin φ ⎞⎟ ⎥⎤ (5–47)
n n ⎣ ⎝ λ ⎠⎦
The mathematics of complete patterns are complex, partly because sx and sy are both
functions of q and f. Thus, the complexity of array factors exists, even for arrays having
element excitations that are separable in the x and y dimensions.
For principal plane patterns, the equations for patterns with separable distributions are
considerably less complex, especially if the phase excitations of all elements are equal.
By setting f = 0° and 90° in (5–46) and (5–47), and with ax = ay = 0, (5–45) for the
principal plane patterns becomes
212 Arrays
2π
F (θ , φ = 0°) = ∑ am exp ⎡⎢ jm ⎛⎜ d x sin θ ⎞⎟ ⎤⎥ × ∑ an ( x-z plane ) (5–48)
m ⎣ ⎝ λ ⎠⎦ n
2π
F (θ , φ = 90°) = ∑ an exp ⎡⎢ jn ⎛⎜ d y sin θ ⎞⎟ ⎤⎥ × ∑ am ( y-z plane ) (5–49)
n ⎣ ⎝ λ ⎠⎦ m
In (5–48) and (5–49), notice the following:
2π
(1) ∑ am exp ⎡⎢⎣ jm ⎛⎜⎝ d x sin θ ⎞⎟ ⎤⎥ is a principal plane pattern for a linear array
m λ ⎠⎦
located along the x dimension
2π
(2) ∑ an exp ⎡⎢ jn ⎛⎜ d y sin θ ⎞⎟ ⎤⎥ is a principal plane pattern for a linear array
n ⎣ ⎝ λ ⎠⎦
located along the y dimension
(3) ∑ am and ∑ an are simply constant multiplicative factors.
m n
Items (1) through (3) above relate to the principal plane patterns for an array with
separable amplitude distribution and ones for which each element has the same phase.
From these items it may be seen that the equations for the principal plane patterns are
much simpler than those of the three-dimensionl patterns. In fact, each principal plane
pattern is the same as that of a linear array.
5.7.4. Patterns Versus q and f: Equal Amplitude, Separable Arrays

Still simpler equations are possible if the arrays have elements of equal amplitude, and
the phases along the x- and y-dimensions are separable. The simplest case is, of course,
where both the element amplitudes and the element phases are equal. By using (5–46)
and letting the amplitude of each array element am be unity, the linear array factor
Fx(q, f) along the x-axis becomes
Fx (θ , φ ) = ∑ exp jm (α s + s x ) = ∑ exp jmψ x (5–50)

m m
where yx = as + sx.
Equation (5–50) is a geometric progression. Its sum is complex (amplitude and phase)
and can be expressed in closed form. This sum, for a total of M identical elements along
the x axis, with the phase reference taken at the center of the line array, and with the
amplitude normalized for a maximum of unity, is derived by Kraus (1988, pp. 138–40),
and is as follows
sin ( Mψ x 2 )
Fx (θ , φ ) = (5–51)
M sin (ψ x 2 )
Then, because Fx and Fy are both geometric progressions of the same general form,
and with N being the total number of elements along the y axis,
sin ( Nψ y 2 )
Fy(θ , φ ) = (5–52)
N sin (ψ y 2 )
Then, according to (5–45), and with all element amplitudes am and an being unity,
F(q, f) for a planar array with separable phase distributions can be expressed as follows:
sin ( Mψ x 2 ) sin ( Nψ y 2 )
F (θ , φ ) = Fx × Fy = × (5–53)
M sin (ψ x 2 ) N sin (ψ y 2 )
where
2π
ψ x = α x + s x = α x + ⎛⎜ d x sin α cos φ ⎞⎟ (5–54)
⎝ λ ⎠
and
2π
ψ y = α y + sy = α y + ⎛⎜ d y sin θ sin φ ⎞⎟ (5–55)
⎝ λ ⎠
The simplest equation occurs for the case where the phases as well as the amplitudes
are equal, which results if each am and an in equations (5–54) and (5–55) is zero. This
simplest condition is the one addressed by (5–13), sec. 5.4.2, but it applies to the case
of equal element amplitudes and phases when the array elements are located in the y-z
plane. Instead, the various equations in secs. 5.7.1 through 5.7.4 are applicable when the
array elements are in the x-y plane.
As seen above, the equations for far-field planar array patterns are not mathematically
simple, even for simpler ones like (5–53) above. The creation of patterns is discussed
further in chapter 6, where example patterns for continuous aperture antennas are included.
Papers on calculations of patterns versus q and f that use Matlab software include
Bregains, Ares, and Moreno (2004) and Sevgi and Uluişik (2005). SM 5.0 included in
the accompanying website gives equations for, and examples of, 3-D plots of patterns
versus q and f using Mathcad software.
5.7.5. Summary of Planar Array Patterns

This section summarizes equations for patterns of planar arrays that are given in secs.
5.7.1 through 5.7.4, where the planar arrays are located in the x-y plane. Previously, in
sec. 5.4, the broadside patterns of planar arrays were discussed, where the array elements
are located in the y-z plane, and the only patterns considered are those with elements
having equal amplitudes and equal phases.
214 Arrays
Equation (5–38) is a general equation for the E-field radiation pattern for a planar
array, where the element amplitudes, phases, and spacing are completely arbitrary. This
equation is written for the radiating far-field; but, as indicated in the text, it can be readily
changed for applicability to the radiating near-field patterns.
Equation (5–41) is a general equation for patterns versus q and f, but it requires there
be equal spacing between elements along the x dimension and along the y dimension.
However, the spacings along the x direction need not be equal to the spacing along the
y direction. The amplitudes and phases of the elements may be arbitrary. Although the
equation is complex, it has wide applicability because, for practical reasons, multielement
arrays are usually constructed with equal spacing between elements.
In addition to having equal spacing between elements, multielement arrays are often
constructed so that the element amplitudes and phases are mathematically separable. This
means that the variations in amplitude and phase along the x dimension are independent
of the variations in amplitude and phase along the y dimension. Equation (5–44) describes
the pattern when the element excitations are separable, but otherwise the amplitudes and
phases of the elements are arbitrary. The requirement of (5–41) for equal spacings along
the x and y dimensions is also applicable to (5–44).
From (5–44) it can be seen that, for separable aperture distributions, the equation for
three-dimensional patterns reduces to the product of two line sources. Equations (5–46)
and (5–47) are the array factors for these line sources, one that has amplitude and phase
variations along the x dimension, and the other has amplitude and phase variations along
the y dimension.
For principal plane patterns, the equations for apertures with separable distributions
are considerably less complex, especially if the phase excitations of all elements are
equal. As may be seen, (5–48) and (5–49) for these principal plane patterns for planar
arrays are of the same form as principal plane patterns for linear arrays.
Equation (5–53) describes the pattern if an array has elements of equal amplitude, and
if the phases along the x- and y-dimensions are separable. The equation is simplified even
further if both the element amplitudes and the element phases are equal. Previously, in
sec. 5.4.2, equation (5–13) was introduced for the simplest case when the amplitudes and
phases of all elements are equal. For (5–13), the array elements are located in the y-z
plane, instead of in the x-y plane for which the equations of secs. 5.7.1 through 5.7.4
apply.
As previously discussed, the array factor F(q, f) is the array pattern calculated under
the assumption that each array element radiates isotropically. In reality, element patterns
are never isotopic, but they may be described by a function f(q, f). If all element patterns
were identical in an actual array antenna, the far-field pattern E(q, f), because of the
principle of multiplication (sec. 5.1.6), would be
E (θ , φ ) = f (θ , φ ) ⋅ F (θ , φ ) (5–56)
Mutual coupling between elements may limit the validity of (5–56) (see secs. 5.1.4
through 5.1.7, and 5.9). Often, however, when (5–56) is normalized to have a maximum
of unity, it provides a practical estimate for actual array patterns. F(q, f) and E(q, f) are
generally complex, and therefore their E-field amplitude patterns are created by plotting
their absolute magnitudes.
5.7.6. Beamwidth and Gain of Broadside Arrays

In principle, beamwidths and directivity can be calculated when a complete, relative
pattern is available, as was illustrated for the short dipole in sec. 4.1. However, whether
such a pattern is available as an equation or is compiled from measured data, the required
calculations for directivity are not simple. Therefore, and because of interdependencies
of directivity on principal-plane beamwidths, approximation formulas for beamwidth and
directivity are useful. Such formulas are available for broadside arrays, which are the
most common type of arrays in narrow-beam high-directivity applications.
The simplest example is a uniform broadside linear array of N isotropic elements
spaced a half-wavelength apart. For values of N larger than 3, the beamwidth is approxi-
mately 102/N degrees. (For N = 3, the beamwidth is 36 degrees, and for N = 2 it is 60
degrees.) The directivity of such an array is equal to N. Since isotropic elements do not
exist, these results are somewhat academic. However, the beamwidth formula can also
be applied to an array of parallel dipoles in the plane perpendicular to the dipole axes.
(This is the type of linear array pictured in Fig. 5–5.) Furthermore, the beamwidth and
gain results for the array of isotropes are directly applicable to an array of many collinear
dipoles (as illustrated by Figs. 5–6 and 5–7), since for a long-enough array the array
pattern alone will primarily determine the beamwidth and directivity. The dipole pattern
will have little effect near the beam center.
When the element spacing of a linear array is approximately a half-wavelength, the
beamwidth formula becomes
51λ
BW = degrees (5–57)
Nd
(which for d = l /2 is 102/N as given above). The variation of the directivity with element
spacing for a given number of elements is not expressible by a simple formula. Curves
showing this variation have been published by Tai (1964) for both uniform and nonuni-
form linear arrays. As mentioned for broadside arrays in sec. 5.2.2, the gain for a given
number of elements increases as the spacing is increased up to a maximum which occurs
at slightly less than one wavelength spacing (about 0.95 l) for a large array (many ele-
ments). For N = 3 the maximum directivity occurs at about d = 0.8 l.
The beamwidth formula, equation (5–57), can also be applied to the horizontal and
vertical beamwidths of a vertical planar array by considering it to be a linear array in
each direction, of Nh elements with spacing dh horizontally, and Nv elements with spacing
dv vertically. Although this formula is for an array without a reflecting screen it also
applies with a reflector if N is large.
Broadside arrays are often made unidirectional by placing a reflecting screen behind
the array, or by using two curtains of arrays with endfire phasing between the curtains.
216 Arrays
When such arrays are large (many wavelengths) the beamwidths and directivity are given
by formulas of the following type:
λ 
BW = k1
 Nd 
(5–58)
4π A
D = k2 2  (5–59)
 λ 
where k1 and k2 are of the order of one or slightly less, if the beamwidth is expressed in
radians and if all array elements are similar and radiating in phase with equal intensity.
If the beamwidth is expressed in degrees, k1 is in the vicinity of 60—somewhat less for
a uniform array and somewhat more for tapered arrays. For a tapered array the constant
k2 in (5–59) is less than one. Commonly, k2 is known as the aperture efficiency.
Equation (5–59) with k2 = 1 represents the maximum directivity that can be obtained
with an ordinary unidirectional planar array of area A. As mentioned, k2 = 1 applies to
a broadside uniformly excited planar unidirectional array (all elements in phase and
radiating with equal intensity). Similarly, the directivity D = N for a broadside half-wave-
spaced uniform linear array of N elements (no reflector) is ordinarily the maximum
obtainable value of directivity for an array of that length. Thus, for array length L and
element spacing l/2, the maximum directivity is D = (2L/l) + 1.
However, it is possible, in an array of closely spaced elements, to achieve directivities
greater than the above mentioned uniform-array maximum values by properly phasing
the elements. An array in which this is accomplished is a supergain or superdirective
array. This possibility, however, is not as attractive as it seems at first. The superdirective
array is characterized by relatively large phase changes within a relatively small distance.
The mutual coupling between elements cause much larger currents fields than in a normal
array. The currents increase enormously if any substantial amount of superdirectivity is
attempted, to the extent that ohmic losses quickly overcome any increase in directivity,
so that gain is not increased.
Except for very small arrays, super directive illuminations (Hansen 1964, pp. 82–91)
have proven impractical, because of the high I 2R losses resulting from these very large
currents. Also, the bandwidth of arrays having superdirective illumination becomes
extremely small as the superdirectivity is increased. Therefore, the supergain effect is
not practical except in very limited degree in special situations. Moderate superdirectivity
is achieved, however, in endfire arrays phased according to the Hansen-Woodyard condi-
tion, but even moderate superdirectivity is impractical in broadside arrays.
Even so, as now discussed, the endfire array is being used successfully to obtain a
higher gain for a given frontal area than available from a broadside planar aperture. An
endfire array may provide higher directivity and gain for a given number of elements
than a broadside array. However, endfire arrays require larger dimensions in the along-
beam direction. The UHF array on the United States Navy E-2C aircraft uses the endfire
principle to advantage (Long, 2004). It has a horizontal array of horizontally polarized
Yagi antennas, which permits a small vertical dimension. The horizontal array provides
a relatively narrow azimuth beam. Of especial importance is that its desired broad eleva-
tion beam is obtained with a much smaller aperture height than would be required of a
broadside planar antenna. The result is an antenna with small vertical dimension having
low aerodynamic drag. Additionally, the antenna gain is about four times the gain that
is available from a continuous aperture of the same frontal area.
The physical aperture Ap is a measure of the physical size of an antenna area over
which radiation is transmitted or received. For a uniformly illuminated aperture, that is,
one for which the amplitude and phase are both constant over the physical area and one
that is lossless, the gain G is as follows:
G = ( 4π Ap ) λ 2 (5–60)
Equation (5–60) was shown by Silver (1949, p. 183) to be true by integrating the
total radiation pattern for a uniformly illuminated aperture and calculating D. The mea-
sured gain of a real antenna is of course less, depending on losses that can occur because
of discrepancies in the antenna pattern, impedance mismatches, and dissipative (I2R)
losses.
In sec. 3.3.4, chapter 3, directivity is defined as D = 4p/W, where W is the imaginary
solid angle that contains all radiated power if it were distributed uniformly and totally
within said solid angle. Thus, directivity and consequently gain are in approximate
inverse proportion to the products of 3 dB principal-plane beamwidths. Theoretically, the
half-power beamwidth qHP for a uniformly illuminated line source of length L
θ HP = 50.8λ L (5–61)
when qHP is expressed in degrees, and l and L are in the same units. Equation (5–59)
can be obtained from the table in sec. 6.10–2, chapter 6.
Now consider a rectangular aperture. Its gain can be determined with (5–61) by
expressing physical aperture Ap in terms of principal plane beamwidths, qHP and fHP, by
using equation (5–60). Then, the approximate gain G for a lossless, uniformly illuminated
aperture is
G ≈ 32,400 (θ HP φ HP ) (G dimensionless, θ HP and φ HP in degs.) (5–62)
Stutzman (1998) investigated equations for a number of theoretical patterns, including

those for this case of a rectangular aperture. He notes that actual patterns due to antenna
fabrication do not ordinarily have perfectly deep nulls; and because of the effects of
installation environments, they do not have side lobes that decrease uniformly at wide
angles, as do theoretical patterns. Obviously, the gain of an actual antenna is also reduced
by lossy materials (I2R losses). Additionally, the radiated power may be reduced by
impedance mismatches. As a general rule if detailed measurement data are unavailable,
Stutzman suggests that a reasonable approximation for gain, for a beam normal to the
aperture, is
218 Arrays
G = 26,000 (θ HP φ HP ) (G dimensionless, θ HP and φ HP in degress) (5–63)
The approximate, “typical” gain of (5–63) is nearly 1.0 dB less than gain of (5–62)
determined by using theoretical beamwidths and for a lossless, uniformly illuminated
rectangular aperture. It is to be underscored that use of (5–63) is recommended only if
the antenna efficiency (“k” in G = kD) is known to be approximately 1. The point being
that directivity and thus gain are reduced by spurious radiation not readily detected and
thus not ordinarily accounted for when determining directivity. For example, if only the
usual principal-plane measurements are made, there may be unsuspected out-of-plane
spurious radiation.
Equation (5–63) is applicable to arrays as well as continuous aperture antennas (chapter
6). Stutzman (1998, pp. 7–11) cautions the reader regarding the use of the following
equation for directivity of arrays that appears often in antenna literature, namely:
D = De Di, (5–64)
where
De = directivity of each element in the array
Di = directivity of the array with isotropic elements
This formula is for large arrays without grating lobes. Even when the transmission effi-
ciency (“k” in G = kD) is large, Stutzman recommends the use of (5–63) instead of (5–64).
This he notes is because the accuracy of (5–64) can vary greatly with the details of array
geometry. For example, small changes in array element position can increase spurious
radiation above what is indicated by a theoretical calculation for directivity (even though
there may be no perceptible change in principal-plane beamwidths), thereby contributing
to a loss in gain.
5.8. Some Other Array Types

5.8.1. Long-Wire Arrays
The directional properties of single long-wire antennas, described in sec. 4.4, can be
augmented by combining two or more of them into an array. The simplest such array is
the V antenna, consisting of two long-wire antennas each with one of its ends at the apex
of the V. If the angle between the two sides of the V is equal to twice the angle that the
cone of maximum radiation of each wire makes with the axis of that wire, the two
cones will “add up” in the direction of the line bisecting the angle of the V, and there
produce a maximum lobe of radiation. If the wires of the V are resonant, there will be
an equal lobe in the opposite (back) direction; that is, a resonant V is bidirectional. A V
of two nonresonant wires will have a single unidirectional beam. Figure 5–22 shows
schematically a resonant V antenna and a typical pattern oriented to correspond to the
schematic of the antenna. The gain is approximately twice the gain of one leg of the
V by itself.
Some Other Array Types 219
Feed line
FIGURE 5–22.
Resonant V antenna and typical pattern.
l l
Feed line
Terminating
resistor
FIGURE 5–23.
Rhombic antenna and typical pattern.
The principle of the V antenna may be extended by connecting two Vs together at

their open ends to form a diamond, or rhombic antenna. This antenna is fed at one end,
as if it were a V antenna; a terminating resistor is usually connected across the far ends
to make the antenna nonresonant and hence unidirectional. The resistor used must be
noninductive at the frequency of the antenna. A value of about 800 ohms is usually
required for proper termination. A schematic diagram of a rhombic and a typical pattern
are shown in Fig. 5–23.
The rhombic is usually used in preference to the V if a nonresonant antenna with uni-
directional pattern is wanted, since the V is much more difficult to terminate. (A ground
connection is needed, as described in sec. 4.4). However, a rhombic may also be operated
unterminated; the pattern will then be bidirectional.
Both V and rombic antennas can be “stacked” vertically or side by side, forming an
array of Vs or an array of rhombics, to achieve additional directivity. Long-wire arrays
are used chiefly in the MF and HF bands, from about 3 to 30 MHz; they are seldom used
above 100 MHz.
If the angle between the legs of a V is made less than twice the single-wire lobe angle,
the maximum of the resulting single lobe will still be in the plane that bisects the V, but
it will be tilted upward out of the plane of the V itself. This result may be desired to
receive waves reflected downward from the ionosphere or to transmit upward at a slight
angle for optimum ionosphere reflection. The same principle can also be applied to
the rhombic.
220 Arrays
lg /2
Beam direction (z-axis)
Slots
C
Input end z
Polarization Resistive
y (x-axis) termination
x
FIGURE 5–24.
Broadside array of longitudinal shunt slots in a waveguide.
5.8.2. Horn and Slot Arrays

A linear or planar broadside array may be formed with horns used as the radiating ele-
ments. If unidirectional horns (e.g., sectoral or pyramidal) are used, the array pattern will
of course also be unidirectional. The pattern is obtained by multiplying a single-horn
pattern by the pattern of the equivalent array of isotropic point sources. Horn arrays are
not in very common use, but they do have some applications.
Waveguide slot arrays, on the other hand, are much used at microwave frequencies.
They consist of a set of slots cut along the face of a waveguide. A typical form of slot
array is shown in Fig. 5–24. The longitudinal shunt slots are cut in the broad face of a
rectangular waveguide operating in the TE10 mode. They are parallel to the longitudinal
center line of the guide and displaced laterally from it. The slots are spaced a half (guide)
wavelength apart along the length of the guide, and they alternate on opposite sides of
the center line; this results in in-phase operation, and hence a broadside pattern. The
polarization is perpendicular to the axes of the slots.
The lateral displacement of the slot from the guide center line determines its degree of
coupling to the wave in the guide, and hence the percentage of the power flowing down the
guide that will be extracted and radiated. Therefore, variation of this displacement provides
a convenient means of controlling the array taper, or amplitude distribution.
The last slot in the array cannot extract all the remaining power. The residue in the
guide is ordinarily absorbed in a resistive termination, to prevent reflection and standing
waves that would upset the intended amplitude distribution and phasing.
Many other forms of slot arrays are possible and useful. The array of longitudinal
shunt slots is perhaps the most common.
5.8.3. Patch or Microstrip Arrays

Microstrip, also commonly known as patch, antennas are described in sec. 4.9. The patch
antenna is made by reliable fabrication techniques, has acceptable bandwidths, and is
General Remarks on Mutual Coupling 221
lightweight and small. For these reasons, patch arrays are widely considered for airborne
and spacecraft applications. In addition, design and construction techniques exist that
exhibit low mutual coupling between array elements, thereby suppressing the likelihood
of blind angles, that is, blindness (Yang, Mosallaei, and Rahmat-Samii 2007).
5.8.4. Phased Arrays for Scanning

Thus far this chapter has discussed antenna arrays that are designed for the major lobe
or beam being pointed in one direction or more. If all the elements of a linear array or
a plane array are in phase, the beam will be directed broadside. On the other hand, if the
phases of the elements of a linear array progressively change by an amount per element
that is equal, in radians, to 2pd/l, where d is the element spacing and l is the wavelength,
the beam direction will be parallel to the line joining the elements (endfire). Furthermore,
by progressive phase changes, beams may be directed at intermediate angles. Advantage
of this fact is taken in the so-called phased arrays, or electronically steered arrays, to
produce a beam that may be directed at various angles by changing the progressive phase
shift along the array. Concepts and techniques of array antennas that are designed to
electronically change beam-pointing direction are discussed in chapter 8, Electronically
Steered Arrays.
5.9. General Remarks on Mutual Coupling

Most basic discussions on array pattern analysis assume that all elements radiate isotropi-
cally, or at least the radiation pattern of each element is identical in shape. These assump-
tions amount to assuming that the mutual coupling between elements is negligible. For
some element configurations, this assumption may provide a useful estimate for an
overall normalized array pattern. However, generally, by neglecting mutual coupling, the
calculated array element input currents and radiated power will be incorrect, and the
validity of the estimated pattern shape may be in question (secs. 5.1.4–5.1.7).
A few comments follow about the environment within which elements are located. In
an array with many elements, those elements near the center are within similar electric
and magnetic environments, with the outermost (edge) elements being subjected to asym-
metrical mutual coupling. However, for purposes of side-lobe control, amplitudes are
tapered to smaller values toward the edge elements. This tapering tends to minimize some
of the effects of asymmetry. Furthermore, often dummy edge elements are included, to
provide a greater uniformity in the mutual coupling environment for the energized ele-
ments that are thusly removed from the aperture edges.
Ideally, an antenna should provide a good impedance match between the transmitter
and free space. If there are impedance mismatches, a loss of input and radiated power
occurs. Thus a mismatch, even at one element, changes the array aperture illumination
and may cause an appreciable pattern change from that which would occur based on the
element’s assumed radiated power. A mismatch causes standing waves, which produce
undesirable, higher voltage levels than a matched condition. Furthermore, internal reflec-
tions may cause the generation of spurious lobes in the overall array radiation pattern.
222 Arrays
In fact, there are conditions, sometimes observed, where a matched condition exists at
the broadside beam direction, but at some other angle most of the power is reflected and
thus a scan “blindness” results. In spite of the difficulty of dealing with mutual coupling,
especially with electronically steered arrays (chapter 8), well-functioning fixed-beam as
well as electronically steered arrays have been developed and operating for a number of
decades (Hansen 2007; Oliner and Knittel 1972).
References
Balanis, C. A., Antenna Theory, Wiley-Interscience, 2005.
Bodnar, D. G., “Materials and Design Data,” ch. 55 in Antenna Engineering Handbook,
4th edi, J. L. Volakis (ed.), McGraw-Hill, 2007.
Bregains, J. C., F. Ares, and E. Moreno, “Visualizing the 3D Polar Power Patterns and
Excitations of Planar Arrays with MATLAB,” IEEE Antennas and Propagation
Magazine, April 2004, pp. 108–112.
Brown, G. H., “Directional Antennas,” Proceedings IRE, vol. 25, January 1937,
pp. 78–145.
Dolph, C. L., “A Current Distribution for Broadside Arrays Which Optimizes the
Relationship between Beamwidth and Side-Lobe Level,” Proceedings IRE, vol. 34,
June 1946, 335–48.
Hansen, R. C., Microwave Scanning Antennas, vol. 1, Academic Press, 1964.
Hansen, R. C., “Phased Arrays,” ch. 20 in Antenna Engineering Handbook, 4th ed.,
J. L. Volakis (ed.), McGraw-Hill, 2007.
Hansen, W. W., and J. R. Woodyard, “A New Principle in Directional Antenna Design,”
Proceedings IRE, vol. 26, March 1938, 333–35.
Harris, F. J., “On the Use of Windows for Harmonic Analysis with the Discrete Fourier
Transform,” Proceedings of the IEEE, January 1978, pp. 51–83.
IEEE Standard Dictionary of Electrical and Electronics Terms, 5th ed., Institute of
Electrical and Electronics Engineers, 1993.
Kraus, J. D., Antennas, 2nd ed., McGraw-Hill, 1988.
Kraus, J. D., and R. J. Marhefka, Antennas, 3rd ed., McGraw-Hill, 2002.
Long, M. W., Airborne Early Warning System Concepts, SciTech Publishers, 2004,
pp. 235–36 and 405–08.
Ma, M. T., “Arrays of Discrete Elements,” ch. 3 in Antenna Engineering Handbook,
R. C. Johnson, 3rd ed., McGraw-Hill, 1993, p. 3–23.
Mailloux, R. J., Phased Array Antenna Handbook, Artech House, 1994.
Oliner, A. A., and G. H. Knittel, Phased Array Antennas, Artech House, 1972.
Reference Data for Radio Engineers, 6th ed., Howard W. Sams & Co, Inc.,
1979.

Sevgi, L., and Çagatay Uluişik, “A MATLAB-Based Visualization Package for Planar
Arrays of Isotropic Radiators,” IEEE Antennas and Propagation Magazine, vol. 47,
February 2005, pp. 156–63.

Stutzman, W. L., “Estimating Directivity and Gain of Antennas,” IEEE Antennas and
Propagation Magazine, August 1998, pp. 7–11.
Tai, C. T., “The Optimum Directivity of Uniformly Spaced Broadside Arrays of
Dipoles,” IEEE Transactions on Antennas and Propagation, July 1964,
pp. 447–54.
Taylor, T. T., “Design of Line-Source Antennas for Narrow Beamwidth and Low
Side Lobes,” IRE Trans on Antennas and Propagation, January 1955, pp. 16–28.
Van der Mass, C. J., “A Simplified Calculation for Dolph–Tchebycheff Arrays,”
J. Appl. Phys., vol. 25, no. 1, pp. 121–124, January 1954.
Yang, F., H. Mosallaei, and Y. Rahmat-Samii, “Low Profile Antenna Performance
Enhancement Utilizing Engineered Electromagnetic Materials,” ch. 34 in J. L. Volakis,
Antenna Engineering Handbook, 4th ed., 2007, p. 34–12.

1. Two dipoles have their centers positioned at R1 and R2 in Fig. 5–1 (and Fig. 5–2),
with their axes parallel to the z-axis (perpendicular to the plane of the paper).
Their separation is d = l /6. (a) What must be their phase difference, a, in order
for the field intensity to be zero in the direction f = 270° (direction of the nega-
tive y-axis)? (b)If E0 in equation (5–4) has the value 12 volt per meter, calculate
E(f), for these values of d and a, and plot the resulting pattern using polar
coordinates. The shape of the resulting pattern should resemble those of compa-
rable phasing and spacing in Fig. 5–3. (Note that the maximum value of this
pattern is less than one; it needs a normalizing factor to make it a proper relative
pattern.)
2. (a) A central transmitting station communicates only with three other fixed sta-
tions, which are all at about the same distance but in directions that are 120
degrees apart. The frequency is in the LF region, and two vertical towers are to
be used as a two-element monopole array for the central station. To provide equal
signal strength at all three of the other stations, and as strong a signal as possible
with a given amount of transmitter power, what spacing and phasing of the
monopole elements (towers) would you employ? (Base your answer on inspec-
tion of Fig. 5–3). (b) What would your answer be if there were four other stations,
instead of three, in directions 90 degrees apart (e.g., one to the north, one east,
one south, and one west)?
3. A uniform linear array consists of four short dipoles, spaced a distance d = l /2
apart, as in Fig. 5–4. The phase difference between elements is a = 0°, so that
it is a broadside array. The short-dipole axes are parallel to the z-axis of the
coordinate system, so that their individual relative patterns are, from equation
(4–2) of chapter 4, Erel = sin q. (This is the complete pattern even though only
one angle is involved.) Write the equation for the complete pattern (versus q and
f) of this dipole array. Three steps are necessary: (a) Modify equation (5–9),
following the instruction given in the text, to convert it to a pattern for an array
224 Arrays
of isotropic elements. (b) Substitute into this expression the appropriate values
for n, d, and a. (c) Apply the principle of pattern multiplication to obtain the
pattern versus q and f of the dipole array.
4. A uniform broadside array of twelve point-source isotropic elements has an
element spacing of 0.65 wavelength (d = 0.65l). (a) What is the approximate
beamwidth? (b) If the spacing is d = 0.5l (instead of 0.65l), what will the
beamwidth be? (c) What is the directivity for d = 0.5l? Will the directivity for
d = 0.65l be less than or greater than this value?
5. A five-element driven endfire dipole array has element spacing d = l /8. (a) What
should the phasing (a) of the elements be to obtain maximum directivity of the
beam? Express the result in both radians and degrees. (b) Using the value of a
found in part (a), calculate E(f) from equation (5–9) for f = 90° and f = 270°
(the two “endfire” directions). What is the front-to-back ratio of this antenna,
expressed as a ratio of electric field strengths? Express it also as a decibel value.
Note: In part (b), you will find that the initial calculation indicates a “negative”
electrical intensity for f = 270°, but recall that absolute-value brackets are used
in equation (5–9) to indicate that the calculated field is regarded as always
positive.)
6. Check the box that you think precedes the correct answer to the following
multiple-choice problems:
(a) A Yagi-Uda antenna has a beam that is
ⵧ unidirectional ⵧ omnidirectional
ⵧ bidirectional ⵧ multilobed (many equal-strength
lobes)
(b) If the total number of dipole elements in a Yagi-Uda antenna is five, the
usual number of these that will be parasitic directors is
ⵧ one ⵧ three
ⵧ two ⵧ four
(c) The principal advantage of a Yagi-Uda antenna over a fully driven endfire
array with the same number of elements is
ⵧ higher gain ⵧ simpler feed system
ⵧ easier adjustment of element ⵧ better polarization characteristics
length and spacing
7. It is desired to obtain a directivity of 1,000 with a large uniform planar dipole
array backed by a reflecting screen. The wavelength is l = 0.5 meter
( f = 600 MHz). (a) What must be the area of this array? If the array is square,
what is the length of each side? (b) If the dipoles of the array have their centers
spaced a half wavelength apart in both vertical and horizontal dimensions, with
dipoles located at each corner of the square, how many dipoles does the array
contain? (Disregard the fact that the edge dimension is not exactly an integral
multiple of a half wavelength; it is very nearly so.)
8. Multiple choice (check one box in each part):

(a) Tapering the currents in a dipole array is done for the following reason:
ⵧ To avoid corona in the ⵧ To reduce the minor-lobe level
end elements
ⵧ To minimize ohmic losses ⵧ To increase the directivity
(b) The absolute minimum side-lobe level is achieved in array design by
using:
ⵧ The Dolph-Chebyschev ⵧ A Taylor distribution
distribution
ⵧ A binomial distribution ⵧ A truncated Gaussian distribution
(c) The Taylor distribution is superior to the Dolph-Chebyschev when:
ⵧ It is applied to a uniform ⵧ It is applied to a linear array of
linear array many elements
ⵧ Minimum beamwidth is ⵧ It is desired to have all minor
desired for a given lobes of equal magnitude
side-lobe level
9. A V antenna has resonant-long-wire legs that are each 5 wavelengths long (10
half wavelengths). What must be the angle at the apex of the V for the beam to
have maximum directivity and to lie exactly on the bisector of the V? For solu-
tion, see equation (4–14), chapter 4.
10. Calculate the patterns shown in Fig. 5–20 for two, 2-m line apertures that operate
at 3 GHz. One aperture has constant amplitude and phase and the other has a
Dolph-Chebychev distribution that gives 26 dB side lobes.
CHAPTER 6
Reflectors and Lenses
It has been shown in sec. 5.1 that an array antenna can be used to achieve a directional
radiation pattern in which the radiated power is concentrated in a beam. This chapter
treats an entirely different method of achieving essentially the same result, by the use of
reflectors and lenses. Antennas using these devices achieve a directional effect most
readily explained in terms of optical principles described in sec. 1.2—the principles of
reflection and refraction.
The branch of optical science that deals with these phenomena in terms of rays and
wave fronts, rather than in terms of electromagnetic-wave theory, is called geometric
optics. The principles of geometric optics can be applied (as mentioned in sec. 1.2) only
when the dimensions of the optical surface are large compared with the wavelength. This
means, for example, that a reflector 3 meters in diameter would not behave in accordance
with a geometric-optics analysis at a frequency of 1 MHz, for which the wavelength is
300 meters. It would do so very well, however, at 10,000 MHz, for which the wavelength
is 3 cm. Generally speaking, therefore, antennas based on geometric-optics principles are
very-high-frequency devices. They are used mainly above 30 MHz, and arguably above
about 1 GHz they are more common than arrays.
Even when the dimensions of the antenna are large compared to the wavelength, the
principles of geometric optics cannot be applied to all aspects of its behavior. For
example, diffraction will occur at the edges of a lens or reflector and at small irregulari-
ties in the structure if such exist. Diffraction is not explainable by geometric optics. A
qualitative description may be given verbally or graphically in terms of Huygens’ prin-
ciple, as was done in sec. 1.2, but an exact description requires the mathematical expres-
sions of electromagnetic theory.
The diffraction effects are of secondary importance, and are ignored here, in the aper-
ture field formulation of lens and reflector antennas. However, they cannot be ignored
when an exact far-field analysis is desired. Diffraction must be considered, for example,
in analysis of the minor-lobe formation, and the main-beam pattern shape and width. The
method of geometric optics (ray-wave front analysis) is basically an approximate rather
than an exact method. But it is so nearly exact for many purposes, and is so much simpler
than the exact mathematics, that it is usually employed wherever possible. Also, it
conveys a useful intuitive conception of some aspects of antenna behavior.
227
228 Reflectors and Lenses
6.1. Focusing and Collimation Plane

Reflector wavefront
An important geometric-optics concept, Collimated
focusing, is familiar in connection with rays incident
on reflector
optical devices such as cameras, projec- Focal point
tors, search-lights (and automobile head- FIGURE 6–1
lights), telescopes, and the like. Antennas Focusing by a curved reflector.
utilizing radio-wave reflectors* and lenses
focus rays in a similar way. Collimation is a special case of focusing.
6.1.1. Focusing by a Parabolic Reflector

If a beam of parallel rays is incident on a suitably curved reflector, the rays will be
brought to focus at a point, as shown in Fig. 6–1. Similarly, if a point source of radiation
is placed at this focal point, the rays from it will be reflected in such a way that they
emerge as a parallel beam. That is, the directions of all ray lines in Fig. 6–1 are reversed.
This is in effect a simplified statement of the principle of reciprocity discussed in
sec. 1.3.
Rays that are parallel are said to be collimated. In a sense, they are focused “at infin-
ity.” It can be shown that this collimation occurs if the shape of the reflector is parabolic.
The automobile headlight is a familiar example of this type of reflector. The giant search-
light, once used for finding enemy bomber aircraft at night (before radar became avail-
able) and still used for some purposes, is another example.
The geometry of a parabola can be expressed in terms of the coordinate system of
Fig. 6–2. The basic coordinates are cartesian (rectangular). The horizontal axis, labeled
z, lies along the axis of the parabola, which is a line of symmetry. (That is, if the lower
half of the parabola were folded over this axis it would lie exactly on the upper half.)
The point at which this axis intersects the parabolic curve is called the vertex. The second
coordinate axis, labeled x, passes through this point also. At a certain distance f from the
vertex, on the parabolic axis, is a point labeled focus. The equation of the parabolic curve,
in terms of these coordinates and the parameter f is
x 2 = 4 fz (6–1)
The focal point, or focus, at distance f from the vertex, is the point at which incoming
collimated rays will converge (as indicated in Fig. 6–1) or from which the diverging rays
of a point source will be collimated.
The bilateral symmetry of the curve is inherent in the x2 term of the equation; that is,
if a given value of z is chosen and the positive value of x is found that satisfies the equa-
tion, it will also be satisfied by minus the same value of x. The curve defined by (6–1)
extends to infinity in the +z-direction (and consequently also in the +x and −x directions).
* The word “reflector” in this chapter implies a nonplanar reflector—either one that is curved or a system of
plane reflectors set at different angles with respect to a reference plane. A simple plane reflector, such as
might be used in conjunction with an array, does not provide a focusing action.
Focusing and Collimation 229
Practical parabolic reflectors, of course, x2 = 4fz

are of finite depth; they terminate, or are +x
Tangent
“cut off,” at some finite value of z, denoted line at A
zap in Fig. 6–2, graphed in decibels.
The open mouth of the parabola is A
γ
known as the aperture. The aperture
Ray
dimension is labeled Da in Fig. 6–2.* It is ρ
γ
apparent that if zap is smaller than f, the Vertex of
parabola α Da
focus will lie outside the aperture instead –z +z
of inside as shown in the diagram. The O Parabolic
zap
Focus
ratio of the focal length to the aperture f axis
size f/D (commonly referred to as the “f
over D ratio”) is an important character-
istic of a parabolic reflector. It is deter-
mined by the depth of the reflector, zap, in
relation to the focal length f.
A parabola whose depth is exactly –x
equal to the focal length (zap = f ) will
have a value of D equal to 4f, and there- FIGURE 6–2
fore for this parabola f D = f 4 f = 14 , or
Geometry of parabolic reflection.
0.25. For reasons that will appear later, it
is generally inadvisable to make a para-
bolic reflector deeper than the focal length (although in special cases it may be done),
and ordinarily the f/D value will range from about 0.25 to 0.5.
The focusing action of the parabolic reflector can be understood by considering a ray
that leaves the focal point at an angle a with respect to the parabolic axis, as shown in
Fig. 6–2. At the point A where this ray encounters the parabolic curve, a tangent to the
curve is drawn, as shown by a dashed line. By the laws of reflection (sec. 1.2), the ray
will be reflected in such a direction that the incident and reflected rays make equal angles
(g) with the tangent line. This causes the reflected-ray direction to be parallel to the para-
bolic axis, regardless of the particular value of a that may be considered. In other words,
all rays emanating from the focal point will be reflected parallel to the parabolic axis.
By the principle of reciprocity, incoming parallel rays will be reflected so that they con-
verge to focus at the focal point of the parabola.
This result can also be shown in another way by considering the total distance that a
ray travels in going from the source (focal point) to the reflection point (A), and thence
to the aperture plane (the dashed vertical line of length Da). By simple geometric analysis
it can be shown that the length of this path is equal to f + zap, for all values of the angle
a. (This result is obvious for the special case of a = 0). This means that the phases of
all waves thus arriving at the aperture plane are the same. Thus a wavefront (a surface
* Since the symbol D is also used to denote the directivity of an antenna, the aperture linear dimension will
be designated Da in applications where it might be confused with directivity symbol. But often it is called
simply D because this symbol has traditionally been used, especially in referring to the f/D ratio.
of constant phase) is created in the aperture plane. Hence the rays are parallel to the axis,
since rays are always perpendicular to a wavefront (Sec. 1–1).
6.1.2. Paraboloids and Parabolic Cylinders

A parabola is a plane curve, that is, it is two dimensional. A reflector, being a curved
surface, is a three-dimensional object. There are two types of surfaces that produce para-
bolic reflection of the type that has been discussed.
One such surface is the paraboloid, formed by rotating the parabola about its axis.
Thus it is a surface of revolution. If a third cartesian coordinate, y, has its axis perpen-
dicular to both x and z in Fig. 6–2, the equation of the paraboloid is
x 2 + y 2 = 4 fz (6–2)
The intersection of any plane containing the z-axis with the paraboloidal surface is a
parabolic curve like the one shown in Fig. 6–2. (The intersecting plane in this figure is
simply the x-z plane.) The intersection of any plane perpendicular to the z-axis with the
paraboloidal surface is a circle. Thus the open mouth of a paraboloidal reflector, the
aperture, is circular (as in a conventional automobile headlight) if the reflector has
the same depth in all planes containing the parabolic axis, that is, the z axis.
It is possible, however, to “cut away” portions of a paraboloidal reflector in such a
way that it does not appear circular when viewed from a point on the parabolic axis.
Such cut paraboloids have certain advantages in some applications, as will be discussed
in sec. 6.3.
The second type of surface is the parabolic cylinder, formed by translating the
parabola of Fig. 6–2 in the direction of the y-axis, that is, by “moving it sideways.” The
intersections of all planes parallel to the x-z plane with the parabolic cylinder are parabo-
las like the one shown in Fig. 6–2. The intersections of all planes parallel to the y-z plane
with the parabolic cylinder are straight lines. If the cylindrical surface has a finite dimen-
sion in the y-direction, the reflector as viewed from a distant point on the z-axis will
appear rectangular, that is, it has a rectangular aperture. The parabolic cylinder has a
focal line, rather than a focal point, and a
vertex line. The approximate appearances
of the full paraboloid, a cut paraboloid,
and a parabolic cylinder are shown in
Fig. 6–3.
6.1.3. Space Attenuation for (a) (b) (c)

Point and Line Sources
FIGURE 6–3
The strength of the illumination on an Three forms of “parabolic” reflectors: (a) full
aperture depends on whether the reflector paraboloid; (b) cut paraboloid; (c) parabolic
or lens feed functions as a point or line cylinder.
Focusing and Collimation 231
source of radiation. The strength of the radiation versus distance is described by a term
called space attenuation. Thus, knowledge of space attenuation is an important aspect of
the design of reflector and lens antennas. As may be seen in Fig. 1–2, chapter 1, rays
emanating from a point source diverge spherically. Thus, the wave energy from a point
source is constant within a tube of solid angle. Therefore, the power density varies in
proportion to 1/r2, where r is the distance from the source. On the other hand, as will
now be discussed, the power density varies in proportion to 1/r for radiation from a line
source.
Consider a line source that radiates radially with equal intensity perpendicular in every
direction from its axis. Now imagine a concentrically placed cylinder that intercepts the
radiated power. The cylinder’s surface area will vary in direct proportion to r, and hence
the power density at the cylinder would vary in direct proportion to 1/r. Therefore, as a
consequence of energy conservation, the power density from a line source varies with
distant r as 1/r; yet it varies as 1/r2 from a point source.
In Fig. 6–1, both focused (collimated) and unfocused rays are shown near the reflector.
The collimated (focused) rays are parallel, and thus the power density for focused rays
is independent of distance r near the parabola’s aperture (open mouth). Refer now to Fig.
6–2 to develop equations for the field strength (illumination) over the apertures of parabo-
loids (parabolas of revolution) and parabolic cylinders. Radiated rays diverge along the
distance r from the focal point to the reflector, but they are focused after reflection.
Consequently, for a paraboloid illuminated by a point source, the power density varies
as 1/r2, where r is the distance between the focal point and the reflector. Similarly, for
a parabolic cylinder illuminated by a line source, the power density varies as 1/r between
the focal line and the reflector.
In Fig. 6–2, note that zap − f is the distance that the aperture point zap is to the right
of the focal point. Recall that, for a focused antenna, all ray path lengths between the
focus and aperture are equal. Then, to determine r versus angle a by using Fig. 6–2, one
observes:
(focus-to-point A ) + (point A-to-aperture ) = (ρ) + ( ρ cos α + zap − f ) (6–3)
In addition, note that along the parabolic axis:
(focus-to-point O ) + (point O-to-aperture ) = f + zap (6–4)
By equating (6–3) and (6–4), it can be seen that 2f = r + r cos a. Then, by using the
trigonometric identity (1 + cos a)/2 = cos2(a/2), one finds
f 1 + cos α α
= = cos2 ⎛⎜ ⎞⎟ (6–5)
ρ 2 ⎝ 2⎠
We now use the fact that, for a point source, power density varies as 1/r2. Then, from
(6–5), for a paraboloid with an isotropic radiator at its focus, power density pa on the
aperture versus angle a is
2
pα ⎛ 1 + cos α ⎞ α
=⎜ ⎟ = cos4 ⎛⎜ ⎞⎟ (6–6)
p0 ⎝ 2 ⎠ ⎝ 2⎠
In (6–6), p0 is power density at the aperture center, where angle a is zero. Similarly, for
a parabolic cylinder, power density pa on the aperture versus angle a is
pα ⎛ 1 + cos α ⎞ 2⎛ α ⎞
=⎜ ⎟⎠ = cos ⎜⎝ ⎟⎠ (6–7)
p0 ⎝ 2 2
In (6–7), p0 is the power density at the center of the parabolic cylinder, where angle a
is zero.
To obtain electric field strength E versus angle a, the fact is used that E varies as
the square root of power density. Then, from (6–6) and (6–7), the variation of E versus
a along the aperture of a paraboloid and a parabolic cylinder can be expressed, respec-
tively, as
Eα 1 + cos α α
= = cos2 ⎛⎜ ⎞⎟ (for paraboloid ) (6–8)
E0 2 ⎝ 2⎠
and
1 + cos α α
= cos ⎛⎜ ⎞⎟
Eα
=
⎝
(for parabolic cylinder ) (6–9)
E0 2 2⎠
Equations (6–8) and (6–9) describe the variations in the aperture amplitude Ea versus
angle a caused by divergence of rays. Thus, (6–8) describes the amplitude on the aperture
of a paraboloid if the feed radiates isotropically, and (6–9) describes the amplitude dis-
tribution if a line source radiates uniformly at all angles about its axis.
Another useful expression is the attenuation, expressed in decibels, between focus or
focal line and the aperture, caused by the divergence of rays. From (6–8), this attenuation
in decibels for a paraboloid is
20 log ( Eα E0 ) = 40 log cos (α 2 ) (for paraboloid ) (6–10)
Similarly, from (6–9), this attenuation in decibels for a parabolic cylinder is
20 log ( Eα E0 ) = 20 log cos (α 2 ) (for parabolic cylinder ) (6–11)
The equations in the present section describe the effects of space attenuation. These
equations provide a necessary step toward accurately determining the aperture strength
versus position along an aperture, and therefore often they are essential for calculating
the side-lobe levels and beamwidths available from reflector and lens antennas. Some-
Beamwidth and Directivity 233
times, however, as discussed in sec. 6.4, approximations are made for antenna designs
that use only the relative field strengths between the aperture center and its edges.
6.2. Beamwidth and Directivity

It has been shown in sec. 6.1 that if a source of radiation is placed at the focus of a para-
bolic reflector, a beam of parallel (collimated) rays will be produced. Since the rays do
not diverge and the wavefront is plane, it might be deduced that the resultant beamwidth
is zero (because a nonzero beamwidth implies that the wavefront is spherical and that
the rays are diverging within a cone).
This reasoning is not correct, and it is an aspect of lens and reflector analysis to which
geometric optics gives incorrect answers. As discussed in sec. 3.2.5, within the radiating
near-field or Fresnel region (see also Fig. G–1 of Appendix G), the wavefront gradually
changes with increased distance from being plane to becoming spherically-shaped on
reaching the radiating far-field or Fraunhofer region. Thus, the radiating near-field region
extends from the plane-wavefront region to essentially the spherical-wavefront region,
and the radiating far-field extends outward to all greater distances. Although the boundary
between radiating near-field and radiating far-field regions is not sharply defined, for
approximate purposes it is usually considered to be at a distance from the antenna equal
to 2D2/l, where D is the largest aperture dimension and l is the wavelength. (The same
units of length are to be used for both D and l in making this computation.) This crite-
rion, as has been indicated, applies to electrically large arrays as well as to reflectors,
and it also applies to lenses and to any other highly directive antenna.
6.2.1. Shape of Aperture

Figures 6–1 and 6–2 illustrate typical shapes of parabolic reflectors in cross section as
viewed from the side, that is, from a direction perpendicular to the optical axis. When
they are viewed along the axis—from “in front”—the area projected onto the aperture
plane is the aperture area, or simply the aperture, which is the area containing the colli-
mated radiation. The shape of this aperture area is an important characteristic of an
antenna. As has been mentioned in sec. 6.1, if a paraboloidal reflector is a figure of revo-
lution the resulting aperture is circular. However, the full circular aperture may not be
desired in some cases, and portions of the reflector can be cut away so that a noncircular
aperture results, as indicated in Fig. 6–3. Commonly used apertures of antennas are
elliptical, square, and rectangular as well as circular, and occasionally odd shaped. The
shape of the aperture affects the solid-angular shape of the beam.
6.2.2. Primary and Secondary Patterns

The feed radiator, considered by itself, has a radiation pattern and a beamwidth that are
different from the pattern and beamwidth of the antenna as a whole. The feed-radiator
beamwidth must be comparable (but not usually exactly equal) to the angle subtended
at the focus by the edges of the reflector in order for the entire surface to be illuminated.
The feed radiator is also known as the primary radiator, and its pattern is called the
primary pattern. The pattern of the entire antenna is called the secondary pattern. When
the term antenna pattern is used, the secondary pattern is meant. The primary pattern
may also be called the feed pattern.
6.2.3. Beamwidth
The secondary radiation pattern of a lens or reflector antenna is very similar to that of a
unidirectional planar array backed by a plane reflector, of the same size and aperture
shape. The beamwidth is given by a formula of the same type used in chapter 5:
λ
BW = k1 (6–12)
Da
where l is the wavelength, Da is the aperture width of the reflector or lens, and k1 is a
constant, depending on the units in which the beamwidth is to be expressed and on some
other factors. If the beamwidth (BW) is desired in degrees, and if both l and Da are
expressed in the same units of length, k1 is of the order of 60 to 70. This means, for
example, that if the aperture width Da is equal to 10 wavelengths, the beamwidth will
be about 6 or 7 degrees.
For a paraboloid of noncircular aperture, the value of Da to be used in (6–12) is the
aperture dimension in the plane in which the beamwidth is to be calculated. If the aperture
has different dimensions in different directions, the beam will have different widths in
these directions.
6.2.4. Directivity
The directivity (D) of a reflector antenna may be expressed in terms of the area (A) of
its aperture. The formula is the same as that for a large planar array:
4π A
D = k2  2  (6–13)
 λ 
The constant k2 will usually have a value between about 0.5 and 0.7, depending on the
shape of the aperture and the characteristics of the source of the radiation used.
The general relationships discussed in chapter 5 between beamwidths, directivity, and
gain are also applicable to continuous apertures. The factor k2 used above that relates
directivity with aperture area is commonly called aperture efficiency.
6.3. Reflector Illumination

A complete parabolic-reflector antenna has two basic components—the reflector and
a source of primary radiation placed at or near the focal point. (When the antenna is
Reflector Illumination 235
used for receiving, this primary “radiator” Reflector Reflector

receives the signal concentrated on it by
the focusing action of the antenna.) The
primary radiator is commonly called Horn Dipoles
simply the feed.
Ideally, the feed should have a pattern Waveguide
of such nature that it radiates toward the
reflector so as to “illuminate” the entire Transmission
surface, with little or no energy radiated line
in other directions. Such an ideal pattern (a) (b)

is of course unattainable, but some types FIGURE 6–4
of primary radiators are much better suited Paraboloidal-reflector antenna fed by (a) horn
than others. Obviously an isotropic radia- radiator and (b) double-dipole endfire array.
tor would be a poor choice. A dipole is
likewise not very well adapted to this task, although dipole feeds are sometimes used.
More often, a dipole is used in conjunction with a small reflector to provide a better
pattern, and sometimes a small dipole array may be used as a feed. The most common
choice of a feed radiator for paraboloidal reflectors is the waveguide horn, which was
described in sec. 4.7. The parabolic-cylinder reflector has special feed requirements that
will be discussed separately.
Practically all the basic radiator types described in chapter 4 may be adapted to feeding
a reflector. When circular polarization is desired, the helix is often used. The conical horn
can also be used for this purpose. Figure 6–4 shows the general appearance and arrange-
ment of a paraboloidal-reflector antenna with two different types of feed. The horn feed
is waveguide-fed. The double-dipole feed consists of two dipoles, spaced and phased to
produce an endfire pattern that illuminates the paraboloidal reflector. Various other feed
arrangements are possible. Representative feed types are described in Silver (1949, chs.
8, 9, and 10).
6.3.1. Primary Pattern Requirements

The problem of feed design for a reflector or lens is to obtain a primary pattern that is
the best compromise between fully covering the surface of the reflector or lens and not
“spilling over” the edge any more than necessary. When the nature of typical patterns is
considered (e.g., chapter 3, Fig. 3–2), it is evident that both requirements cannot be satis-
fied simultaneously. In practice, some spillover is accepted. This spillover represents a
loss of power from the main-beam. However, if the pattern of the feed radiator were
narrowed appreciably in order to avoid spillover, the power density at the edges of the
reflector or lens would be reduced greatly compared to that at the center. The effect would
be very similar to that in an array with an excessive “taper” of the currents in the ele-
ments. As discussed in sec. 5.6, an excessive taper causes severe gain reduction and
widening of the beam, although it also has the beneficial effect of reducing side-lobe
amplitudes.
Parabolic The radiation from the feed is called

reflector
Focal point
the illumination of the reflector. The vari-
b ation of the intensity of the radiation over
Parabolic axis
the aperture is called the aperture distri-
bution. The edge taper is the ratio of the
power density at the edge of the aperture
to that at the center, or the decibel equiva-
FIGURE 6–5 lent of this ratio. An excessive taper
Half-angle subtended at focal point by causes loss of gain and widening of the
parabolic aperture. beam due to inefficient use of the avail-
able lens or reflector area; it results in a
larger-than-necessary value of k1 in (6–12) and a smaller-than-necessary value of k2 in
(6–13). Too little taper causes loss of gain due to excessive spillover and it also causes
larger-than-necessary side lobes and back lobes. Exactly what amount of illumination
taper will be employed in a given case depends on the relative importance, for the appli-
cation at hand, of achieving maximum gain and reducing side lobes and back lobes to
the minimum practical level.
The problem is somewhat complicated by the fact that a certain amount of illumination
taper will occur even if the primary pattern has a uniform intensity within the solid angle
subtended by the reflector. This inherent taper is due to the way in which the curvature
of the reflector redistributes the power density in the aperture, compared to its original
distribution in the primary radiation from the feed. The amount of taper thus produced
may be calculated by geometric analysis. The inherent taper is greater with small f/D
ratios than with large f/D.
For the paraboloidal reflector the inherent taper (space attenuation) may be expressed
in terms of half the angle subtended by the reflector aperture edge at the feed, that is, at
the focal point. This half angle is designated b in Fig. 6–5.
From (6–6) of sec. 6.1.3, the ratio of the power density at an angle a (Fig. 6–2) to
that at the center, for a uniform-intensity primary pattern, is given by cos4(a /2). Since
angle a equals b at the reflector edge (Fig. 6–5), the edge taper (space attenuation)
expressed in decibels is, therefore,
β β
τ1 = 40 log cos = − 40 logsec (6–14)
2 2
The taper due to the primary pattern, in decibels, is approximately given by the
formula:
β 2
τ 2 = −12 
 BW 
(6–15)
where BW is the half-power beamwidth (as defined in sec. 3.5) of the feed radiator. (The
tapers are here expressed as power ratios less than one, or negative decibel values.)
The value of b for a given paraboloidal Aperture w

outline
reflector may be expressed in terms of its
f/D ratio. The formula is
hc h
he
2f 1
cot β = − (6–16)
D 8 ( f D)
The following table of values of b was FIGURE 6–6

calculated by using (6–16). It gives the
Equal-width vertical strips on a tapered
value 2b rather than simply b because 2b aperture.
is the total angle subtended by the aper-
ture (see Fig. 6–5) and is thus a rough
indication of the feed beamwidth required.
f/D 2b
0.10 273°
0.25 180°
0.50 106°
1.00 56°
(This table indicates why the f/D ratio is seldom smaller than 0.25 or larger than 0.50,
i.e., the required feed beamwidths become impractically large or small, respectively.)
A further tapering effect occurs if the aperture is not rectangular. Consider, for example,
an approximately hexagonal aperture as illustrated in Fig. 6–6, with height hc at the center
and he at the edge. Consider vertical strips of this reflector, of width w and height h. If
the illumination of the aperture is uniform, the total power radiated from a given strip is
proportional to its area, hw. Therefore the ratio of the power radiated from the center
strip to that from the edge strip is hc /he. This is equivalent to a horizontal illumination
taper (in the direction perpendicular to the strips) given in decibels by
hc
τ 3 = −10 log (6–17)
he
The analysis is of course more complicated for an aperture having a curved outline, or
for nonuniform illumination in the vertical direction, but the principles are the same. The
total illumination taper, expressed as the ratio of the edge illumination to the center illu-
mination in decibels, is the sum of the three component tapers, t1, t2, and t3.
6.3.2. Taper and Side Lobes for Elliptical Apertures

The total taper that should be used for a given case depends on the side-lobe level
required. A published study (Adams and Kelleher 1950) for elliptical apertures (including
circular apertures) gives the following indicative values for the sum of t1 and t2. (The
factor t3 is implicitly included by the stipulation that the aperture is elliptical.)
t1 + t2, Decibels Side-Lobe Level, Decibels

−6 −20
−11 −25
−14.5 −30
−17.5 −35
An example illustrates the use of these principles and results. Suppose that it is desired
to design a circular-aperture paraboloidal-reflector antenna for a 30 dB side-lobe level.
A horn is to be used as the feed radiator, and the f/D ratio is 0.4.
From (6–16) it is first calculated that the value of b is given by
1
cot β = 0.8 − = 0.4875
3.2
∴ β = 64°
Hence the primary radiator must illuminate a reflector that subtends an angle of 128
degrees.
The taper due to reflector curvature is calculated from (6–14):
τ1 = 40 log10 cos (32°) = 40 log10 0.848

= −2.9 dB
The table above indicates that for −30 dB side lobes a total taper of −14.5 dB is required.
The feed pattern must therefore provide a taper of −14.5 + 2.9 = −11.6 dB. This is the
required value of t2 in (6–15).
The next step is to determine the required primary-pattern beamwidth, BW. That is,
(6–15) must be solved for the value of BW corresponding to the given value of t2, which
is −11.6, and b = 64°. The result is
BW = 64 12 11.6 = 65°
This means that a feed horn with a 65-degree half-power beamwidth is required.
Equations (4–23a) and (4–23b) in chapter 4 provide for the design of a horn of a given
beamwidth. It is thus found that an optimum-flare pyramidal horn must have an E-plane
mouth dimension (dE) of 0.86l, and an H-plane dimension (dH) of 1.03l, where l is the
wavelength.
6.3.3. Optimum Illumination Taper

The preceding discussion of pattern requirements and taper of the total illumination in
relation to side-lobe level, beamwidth, and gain is for the special case of elliptical-
aperture paraboloidal reflectors. As has already been shown, this taper is affected by both
the primary (feed) radiation pattern and by the geometry of the surface of the reflector,
and also by the shape of the aperture outline. The formulas given are based on experi-
mental investigation rather than theoretical analysis.
Aperture tapers for arrays were discussed in chapter 5, where the similarities in the
effects of tapers on arrays and continuous apertures were briefly discussed. In addition,
Tables 6–1 and 6–2 in sec. 6.10 include a number of different continuous aperture dis-
tributions, and the beamwidths and relative gains that result therefrom.
In sec. 5.6.5 the Taylor distribution is said to be widely used in the design of efficient
and low side-lobe-level arrays, providing there are 20 or more elements (Taylor 1955).
Although widely used in the design of arrays having a large number of elements, the
Taylor distribution is basically a design for continuous apertures, and thus it provides an
efficient aperture design for continuous distributions. In other words, the Taylor distribu-
tion is generally considered to be an optimum aperture distribution that is applicable both
to arrays and continuous aperture antennas.
6.3.4. Line-Source Feeds

The foregoing example of a required primary-radiator beamwidth calculation applies
specifically to the feed for a paraboloidal reflector, but the principles apply to other types
of antennas characterized by an illuminated surface, such as parabolic-cylinder reflectors
and lenses. The paraboloidal reflector requires a feed that approximates a point source,
which is located at or near the focal point. As mentioned, the paraboloid is geometrically
a figure of revolution. But the parabolic-cylinder reflector, and also lenses whose surfaces
are cylindrical, are characterized by a focal line. The feed must therefore be a line source,
rather than a point source.
The simplest example of a line source is the linear array described in sec. 5.2. (At
microwave frequencies a commonly used line source is a waveguide-slot array.) When
the linear array is used with a cylindrical reflector, the length of the array is made approxi-
mately equal to the cylindrical length dimension of the reflector. The line of the array is
positioned on the focal line.
The beamwidth and pattern in the plane containing the line feed and the parabolic axis
are determined entirely by the array, according to the principles discussed in sec. 5.2. (A
parabolic-cylinder reflector functions in this dimension as if it were a plane reflector.) In
the perpendicular plane the pattern is determined by the aperture dimension, and the
beamwidth is given by (6–12). Thus a cylindrical reflector is, with this type of feed, a
hybrid antenna—a sort of cross between an array and an “optical” antenna.
The total illumination taper in the cylindrical direction is that of the array (or other
feed); but in the other direction there is space attenuation introduced because of path-
length differences. The formula for this taper (space attenuation) with a parabolic-
cylinder reflector is not the same factor that applies for the paraboloidal reflector. From
(6–7) of sec. 6.1.3, the ratio of the power density at an angle a (Fig. 6–2) to that at the
center, for a uniform-intensity primary pattern, is given by cos2(a /2). Since angle a
equals b at the reflector edge (Fig. 6–5), the edge taper (space attenuation) expressed in
decibels is, therefore,
β
Parallel
τ1 = 20 log cos (6–18)
B
plates
2
To this must be added the feed-pattern

taper and any other aperture-shape taper,
Waveguide
as with a paraboloidal reflector, to obtain
Parabolic Horn
the total taper.
strip Arrays are not the only possible form
A of line-source feed. At microwave fre-
FIGURE 6–7 quencies a “pillbox” feed may be used,
Pillbox antenna (line source of radiation). like the device sketched in Fig. 6–7. It
consists of a parabolic strip of metal,
bounded on both sides by flat metal plates.
A feed horn placed at the focal point illuminates the parabolic strip so that parallel
reflected rays exit the open mouth of the pillbox. Since the parabolic strip is narrow
compared to the parabolic aperture (AB), the radiation from the pillbox is distributed
along a long narrow opening—in effect, along a line.
The waves propagating in the space between the plates are guided waves. They propa-
gate in waveguide fashion and are subject to waveguide effects discussed in Section 2.5.
Depending on the orientation of the E and H fields within the parallel plates, the phase
velocity of the waves may not be that of free space; and it may vary with frequency. For
the transverse electromagnetic (TEM) mode with the E field perpendicular to the plates
and H field parallel thereto (i.e., E and H fields perpendicular propagation direction),
phase velocity is that of free space. With the transverse electric (TE) mode having its E
field parallel to the plates, a low-frequency cutoff exists for a given plate spacing. Below
the cut-off frequency, waves will not propagate. Above the cut-off frequency, phase
velocity depends on operating frequency and the spacing between the plates. Because
the E field is parallel to the plates in the TE mode, a few metallic or dielectric support
pins may be used between the plates. This is not possible with the TEM mode, and for
large pillboxes adequate plate stiffness could be of concern.
Special techniques may be employed to avoid the “blocking” that results due to the
presence of the feed horn in the aperture. In one technique a “double pillbox” is used in
which there are two parallel-plate spaces, coupled by an opening at the parabolic surface
in such a way that the illuminating rays propagate in one parallel plate region but are
reflected into the other. The pillbox aperture allows these reflected rays to escape, unob-
structed by the horn feed. A cross section of this arrangement is sketched in Fig. 6–8.
Another technique is the use of an “offset” parabolic section. This technique is described
in this section in connection with paraboloidal reflectors (see Fig. 6–10), where it is also
used to avoid aperture blocking by the feed radiator.
The pillbox, like other feed radiators, is sometimes used by itself as a complete
antenna. It provides a radiation pattern that is relatively narrow in the plane of the parallel
plates and broad in the perpendicular direction. It is in effect a parabolic-cylinder antenna
of very short cylindrical length. When it is used as the line-source feed of another
parabolic-cylinder reflector, the final pattern is the result of two parabolic reflections, one
in each of the perpendicular (principal) Parallel Collimated

plates output
planes of the pattern. It might seem that
radiation
it would be simpler just to use a parabo- Parabolic
loidal reflector in the first place. But the reflector
double-cylindrical-reflector arrangement Waveguide
Horn
is sometimes advantageous in that it
allows the pattern to be controlled in each
plane somewhat independently of the Input
other plane. FIGURE 6–8
Cross section of double-pillbox arrangement.
6.3.5. Feed-Positioning Requirements
If a parabolic reflector is to be used to obtain a beam pattern that is maximized on the
axis of the antenna, the feed must be located at the focal point, or for the parabolic-
cylinder reflector or cylindrical lens, on the focal line. If the feed is moved a small amount
laterally—that is, perpendicularly to the parabolic axis so as to keep it in the focal
plane—the beam will still be formed but its direction will be changed. This effect may
be used (as will be discussed in sec. 6.9) to cause the beam to change direction without
moving the entire antenna. However, if the feed is moved too far laterally from the focus,
the beam will deteriorate, and therefore the amount of beam motion that can be obtained
in this way is limited.
If the feed is moved off the focal point or line in a direction along the axis, the radia-
tion pattern will be broadened, since the rays will no longer be collimated (parallel).
Therefore it is important to position the feed correctly at the focal point (or, when it is
desired to direct the beam off axis, in the focal plane).
The focal point (or line) may be located by simple geometric measurement—that is,
with a steel tape, or by the use of a surveyor’s transit if the reflector is very large. The
theoretical location is of course determined by the parameter f in (6–1) (or 6–2).
The question arises, exactly what point of the feed structure is to be positioned at the
focus of the reflector? Consider, for example, a horn radiator. Should the mouth of the
horn be at the focus, or should the point at which the flare starts, or some other point?
There is a similar question if the feed consists of a dipole backed by a parasitic reflecting
element or a small sheet reflector.
To answer this question it is necessary to introduce the concept of the phase center
of a radiating source. This concept is based on the fact that in the far-field region of any
radiator, the wavefront is spherical (even though, as has been discussed in this chapter,
the aperture wavefront may be plane). In this spherical-wavefront region of the main
beam (i.e., in the far field), if the ray lines are drawn perpendicular to the wavefront,
they will constitute a set of diverging (nonparallel) rays. They will appear to be emanat-
ing from a point. In the language of geometry, this point is the center of curvature of the
wavefronts. This point is also called the phase center of the antenna. It is the point from
which the radiation appears to be originating.
The calculation of the phase-center location for a given feed radiator may be a complex
problem; it is often simpler to adjust its position experimentally until the best pattern is
obtained (i.e., narrowest beam in the desired direction, and greatest gain). Methods of
performing the necessary pattern and gain measurements are discussed in Ch. 9. A knowl-
edge of the approximate location of the phase center of the feed is of course helpful in
this procedure. In most cases it is at or near the geometric center of the feed. For a low
flare angle horn, it is near the center of the aperture plane (mouth). However, as the flare
is increased, the phase center moves inside the horn. For some types of feed radiators
the phase center may change if the frequency of operation is changed; thus this parameter
becomes another factor in the determination of the bandwidth of the antenna. (Certain
types of log-periodic radiators, discussed in sec. 7.2, are especially subject to this effect,
so that their otherwise excellent bandwidth properties may be severely limited when they
are used as feed radiators for reflectors and lenses. However, this is not true of all log-
periodic types of radiators.)
In addition to the problem of knowing where to position the feed, there is the com-
panion problem of insuring that it will remain at the correct position when subjected to
various stresses that may occur due to wind, gravity, ice loading, and so forth. The
obvious method of accomplishing this is to support the feed with a strong and rigid
mechanical structure. Unfortunately, in the case of a reflector antenna, there is the con-
flicting requirement that the feed support must not block too many of the reflected rays;
it must not be opaque if it is bulky, and if it is opaque it must not be bulky. A common
method of support in large paraboloidal reflectors is a tripod or quadrupod attached to
the reflector with its apex somewhat beyond the focal point, so that the feed may be
nested under the apex. The support booms may be either hollow metal tubes or solid
pieces of insulating material, such as fiberglass. When the reflector is small and of short
focal length, the waveguide or transmission line may furnish adequate support. Reflectors
with feeds supported by these methods are shown in Fig. 6–9.
The fact that the feed support (and the feed itself) blocks the aperture is a disadvantage
of many reflector antennas. However, if a paraboloid or parabolic cylinder is cut so that
the collimated rays do not “see” the feed radiator, the problem of feed blockage is thereby
solved. Figure 6–10 shows how this is done. The parabolic section actually used is shown
by the solid curve, while the full (symmetrical) parabolic section is shown dotted. This
arrangement avoids blockage but somewhat complicates providing the desired primary
pattern taper.
The feed-support is most severe for paraboloidal reflectors when the beam direction
must be steerable in any direction, or over a large range of angles. This means that some
method of moving the reflector (often called a “dish”) must be provided, and the feed
must be moved at the same time so as to keep it at the focal point; that is, the feed and
the dish must be mechanically connected by a rigid structure. The arrangements of Fig.
6–9 meet this requirement. Feed support is simplified when the antenna need not be
moved, that is, when the beam is always to point in a fixed direction. The feed support
can then be attached to the antenna base (e.g., the ground) rather than to the dish, and
light weight is not as important. This situation occurs, for example, with large reflector
antennas used for tropospheric-scatter point-to-point communication at VHF or UHF.
When the feed is located in the aperture of the dish, so that it blocks some of the
reflected radiation, the antenna pattern is adversely affected in two ways. The gain is
Tripod
Waveguide-
boom
supported
horn feed
Feed
(waveguide runs
large trussed down boom)
mesh-surface
reflector
Small
solid-sheet
dish
FIGURE 6–9
Two methods of reflector feed support.
somewhat reduced, because some power Collimated rays

is reflected into directions other than the
Section of full
forward direction. The side-lobe level is parabola
also increased. These are the principal Aperture
reasons that blocking of the aperture by size
the feed is undesirable, and for employing
a cut reflector, as in Fig. 6–10, if the feed
is large, to minimize or eliminate block-
ing. Another undesirable effect occurs Feed Focus of
parabola
also when the feed is in the collimated
beam. Some of the radiation blocked by
the feed is reabsorbed—for example, if
the feed is a horn, it enters the horn mouth
and propagates back down the guide. This
in effect creates a reflection mismatch, FIGURE 6–10
even though the horn radiating into empty Offset or “cut” parabolic reflector, showing that
space would properly match the guide. collimated rays are not intercepted (blocked) by
This effect can be compensated for by feed.
impedance-matching devices such as
those described in sec. 2.3, but this method
has the disadvantage of reducing the bandwidth of the antenna. The method of using an
offset feed with a cut paraboloid is better; but if the feed is small compared to the total
aperture of the antenna, the effects of aperture blocking will be small and can usually be
ignored. This is a reason for avoiding paraboloids with large f/D ratios, since a large f/D
means that a relatively narrow primary pattern is required, and this in turn requires the
feed horn or other radiator to be relatively large.
6.3.6. Cassegrain Feeds

In some applications of paraboloidal reflectors it is important to minimize the length of
the transmission line between the primary radiator and the first stage of the receiver, or
the transmitter output. When the feed is placed at the focus of the paraboloid, the trans-
mission line must usually be at least long enough to reach transmitting or receiving
equipment (or both, in a radar antenna) located behind or below the dish. This can mean
a run of 30 to 60 meters or more of line or guide, and the losses in such lengths may be
more than can be tolerated, especially in low-noise receiving systems (sec. 7.8).
One solution to the problem is to locate the first amplifying stage of the receiver very
close to the feed, within at most a meter. This eliminates the losses on reception, but it
is not a practical solution for transmitting, since the transmitter power amplifier is usually
too large and too heavy to locate at the feed. Even for receiving it may sometimes be
impractical to locate a preamplifier at the focus because of the difficulty of performing
servicing and adjustments. Requirements for cooling and weatherproofing may also be
difficult to meet. Finally, some of the very low-noise microwave preamplifiers, such as
the maser, may be too large and complicated for this location, depending on the size of
the reflector and the strength of the feed support structure.
An alternative procedure is the use of a Cassegrain feed, in which the primary radiat-
ing source is located in or just behind a small opening at the vertex of the paraboloid,
rather than at the focus. It is aimed at a small secondary reflector or subreflector located
between the vertex and the focus. The arrangement is shown diagrammatically in
Fig. 6–11.
As shown, the rays from the primary radiator are reflected from the Cassegrain sub-
reflector and then illuminate the main reflector just as if they had originated at the focus.
They are then collimated by the main reflector in the usual way. As indicated, the sub-
reflector is required to have a hyperboloidal curvature to reflect the rays from the source
in the correct manner, so as to produce a
“virtual source” at the paraboloidal focus.
Line or Since the transmitter and receiver are
waveguide Hyperboloidal now located out of the ray paths and
subreflector
Paraboloid Blocked
at a location where mechanical support is
focal point rays not usually a problem, they may be as
Feed
large and heavy as required. Also, they
Paraboloidal
reflector are relatively accessible for servicing and
adjustment.
However, aperture blocking is now
FIGURE 6–11 produced by the Cassegrain subreflector.
Cassegrain feed for a paraboloidal reflector. This blocking may be tolerated in many
Radiation Patterns of Horn Antennas 245
large-antenna applications in order to gain the advantages of an accessible primary

radiator, but it is a defect of the Cassegrain scheme that may be serious. One way of
avoiding it is by using an offset reflector, as is sometimes done with focal-point feeds.
Another method is a polarization-twisting scheme (Hannan, 1961), if the primary radiator
is linearly polarized. The Cassegrain reflector is made in the form of a wire grating rather
than a solid metal sheet. The grating wires are parallel to the primary polarization so that
total reflection occurs, as required. On the surface of the main paraboloidal reflector
there is a special polarization-rotating structure, such that the collimated rays have a
polarization rotated 90 degrees with respect to the primary polarization. The Cassegrain
grating reflector is transparent to this polarization and hence does not block the radiation.
(This polarization-rotation technique can also be used with ordinary paraboloidal
feed systems to eliminate mismatch effects resulting from reentry of the reflected
waves into the primary radiator. When the polarization is rotated, this reentry will not
occur if the feed radiator is insensitive to the reflected polarization, which it ordinarily
will be.)
The Cassegrain system is directly copied from the reflector system used in large astro-
nomical optical telescopes, so as to permit the eyepiece to be placed below the main
reflector. Fittingly, Cassegrain antennas have often been used as “radio telescopes” for
radio astronomy.
6.4. Radiation Patterns of Horn Antennas

6.4.1. Horn Patterns, in General
Waveguide horns are commonly used as feeds for reflectors and lenses. The radiation in
front of a horn appears essentially as if radiated from a specific point, that is, the phase
front at a fixed distance is almost spherical. Consequently, horns are excellent primary
feeds for reflectors and lenses. A good approximation is that the phase center is in the
plane of the aperture (the horn’s mouth), but waveguide flaring outward toward the mouth
may cause the phase center to be slightly behind the aperture plane (Muehldorf 1980).
Beamwidth, and thus gain, of a horn depends mainly on its aperture size relative to
wavelength, but beamwidth and gain depend also on the amount of flaring (tapering) of
the horn. For a given aperture size, the phase variation across a horn aperture is least,
and the gain is maximum, for a long horn with small flare angles. However, for conve-
nience in handling, a horn should be short. An optimum horn is one that compromises
in aperture phase error to maximize its gain for its length, and therefore its flare angles
are intermediate between slight and abrupt.
Without having specific feed design data available, 3 dB beamwidths for “optimally”
flared horns may be used as first estimates. Those beamwidths, repeated from sec. 4.7,
are:
56
θE = degrees (6–19)
dE
and
67
θH = degrees (6–20)
dH
where E and H refer to the horn’s E and H plane patterns. The symbols dE and dH are
the aperture dimensions (widths), expressed in wavelengths along the E and H planes.
Graphs of radiation patterns, developed from aperture analyses, known as universal
patterns, are available for a variety of horn types and dimensions (Balanis 2005; Love
1993, pp. 15–6–15–9, 15–35). These “universal” patterns provide details of the major
lobe shapes for aperture widths as small as 1.5 or 2l. However, designs for center-fed
reflectors and lenses often require feed-horn patterns with dimensions even smaller than
1.5l. Section 6.4.3 provides guidance on feed design for aperture widths less than 2l.
Sometimes identical E- and H-plane feed patterns are desired. However, for a conven-
tional tapered waveguide horn, these patterns are unequal and the side lobes are higher
in the E plane. This is because the E- and H-plane amplitude distributions of a waveguide
are different, with no amplitude taper in the E-plane. Horns designed with corrugated
inner walls near the horn’s mouth (aperture) reduce differences in E- and H-plane patterns
(Love 1993, pp. 15–43), but corrugated horns are heavier, larger, and more expensive
than smooth-walled horns.
6.4.2. Shapes of Major Lobes

The design of a wide-angle feed pattern is often critical, because reflector and lens pat-
terns are highly sensitive to the strength of the primary feed pattern at the reflector or
lens aperture edges, where the amplitude versus angle of the horn pattern is steepest and
thus hardest to control. Therefore, detailed knowledge of the primary feed pattern shape
is desired, but sometimes only the horn beamwidth is known. Following (6–15), a good
estimate of normalized pattern amplitude, within a main lobe, in terms of 3 dB beam-
width BW3 is
2
 β 
20 log E (β ) = −12 
 BW3 
(6–21)
Thus, solving for E(b ), one obtains
2
 β 
−0.6
 BW3 
E (β ) = 10 (6–22)
Note that BW3 is the feed beamwidth between half-power points, but b is measured from
the beam center.
For primary feed patterns, antenna designers usually prefer working with the beam-
widths between the −10 dB points, instead of the 3 dB beamwidths. Reasons include
Radiation Patterns of Horn Antennas 247
(i) the −10 dB levels are nearer the reflector or lens edges, and the antenna
side-lobe levels are highly sensitive to the edge illumination, and
(ii) near the −10 dB angles the primary feed (horn) pattern changes more rapidly
with angle than at the −3 dB angle, causing overall antenna performance to
be more sensitive to the −10 dB locations.
Now let BW10 be defined as the full −10 dB beamwidth and let b0.1 be the angle
between the feed’s beam center and a −10 dB direction, that is, BW10 = 2b0.1. Then, the
relationship between BW10 and BW3 is as follows:
2
 β 0.1 
2
 0.5 ⋅ BW10 
20 log E (β 0.1 ) = −10 = −12  = −12 
 BW3 
(6–23)
 BW3 
or
BW10 10
=2 ≈ 1.83 (6–24)
BW3 12
Finally, the amplitude E(b ) at angle b, measured from the beam center, in terms of the
full −10 dB beamwidth BW10 can be expressed as
2 2
 β   B 
20 log E (β ) = −12   = − 40 
 BW10 
(6–25)
 BW3 
or
2
 β 
−2
 BW10 
E (β ) = 10 (6–26)
6.4.3. Measured Wide-Beamwidth Horn Patterns

As already mentioned, when only major lobe shapes are needed, published universal
patterns may be adequately reliable for dimensions as small as 1.5 or 2l. In reality, pat-
terns from small horns can be affected significantly by radiation from aperture edges.
Therefore, for horns having widths of less than 2l, measured patterns should be used
whenever possible.
Figure 6–12 shows −10 dB beamwidth data measured on waveguide horns with small
width D to wavelength l ratios (note that the abscissa is l/D, not D/l.). They were
obtained from a large number of measurements made over several years at the Radiation
Laboratory during WW II (Risser 1949, ch. 10, pp. 341–66). The flare angle of the
average horn is about 20°, which ordinarily indicates small phase variations across the
mouths of the horns. Thus, although specific information on phase is unavailable,
Fig. 6–12 is useful for providing first approximations for pattern beamwidths for
dimensions smaller than 2l. For brevity, 200°

BW10 is used herein to denote −10 dB
beamwidth (the angle in a principal plane
between the two −10 dB directions). 150°
10-dB Beamwidth, degrees

Without other measured pattern data
available, Fig. 6–12 can serve as a useful
starting point for estimating the shape of
the major lobe with (6–26), at least for the 100°
amplitude at angles between the beam
center and the −10 dB points.
50°
6.4.4. Summary
• Equations (6–19), (6–20), and
(6–22) are useful for estimat- 0°
0.5 1.0 1.5 2.0
ing horn beamwidth and major Wavelength/Aperture
lobe shape, if the required in-
FIGURE 6–12
plane horn width exceeds 2l.
• When more detailed informa- Measured −10 dB beamwidths of waveguide
tion and level of the larger side horns having small phase variations over the
apertures. —-— E-plane of sectoral horns,
lobes are needed, horn dimen- —— H-plane of pyramidal horns with E-plane
sions can be obtained from aperture ≥l, ——— H-plane of sectoral horns.
published universal patterns, From Risser (1949, ch. 10, p. 364).
provided the required in-plane
horn width exceeds 2l.
• When patterns are needed that require feed dimensions smaller than 2l, a
possible step is to estimate aperture dimensions from the −10 dB beamwidth
data of Fig. 6–12 and to estimate major lobe shape by using (6–26).
• Primary-feed patterns are critical to antenna design and ordinarily, after initial
calculations are made, pattern measurements are necessary if the overall
design goals are to be assured.
• In cases involving very low side lobes or patterns of unusual shapes, an overall
antenna development may entail several iterative steps including measure-
ment, modification, and analysis.
6.5. Pattern Calculation and Reflector Antenna Design

6.5.1. General Remarks
Calculations for radiation patterns of continuous aperture antennas are in many ways like
those for arrays. Basic to the likeness of the two aperture types is a similarity of their
main lobe patterns and of details of their patterns at angles out to nearly 90° from broad-
side (see sec. 5.5.5). Specifically, nearly identical patterns exist if three conditions are
met, namely; (1) the aperture widths are at least 5l in the plane of the patterns, (2) the
separation of array elements is l /2 or less, and (3) the amplitude and phase distribution
Pattern Calculation and Reflector Antenna Design 249
on the continuous aperture are, at the coordinates of the array elements, the same as those
of the array elements.
A major difference in pattern design between arrays and reflector antennas is, however,
that reflectors are illuminated by a primary feed; thus, feed pattern shapes critically affect
the secondary (overall) patterns of reflector antennas.
The table of 2b (the total angle a paraboloid subtends) versus f/D ratio, that follows
equation (6–16), illustrates the range of feed pattern widths required to subtend the aper-
ture dimension D of a paraboloid versus its f/D ratio. However, as can be seen from
Fig. 6–10, smaller feed beamwidths are needed if a sectionalized (cut) parabolic cylinder
or a “cut” paraboloid is used. Larger f/D ratios (requiring smaller beamwidths) are usually
avoided because, for mechanical rigidity, short focal lengths are desired. On the other
hand, short focal lengths require wide feed patterns that require electrically small feeds;
and, as discussed in sec. 6.4.3, their patterns are generally less predictable than those of
larger feeds.
There are two general types of primary feeds: (a) the point source and (b) a line source.
A point source feed is used for illuminating a reflector having axial symmetry (e.g., a
paraboloid). Because of this symmetry and for effective use of the cross-sectional area,
the limits on compatible feed beamwidths relative to aperture widths constrain the ratio
of maximum-to-minimum principal plane patterns to approximately two or three. Con-
sequently, when point-like feeds are used, the range of possible azimuth-to-elevation
beamwidth ratios is quite limited. On the other hand, line-source feeds are usually used
where azimuth and elevation beamwidths differ greatly. For example, line-source feeds
for antennas having beamwidth ratios as large as 20 : 1 are commonly used with surveil-
lance radar.
As discussed for arrays in chapter 5, a pattern versus q and f is relatively easy to
determine, if the aperture distributions are separable in the principal planes and the linear
aperture patterns for the two aperture distributions are known. Then, the patterns may be
determined from the product of the two linear aperture patterns. Otherwise, as with
arrays, the complete pattern calculations require computational intensive, point-to-point
numerical calculations. For apertures with nonseparable distributions, patterns for the
principal planes are much simpler to calculate than the complete patterns. Then, also as
with arrays, the principal plane patterns may be calculated by using an equivalent line
aperture for each principal plane and integrating each line aperture distribution over the
relevant line aperture.
An example follows in the next section for calculating the illumination along the posi-
tive x-axis (in the upper half) of a parabola. In addition, the far-zone pattern in the x-z
plane is calculated, based on the calculated x-axis amplitude distribution. In this example,
the feed for the parabola is a line source. Thus, the space attenuation between feed and
reflector is caused by two-dimensional spreading of the radiation between feed and reflec-
tor. A line-source feed is commonly used to illuminate a parabolic cylinder. Then, the
illumination is separable because the x-axis illumination of the parabola’s aperture will
be independent of the illumination of the parabolic cylinder along the length of the line-
source feed (y-axis). Thus, as in the case of separable distributions for arrays, the normal-
ized far-zone pattern in the x-z plane is controlled by the x-axis distribution and the
pattern in the y-z plane is controlled by the distribution by the y-axis distribution. Addi-
tionally as described in chapter 5 for arrays having independent (mathematically separa-
ble) x- and y-axis distributions, the pattern versus q and f may be determined as the
product of the two patterns created by the x- and y-axis distributions.
For a paraboloid or a sectional surface area thereof, the steps in calculating detailed
patterns are different than for a parabolic cylinder. This is so because the axial symmetry
of the reflector surface shape prohibits the illumination of the principal planes being
independent of one another. Therefore, as in the case of arrays having nonseparable dis-
tributions, a detailed principal plane pattern calculation uses an equivalent line aperture
distribution, and each line aperture distribution is integrated over the relevant line aper-
ture. Principal plane patterns are of course much simpler to determine than complete
patterns, which require point-to-point numerical calculations over the whole aperture.
6.5.2. Aperture Distribution and Pattern of a Half Parabolic Cylinder

This section describes results of pattern calculations for the top half of a parabolic cyl-
inder that are included in SM 4.9 of the accompanying website. The calculations are
made using Mathcad software, and they determine the amplitude distribution for a half
parabolic cylinder and its far-zone pattern. The coordinates of Fig. 6–2 are used. In the
x-z plane the reflector is a parabola with vertex at x = z = 0, with focus on the z-axis, and
located only in x-z plane at x ≥ 0. Additionally, the feed for the reflector is a broadside
line source located in the y-z plane and extends from the parabola focus in the negative
y direction (into paper). Thus, since the feed radiation versus distance r from the focal
line diverges only in angle a, the power density and the amplitude of the feed radiation
is directly proportion to 1/r and 1/ ρ , respectively.
The problem is to determine the amplitude distribution, including the effects of space
attenuation, alone the x-axis and the far-zone radiation pattern in the x-z plane. As an
example, let the aperture dimension be 1 m in the x-z plane, focal length be 0.5 m, and
operating wavelength be 0.1 m. In addition, let the reflector be illuminated with a horn
feed located along the focal line that has a 45° half-power width in the x-z plane. The
horn feed is set in its elevation angle so that its pattern is centered so that the strength
of the feed horn pattern, after including effects of space attenuation, is equal at the top
and bottom reflector edges. By trial and errors, it is found that equal edge amplitudes
exist if the horn beam center at angle a that is 48° above the z-axis.
Figure 6–13 includes the relative feed amplitude pattern, the space attenuation and the
resulting aperture distribution; each expressed in terms of the reflector x coordinate. From
these results, the reader will notice that the aperture amplitude illumination is equal at
both the top and bottom reflector edges. However, the amplitude is asymmetrical along
the x coordinate, and the peak amplitude is not at the center of the parabola’s x coordi-
nates. As expected and may be seen in Fig. 6–13, because of path length differences, the
space attenuation is asymmetrical along the x coordinate.
Figure 6–14 is a graph of the far-zone pattern, of the illuminated half parabolic cylin-
der, in the x-z plane. Note that the pattern lacks the deep nulls that are ordinarily seen in
calculated far-zone patterns, and the first side lobes are almost indistinguishable from
Pattern Calculation and Reflector Antenna Design 251
the outer edges of the major lobe. This 0

loss in null-depth is caused by the asym-
–4
Relative Strength in dB
metry in an aperture amplitude distribu-
tion along the x dimension.
–8
–12
6.5.3. Steps for Reflector
Antenna Design –16
Steps for developing a parabolic reflector
antenna follow. Similar steps are appro- –20
0 0.25 0.5 0.75 1
priate for other reflector shapes and Aperture X-Axis Coordinate
lenses.
Normalized Feed Pattern
A. Choose approximate overall
Space Attenuation
reflector aperture size. As first
Field Strength on Aperture
approximations:
• BW = 70l/D for overall FIGURE 6–13
aperture dimensions. BW is Relative feed pattern amplitude, space
a desired half-power prin- attenuation, and field strength in decibels
versus aperture x-axis coordinate.
cipal plane beamwidth in
degrees, D is antenna
dimension in plane of BW,
and l /D is dimensionless.
Normalized Pattern in dB
0
• G = 26,000/q1q2 for gain, –10
where q1 and q2 are princi- –20
pal plane beamwidths in –30
degrees. Special beam –40
shapes will result in smaller –50
–60
gain. For example, a cose- –30 –20 –10 0 10 20 30
cant-shaped pattern typi- Observation Angle Relative to Broadside
cally causes gain to be
FIGURE 6–14
reduced by factor of 0.7
(1.5 dB). Normalized far-zone pattern in x-y plane.
• Allow for increased
antenna size needed for structural integrity.
B. Select a general reflector shape: e.g., a paraboloid or a parabolic cylinder
• A simple pyramidal feed horn usually suffices for a paraboloid, if focal-
length to aperture-dimension ratios are within acceptable bounds, for
example, see table that follows equation (6–16). Consequently, for a parab-
oloidal reflector, the ratio of large-to-small principal-plane beamwidths is
ordinarily limited to less than 2 : 1.
• Parabolic cylinders require line source feeds, which permit large ratios of
principal plane beamwidths and wide-angle scanning. Line source feeds
are, however, more difficult to design and fabricate than horns.
C. Decide where to locate the feed with respect to the aperture, i.e., whether or
not to use an offset feed for reducing aperture blockage and minimizing
impedance mismatch.
D. Perform trades between focal length and feed pattern beamwidths to furnish
reflector edge illumination that is at least 10 dB below reflector peak illumi-
nation, when effects of space attenuation are included. A larger edge illumina-
tion results in generally unacceptable reflector spillover and loss of gain. A
lower edge illumination results in lower side lobes, slightly wider beam-
widths and lower gain.
E. Choose feed configuration and dimensions based initially on the reflector
edge illumination. Based on estimated feed beamwidths, calculate expected
shape of feed’s major lobe versus angle, by using either (6–22) or (6–26). It
may be necessary to make feed measurements, especially if −30 dB or lower
side lobes are required or if the estimated feed dimensions are less than 2l.
F. Calculate illumination versus aperture coordinates based on expected shape
of feed’s major lobe and reflector geometry.
G. Calculate patterns in principal planes. Special requirements may also require
pattern specifications in other planes.
H. Continue the iterative process of selecting focal length and feed configuration
until acceptable secondary patterns are calculated. Before desired patterns
are obtained, it may be necessary to revise the original estimate of overall
reflector dimensions.
I. Patterns of the actual antenna should be measured for assurance, especially
if the side-lobe or beam-shape requirements are stringent.
6.6. Reflector Construction

The term “parabolic” is here used to encompass both paraboloidal and parabolic-cylinder
reflectors. Their sizes may range from tiny microwave dishes whose apertures are mea-
sured in centimeters and up to 100 m. Possibly the largest reflector antenna is located at
the Arecibo Ionospheric Observatory at Arecibo, Puerto Rico, where it is used as a sci-
entific instrument for probing the earth’s ionosphere, principally at 435 MHz. It is not
parabolic but spherical in shape, and uses a specially designed feed for correcting phase
errors (spherical aberrations) caused by its shape. The diameter and depth are 305 m and
51 m, respectively; and its surface contains forty-thousand aluminum panels. The reflec-
tor was fashioned in a natural “bowl” in the earth and the reflector axis points vertically.
The beam may be steered within a moderate angular range about the vertical axis by
moving the feed, which is suspended by a system of towers and cables high above the
reflector. The reflector itself is of course fixed in position.
The largest fully steerable radio-telescope antennas have diameters of 100 m, and there
are two. One is in Effelsberg, Germany, which began operations in 1972. Afterwards,
another fully steerable 100-m reflector was constructed at U.S. National Radio Astron-
omy Observatory at Green Bank, West Virginia (Imbriale and Jones, 2007). The third
largest is the 76.2-m diameter radio telescope antenna at Jodrell Bank, in northwest
Reflector Construction 253
England. When completed in 1957, it contained the largest fully steerable reflector. There
are several in the 45- to 60-m size range, and dishes of 20 and 25 meter diameter are
fairly common. However these are still “huge” antennas, especially if they are
steerable.
6.6.1. Fabrication Methods

The smaller dishes are generally made of aluminum. They are formed by spinning or
stretching aluminum on a solid form, which may be made of wood by standard wood-
working techniques, using a parabolic template to check the shape. The template is in
the form of a parabolic curve conforming to equation (6–1) for the desired value of f,
the focal length. It may be made of metal, plastic, or any stiff material. If a paraboloid
is being formed, the template is arranged to rotate about the desired parabolic axis. For
a parabolic cylinder, it is provided with a translational axis on which it can be moved
along a line parallel to the desired focal line. This technique can be employed for fabri-
cating dishes up to about 3 meters in diameter. Very small reflectors may be cast and
then machined if a precise surface is required.
For larger dishes the parabolic reflector must usually be supported by structural stiff-
ening members attached to its back surface. The usual fabrication procedure is to con-
struct this backup structure first. Radial members that are parabolically curved to the
correct shape are joined by circular members to form a paraboloid, or similar pieces are
connected by straight pieces to form a parabolic cylinder. Additional interconnections
may be made to provide the required strength and stiffness. Finally, the reflecting surface
is laid over and attached to this structural support. Again, the parabolic shape may be
checked for correctness by means of a template. Shims may be added or removed to
make final corrections. The surface may be solid sheet metal, perforated sheets, or a
metallic mesh, in the form of panels of convenient size. If each panel is large compared
to the wavelength of operation, it is ordinarily not necessary to provide electrical bonding
of adjacent panels at the adjoining edges, although it may be mechanically desirable to
do so. The purpose of using perforated sheets or mesh material instead of solid sheets is
to reduce the total weight of the dish and to make it have less wind resistance. The per-
forations or mesh openings must be small compared to the wavelength, preferably less
than 1/16 wavelength in the direction perpendicular to the polarization of the radiation
if the polarization is linear. In the other direction the opening size is noncritical.
These descriptions of fabrication techniques are not detailed, and do not cover all
possible techniques and materials. The purpose here is solely to convey a general idea
of the methods used. Furthermore, it is to be recognized that the criticalness of reflector
surface details is dependent on possible special patterns requirements, such as very low
side lobes and/or back radiation.
6.6.2. Reflector Surface Accuracy Requirements

The surface of a reflector must conform to the true parabolic curve described by equa-
tions (6–1) or (6–2) to within a fraction of a wavelength in order to provide the desired
plane wavefront of the reflected radiation. But perfect accuracy is unattainable, and
construction of a precise surface is very expensive. It is desirable, therefore, to know
what degree of inaccuracy can be tolerated without serious sacrifice of performance.
Theoreticians have studied the question of how much deviation from a true parabolic
surface is permissible, assuming that the deviations are of a random nature—that is, that
they are not of a regular nature. An intuitive understanding of the effect of surface irregu-
larities can be obtained from Fig. 1–6 (chapter 1). As shown, the total reflection consists
of a specular (desired) component and a scattered or diffuse (undesired) component.
The generally accepted rule-of-thumb figure is that the permissible deviation is 1/32 to
1/16 wavelength. Thus the irregularity of the reflector surface in effect determines the
maximum frequency at which it can be used successfully; the higher the frequency of
operation desired, the more precisely accurate the surface must be. Appendix I discusses
relatively simple calculations on the effects of random phase errors on antenna patterns.
An analysis performed by Ruze (1966) gives detailed information on this matter. The
random deviations of the surface from a true parabolic shape cause diffuse reflection that
increases the side-lobe level and reduces the gain. Ruze has shown quantitatively how
these effects are related to the root-mean-square (rms) deviation of the reflector from the
exactly correct surface. He showed that a given rms deviation has a greater effect on
side-lobe level in a small reflector than in a large one, and also that the effect depends
on the correlation interval of the irregularities. Correlation interval is a statistical concept.
If the individual “bumps” of the surface extend over a small area, the correlation interval
is small. If they extend over a larger area, as would be true of a “wavy” rather than a
“bumpy” surface, the correlation interval is larger. In terms of these concepts, Ruze
showed that a given magnitude of rms deviation from the true surface causes a greater
increase of side lobes and reduction of gain when the correlation interval is large than
when it is small.
His paper contains curves giving the results for specific cases. As an illustration, a
paraboloidal reflector of circular aperture, having a diameter of 25 wavelengths and
typically tapered illumination, will have an additional side-lobe level of −20 dB when
the rms deviation of the reflector is 1/32 wavelength if the correlation interval is
one wavelength. But if the correlation level is only a quarter of the wavelength, the
additional side-lobe level will be only about −35 dB. By “additional side-lobe level” it
is meant that the relative power of the side lobes due to reflector inaccuracies will be
added to the relative power of whatever side lobes would be present if the reflector were
perfect. As an example of how this “addition” would be computed, suppose that the side-
lobe level with a perfectly accurate reflector would be −20 dB, which corresponds to a
power level of 1/100 or 0.01 of the main-beam level. Then an additional −20 dB side-
lobe level would add another relative power level of 0.01, giving a total level of 0.02.
Thus the combined effects give a net side-lobe level of −17 dB. On the other hand, an
additional −35 dB side-lobe contribution corresponds to an additional 1/3000, or roughly
0.0003, so that the total in this case becomes 0.0103, which is negligibly different from
−20 dB.
Another factor of great importance is the rigidity of the reflector—its resistance to
deformation due to wind or gravitational forces (including additional loading that may
Corner-Reflector Antennas 255
be imposed by ice in the colder climates). Gravitational forces are usually important only
in steerable reflectors that may point sometimes at the horizon and sometimes at the
zenith or intermediate elevation angles. If the reflector shape is paraboloidal at one of
these angles, the loading at other angles may be such that it will deform, that is, the edges
of the dish may sag differentially. This effect is important only for large heavy reflectors.
Deviations of this type will have a serious effect on the pattern, however, since they are
of the “large correlation interval” type. A related effect is the possible variable effect of
gravity on sag of the feed support structure, for different elevation angles of the antenna.
The net effect of reflector deformation and feed-support deformation can be separated
into two parts. One is a deterioration of the pattern—increase of side-lobe level and
reduction of gain; the other is a shift in the direction of the maximum of the beam, which
may cause angle-measurement errors in a radar or radio-astronomy system unless the
shift is known and corrected in interpretation of data.
Distortion of the antenna by wind forces can also be serious, and resulting shifts of
the beam direction cannot be corrected by a “calibration” as can be done for gravitational
shifts. Wind forces, as previously mentioned, can be reduced by using perforated or mesh
material for the reflector surface (“skin”). However, the most serious consideration in
the reduction of wind force is usually that the antenna may be overturned in a high wind,
or that the steering motors may be unable to overcome an opposing wind force. Most
large steerable antennas, in fact, will not operate in very high winds, and are usually
stowed (locked in a fixed position) when winds exceed 30 or 40 mph. Some types of
moderate-sized steerable antennas, however, such as those of radars for air traffic control
or for military and naval applications, are capable of operating in fairly high winds.
6.7. Corner-Reflector Antennas

Although the reflectors considered in this chapter are basically restricted to those having
curved surfaces to produce a focusing or
collimating action, thus eliminating plane
reflectors from consideration, there is a
type of reflector that represents an inter- l
mediate case, namely, the corner reflec-
tor. It is somewhat analogous to a parabolic Feed
cylinder in that it is a “surface of transla- d
tion,” requiring a line source of illumina- Da
90°
tion. It is formed by the intersection of
two plane reflectors, usually at right
Perspective view of
angles (“square corner”). It is most often corner reflector antenna
fed by a dipole or a collinear dipole array. with 3–dipole collinear
array feed
A cross section of a typical arrangement
Side sectional view (b)
is shown in Fig. 6–15a, and a perspective
(a)
view is given in Fig. 6–15b.
The aperture dimension, Da, may be FIGURE 6–15
between one and two wave-lengths. The Corner-reflector antenna.
feed-to-vertex distance, d, is made equal to half the side length, l. Thus the design equa-
tions are
l
d= (6–27)
2
Da = 1.414l = 2.828d (6–28)
(The second equation is obtained from fact that the hypotenuse of a right triangle is the
square root of the sum of the squares of the other two sides.) The distance d is generally
made to be between 1/3 and 2/3 wavelength. A thorough discussion of corner-reflector
antenna design is given by Kraus and Marhefka (2002).
These equations are valid only within the stated range of values of Da. Within
this range, increasing the size does not greatly affect the beamwidth and gain, but the
radiation resistance and the bandwidth are increased. It is apparent from the fact that the
maximum allowable value of Da is about two wavelengths that the beamwidth in this
dimension is fairly broad; it is about 40 degrees. In the other dimension, as with the
parabolic cylinder the beamwidth is determined entirely by the feed radiator, which may
be a linear array with a fairly narrow beam. If a beamwidth in the corner-aperture dimen-
sion of appreciably less than 40 degrees is desired, a parabolic-cylinder reflector should
be used instead of the corner. (The same feed may be used, positioned on the focal line
of the parabolic cylinder.) The directivity of a corner-reflector antenna is approximately
10 dB greater than that of a dipole or of a linear dipole array by itself.
The surfaces of a corner reflector are frequently, in fact usually, made of spaced wires
or tubes parallel to the vertex, rather than solid sheet metal. The spacing is a small frac-
tion of a wavelength, and for polarization parallel to the wires the reflection is practically
as good as that from a solid reflector. If the feed is a dipole or dipoles parallel to the
vertex, as is almost always the case, the polarization will be correct. The use of spaced
wires or tubes reduces the weight and the wind resistance.
Corner reflectors with corner angles other than 90 degrees are sometimes used. A
60-degree corner has a slightly higher directivity than a 90-degree corner, but the
sides must be longer to realize the increase. (The
vertex-to-feed distance must also be greater.) It is
likely that the greater directivity may be obtained
Dipole with less total reflector surface by using a parabolic
cylinder, or by stacking two 90-degree corners as
Reflectors shown in Fig. 6–16. This arrangement is in effect an
Direction array of corner reflectors. The two line sources must
of beam
of course be fed in phase with each other. (Inciden-
Dipole
tally, parabolic cylinders may also be stacked in this
way. The gain will be about the same as that of a
single reflector of the same total aperture, but there
FIGURE 6–16 may be some advantages in reduction of reflector
Stacked corner reflectors. depth.)
Lens Antennas 257
6.8. Lens Antennas

Lenses are used in approximately the same applications as are paraboloidal and parabolic-
cylinder reflectors, though the frequency range of common use is somewhat higher,
beginning at perhaps 1 GHz. Their greatest use is at 3 GHz and above. At the lower fre-
quencies they become too bulky and heavy.
6.8.1. Optical Lenses

The collimating action of a simple optical lens is shown in Fig. 6–17. The behavior is
shown for a source on the axis at the focal distance from the lens, resulting in collimated
rays (plane wavefronts) on the other side of the lens. These collimated rays are parallel
to the lens axis, an imaginary line through the optical center of the lens from the focal
point.
The rays from the point source at the focus diverge uniformly in all directions. A cone
of these rays is intercepted by the lens. The material of the lens has an index of refraction
that is (in the usual case when the external medium is air or vacuum) greater than that
of the surrounding medium. (Optical lenses are of course usually made of glass, although
transparent plastics are sometimes used.) In accordance with the principle illustrated by
Fig. 1–3, chapter 1, the lens bends the rays that impinge upon it at any angle other than
perpendicularly to the surface. That is, all rays are bent except the one from the focus
through the center of the lens. Unlike the flat surface shown in Fig. 1–3, the lens surfaces
are curved, so that different parts of the lens produce different amounts of bending. The
result is that the rays from the focal point, upon emerging from the opposite side of the
lens, are parallel rather than divergent. For a simple thin optical lens, the surface curva-
ture to produce this result is spherical; that is, the lens surface conforms to a portion of
the surface of a sphere.
This collimating action does not occur if the source is placed closer to or farther from
the lens than the focal point. Then the rays on the emergent (right-hand) side of the lens
will (respectively) diverge or converge. Both these types of behavior are useful in optical
applications of lenses, but in radio applications the collimating behavior depicted in
Fig. 6–17 is practically always desired. The rays on the right-hand side of the lens, in
Fig. 6–17, are in effect “focused at infinity,” that is, parallel rays may be regarded as
converging to a point at an infinite distance. This description is, however, the geometric-
optics approximation of the actual behavior, as has been noted in sec. 6.2.
A lens, like a reflector, is a reciprocal device. A source at the focus, on (say) the
left-hand side of the lens, produces parallel rays on the right-hand side. Conversely,
incoming parallel rays on the right-hand side will converge to a point on the left-hand
side, at the focus. These two cases correspond to the use of a lens antenna for transmit-
ting and receiving, respectively. They also illustrate the general principle of reciprocity,
discussed in sec. 1.3.
Many of the principles and practices discussed in connection with parabolic reflectors
also apply to lenses used as antennas. In particular, this is true of the principles of illu-
mination taper, aperture shape, and taper due to surface curvature in relation to side-lobe
Collimated (parallel) rays

Spherical
wavefront Lens Plane wavefront
Source
(focal point) Lens axis
FIGURE 6–17.
Collimation of rays by a simple converging lens.
levels, and of the relations between aperture size, beamwidth, and gain. Feed radiator
requirements are in general quite similar, except that aperture blocking effects do not
occur, as already noted. As will be discussed in sec. 6.7 in more detail, a collimated beam
may also be formed in a direction not parallel to the lens axis, by moving the source
(feed) away from the focus in the focal plane—that is, perpendicularly to the lens
axis—as was also indicated to be true for parabolic reflectors, for limited deviations from
the optical axis.
6.8.2. Lens Surface Configurations

The foregoing discussion of the optical properties of lenses considers the ray behavior
in a single plane. If the surfaces of the lens have the same shape in all axial planes—that
is, if they are “surfaces of revolution”—the same ray behavior is observed in all planes
through the lens axis, and rays from a point source will be collimated (as in Fig. 6–17).
A lens of this type is known in antenna applications as a rotational lens. If, on the other
hand, the lens surfaces are curved in one direction only and are straight in the perpen-
dicular direction, the focusing or collimating action will take place in one plane only,
so that a line source of radiation is required. This type of lens is a cylindrical lens. The
two types are analogous in their behavior and feed requirements to the paraboloidal
and parabolic-cylinder reflectors. Lenses may also have apertures that are circular, ellip-
tical, rectangular, or other shapes.
The surfaces of simple optical lenses are spherical. Those of lenses used for radio
applications generally have a more complicated curvature. One surface of a lens may be
curved and the other plane, or both surfaces may be curved. Whereas the refracting mate-
rial (e.g., glass) of optical lenses usually has an index of refraction (n) greater than one,
the refracting region of a radio lens may have n either greater or less than one. Therefore
the curved surfaces of convergent radio lenses may be either convex or concave. (A glass
optical lens, to be convergent—that is, capable of collimating rays from a point or line
Lens Antennas 259
Convex-plane Convex-convex Convex-concave

(a) (b) (c)
FIGURE 6–18.
Lens surface curvature combinations for n > 1.
Concave-plane Concave-concave Concave-convex

(a) (b) (c)
FIGURE 6–19.
Lens surface curvature combinations for n < 1.
source—must have at least one convex surface.) If the index of refraction is less than
one (n < 1), one or both of the curved surfaces will be concave, whereas for n > 1 one
or both will be convex. It is also possible to have one surface convex and one concave,
provided that the convex surface has the greater curvature for n > 1 and vice versa for
n < 1. The various possibilities are shown in Figs. 6–18 and 6–19. The curvatures are in
all cases calculated to produce the right amount of total ray bending at the two surfaces
so that the rays from a focal point or line will issue from the opposite lens surface all
parallel, that is, collimated.
Lenses having simple curved and plane surfaces may be very thick, and therefore
excessively bulky and heavy. Also, waves passing through the thickest portions of the
lens may suffer considerable loss of power, since the lens structure or material may be
dissipative. To offset these effects, lens surfaces can be zoned. That is, at definite dis-
tances from the center of the lens, the surface is “stepped” so that its total thickness is
reduced. The depth of each step is such that the rays passing through the lens on each
side of the step have path lengths that differ by a full wavelength; hence they are in-phase.
The wavefront issuing from the lens is therefore the same as if zoning were not used.
Figure 6–20 shows cross sections of zoned lenses corresponding to the unzoned types
of Fig. 6–18a. Although zoning provides the benefits mentioned, it also introduces
discontinuities in the wavefront, which

sometimes lead to increased side lobes
and reduced gain. By careful design and
special techniques these effects can be
minimized.
6.8.3. Lens Refracting Media

Small microwave lenses, mainly at fre-
quencies above 10 GHz, can be made of
solid dielectric materials having indexes
of refraction greater than one (n > 1). For
the larger lenses, however, these materi-
als would result in a lens of excessive (a) Curved surface zoned (b) Plane surface zoned
weight and cost. It is therefore customary FIGURE 6–20.
to employ, instead, artificial dielectrics Two methods of zoning the lens of Figure
that consist mainly of air spaces between 6–18a. Dashed lines show contour of unzoned
metallic conducting pieces. These pieces equivalent lens.
can be configured and spaced so as to
produce a phase velocity of the waves that
is either greater or less than free-space velocity. If the phase velocity is greater than that
of free space, the effective index of refraction is less than one, and vice versa.
Artificial-dielectric media may be divided into three classes: (1) path-length media;
(2) metallic-obstacle delay media; (3) metal-plate waveguide media. The first two types
produce the effect of an index of refraction greater than one; the last type has a phase
velocity greater than that of free space (just as ordinary waveguides do, as discussed in
sec. 2.5), and hence an index of refraction less than one.
A path-length medium is created by forcing the waves to follow a path within the lens
that is longer than a straight-line path. Thus, even though the actual velocity in the
medium may be the same as in free space, the effect is that of a lower phase velocity.
One way of accomplishing this, for example, is to force the waves to travel between
parallel metal plates that are formed into a special tin-hat shaped “geodesic” lens (see
sec. 6.8.8).
The metallic-obstacle delay medium consists of an array of small metallic objects,
such as spheres or cubes, separated by a nonconducting low-dielectric-constant material
(such as “polyfoam”) that is used primarily to support the metallic objects. (This type of
lens would function equally well with air between the metal pieces if they could somehow
be supported by the air.) The size, shape, and spacing of the metal pieces determine the
effective dielectric constant and refractive index.
The metal-plate waveguide refracting medium consists of thin metal plates parallel to
one another in the direction of propagation through the lens. The velocity of propagation
(phase velocity) through the space between the plates is determined by the plate spacing
just as it is for ordinary waveguides. Such a parallel-plate region is like a waveguide
with the dimension b of Fig. 2–11 (chapter 2) of very large value, and with the dimension
Lens Antennas 261
a close to the cutoff value (half of the free-space wavelength) so that the phase velocity
will be high. The plates must operate in the fundamental TE mode (electric field parallel
to the plates), and thus such a lens will function properly only for a proper polarization
of the feed. However, it is also possible to have two sets of plates running in perpendicular
directions so that a set of square waveguide channels is formed. This is known as “egg
crate” construction, and it may be used with waves of any polarization.
6.8.4. Variable-Index-of-Refraction Lenses

The lenses considered thus far depend for their action on the curvature of their surfaces.
The medium between these surfaces must have an index of refraction different from that
of the surrounding medium, which is generally air with n ≈ 1. Within the lens, however,
the index has been assumed to be the same everywhere. It is entirely possible, however,
to construct a lens in which the index of refraction varies within the lens. It would be
possible in this way to construct a converging lens whose thickness was constant, that
is, with plane surfaces. Alternatively, the variable-index construction may be used to
make lenses with most unusual properties.
An example that has found important applications is the Luneberg lens, which is a
sphere whose index of refraction (n) varies as a function of the radial distance from the
center of the sphere according to the formula:
n (r ) = 2 − (r R )2 (6–29)
where R is the radius of the sphere and r is the radial coordinate of any point within the
sphere. It is apparent that n = 1 at the surface of the sphere (r = R) and has a maximum
value of 2 at the center of the sphere (r = 0). Such a lens has an interesting and useful
property; it will collimate the rays from a feed source placed anywhere on its surface.
The collimated rays will emerge on the opposite side of the sphere from the feed point,
traveling in the direction of the line from the feed point through the center of the sphere.
Thus a beam can be caused to point in any direction by moving the feed to an appropriate
point on the lens. The behavior of the rays in a Luneberg lens is shown in Fig. 6–21. A
sphere having the required variation of refractive index can be made using either an
artificial dielectric medium or concentric spherical shells of solid dielectric material of
different indexes of refraction. The effective refractive index of a solid dielectric material
can be reduced to a desired value for this purpose by manufacturing it in the form of a
“foamed” solid containing air spaces.
There are also two-dimensional instead of spherical (three-dimensional) Luneberg
lenses. Two-dimensional variable-index-of-refraction lenses focus in only one plane
(Johnson 1993, p. 16–26). They have dielectric properties like a thin cylindrical slice cut
through the center of a 3-D Luneberg lens. An early type consisted of nearly flat plates
partially filled with dielectric material and it operated in the TE10 waveguide mode. In
addition, there have been flat-plate lenses that operated in the TEM mode. One of these
used concentrate rings of different dielectric constants and another used partially filled
Collimated rays
Lens
Plane
Radiating source wavefront
(feed)
FIGURE 6–21.
Rays in a Luneberg lens.
dielectric plates that are tapered to zero thickness at the lens edges. Another 2-D type
lens is the geodesic Luneberg lens, described in sec. 6.8.8.
6.8.5. Focusing with Dielectric and Metal Plate Lenses

A number of lens shapes for dielectric and metal-plate lenses are shown, respectively, in
Figs. 6–18 and 6–19. The index of refraction for dielectric lenses (Fig. 6–18) is greater
than unity, and less than unity for TE10-mode waveguide lenses (Fig. 6–19). Only the
simple lenses of Figs. 6–18(a) and 6–19(a) will be addressed here. They have one refract-
ing surface, the other being normal to the rays. The other lens shapes of Figs. 6–18 and
6–19 have two refracting surfaces. Although more complex, the techniques for designing
lenses with double refractive surfaces are also well established.
Discussions on lens focusing are limited in this book to determining lens shapes
needed to equalize electrical path lengths. A proper lens design includes other specifica-
tions. For example, beamwidths and side lobes depend on the amplitude distribution
across the aperture. However, control of the aperture amplitude requires an understanding
of how refraction within a lens modifies the amplitude across the aperture. Also, to reduce
weight and size, lenses are shaped by zoning as illustrated in Fig. 6–20. Lens designs
also must address losses caused by air-lens reflections and absorption, and the effects of
lens dimensions on the operating bandwidth. Kock (1946), Risser (1949, ch. 11), and
Bodnar (2007) are recommended additional reading on the design of lenses.
6.8.6. Dielectric Lenses

Figure 6–22 shows ray paths through a plano-convex lens, which is symmetrical about
the line FQ2. Assume that a focused antenna consists of a primary feed at point F and
the lens. Then the phase along the aperture of dimension D is constant, and the electrical
lengths FPP2 and FQQ2 are equal. The phase delay for a path length l, expressed in
radians is (2pl)/l. Then, setting the electrical lengths FPP2 and FQQ2 equal gives
Lens Antennas 263
r PP L QQ1 Q1Q2
+ 2 = + + (6–30)
λ0 λ d λ0 λd λd
te
pto
ym
As
Note from Fig. 6–22 that PP2 = Q1Q2
and r cosb = (L + QQ1). Therefore, from r
P P2
(6–30), the electrical lengths of paths Primary b0

r
FPP2 and FQQ2 are equal when Feed b
F
Q Q1 Q2
r L r cos β − L L
= + (6–31) D
λ0 λ0 λd
where
l0 = wavelength in free space (air or
vacuum)
FIGURE 6–22.
ld = wavelength in the dielectric lens
Path lengths in a dielectric lens.
From equation (1–19), chapter 1, the
index of refraction n of a lossless sub-
stance is expressed as
c µε
n= = (6–32)
v µ0 ε 0
For most dielectric materials m = m0, and thus for a lossless material usually
n = ε ε0 (6–33)
where e/e0 is the relative permittivity of the material having index of refraction n.
Now, from the general relationship fl = v and (6–32), for a material having index of
refraction n, and with its wave velocity vd and wavelength ld
c f λ0 λ0
n= = = (6–34)
vd f λ d λ d
Then, multiplying (6–31) by l0, and solving for r the following is obtained
(n − 1)L
r= (6–35)
n cos β − 1
Equation (6–35) gives the shape of the lens, and it describes a hyperbola having
asymptotes at an angle b0 with respect to the axis of symmetry. With r very large, it
can be seen from (6–35) that ncosb is then approximately one, and thus cosb0 ⬵ (1/n)
and
β0 = arc cos (1 n ). (6–36)
For convenience, a lens is said to be (a) “spherical” if the surfaces are generated by
a rotation about the line FQ2 and (b) “cylindrical” if it has symmetry along a focal line.
Risser (1949, ch. 11) addresses both “spherical” and “cylindrical” lenses. Parenthetically,
the F number of a lens is the ratio of the focal distance L to the lens diameter D; thus,
F number = L/D.
Effects of aperture tapering by primary feeds, and losses caused either by air-dielectric
reflections and absorption for dielectric lenses are addressed by Risser (1949, ch. 11).
6.8.7. Metal Plate Lenses

Ordinarily the electric field of a metal plate lens is oriented parallel to the planes of the
plates having spacing of dimension a. Then the plates are a guide for the waves providing
that a ≥ l0/2. If a ≤ l0/2, the spacing is said to be below critical and the plates are then
opaque. In accordance with equations (2–70) and (2–72), chapter 2, the wavelength of
the waves propagating within air-filled plates is
λ0
λg = (6–37)
1 − (λ 0 2 a )2
Thus, following (6–34),
λ0
n= = 1 − (λ 0 2 a )2 (6–38)
λd
Figure 6–23 illustrates a cylindrical metal-plate lens illuminated by an E-plane line

source feed. The plates are each identically shaped and spacing between plates is a con-
stant value a. Therefore n is determined
by (6–38) above.
Figure 6–24 shows ray paths through a
metal plate lens, which is symmetrical
about the line FQ1. Assume a focused
E antenna consists of a primary feed at
point F and the lens. Then the phase
along the aperture of dimension D is
a constant, and the electrical lengths
e e d FPP1 and FQQ1 are equal. Recall that
F
ane
E-pl electrical phase of a path of length l
FIGURE 6–23. expressed in radians is 2pl/l Thus, by
Cylindrical metal waveguide lens with plate equating the electrical lengths of paths
spacing a and illuminated by a vertically FPP1 and FQQ1 and dividing by 2p,
polarized line-source feed. one finds
Lens Antennas 265
L QQ1 r PP
+ = + 1
λ0 λg λ0 λ g
or
L r PP − QQ1 r L − r cos β
= + 1 = + (6–39)
λ0 λ0 λg λ0 λg
Finally, by substituting n for l0/lg in (6–39) and solving for r, one obtains
(1 − n ) L
r= (6–40)
1 − n cos β
where
l0 = wavelength in free space (air or vacuum)
lg = wavelength between the metal plates
The reader will notice that (6–40) for the metal plate lens is equal to (6–35) for the
dielectric lens. However, because n < 1, to make both the numerator and denominator
of (6–40) positive, one had to first multiply both by −1. As noted previously with n > 1,
(6–35) has the shape of a hyperbola. However, with n < 1, (6–40) has the shape of an
ellipse.
Generally, a metal plate lens is more frequency sensitive than a dielectric lens, in that
it focuses only for a relatively small bandwidth. Recall that (6–40) must be satisfied for
the lens to be perfectly focused. However, for waveguide, n is frequency dependent.
Thus, a metal plate lens can be perfectly focused only at one frequency. As with dielectric
lenses (Fig. 6–20), metal plate lenses are zoned (stepped) to reduce weight and to increase
bandwidth. Even so, depending on the desired side-lobe level, good metal-plate lens
performance is limited to bandwidths of
only a few percent (Kock 1946).
P P1
r
6.8.8. Geodesic Luneberg
Lenses
b
Fermat’s principle of least time is a fun- Q Q1
F
damental law of optics. It may be
expressed as follows: “When a ray passes L
from one point to another by any number D
of reflections or refractions, the path taken

by the ray is the one for which the corre-
sponding time of traversal is the least.”
Another relevant definition is that of a FIGURE 6–24.
geodesic: which is the shortest line Air and waveguide paths within a metal plate
between two points, where the line is lens.
constrained to follow a surface. Thus, the shortest path that a person can travel between
New York and London is a geodesic, because the shortest path is constrained to follow
the earth’s surface. Therefore, for a medium that provides a constant propagation velocity,
a natural result of Fermat’s principle is that the path of an electromagnetic wave is a
geodesic. Consequently, focusing occurs by forcing electromagnetic waves in air to travel
along curved surfaces.
To appreciate what a geodesic lens accomplishes, we first consider the length required
of a sectorial H-plane horn (Fig. 4–21, chapter 4), if it has an aperture width dH of 50l
at l = 0.1 m. The far-zone equation L = 2D2/l of Appendix F, that results in 22.5° phase
error across an aperture, indicates a required horn length L of 500 m. Obviously, this
length is impractical. However, as will now be discussed, a geodesic lens of 50l i.e., 5 m
diameter can provide a 5 m aperture with no phase error.
Geodesic lenses focus in one plane by having equal physical path lengths between a
point and a line, as now described. Advantages over dielectric lenses include very wide
bandwidth, and excellent impedance properties. Additionally, mechanical tolerances are
not severe, and appropriate manufacturing techniques coupled with good shop practice
can produce effective geodesic lenses at moderate cost.
A basic geodesic lens consists of two evenly spaced, rotational symmetric, conducting
surfaces. Ordinarily, propagation within the surfaces is in the TEM mode for which wave
velocity is that of free space. Thus, the focusing between a feed point and the aperture
results because all path lengths are equal, and consequently such a lens is focused over
a very wide bandwidth. Figure 6–25 shows an example lens with dashed lines indicating
two of the equal path lengths between a point and a straight line. The paths of the rays
as seen from a top view, when projected onto a planar surface, are the same as the paths
within a thin cylindrical slice cut through the center of a three-dimensional Luneberg
lens. Consequently, a geodesic lens focuses in only one plane.
Figure 6–26 shows cross sections of previously developed geodesic Luneberg lenses.
Note the input and output lips designated oa and ob at the edges (rims) of the lenses.
The tin-hat lens is symmetrical (ao and ob are identical) and has horizontal lips. Thus
focusing is maintained when a feed is rotated along the rim through 360°. However, the
input and output lips (ao and ob) of the helmet and clam-shell lenses are different and
thus destroy lens symmetry. These lips are provided to allow beam scanning with multi-
ple, rotating feeds over a limited angular
sector.
Line of
Constant Phase Figure 6–27 illustrates the use of a geo-
desic lens for exciting a line source reflec-
Focal tor feed for a scanning antenna. The
Point geodesic lens and nonfocusing flat-plate
extensions transform a point source at the
Conducting Surfaces lens feed horn into a linear wave front
FIGURE 6–25. along the reflector feed assembly. The
Paths through a geodesic lens for two of the linear wave front from the lens is directed
rays of a collimated beam (from Johnson diametrically opposite the lens feed point.
1962), courtesy of The Microwave Journal. Therefore, the angle of incidence at the
Lens Antennas 267
LENS AXIS
LENS AXIS
o o
o o
b b
a
a
TIN-HAT LENS HELMET LENS
LENS AXIS
o o
a b
CLAM-SHELL LENS
FIGURE 6–26.
Sectional views of mean surfaces for common types of geodesic lenses (from Johnson 1962),
courtesy of The Microwave Journal.
reflector feed assembly (and the scan LENS FEED SECTOR
angle of the radiated beam) is changed

by moving the lens feed-horn along the LENS FEED
periphery of the input lips. HORN
LENS
Geodesic Luneberg lenses have been RAYS
used with centimeter and millimeter θ
FLAT-PLATE
wavelength, rapid mechanical-scan anten- EXTENSIONS
nas. In a 16-GHz dual-beam radar appli-
cation (Hollis and Long, 1957), two θ
identical 127 cm clam-shell lenses were
housed in an antenna that scanned a 40°
azimuth sector alternately, each beam at
LINEAR WAVE REFLECTOR
a rate of 17 scans per second (see Fig. 7– FRONT FEED HORN
27, sec. 7.7.2). Half-power azimuth beam- FIGURE 6–27.
widths were 1.07°. The two beams were
Projected ray paths from geodesic lens to the
separated 1.85° in elevation and had 0.76° linear aperture of reflector feed horn.
half-power beamwidths. Focusing in ele-
vation was accomplished with one parabolic cylinder reflector, illuminated with two
vertically offset line source feeds.
LENS AXIS
MAIN SPINNINGS
REINFORCING RINGS
INSERT
SPINNINGS
SP ACER BUSHING
FEED ARC
FIGURE 6–28.
A sectional view through a 16 GHz geodesic Luneberg lens (from Hollis and Long, 1957;
© 1957, IEEE).
Figure 6–28 shows a cross section of one of the 16 GHz clam-shell lens. Each lens
consists of two closely nested aluminum spinnings; each spinning is made of two sec-
tions: a main spinning, and an insert that replaces a section of the main spinning between
the periphery and the reinforcing ring. This insert forms an internal feed arc that permits
sectoral feed horns to rotate in a plane about the lens axis. The insert subtends 110° on
the lens periphery, to permit the lens to focus for all feed positions over the 40° scan
sector. The fabrication tolerance for the lens surfaces was set at 32 1 inch. Although this
tolerance was maintained over most of the lens surface areas, greater deviations did exist.
Pattern measurements indicated a maximum side-lobe level of −25 dB.
A narrow-beam geodesic Luneberg lens was used in the AN/MPS-29 (XE-1), a 70-
GHz radar (see sec. 7.7.2). The lens diameter is 350l, that is, 60 inches (152 cm), that
provides a 0.2° scanned beam (Long, Rivers, and Butterworth 1960). Spacing between
the lens conducting surfaces is about l.4l. This dimension is oversized for the TEM
mode, yet it was successfully used to minimize lens attenuation. Measured side-lobe level
of the lens-reflector combination did not exceed −22 dB for any of the feed horns or scan
directions. The lens was spun of mild steel, machined, copper-plated, and protected with
a thin coating of Irilac, a low loss plastic surface coating. Templates for the conducting
surfaces were made on a digitally controlled profiler, and the surfaces were machined on
a tracer controlled boring mill.
A folded geodesic Luneberg lens is an entirely different configuration than the lenses
of Fig. 6–26 (Goodman et al. 1967). Its height-to-diameter ratio is reduced by “folding”
the lens surfaces. A helmet-type lens (Fig. 6–26) with one fold, located symmetrically
about the lens axis, was made for 70 GHz. To affect the height reduction, the upper-dome
sections of the lens were reversed to become bowls, with their troughs along the center-
line of the domes. The lens aperture was approximately 23 inches (58 cm) and its
maximum mean-surface height was 3.6 inches (9.4 cm). Measured side lobes of the
folded lens were, surprisingly, smaller than for the unfolded lenses. The side lobes were
“lost” in the receiver noise, which was at least 36 dB below the peak of the main lobe.
The −3 dB beam width was 0.56°, the same as expected for an unfolded lens. However,
the folding broadened the base of the main lobe at angles wider than the −10 dB angles.
Thus, the folded geodesic lens provided reduced profile height and lowered side lobes,
Pattern Calculations for Continuous Apertures 269
and these were achieved with negligible degradation of antenna gain or beamwidth. This
lens was installed in an azimuth scanning, 70 GHz armored-vehicle mounted radar that
provided adequate vision for night driving (Dyer and Goodman 1972).
Geodesic Luneberg lenses have been used in other types of scanners. For example,
Allen (1956) designed a tin-hat lens that was sector-scanned with multiple waveguides
in a configuration called organ-pipe feed; and Schaufelberger (1960) employed the ultra-
wide-band millimeter wavelength feature of a geodesic Luneberg lens to provide the
detection sensitivity needed for target detection by radiometry.
6.9. Beam Steering by Feed Offset

The Luneberg lens described above is an ideal device for producing a beam that changes
direction when the feed alone is moved, without moving the entire antenna. As indicated,
the beam direction will be along the line from the feed phase center through the center
of the lens. The quality of the beam will be equally good for all feed positions on the
lens surface; that is, the gain, beamwidth, and side-lobe level will remain constant for
all beam positions.
Obviously, much less mechanical energy is required to move a small feed horn than
to move an entire antenna. This requires, however, a flexing or rotating joint in the
waveguide so that the feed horn can be moved while the transmitter remains stationary.
An alternative is to place a number of horns at fixed points to provide fixed beams in
desired directions, and then to switch from one to another in accordance with the require-
ments of the particular application.
A beam may be similarly steered with an ordinary lens or with a parabolic reflector,
by moving the feed off the lens or reflector axis but keeping it in the focal plane. The
off-axis collimation of a beam of rays using an offset feed with an ordinary lens is shown
in Fig. 6–29a. The same effect with a parabolic reflector is shown in Fig. 6–29b. However,
the angular deviation q of the beam from the axial direction that can be obtained in this
way is much more limited with ordinary lenses than with the Luneberg lens, and still
more limited with a parabolic reflector. As the steering angle is increased, the beam shape
deteriorates, resulting in increased beamwidth, decreased directivity, and increased side-
lobe level. At large values of offset of the feed, the beam deterioration becomes excessive;
in fact, the pattern can hardly be referred to as a beam, and it is not usable as such.
The permissible scan angle (or steering angle) by the feed-offset method is roughly
proportional to the on-axis beamwidth and is generally of the order of a few beamwidths.
The maximum scan angle is greater with large than with small f/D ratios. It may also be
increased with lenses by the use of special surface curvatures, and with reflectors by
combining special curvature with special feed design.
6.10. Pattern Calculations for Continuous Apertures

The aperture of an antenna is an area through which the radiation passes, and it is a plane
surface near the antenna and perpendicular to the direction that the maximum radiation
passes. Pattern calculations for continuous apertures are commonly made with equations
Lens
q
Focal Lens
point axis
Plane
wavefront
Source Collimated rays
Spherical
wavefront
(a)
Plane
wavefront
Focal point
Reflection
q Parabolic
axis
Collimated rays
Focal plane Source

(offset feed)
(b)
FIGURE 6–29.
Off-axis beam formation by (a) a lens and (b) a reflector.
that are based on sound electromagnetic principles, yet certain assumptions called high-
frequency, scalar approximations are used (Silver 1949, pp. 169–99). High frequency
refers to the aperture being large enough that the effects of radiation from aperture edges
are negligible; and scalar refers to the field over the aperture being almost completely
linearly polarized, with only a small fraction of the energy being cross-polarized.
6.10.1. Historical Note on Aperture Analyses

In his classic antenna book, Silver (1949) describes microwave antenna theory and hard-
ware developments made during World War II at the MIT Radiation Laboratory. There
he references (Silver 1949, pp. 182–95) research reports on the application of mathemati-
cal techniques for calculating aperture antenna patterns. By 1946, antenna specialists
knew that far-zone radiation patterns could be calculated, in principle, using the Fourier
integral of aperture distributions. At that time, however, user-friendly computer means

for doing aperture calculations and pattern plotting were not available.
Beginning in the 1950s, antenna developments within the Engineering Experiment
Station, Georgia Institute of Technology (now Georgia Tech Research Institute), benefited
from the use of an analog Fourier integral computer. Searcy Hollis built the computer
and designed it specifically for antenna analyses (Hollis, 1956). A paper by Clayton and
Hollis (1960) describes the development of this Fourier integral computer, its use in cal-
culating and plotting antenna patterns, and how linear aperture distributions can be used
to determine patterns versus q and f. The paper also describes a version of the Fourier
integral computer manufactured and sold by Scientific Atlanta, Inc. That computer used
a phototube reader that inputted data from graphs of aperture amplitude and phase dis-
tributions, and computed patterns were printed on a paper recorder.
Later, calculated patterns versus q and f became more easily attainable through use
of general-purpose digital computers. Often the pattern calculations incorporated previ-
ously developed mathematical equations, for approximating the patterns of continuous
line sources. Nowadays, generally available personal computers (PCs) can readily provide
detailed pattern calculations using point-to-point amplitude and phase data over an entire
aperture. Consequently, today there is an extensive body of knowledge on using PCs for
antenna aperture analyses (Diaz and Milligan 1996).
6.10.2. Continuous Line Source

As with planar arrays, radiation pattern calculations for line sources are basic to the
understanding of pattern calculations for continuous planar apertures. Figure 6–30
describes the coordinate system used for the material that follows. Based on equations
z
x
b
q
q
a
y a z
f
b y
f
x
(a)
(b)
FIGURE 6–30.
Coordinates for a rectangular aperture located in the x-y plane.
given in Johnson (1993, p. 2–15), the normalized radiation pattern in the x-z plane of a
continuous line source on the x-axis may be expressed as
2π x sin θ ⎞ ⎤
A ( x ) exp ⎡⎢ j ⎛⎜
1 L2
E (θ ) = ∫ ⎟⎠ ⎥ dx (6–41)
L − L 2 ⎣ ⎝ λ ⎦
where E(q) = electric far-field pattern versus observation angle q, normalized for peak
amplitude of unity
q = observation angle measured from normal to aperture
L = overall length of aperture
x = position along the aperture: −L/2 ≥ x ≤ L/2
A(x) = a(x)eja(x)
a(x) = aperture amplitude versus position x
a(x) = aperture phase versus position x
The analogy between equations for linear arrays and linear continuous apertures may be
apparent. Equation (6–41) is a summation of amplitude and phase contributions, at
each observation angle q from aperture currents of infinitesimal length dx. For mathe-
matical convenience, solutions are usually expressed in terms of the variable u, where
πL
u= sin θ .
λ
The factor 1/L normalizes (6–41), so that the maximum of E(q) is unity. Thus,
an unnormalized pattern g(q) in the x-z plane for a continuous line source along the
x-axis is
2π x sin θ ⎞ ⎤
A ( x ) exp ⎡⎢ j ⎛⎜
L 2
g (θ ) = ∫ ⎟⎠ ⎥ dx (6–42)
−L 2 ⎣ ⎝ λ ⎦
To provide a feeling for the mathematics of (6–42), its solution will be obtained
when its integrand is simplified by letting A(x) = 1. Thus, the radiation pattern
will be determined for a line source having constant amplitude of unity and zero
phase along its length. To accomplish this, two mathematical relationships are used;
namely,
1
∫ exp (cz ) dz = c exp (cz )
and
[exp ( jq )] − [exp (− jq )] = [(cos q + j sin q )] − [(cos q − j sin q )] = 2 j sin q

Then with A(x) = 1, (6–42) becomes
L 2   2π x sin θ   dx = 1 exp  2π  L sin θ − exp  − 2π  L sin θ 

∫− L 2 A ( x ) exp  j  λ
  π   λ  2  
λ 2 
 j   sin θ
2
 λ 
πL
sin  sin θ 
1 2 j  2π  L  =L   λ   = L sin u
=  sin   sin θ   (6–43)
π  
 λ 2 π
j   sin θ     sin θ
2 L u
 λ   λ 
πL
where u = sin θ . Now, since sin u/u equals unity when u is zero, the normalized
λ
amplitude pattern for a uniformly illuminated line source is
sin u
E (u ) = (for constant amplitude and phase ) (6–44)
u
Another line amplitude distribution frequently used is cosine amplitude and constant
phase. In this case, the amplitude is made unity at the aperture center and zero at its
edges. Then A(x) = a(x) = cos(px/L) for −L/2 ≤ x ≤ L/2. From Silver (1949, p. 187) and
other references, the pattern obtained using this cosine distribution with (6–41) is
cos u
E (u ) = 2
(for cosine amplitude distribution ) (6–45)
1−  
2u
π
πL
where u = sin θ .
λ
Graphs of E(q) from (6–44) and (6–45), expressed in decibels, for the uniform and
cosine amplitude distributions are included in Fig. 6–31. There the first side lobes for
the uniform distribution are only 13.2 dB less than the peak of the major lobe. On the
other hand, the peak side lobes for the 0
Pattern strength (dB)
cosine distribution are smaller, being

23 dB below the major lobe. –10
Mathematical solutions for the radia- –20

tion patterns of a variety of continuous
–30
linear aperture distributions are tabulated
in various books (see, e.g., Silver 1949, –40
–10 –5 0 5 10
p. 187). Table 6-1 includes the half-power u
beamwidths and maximum side-lobe FIGURE 6–31.
levels for several aperture amplitude dis-
Normalized pattern amplitudes expressed in
tributions for which the phase is assumed decibels versus u = p(L/l)sin q. — Constant
constant over the aperture. In Table 6-1, aperture amplitude distribution; . . . . . . Cosine
aperture dimensions are normalized so aperture amplitude distribution.
TABLE 6-1 Beamwidth, Maximum Side Lobes, and Relative Gain Versus Linear Aperture
Amplitude Distribution (from Silver 1949, p. 187)
Type of Distribution Half-Power Intensity 1st side

−1 ≥ x ≤ 1 Beamwidth (degrees) lobe (dB below max) Relative Gain
a(x) = 1 50.8 l/L 13.2 1.00
a(x) = 1 − 0.2x2 52.7 l/L 15.8 0.99
a(x) = 1 − 0.5x2 55.6 l/L 17.1 0.97
a(x) = 1 − x2 65.9 l/L 20.6 0.83
a(x) = cos(px/2) 68.8 l/L 23.0 0.81
a(x) = cos2(px/2) 83.2 l/L 32.0 0.67
a(x) = 1 − |x| 73.4 l/L 26.4 0.75
that coordinate x is between −1 and +1. Notice that the first side lobes are appreciably
reduced by the tapering of amplitudes to small values at the aperture edges. The relative
gain in Table 6-1 is commonly called aperture efficiency. It is the gain relative to the
gain available if the aperture were illuminated with constant amplitude and constant
phase.
In the previous discussions, the phase over the aperture is assumed constant. A major
effect of aperture phase errors is to reduce the gain and broaden the major lobe. In addi-
tion, phase errors reduce the depth of pattern minima between the lobes and raise the
side lobes. Very large phase errors over the aperture can cause the main lobe to be split
into parts and cause large side lobes. For a detailed discussion on phase errors along a
continuous aperture, the reader is referred to Johnson (1993, pp. 2-21–2-27).
6.10.3. Aperture Analysis: General

A discussion follows on the calculation of patterns for a planar aperture located in the x-y
plane of Fig. 6–30. Let the excitation on the aperture be A(x, y) = a(x, y) exp[ ja(x, y)],
where a(x, y) describes the amplitude and exp[ja(x, y)] the phase. The beam points in
the positive z direction. Equation (6–46) below is the usual starting point for narrow
beam antennas. More exact patterns can be obtained simply by multiplying the patterns
from (6–46) by (1 + cos q)/2, the obliquity factor (see Appendix G or Silver 1949, p.
173). Usually, however, the obliquity factor is assumed to be unity. This is because the
primary interest in aperture analysis is for narrow beam antennas and small q values, so
that the obliquity factor is essentially unity.
The unnormalized electric field pattern for a rectangular aperture can be expressed
as
2π
a2 b2 j sin θ ( x cos φ + y sin φ )
g (θ , φ ) = ∫ A ( x, y ) e
a 2∫ b 2
λ dxdy (6–46)
− −
and it expresses the relative electric field strength at a fixed distance from the aperture
origin versus the angles q and f. Here a is the aperture length along the x-axis, and b is the
length along the y-axis. Note that the aperture distribution A(x, y) is a function of two vari-
ables, and thus it describes the amplitude and phase at the aperture’s x and y coordinates.
With reflector and lens antennas, often the dependencies of A(x, y) on x and y are interde-
pendent so that it is difficult or impossible to describe A(x, y) with a simple equation. Then,
it may be necessary to solve (6–46) by point-to-point integrations over the aperture.
The simplest g(q, f) calculation is one for a line source along either the x- or the y-
axis. Then from (6–46), for an x-axis continuous distribution
2π
a2 j x sin θ cos φ
g (θ , φ ) = ∫ A(x)e λ dx (6–47a)
−a 2
because all y values are zero. Notice that the integrand in (6–47a) is the same as in (6–42),
except that the product sin q cos f replaces sin q. Consequently, solutions of (6–47a) are
the same as solutions for (6–42), except that their solutions in terms of the variable u,
where u = (pL/l) ⋅ sin q, are replaced by
Ux = (π a λ )⋅ sin θ cos φ (6–47b)
Likewise, for a y-axis continuous distribution, for which all x coordinates are zero
2π
b2 j y sin θ sin φ
g (θ , φ ) = ∫ A ( y) e λ dy (6–48a)
−b 2
Then, similar to solutions for (6–47a), solutions of (6–48a) are the same as solutions for
(6–42), except that the solutions in terms of the variable u, where u = (pL/l) ⋅ sin q, are
replaced by
U y = (π b λ )⋅ sin θ sin φ (6–48b)
6.10.4. Apertures Having Separable x and y Distributions

If the aperture distribution A(x, y) at the x coordinates is independent of the y coordinates
and visa versa, it can be expressed as
A ( x, y ) = A1( x )⋅ A2 ( y ) = a1( x ) exp [ jα1( x )]⋅ a 2 ( y ) exp [ jα 2 ( y )] (6–49)
Equation (6–49) is said to be mathematically separable or factorable. Separable equations

are applicable to the case of a line source feed that illuminates a parabolic cylinder or a
cylindrical lens. For example, suppose the line source and the focal line of the cylinder
coincide along the y-axis. Then the excitation along the line source defines the term A2(y).
Now suppose the radiation pattern about the y-axis is shaped by a flared line source feed
horn. Then the shaped radiation pattern and the space attenuation determine the excitation
A1(x). Thus, A(x, y) is separable and (6–46) can be simplified to become
2π 2π
a2 j x sin θ cos φ b2 j y sin θ sin φ
g (θ , φ ) = ∫ A1( x ) e λ
∫
dx ⋅ A2 ( y) e λ dy (6–50)
−a 2 −b 2
Equation (6–50) is simplified further if patterns only in the x-z and y-z planes are
considered. Then, the two principal plane patterns are dependent on aperture distributions
along the x- and y-axes, as follows:
2π
x sin θ
g (θ , 0 ) =  ∫
j
A2 ( y )dy  ⋅ ∫
b2 a2

A1( x ) e λ dx ( X -Z plane ) (6–51a)
 −b 2  −a 2
2π
y sin θ
g (θ , 90 ) =  ∫
j
A1( x )dx  ⋅ ∫
a2 b2
A2 ( y )e λ dy ( y-z planee ) (6–51b)
 −a 2  −b 2
Notice that the bracketed terms of (6–51a) and (6–51b) are constants. Also, notice that
the integrals outside the bracketed terms are similar in form to (6–42). Thus, except for
a multiplicity factor, (6–51a) and (6–51b) for planar apertures are describable by radia-
tion patterns of line sources. Therefore, knowledge of linear aperture patterns is often
beneficial to an understanding of the patterns of continuous planar apertures.
The simplest planar aperture distribution has unity amplitude and zero phase, and its
pattern versus q and f can be determined with relative ease. For this case,
A ( x, y ) = 1⋅ exp 0 × 1⋅ exp 0 = 1.
Then, since A(x, y) is separable, g(q, f) is the product of two integrals and each is the
pattern of a line source of constant amplitude and phase. Thus, from (6–47b), (6–48b),
and (6–50), the radiation pattern for a distribution of constant amplitude and phase
becomes
 sin  π a sin θ cos φ    sin  π b sin θ sin φ  

 sin U x   sin U y    λ    λ 
g (θ , φ ) = A  ⋅   = A   ⋅   (6–52)
 U x   U y   π a sin θ cos φ   π b sin θ sin φ 
 λ   λ 
where A is the aperture area (a times b).

Figure 6–32 is a plot of 20 log |g(q, f)| versus q and f) of (6–52), normalized to zero
dB. From this figure it can be seen that the patterns in the principal planes (x-z and y-z)
are similar, and they are identical if the aperture dimensions a and b are equal. From
(6–44), it is known that each principal plane pattern, when normalized to unity, has the
form sin u/u. Here u is (p w/l)sin q with w being the aperture width within the plane of
the pattern. As previously noted; the function sin u/u is at its maximum of unity at u =
0 and its first side lobes are relatively large, being only 13.2 dB below the peak of the
major lobe.
6.10.5. Equivalent Linear Aper-

tures in Principal Planes
Equations (6–51a) and (6–51b) give prin-
cipal plane patterns for a planar aperture,
expressed as functions of separable linear
aperture distributions. In theory, principal
plane patterns for nonseparable aperture
distributions also may be expressed in a
closed, integratable mathematical form.
However, in practice, principal plane pat-
terns often are determined by numeri-
φ = 0° φ = 90°
cal integration of measured or estimated
aperture distributions. The concept of
equivalent linear aperture distributions, φ = 45°
now discussed, is applicable to antennas FIGURE 6–32.

having nonseparable aperture distribu- Radiation pattern for a uniformly illuminated
tions. It can provide insight into how a square aperture, graphed in decibels. Courtesy
principal plane pattern is affected by of J. W. Cofer, Georgia Tech Research
changes in the amplitude and phase dis- Institute, calculations made circa 1969.
tribution of a planar aperture.
The pattern g(q, f = 0°) in the x-z principal plane for an equivalent line source along
the x-axis can be obtained by rearranging the terms in (6–46) as follows:
2π
x sin θ
g (θ , φ = 0°) = ∫
a2
 b 2 A ( x, y ) dy  ⋅ e j
− a 2  ∫− b 2
λ dx (6–53a)

Similarly, the pattern g(q, f = 90°) in the y-z principal plane for an equivalent line source
along the Y-axis, can be obtained by rearranging the terms in (6–46) as follows:
2π
y sin θ
g (θ , φ = 90°) = ∫
b2
 a 2 A ( x, y ) dx  ⋅ e j
− b 2  ∫− a 2
λ dy (6–53b)

Now compare (6–53a) with (6–42) for the x-z plane pattern of an x-axis linear aperture.
Note that the bracketed integral of (6–53a) functions for (6–42) as the equivalent linear
aperture distribution along the x-axis. However, this bracketed integral (the equivalent
linear aperture distribution) is a function of x and y. Thus, to evaluate the bracketed
integral at each x coordinate, A(x, y) is integrated over all y coordinates between −b/2
and b/2. Likewise, the pattern in the f = 90° plane is attained from (6–53b). The principles
of evaluating (6–53a) and (6–53b) are beneficial in envisioning the effects of aperture
tapers, as is discussed below.
A simple mathematical example is now given, which is that of determining the equiva-
lent linear aperture distribution and the pattern for a uniformly illuminated circular
aperture. Then, the aperture amplitude may be expressed as A(x, y) = 1 over the circular
aperture, but it is zero beyond. For specificity, let the radius of the aperture be unity.
Then, the x and y coordinates of the outer aperture edges are interrelated by x2 + y2 = 1.
In other words, at each x-value, the y coordinate at the aperture edge is ye = (1 − x2)1/2.
Thus, for the bracketed term of (6–53a), the limits of integration b/2 and −b/2 may be
replaced by ye/2 and −ye/2. Consequently, the bracketed term of (6–53a) then becomes
1⋅ dy = ye = (1 − x 2 )
b2 ye 2 12
∫−b 2 A ( x, y ) dy = ∫− y e 2
(6–54)
Thus, even though the amplitude A(x,y) is unity when not zero, it is in fact a nonseparable
amplitude distribution. This is because the limits of integration, and therefore the inte-
grated amplitude along a y coordinate, are dependent on its x-value.
From (6–54) above, it is seen that the effective illumination along the x-axis is math-
ematically tapered toward each aperture edge. In other words, the bracketed term of
(6–53a), that is, equivalent linear aperture distribution, is (1 − x2)1/2. Thus, (6–53a)
becomes
2π
x sin θ
(1 − x )
a2 j
2 12
g (θ ) = ∫ e λ dx (6–55)
−a 2
Consistent with the original assumption, g(q) of (6–55) is the equation for a principal
plane pattern of a uniformly illuminated circular aperture. The beamwidth and maximum
side-lobe level for this aperture distribution are included in Table 6-2 (distribution:
f(r) = 1) of the next section. Parenthetically, by using variable instead of fixed integration
limits when determining an equivalent linear aperture distribution, the above example
shows that g(q) can be determined using (6–53a) or (6–53b) for planar apertures that are
not rectangular.
The determination of equivalent line-source apertures for nonseparable distributions
is usually more difficult than in the above example. However, the concept of equivalent
linear apertures helps to show how aperture amplitude shaping of planar apertures affects
side lobes and beamwidths, similar to how amplitude tapering affects the pattern of a
linear aperture. Relationships between the physical shaping of a planar aperture and the
tapering of an effective linear aperture are discussed also in connection with Fig. 6–6.
TABLE 6-2 Beamwidth, Maximum Side lobes, and Relative Gain Versus Circular
Aperture Distribution (from Silver 1949, p. 194)
Type of Distribution Half-Power Intensity 1st Side Lobe

0≤r≤1 Beamwidth (degrees) (dB below max) Relative Gain
f(r) = 1 58.9 l/D 17.6 1.00
f(r) = (1 − r 2) 72.7 l/D 24.6 0.75
f(r) = (1 − r 2)2 84.3 l/D 30.6 0.56
6.10.6. Circular Apertures

Beamwidth, maximum side lobes, and
relative gain for three circular apertures
are given in Table 6-2. Each aperture dis-
tribution is a function of radius r, where
r is normalized to unity. Equations of the
patterns for these distributions have been
solved in closed analytical form (Johnson,
p. 2–21). Figure 6–33 shows the ampli-
tude of the radiation pattern, expressed
in decibels, of a circular aperture that φ = 90°
φ = 0°
is illuminated with constant amplitude
and phase. There are major differences
φ = 45°
between this pattern and the pattern for a
uniformly illuminated square aperture of FIGURE 6–33.
Fig. 6–32. The first side lobes for the cir- Radiation pattern for a uniformly illuminated
cular aperture are −17.6 dB with respect circular aperture, graphed in decibels. Courtesy
to the major lobe peak and the −3 dB of J. W. Cofer, Georgia Tech Research
Institute, calculations made circa 1969.
beamwidth is 58.9l /D degrees. On the
other hand, for a square aperture, the
comparable numbers along the principal planes are −13.2 dB and 50.8l/L degrees. Thus,
the effect of the circular shape, relative to the square shape, is to decrease (taper) the
aperture amplitude toward the outer edges.
6.10.7. Patterns Versus q and f: A Summary

As already discussed, a pattern versus q and f is relatively easy to determine, if the
aperture distributions are separable and the linear aperture patterns for both the x- and y-
axis distributions are known. Then, as illustrated in the remainder of this section, patterns
are determined as the product of two line-source patterns, in accordance with (6–50).
Otherwise, calculations of patterns versus q and f require computational intensive, point-
to-point numerical calculations in accordance with (6–46). For apertures with nonsepara-
ble distributions, patterns for the principal planes are much simpler to calculate than
patterns versus q and f. These principal plane patterns are calculated by obtaining an
equivalent line aperture for each principal plane and then integrating each line aperture
distribution over the relevant line aperture following (6–53a) and (6–53b).
The remainder of section 6.10.7 focuses on patterns for separable distributions. Previ-
ously, (6–52) was derived to show how the simplest of patterns versus q and f is obtained.
Its aperture distribution is separable, because the amplitude and phase are constant. Another
separable planar aperture distribution is now discussed, that includes constant amplitude
a1(x) along the x-axis and a cosine-squared amplitude distribution a2(y) along the y-axis.
The radiation patterns for the two distributions, A1(x) and A2(y), are now used to
obtain the pattern versus q and f. The normalized pattern for a uniform line-source
distribution, from (6–44), is
E1(u ) = 
sin u 
 u 
(6–56)
πa
with u = sin θ for a pattern in the plane of the line source.
λ
The cosine-squared distribution along the y-axis is unity at the aperture center and
zero at each end, and the phase is assumed to be zero. Then, A2( y) = cos2(p y/b), with
−b/2 ≥ y ≤ b/2; and from Johnson (1993, p. 2–16), the normalized pattern for a line-source
with a cosine-squared distribution is
sin u   π 2 
E 2 (u ) = 
 u   π 2 − u2 
(6–57)
πb
with u = sin θ for a pattern in the plane of the line source.
λ
To obtain the pattern E(q, f) for a planar aperture having line source distributions in
accordance with (6–56) and (6–57),
−u in E1(u ) of (6 −56 ) is replaced by U x to obtain E1(U x ) ,
and
−u in E 2 (u ) of (6 −57) is replaced by U y to obtain E 2 (U y ) .
Then, the pattern E(q, f) is
E (θ , φ ) = E (U x , U y ) = E1(U x )⋅ E 2 (U y )
(6–58)
where Ux = (pa/l) ⋅ sin q cos f and Uy =

(pb/l) ⋅ sin q sin f.
Figure 6–34 includes a pattern graphed
in decibels for an aperture having a
uniform distribution along the x-axis and
φ = 0° φ = 90° cosine-squared distribution along the y-
axis. Notice that the side lobes of the f =
φ = 45° 45° pattern are much smaller than along
either the f = 0° or f = 90° pattern cuts,
FIGURE 6–34.
and this also was the case for the pattern
Radiation pattern graphed in decibels, with of Fig. 6–32. Generally, however, the
uniform amplitude along one axis and cosine-
squared tapered amplitude along the other axis. side-lobe levels in a f = 45° pattern
Courtesy J. W. Cofer of Georgia Tech Research depends on the relative a and b widths of
Institute, calculations made circa 1969. the aperture, in addition to the detailed
Comparison of Directive Antenna Types 281
shapes of the aperture distributions. Consequently, the side lobes for f = 45° patterns are
not always smaller than those of both the f = 0° or f = 90° patterns. However, for some
operational scenarios, the effect of a side-lobe reduction in a plane, without loss of gain,
sometimes may be obtained by rotating the aperture 45° about its z-axis.
6.11. Comparison of Directive Antenna Types

Two somewhat different ways of achieving directional antenna patterns have been
described—the method of arrays in chapter 5, and the method of ray collimation by
reflectors and lenses in the present chapter. Each of the two methods has advantages and
disadvantages; in some applications one is preferred, in other applications the other is
better. Arrays have been more commonly used at lower frequencies, and now they are
being used more or less commonly at frequencies up to 10 GHz. Although arrays are not
commonly used above about 30 GHz, they may be used up to at least 100 GHz (Emrick
and Volakis 2007). Reflectors and lenses are most common above 3 GHz, although para-
bolic reflectors are fairly common down to at least 100 MHz. Lenses are basically micro-
wave devices, not ordinarily used below about 3 GHz.
In the frequency region from about 300 to 3,000 MHz, and exclusive of the need for
rapid electronic beam positioning or shaping, the choice between an array and a parabo-
loidal reflector may sometimes be difficult. Reflectors are favored when the desired
beamwidth is less than a few degrees, and arrays are favored when high aperture effi-
ciency is desired. Additionally, reflectors are usually indicated if broadband operation or
low noise temperature (see secs. 7.1 and 7.8) are required.
Electronic scanning arrays are, generally, expensive but they may be the only solution
for applications requiring rapid beam pointing, adaptive beam shaping, or a combination
of rapid beam pointing, beam shaping, and signal processing (see chapter 8). However,
a combination of electronically controlled primary feed and a reflector or a lens may be
more cost-effective and lighter. For beam scanning only in an angular sector at speeds
up to at less 20 scans per second, mechanically moving feeds in combination with reflec-
tors or lenses have been successful (see sec. 7.7.2).
The principal advantage of lenses over reflectors is that the feed and feed-
support structures do not block the aperture, since the rays are transmitted through the
lens rather than returned toward the feed. Lenses can also be designed to operate satis-
factorily with the feed farther off the optical axis, and hence are useful in applications
requiring a beam that can be moved angularly with respect to the axis (although, as has
been mentioned in sec. 6.3.5, this is possible in a moderate degree with reflectors). A
further advantage is that permissible mechanical tolerances are somewhat greater for
lenses than for reflectors. On the other hand, lenses are usually bulkier, heavier, more
complicated to design and construct, and more expensive for the same gain and beam-
width than are reflectors. These factors become somewhat less significant at the very
short wavelengths, above 10 GHz, and it is in this region that lenses are most commonly
used.
References
Adams, R. J., and K. S. Kelleher, “Pattern Calculation for Antennas of Elliptical Aperture,”
Proceedings IRE, vol. 38, September 1950, pp. 1052 ff. See also K. S. Kelleher,
“Reflector Antennas,” ch. 12 in Antenna Engineering Handbook, H. Jasik (ed.),
McGraw-Hill, 1961.
Allen, C. C., “Geodesic Lens With Organ Pipe Feed,” Record of the Georgia Tech-SCEL
Symposium on Scanning Antennas,” Georgia Institute of Technology, 18–19 December
1956, pp. 297–314.
Balanis, C. A., Antenna Theory: Analysis and Design, 3rd ed., John Wiley & Sons, 2005,
ch. 13.
Bodnar, D. G., “Lens Antennas,” ch. 18 in Antenna Engineering Handbook, 4th ed., J. L.
Clayton, L., Jr., and J. S. Hollis, “Calculation of Microwave Antenna Radiation Systems
By the Fourier Integral Method,” Microwave Journal, Sept. 1960, pp. 59–66.
Diaz, L., and T. Milligan, Antenna Engineering Using Physical Optics, Artech House,
1996.
Dyer, F. B., and R. M. Goodman, Jr., “Vehicle Mounted Millimeter Wave Radar,” 18th
Tri-Service Radar Symposium Record, University of Michigan, 1972. Reprinted in
S. L. Johnston, Millimeter Wave Radar, Artech House, pp. 485–506, ch. 19.
Emrick, R., and J. L. Volakis, “Millimeter-Wave and Terahertz Antennas,” ch. 23 in
Goodman, R. M., Jr., R. C. Johnson, H. A. Ecker, and W. K. Rivers, Jr., “A Folded
Geodesic Luneberg Lens Antenna,” presented at the 17th Annual Symposium on USAF
Antenna Research and Development at the University of Illinois, November 1967.
Hannan, P. W., “Microwave Antennas Derived from the Cassegrain Telescope,” IRE
Transactions on Antennas and Propagation, March 1961, 140–53.
Hollis, J. S., “A Fourier Integral Computer for Calculation of Antenna Radiation Patterns,”
MSEE Thesis, Georgia Institute of Technology, 1956.
Hollis, J. S., and M. W. Long, “Luneberg Lens Scanning System,” IRE Transactions on
Antennas and Propagation, vol. AP-5, January 1957, pp. 21–25.
Imbriale, W. A., and D. L. Jones, “Radio-Telescope Antennas,” ch. 49 in Antenna
Johnson, R. C., “The Geodesic Luneberg Lens,” Microwave Journal, vol. 5, August 1962,
pp. 76–85.
Johnson, R. C., Antenna Engineering Handbook, 3rd ed., McGraw-Hill, 1993.
Johnson, R. C., and R. M. Goodman, Jr., “Geodesic Lenses for Radar Antennas,” Record
of IEEE EASCON ’68, Washington, DC, 1968, pp. 64–69.
Kock, W. E., “Metal Plate Antennas,” Proceedings of the IRE, November 1946, pp.
828–36.
Long, M. W., W. K. Rivers, Jr., and J. C. Butterworth, “Combat Surveillance Radar
AN/MPS-29 (XE-1),” Record of the Sixth Annual Radar Symposium, University of
Michigan, 1960. Reprinted in S. L. Johnston, Millimeter Wave Radar, Artech House,
pp. 461–78, 1980.
Love, A. W., “Horn Antennas,” ch. 15 in R. C. Johnson, Antenna Engineering Handbook,

3rd ed., McGraw-Hill, 1993.
Kraus, J. D., and R. J. Marhefka, Antennas, 3rd ed., McGraw-Hill, 2002, pp. 352–64.
Muehldorf, E. I., “The phase center of horn antennas,” IEEE Transactions on Antennas
and Propagation, vol. AP-18, November 1980, pp. 753–60.
Risser, J. R., “Waveguide and Horn Feeds,” ch. 10, in S. Silver, Microwave Antenna
Theory and Design, McGraw-Hill, 1949.
Risser, J. R., “Dielectric and Metal Plate Lenses,” ch. 11 in S. Silver, Microwave Antenna
Theory and Design, McGraw-Hill, 1949.
Ruze, J., “Antenna Tolerance Theory: A Review,” Proceedings of the IEEE, April 1966,
633–40.
Schaufelberger, A. H., “An Airborne High Speed Scanning Antenna for a Millimeter
Radiometer,” Symposium Record on the USAF Antenna Research and Development
Program, Vol. II, University of Illinois, Monticello, Illinois, 3–7 October 1960.
Taylor, T. T., “Design of Line-Source Antennas for Narrow Beamwidth and Low Side
Lobes,” IRE Transactions on Antennas and Propagation, January 1955, pp. 16–28.

1. A circular-aperture paraboloidal reflector has a focal length of 10 feet. The depth
of the reflector (zap in Fig. 6–2) is 6.4 feet. (a) What is the aperture diameter,
Da? (b) What is the f/D ratio? (Hint: Use equation (6–1), noting that when
z = zap, x = D/2.)
2. A paraboloidal reflector of circular aperture has a diameter Da = 30 feet. The
frequency of operation is 450 MHz. (a) If the constant k1 of equation (6–12) is
70 degrees for this antenna, what is the beamwidth? (b) If k2 in equation (6–13)
is 0.6, what is the directivity, D? Express this result as both a power ratio and
in decibels. (Note: All length and area quantities in equations (6–12) and (6–13)
must be expressed in a consistent system of units.)
3. A paraboloidal reflector has an aperture dimension Da = 40 feet, in a particular
plane. The focal length is 12 feet. (a) What is the angle (2b) subtended by the
reflector as viewed from the focus, in this plane? (b) If a primary horn feed of
beamwidth 100 degrees in this plane is placed at the focus of this reflector,
approximately what will be the illumination edge taper (t2)? (c) Approximately
what will be the side-lobe level, in decibels?
4. A rectangular-aperture parabolic-cylinder reflector is to be operated with a uniform
collinear-dipole array as the feed radiator. The cylindrical axis and the array are
aligned horizontally. It is desired to obtain a horizontal beamwidth of 2 degrees
and a vertical beamwidth of 20 degrees. (Note: The horizontal beamwidth is
determined by the array, whereas the vertical beamwidth is determined by
the parabolic aperture.) The wavelength of operation is l = 0.25 meter
( f = 1200 MHz). The array spacing d of equation (5–57) is 0.7l, and the constant
k1 of equation (6–12) is 60 degrees. It is deemed advisable to extend the horizontal
dimension of the reflector a distance of one wavelength beyond the centers of

the end elements of the array. What must be the dimensions of reflector aperture?
Hints: The horizontal beamwidth will be the same as that of a uniform linear
array of isotropic elements; the element pattern in this case has virtually no
effect, and the reflector does not affect the horizontal beamwidth. The length of
an array of n elements, between the centers of the end elements, is (n − 1)d. In
solving equation (5–57) to find the required value of n, the initial solution will
not be an exact integer; but of course n must be an integer. Therefore, take for
n the nearest integer to the nonintegral value found from equation (5–57).
5. A 90-degree corner-reflector antenna operating at a wavelength of one meter
( f = 300 MHz) has a dipole feed positioned one-half wavelength from the vertex
of the corner. Approximately what should the side length (l) and the aperture
dimension (Da) be? Express the result in meters, and also in feet and inches.
6. State whether you would use an array, a paraboidal reflector, or a lens to meet
the following antenna requirements: (a) A beamwidth of 50 degrees in both
principal planes, at a frequency of 50 MHz. (b) A beamwidth of approximately
one degree in both principal planes, at a frequency of 30 GHz, with considerable
capability of angular beam motion by off-axis feed positioning, and freedom
from aperture-blocking effects. (c) A beamwidth of approximately one degree in
a fixed direction in both principal planes, at a frequency of 5 GHz, with the
lightest possible weight, lowest cost, and good bandwidth properties.
7. List three advantages of a paraboloidal-reflector antenna over a lens antenna, for
the same aperture size.
8. List three advantages of a lens antenna over a reflector antenna, for the same
aperture size.
9. A rectangular aperture in the yz coordinate plane has a y-dimension Dy = 10l.
The field in the aperture is of uniform intensity and phase. At what positive and
negative values of the azimuth angle f, in the xy-plane, will the first nulls of the
pattern occur?
10. Show that for the aperture of Problem 9 (Dy = 10l), the relative value of E(f)
is approximately 0.707 when f has the value 25.4l /Dy degrees. This result con-
firms the fact that k1 in equation (6–12) has the value 50.8 (twice 25.4) for this
particular case, since the E(f) = 0.707 point on the beam pattern corresponds to
the half-power-density points, which define the beamwidth.
CHAPTER 7
Antennas with Special

Properties
Chapter 7 discusses properties of antennas and antenna analysis techniques not addressed
in other chapters. Antennas are used in many different applications, and consequently
there are numerous special properties needed to satisfy their requirements. The range of
topics addressed in this chapter is wide, and includes techniques for providing wide
bandwidths, multiple polarizations, low receiver noise, and extremely low side lobes. A
specific application may, of course, employ more than one of these techniques. There is
also a section solely on antennas for direction finding and another on antennas that
accomplish beam scanning mechanically. The concept of synthetic-aperture antennas,
although perhaps more of a signal processing than an antenna topic, is also discussed
here, because it is a uniquely effective way of attaining very narrow antenna beamwidths
through use of a relatively small antenna on a moving platform. Finally, the chapter closes
with a section on seemingly unrelated subjects, namely, geometrical theory of diffraction
(GTD), method of moments (MoM), and fractals. These topics are, however, connected
by the subject of radiation from sharp edges (curvatures small compared to a wavelength),
which currently cannot be analyzed by exact theoretical methods.
GTD and MoM analyses are computer programmed for solutions to complex antenna
configurations. GTD uses ray-optics representation of electromagnetic propagation, and
it also incorporates diffraction theory and surface waves for surfaces large compared to
a wavelength. MoM, on the other hand, is applicable primarily to wire antennas (e.g.,
dipoles and longer wire radiators), but it has also been used for patch antennas. MoM
programs compute currents, from which radiation can be calculated. In contrast, fractals
are computer-generated random shapes proven useful as novel antenna configurations,
and for which MoM programs are used to analyze the radiation from their sharp edges.
7.1. Broad-Band Antennas

As discussed in sec. 3.9, the bandwidth of an antenna can be defined in various ways,
depending on the requirements. It is customary to speak of the impedance bandwidth,
gain bandwidth, polarization bandwidth, and the pattern bandwidth separately when a
precise description of antenna bandwidth is required.
285
286 Antennas with Special Properties
Bandwidths are sometimes specified in terms of percentage of the center frequency

and sometimes as the ratio of the maximum to the minimum frequency. Which method
of specification is used depends on the magnitude of the bandwidth. If the ratio of
maximum to minimum frequency is less than 2 : 1, the percentage figure is often given
(a 1.1 : 1 ratio, for example, corresponds to roughly 10% bandwidth). The higher band-
width ratios are generally given as ratios. Bandwidths of 10 : 1 and higher are achievable
with special types of antennas. On the other hand, a 10 percent bandwidth or even less
may represent a great engineering achievement in some applications—for example, in a
large array antenna. Since broad bandwidth is often a necessary property of an antenna,
much research and development have gone into improving the techniques of designing
antennas for broad-band operation, and much progress has been made.
7.1.1. Basic Broad-Band Principles

Conventional antennas are essentially resonant structures. They have properties that go
through maximum and minimum values at particular frequencies. These properties are
the input impedance and the various pattern properties that have been mentioned. A
center-fed linear wire is an example. At the frequency at which its length is (approxi-
mately) a half wavelength, the reactive component of the input impedance is minimum
(or zero), and the radiation is maximum in the direction perpendicular to the dipole, as
was discussed in sec. 4.3. As the frequency is increased indefinitely, the impedance and
the radiation in the perpendicular direction go through successive maxima and minima,
although the two properties do not go through these in step with each other. This fact
indicates why, in this case, the impedance bandwidth and the pattern bandwidth have to
be considered separately.
It is well known that if a resonant circuit is “loaded” by connecting or coupling a dis-
sipative element (e.g., a resistor) to it, the sharpness of the resonance, and also the
amplitude of variation of impedance and other properties, will be reduced. This is another
way of saying that the bandwidth will be increased. This fact provides one approach to
broadening the impedance bandwidth of antennas, at the sacrifice of efficiency. This
method is used in the terminated rhombic and V antennas (sec. 5.8.1), and it is the basis
of the well-known excellent bandwidth properties of these antennas. Resistive loading
can also be used with short-monopole VLF antennas to broaden the bandwidth when
wide-band modulation of the transmitted signal makes increased bandwidth necessary.
Without such loading these antennas may be very narrow-band because of the large
reactive component of the input impedance; when this reactance is tuned out by means
of a large inductance, a “high Q” resonant circuit may result. (High Q implies narrow
bandwidth.) However, a resistive loading is unnecessary when ground losses and other
conductor losses reduce the Q sufficiently.
At higher frequencies, where antennas are usually full length, conductor losses may
be very low compared to their radiated power. Consequently, simple antennas may be
found to have inadequate bandwidth. But at these frequencies, deliberate loading with
resistance is seldom practiced. In the first place, the extremely narrow bandwidth some-
times encountered at VLF is not usually found at higher frequencies, because there is a
Broad-Band Antennas 287
better ratio of radiation resistance to reactance. In effect, the radiative loading lowers the
Q. Moreover, other methods of improving bandwidth have been discovered. These
methods may be separated into two classes—those that apply to basic radiators, and those
that apply to complex antennas, such as arrays, reflector-feed combinations, and lens
antennas (in which the refracting element as well as the feed may be frequency-sensitive).
This separation is not absolute, but it is of some value.
The methods that apply to basic radiators, such as dipoles, are mostly geometric in
nature. That is, the employment of certain shapes, and certain size ratios, have been found
to be beneficial. A dramatic extension of this class of methods has been made in the
development of the spiral and log-periodic antennas, or frequency-independent antennas
as they are also called.
Other methods apply to complex antennas. In some cases it is possible to combine
elements whose variations of impedance (or other properties) are complementary in such
a way that they tend to offset each other, as the frequency is varied. The bandwidth of
an array is, naturally, improved by employing broad-band radiators as the elements. It is
important to note, however, that the element bandwidth in the array environment may
be less than it would be when isolated, because of mutual coupling effects.
Another important factor in broad-band array design is the transmission-line or wave-
guide feed system that distributes the rf power of the transmitter, to provide the required
amplitude and phase to each array element. Certain possible arrangements are much less
frequency-sensitive than others. The two basic methods are shown in Fig. 2–8, chapter
2. For broadside arrays, the method of Fig. 2–8a is much better than that of Fig. 2–8b,
because it preserves the same total length of line from the common input point to each
element of the array; therefore all elements will be in-phase regardless of the frequency.
But with the arrangement of Fig. 2–8b, the elements will be in-phase only at frequencies
for which the interconnecting line length, d, is an integral multiple of a wavelength. The
arrangement with all input-to-element path lengths equal is called a corporate structure
feed system.
It is often necessary to provide impedance-matching transformers at an antenna’s input
terminals or at the elements of an array. Baluns (sec. 4.11) may also be required. The
bandwidth of the transformers and baluns is an important factor in determining the overall
impedance bandwidth. It is desirable to employ a transformer and balun whose reactance
variations with frequency are in such a direction that they tend to cancel the reactance
variation of the antenna or the array element. For example, a dipole fed at the center has
inductive reactance at frequencies above the resonant frequency, and capacitive reactance
at lower frequencies. A possible method of adjusting the dipole input impedance is to
make it shorter than the resonant length, so that it is capacitive; then a short-circuited
section of transmission line (shorted stub) is connected across the antenna input termi-
nals, and its length is adjusted until it has an inductive reactance just sufficient to coun-
teract the capacitive reactance of the dipole. For example, the required stub length is
obtainable by rearranging equation (2–36), chapter 2. The resulting input resistance will
have different values for different amounts of shortening of the dipole from the resonant
length. The reactances of the stub and of the dipole will tend to cancel as the frequency
is varied up or down. The cancellation will be imperfect, but there will be less net
reactance change with frequency than there would be for a resonant dipole without the
stub. This is a relatively simple illustration of a principle. More elaborate and effective
methods may be used in practice.
These observations serve mainly to convey some of the general ideas and a few par-
ticulars of the art of making complex antennas broad-band. The remainder of this section
is devoted to a discussion of the design of individual radiators of broad bandwidth.
7.1.2. Broad-Band Dipoles

Since dipoles are the most useful of radiators over a large part of the radio spectrum,
considerable study has been devoted to designing them for broad bandwidth. The prin-
cipal factor determining the bandwidth of a dipole has been found to be its thickness
relative to its length. A “fat” dipole has much broader bandwidth than a thin one.
This principle can of course be more readily exploited at the moderately high frequen-
cies where the dipole length is not too great—generally at frequencies above about
100 MHz, occasionally at somewhat lower frequencies. Dipoles are commonly employed
as radiators up to about 1 GHz, occasionally up to a few gigaHertz. Therefore the practi-
cal region of application of the principles to be discussed is about 30 MHz to 3 GHz.
The discussion of dipole properties in sec. 4.3 is based on the assumption that the
length-to-diameter ratio is very large, that is, that the dipole is thin. Fat dipoles do not
have the same input impedance and current distribution as thin dipoles. Their radiation
patterns are similar, although not identical. The important property of the thick dipole is
that its impedance variation with frequency is much less than for a thin dipole.
It is common practice, therefore, in the frequency region mentioned, to make dipoles
using large-diameter tubing or pipe rather than thin wire. This results in a cylindrical
dipole, as illustrated in Fig. 7–1a. A length-to-diameter ratio as low as 10 is not uncom-
mon. The shape is sometimes modified as
L L shown in Fig. 7–1b and 7–1c.
The variation of the input impedance
of the simple cylindrical dipole for three
D different values of the length-to-diameter
ratio (L /D) is shown in Figs. 7–2 and 7–3.
(a) (b) These curves are based on calculations of
L Kennedy and King (1953). Because of
effects at the point of connection of the
transmission line, these values may not
be observed exactly in practice, but they
serve as a guide to the approximate values
and the nature of the variation. It is appar-
(c) ent from these curves that the variations
FIGURE 7–1. of resistance and reactance are much less
Several forms of fat dipoles. (a) Cylindrical for fat dipoles (L /D = 17 and L /D = 122)
dipole. (b) Cylindrical with conical ends at than for thin dipoles (L /D = 11,000). The
gap. (c) Prolate spheroidal antenna. frequency f0 is the value at which the
dipole is a half wavelength—that is, 10,000

∼ 11.000
L = l/2 when f = f0. The resulting values L/D =
of the voltage standing-wave ratio
Resistive component, ohms

(VSWR) on a transmission line for the 1000
various values of f/f0 could be calculated L/D ∼

= 122
by taking the values of R and X given by ∼ 17
100 L/D =
these curves to determine the quantities
RL /Z0 and XL /Z0 for use in equation
(2–45), chapter 2, from which the line
10
voltage variation may be plotted. Alterna-
tively, the Smith Chart, described in sec.
9.12, could be used to determine the
1
VSWR directly from the values of RL /Z0 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
and XL /Z0. f/f0
FIGURE 7–2.
7.1.3. Biconical Antennas Resistive component of input impedance of a

center-fed cylindrical dipole as a function of
A radiator generally regarded as a form frequency, for three values of length-diameter
of dipole is the biconical antenna, shown (L /D) ratio. Plotted from values calculated by
in Fig. 7–4. This type of antenna is usually Kennedy and King (1953).
analyzed in terms of the cone half-angle
q together with the parameters s and h. As
10,000
seen by comparison of Fig. 7–4 with Fig.
L/D ∼
= 11.000
4–22 of chapter 4, it is virtually identical ∼ 11.000
L/D =
to the biconical horn. Whether it is
Reactive component, ohms (±)
1000 L/D =∼ 122

regarded as a dipole or a horn is partly a L/D ∼
= 122
L/D ∼
= 17
matter of preference in some cases. Con-
ventionally it may be considered a dipole ∼ 17
L/D =
100
if q is less than 45 degrees and a horn if
q is greater than 45 degrees. The mode of
excitation also affects this distinction.
10
The biconical antenna can be regarded as
a dipole only if it is excited to produce
polarization parallel to the conical axes
1
(TEM mode). In either case, however, the 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
f/f0
antenna has broad-band properties. The
broad-band qualities of horns have already FIGURE 7–3.
been mentioned. As a dipole, the biconi- Reactive component of input impedance of a
cal antenna has a bandwidth roughly center-fed cylindrical dipole as a function of
equivalent to that of a cylindrical dipole frequency, for three values of length-diameter
(L/D) ratio. Dashed-line portions of curves
whose diameter is somewhat less than the
represent negative (capacitive) reactance, and
maximum cone diameter. Biconical anten- solid-line portions represent positive (inductive)
nas have been extensively analyzed by reactance. Plotted from values calculated by
Schelkunoff (1943). He has also used the Kennedy and King (1953).
biconical antenna as a theoretical basis for deducing properties of cylindrical antennas,

with results similar to those of Figs. 7–2 and 7–3.
7.1.4. Monopoles
Dipoles, whether cylindrical, biconical, or of other shape, are symmetrical radiators when
fed at their centers. All dipole types can also be used as monopoles, which were discussed
for specific dipole forms in secs. 4.1 and 4.3. A monopole is half of a dipole operated in
conjunction with its image in a conducting ground plane perpendicular to it. Monopole
designs corresponding to the various broad-band dipoles that have been discussed are
shown in Fig. 7–5. There are innumerable variants of the shapes shown. The feed-point
(input) impedance of a monopole is exactly half the value of the corresponding dipole,
if the ground plane is a perfect conductor
and of infinite extent, and the pattern cor-
responds exactly to half the pattern of
the dipole. Since for a given input power
the radiation is concentrated into half the
q solid angle occupied by the dipole pattern,
Transmission the directive gain of the monopole is
line twice that of the dipole in free space. (See
h
discussion of directivity in sec. 3.3 and of
vertical radiators in secs. 4.1 and 4.3.) In
s
practice, of course, the ground plane will
not be a perfect conductor of infinite
extent, but the properties of the monopole
will approximate those just described if
the ground plane is a good conductor
FIGURE 7–4. extending to a distance several times as
Biconical antenna. great as the height of the monopole.
Ground
plane
Coaxial feed lines Coaxial feed lines

FIGURE 7–5.
Monopole antennas corresponding to the dipoles of Figures 7–1 and 7–4.
Monopoles are especially useful as vehicle antennas where the ground plane is the
“skin” of the vehicle (ship, boat, automobile, tank, aircraft, spacecraft). On aircraft the
monopole may for aerodynamic reasons take the shape of a streamlined “blade” rather
than the circular-cross-section shapes commonly used in other applications. Another
advantage of the monopole is that it is conveniently fed by a coaxial line. (The advantages
of coaxial lines are discussed in sec. 2.4.) A monopole has the same bandwidth properties
as the corresponding dipole.
7.1.5. Folded Dipoles

A very important variation of the simple dipole is the folded half-wave dipole, depicted
in Fig. 7–6. A principal advantage of the folded dipole is its higher input impedance that
can be more readily matched to the characteristic impedance of a two-wire line given by
equation (2–62) of chapter 2. The folded construction results in a step-up of the approxi-
mately 70-ohm simple-dipole input impedance by a factor of 4. Hence the input imped-
ance of a basic folded dipole is about 300 ohms (actually slightly less). Folded dipoles
are often used as elements of television receiving antennas and as the feed element for
Yagi-Uda arrays.
This statement applies only to a folded dipole by itself. When it is an element of an
array, or is used with a reflector, its input impedance may no longer have this value, and
an impedance-matching device may then be required. Even so, the matching to a two-
wire line is easier to accomplish than it would be with a simple dipole.
The folded dipole also has somewhat better intrinsic bandwidth properties than a
simple dipole of the same size conductors. The folded construction tends to produce the
reactance-canceling action of a short stub line in parallel with the simple-dipole reac-
tance, at frequencies off resonance. The great difference between the impedance behavior
of the folded dipole and the ordinary dipole may be seen by considering a very short
folded dipole (short compared to a half wavelength) in comparison with a very short
ordinary dipole; the short folded dipole will have a very low inductive impedance
whereas the short ordinary dipole will have a very high capacitive input impedance.
For longer-than-half-wave dipoles the opposite comparison of reactances applies; the
folded dipole is capacitive, and the simple dipole is inductive. The folded dipole has
bandwidth properties equivalent to those
of a fat dipole of diameter somewhat less l/2
than the spacing of the folded-dipole
conductors.
Additional step-up of the simple-
dipole impedance may be obtained by
adding more “folded” conductors. A
three-wire folded dipole is shown in
Fig. 7–7. Its input impedance is approxi-
mately nine times the normal 70-ohm
impedance, or about 630 ohms. Still FIGURE 7–6.
more conductors can be added to obtain Folded dipole.
l/2 even higher impedance, though an input

impedance higher than about 600 ohms is
seldom desired. The step-up ratio may
also be altered by making the folded
conductors of different size from the
basic conductor. Many other variations of
the folded-dipole principle have been
devised.
7.1.6. Superturnstile Antenna

A form of broad-band dipole much used
FIGURE 7–7. as the element of a TV transmitting array,
Three-wire folded dipole. known as the superturnstile, is shown in
Fig. 7–8. This element is partly a dipole
and partly a slotted-sheet antenna. The
∼0.5λ
slot is vertical and the dipole horizontal;
hence the polarization is horizontal. It
Opening (slot) is fed by a balanced line at the center of
the slot. The input impedance is about
Conducting 75 ohms. This radiator has a 30 to 35
0.65 sheet or grid
to percent impedance bandwidth for a
0.7λ VSWR not exceeding 1.1 : 1. Its broad-
band properties stem partly from the large
Transmission
line
dipole width and partly from the fact that
the dipole and the slot have “complemen-
FIGURE 7–8.
tary” impedance variation characteristics
that tend to offset each other as the fre-
Element of a superturnstile antenna.
quency is varied.
To reduce wind resistance, the solid
sheets are replaced by horizontal conducting rods, as depicted in Fig. 7–9. The dimen-
sions given in Fig. 7–8 are approximate for the center frequency, and they will depend
somewhat on the spacing of the rods. In a typical VHF TV transmitting array, the antenna
elements are mounted in pairs on a vertical steel mast, the pairs stacked one above the
other, as sketched in Fig. 7–9. The mast runs “through” the slot. Each pair is fed 90
degrees out of phase, resulting in approximately circular coverage in the horizontal plane.
The stacked pairs are fed in phase, to produce array gain horizontally (in elevation).
7.1.7. Miscellaneous Broad-Band Dipoles and Other Radiators

Innumerable variations of the basic dipole types, previously described, have been devised.
Some of the more common ones are shown in Fig. 7–10.
Broad-band dipoles have been discussed extensively because of their importance in
many applications, and also because many of the principles discussed apply to other
Frequency-Independent Antennas 293
radiator types as well. It is not possible

here to discuss the broad-banding of every
type of radiator, but some general obser- 90°
Elements
vations can be made.
Horns and axial-mode helixes are basi-
cally broad-band radiators, since their
operation does not depend on any criti-
cally resonant element or elements. They
do of course have maximum and minimum
frequencies. For the rectangular horn
these are usually the frequency limits of
the rectangular waveguide, which is a 2 : 1
Mast
range. (See sec. 2.5.) Slot radiators have
bandwidth properties similar to those of
dipoles. The various nonresonant anten- Elements
nas described in sec. 4.4 have excellent
bandwidths, but the resonant long-wire
types do not.
In all this discussion, and in what
follows, the bandwidth is considered for
operation of the antenna without any
adjustments when the frequency is
changed. It is of course possible to operate
an essentially narrow-band antenna over
a considerable frequency range if some
provision is made for adjustment of criti- FIGURE 7–9.
cal lengths, or for adjusting an imped- Two vertically stacked “bays” of a
ance-matching transformer. In some superturnstile TV transmitting antenna.
applications this type of operation is pos-
sible, and is employed. Possibly the most common example is the adjustable “rabbit ears”
indoor antenna sometimes used with television receivers, the dipole length being adjust-
able by means of a telescoping construction.
7.2. Frequency-Independent Antennas

A very sophisticated and effective approach to achieving broadband antenna design has
evolved from the initial work of a number of researchers at the University of Illinois
(Rumsey 1957, 1966; DuHamel and Isbell 1957). The class of antennas that has resulted
are called frequency-independent or log-periodic antennas. (The term log-periodic is
perhaps too restrictive since it refers to only one aspect of their behavior.) These antennas
achieve bandwidths of 10 : 1 with ease, and much larger values are possible with careful
design. Their broad-band behavior includes both impedance and pattern characteristics.
They are a class rather than a type of antenna, because there are many different types.
They have a wide variety of physical appearances, some of them quite bizarre in
l /2
l /4
l /4
(Flat
sheets)
(a) (b) (c)
l /4
l /4 (Disc size
0. 3l variable)
∼30°
0.3 l
l /3
Coaxial
feed line
Coaxial Coaxial
feed line feed line
(d) (e) (f)
FIGURE 7–10.
Miscellaneous dipoles and related types. (a) Triangular dipole. (b) Fan dipole. (c) Ellipsoidal
dipole. (d) Folded monopole with simulated ground plane. (e) Conical skirt monopole.
(f) Discone antenna.
comparison with more conventional antennas. They may have unidirectional or bidirec-
tional patterns of low to moderate directive gain. Higher directive gain may be obtained
by using them as elements of an array or as feeds for parabolic reflectors and lenses, in
which use, however, a design must be chosen that maintains a constant or nearly constant
phase center. (Many log-periodic types do not.) Log-periodic array design presents
special problems not encountered in ordinary arrays. Excellent discussions of the subject
have been published by Deschamps and DuHamel (1961), Jordan, Deschamps, Dyson,
and Mayes (1964), and Rumsey (1966).
7.2.1. The Log-Periodic Principle

All the frequency-independent antennas have a special kind of repetitiveness in their
physical structure, which results in a repetitive behavior of the electrical characteristics.
That is, the design involves a basic geometric pattern that is repeated, but with a chang-
ing size of the pattern. The pattern size changes with each repetition by a constant scale
factor so that the structure “grows” (or shrinks, depending on the direction considered).
These principles are illustrated by the log-periodic “dipole array” of Fig. 7–11, which is
Beam direction
R6
R5
l6
l5
Transmission
line
Dipoles
FIGURE 7–11.
Log-periodic dipole array.
perhaps the nearest that a log-periodic design can come to resembling a conventional
antenna. (There is a marked similarity to the end-fire dipole array described in
sec. 5.2.)
This particular form of dipole array was devised by Isbell (1960). It consists of a
number of dipoles of different length and spacing, fed by a two-wire line that is trans-
posed between each adjacent pair of dipoles. The array is fed at the small end of the
structure, and the maximum radiation is toward this end. The lengths of the dipoles and
their spacing are graduated in such a way that certain dimensions of adjacent elements
bear a constant ratio to each other; these are the quantities designated R and l in Fig. 7–11.
That is, if this design ratio is designated by t (a number less than one), then
R2 R3 R4 1 l2 l3 l4
= = = = = = (7–1)
R1 R2 R3 τ l1 l2 l3
It is apparent that these conditions cause the ends of the dipoles to lie along straight lines
that meet at an angle designated a. It is a characteristic of the frequency-independent
antennas that certain aspects of their structures can be specified in terms of an angle or
angles.
The result of these structural conditions is that if a plot is made of the input impedance
as a function of frequency, the variation will be found to be repetitive. If the plot is made
against the logarithm of the frequency, rather than the frequency itself, this variation will
be periodic—that is, the impedance will go through cycles of variation in such a way
that each cycle is exactly like the preceding one. A typical plot is shown in Fig. 7–12. It
is this behavior that gives rise to the name “log periodic”; the impedance is a logarithmi-
cally periodic function of the frequency. (It is not implied, however, that the variation is
necessarily sinusoidal.) Moreover, all the electrical properties of the antenna undergo a
similarly periodic variation. In particular,

the radiation pattern varies in this way,
Input impedance
along with such parameters as the direc-

tive gain, beamwidth, side-lobe level, and
Logarithm of frequency beam direction.
If the structure defined by (7–1) were
FIGURE 7–12. extended all the way to the apex, that is,
Typical behavior of log-periodic antenna input to the vertex of the angle a, and also
impedance. indefinitely in the other direction, this
periodicity would extend to zero fre-
quency in one direction and to infinite frequency in the other. Of course neither of these
extensions is physically possible, because the extension to zero frequency (infinite wave-
length) would require an infinitely large structure, whereas extension to infinite frequency
(zero wavelength) would require a structure of microscopic fineness at the apex; and the
transmission line feeding would have to be of infinitesimal conductor size and spacing—
an obvious impossibility. Therefore, the structure must be terminated at some point in
each direction, and these points determine high and low cutoff frequencies, beyond which
the log-periodic property does not extend.
This periodicity does not by itself insure that the antenna will be broad-band. But if
the impedance and pattern variation over one cycle of the periodic variation can be made
sufficiently small, so that some bandwidth criterion is satisfied, the same criterion will
be satisfied over all periods. The requirement, therefore, is to find structural patterns that
do provide the necessary behavior over one log-periodic cycle. The possible bandwidth
is then limited only by the number of cyclic structures employed in the antenna. Fortu-
nately, many such structural patterns have been found.
The magnitude of the log-frequency periods is determined by the design ratio, t, and
is in general of magnitude log (1/t).* That is, if two successive maxima of the impedance
or pattern variation occur at frequencies f1 and f2, it will be found that they are related
by the formula
f2 1
log f2 − log f1 = log = log (7–2)
f1 τ
from which it follows that f2/f1 = 1/t, or f1 = t f2. Therefore, if the log-periodic antenna
has certain measured properties (e.g., impedance, gain) at any particular frequency f, it
follows that it will have exactly the same properties at frequencies t f, t 2f, t 3f, and so
on, and also at f/t, f/t 2, f/t 3, . . . , provided that these frequencies are all within the cutoff
limits.
The word “exactly” in the foregoing statement applies only if the structural scaling
by the design ratio t is exact and is applied to every detail of the structure. For example,
in the antenna of Fig. 7–11 not only the dimensions R and l must be scaled but also the
sizes of the conductors. If, for example, the dipoles are made of tubing, the tubing must
* In certain cases the impedance (but not the pattern) may vary with a period half as large (that is, with
1 1
period 2 log (1/t)). But these cases do not violate Eq. 7–2, since a variation with periodicity 2 log (1/t) is
also periodic with period log (1/t).
be of properly increasing thickness as the distance from the apex of the structure increases.
The width of the gaps at the dipole centers must also increase. However, these relatively
minor factors can be ignored without serious effect if the overall bandwidth (ratio of
maximum to minimum cutoff frequencies) does not exceed about 10 : 1.
When a log-periodic antenna is fed at a particular frequency, it is found that the radia-
tion occurs from a certain portion of the structure, and that other portions do not radiate.
This “active” region of the antenna of Fig. 7–11 is the region in which one or more of
the dipoles is nearly half a wavelength. The cutoff frequencies are those at which the
shortest and longest dipoles are approximately half a wavelength. Thus the active region
of the antenna is near the apex for the highest frequencies radiated, near the large end
for the lowest frequencies, and at intermediate positions for frequencies in between. This
behavior may also be expressed by saying that the phase center of the antenna progresses
from the large end to the small end as the frequency goes from minimum to maximum
of the band-width range.
Log-periodic antennas may be either unidirectional or bidirectional in their pattern
characteristics. The bidirectional types are characterized by two active regions that move
apart as the frequency is decreased; in these types the phase center remains fixed and
coincides with the feed point.
For the antenna of Fig. 7–11 typical values of the design ratio and apex angle are
t = 0.8 and a = 30°. For any given value of a, there is a minimum permissible value of
t (which is always a number between 0 and 1). Larger values of a and smaller values
of t go together and result in more compact design for a given bandwidth. On the other
hand, smaller values of a and larger values of t result in improved performance—smaller
variation of impedance and pattern, and higher gain, at the cost of a larger structure. The
reason for this effect may be intuitively understood by realizing that for small a and large
t, at any given frequency, there are more dipoles of nearly half-wavelength; hence the
active or radiating region of the antenna encompasses more dipoles than for large a and
smaller t. As is always true, the gain increases with the number of radiating elements
and with the size of the radiating region. Also, the impedance and pattern variations are
smaller because of the smoother transition in the configuration of the radiating region
when it encompasses several dipole elements.
7.2.2. Practical Log-Periodic Designs

As mentioned, numerous structural forms have been found to be suitable for log-periodic
antennas. Three representative ones are shown in Fig. 7–13.
First we refer to the wire-trapezoidal-tooth antenna of Fig. 7–13b, which is frequently
used as a moderately directive steerable (rotatable) HF communication antenna. It has
two arms or branches originating at the apex of the antenna. Each branch has a charac-
teristic angle a that plays the same role in the antenna design as the angle a in the log-
periodic dipole array (Fig. 7–11). The other angle, g, affects the directional characteristics.
When g is 180 degrees, which it is for the trapezoidal-tooth structure shown in Fig. 7–13a,
the radiation is bidirectional and perpendicular to the plane of the antenna. When g is an
acute angle, the radiation is unidirectional and toward the apex of the structure. The
dipole array, Fig. 7–11, at first glance does not seem to conform to these observations.
Sheet
metal
Feed
line
a
(a)
Support
Beam
direction
g
Wires
Feed
line Support
(b)
Beam direction
q Wires
Feed
line
(c)
FIGURE 7–13.
Some practical log-periodic antenna designs. (a) Trapezoidal tooth structure. (b) Wire
trapezoidal tooth antenna. (c) Wire triangular tooth antenna.
It can be shown, however, that actually it is a specialization of the trapezoidal-tooth

structure (Fig. 7–13a), in which the width of the teeth is reduced to nearly zero, and the
angle g has been reduced from 180 degrees to zero. That is, the two branches have been
folded over until they are in the same plane, but not in electrical contact. The wire trian-
gular tooth antenna (Fig. 7–13c) is another plane structure that results in a unidirectional
pattern.
A variation of the Isbell log-periodic dipole array, developed by Mayes and Carrel
(see Mayes, 1982) at the University of Illinois, employs resonant-V dipole elements
instead of conventional straight dipoles, with the acute angle of the V toward the small
end of the array (in the direction of the beam). The dipole elements of this array operat-
ing as half-wave radiators provide a moderate bandwidth. Slightly beyond the upper limit
of this basic band, the elements again become active in a 3/2-wavelength mode, and at
still higher frequencies they operate in a 5/2-wavelength mode. The V configuration
provides good directivity at the higher frequencies, and the multiple-mode operation
provides a large operating bandwidth (as much as 20 : 1) in a relatively compact antenna.
This design has been in wide use as a TV
receiving antenna for covering the entire
Input
spectrum from the lowest TV VHF
channel (54 MHz) to the previous highest
TV UHF channel (890 MHz), although
the current highest U.S.-allocated TV
UHF frequency is 698 MHz (see sec.
1.1.2). Other modifications of the Isbell
array employ folded-dipole and folded-
monopole elements (the latter being oper-
ated against a ground screen or the actual
ground).
7.2.3. Equiangular Log-Periodic

Antennas
It has been noted that certain features of
the structure of a log-periodic antenna
may be specified in terms of an angle or
angles (Dyson 1959, 1965). When the
structure can be entirely specified in terms
of angles, it is found that the impedance
variation is very small over the log-peri-
odic frequency cycle. Hence it is highly
desirable to use structures that meet this FIGURE 7–14.
criterion. An example of such a structure,
Sketch indicating general nature of a conical
the conical spiral antenna shown in logarithmic-spiral antenna. (A practical antenna
Fig. 7–14, is characterized by the half would have a smaller pitch angle—i.e., more
angle of the cone and by the angle that the turns per unit length of cone.)
spiral makes with a circumference, or its complement that the spiral makes with a radial
line from the apex. There are two interleaved spiral conductors lying on a conical (non-
conducting) surface. Each spiral is connected at the apex to one side of the feed line.
The radiation is unidirectional, going in the direction toward which the apex of the cone
points. The peak of the beam is polarized elliptically, becoming virtually circularly polar-
ized as the number of turns per unit length is increased.
7.2.4. Self-Complementary Antennas

As mentioned in sec. 4–8, a certain relationship exists between the patterns and imped-
ances of complementary antennas (e.g., a “flat” dipole in free space and a slot of the
same size and shape in an infinite conducting plane). Booker (1946) showed that if an
antenna consisting of both conducting-sheet elements and openings between them (e.g.,
slots) can be so configured that the conductors and the open spaces are of identical shape
and size, its input impedance is independent of frequency. An antenna having this geo-
metric property is called self-complementary. The test of self-complementarity is that if
the conductors are replaced by open spaces and the open spaces by conductors, the
appearance remains the same except for a rotation through an angle.
A simple example of a self-complementary structure is a pair of triangular sheets, as
in Fig. 7–15, of 90-degree flare angle. The spaces between them will of course also have
90-degree flare angles; thus if the sheets are replaced by spaces and the spaces by sheets,
the structure is unchanged except for a rotation through 90 degrees. The self-comple-
mentarity holds only if the sheets extend to infinity; therefore, this arrangement does not
constitute a practical self-complementary antenna. It is shown here solely to explain the
self-complementarity principle.
If the log-periodic principle is combined with the self-complementary principle, the
structure need not extend to infinity since
∞ it is only necessary for self-complemen-
tarity to exist in the active regions of
the structure. A sheet-logarithmic-spiral
90° structure, shown in Fig. 7–16, which is
also an equiangular structure, can be made
self-complementary. It provides bidirec-
Metal Feed tional circularly polarized radiation in
sheets 90°
line directions perpendicular to the plane of
the sheet, with excellent impedance char-
acteristics; that is, the impedance is virtu-
ally constant over the frequency range
between the log-periodic cutoff points.
Other self-complementary designs have
∞
also been devised. The impedance of a
FIGURE 7–15. two-terminal self-complementary antenna,
Triangular sheet self-complementary structure. according to Booker’s principle, is 60p
(about 189) ohms. (It is also possible to

have structures of more than two termi-
nals, i.e., with more than twofold angular
symmetry, and such structures have other
values of impedance.)
This discussion of frequency-
independent antennas has presented the Input
terminals
elementary principles and some of the
highlights of current designs. Such
antennas are practical in the frequency
region of perhaps 10 MHz to 10 GHz.
The theory and practice are in relative
infancy, although great strides have been
made.
FIGURE 7–16.
Central portion of a logarithmic-spiral self-
complementary antenna. Shaded portions are
7.2.5. Spiral Antennas sheet metal. Note that rotation of the figure
through 90° carries the shaded portions into the
Spiral antennas consist of conducting clear areas and vice versa and rotation through
wires or tapes arranged in spiral shapes, 180° leaves the figure unchanged.
on either a conical or planar surface
(Filipovic and Cencich 2007). Spiral
antennas are inherently circularly polarized (CP) and have relatively constant input
impedance and radiation patterns over a wide range of frequencies. Without modification,
planar spiral antennas are bidirectional with opposite circularity in the two directions
perpendicular to the planar surface. Planar spiral antennas are made unidirectional by
placing either absorbing materials or a reflecting surface on one side of the planar surface.
The reflecting plate is spaced at l/4 distance, and then the wave to be rejected is polar-
ization reversed and adds constructively to wave of the other direction, causing unidi-
rectional performance. The conductors of conical spiral antennas are wound along a cone
and their patterns are unidirectional, with the beam maximum directed toward the apex
of the cone.
The conducting spirals are called arms. Customarily, the shapes of the spiral arms have
been either Archimedean or equiangular. The Archimedean spiral has a constant arm
width and constant separation between arms throughout the spiraling aperture. The equi-
angular spiral has progressively increased arm width and separation between the arms
until they open to the outside aperture dimension. Although the antenna of Fig. 7–16 is
a self-complementary spiral, the spiral shape is equiangular.
Spiral antennas may be made in a number of different geometries not only for
improved electrical performance, but also for better packaging, arraying, and mounting.
A square shape may allow closer spacing for array elements, and wiggly or zig-zag shapes
permit longer conductor lengths and thereby enhance low-frequency performance. Spiral
antennas can be flush mounted, made on antenna reflectors, or placed on aircraft and
other vehicle surfaces.
7.3. Antennas for Multiple Polarizations

7.3.1. Antennas for Circular Polarization
An antenna that will radiate and receive circularly polarized waves may be required for
certain applications. This is especially true in uses involving transmission through the
ionosphere at frequencies where the ionosphere may produce rotation of the wave polar-
ization (Faraday effect). This may occur, for example, in communication or radar trans-
mission between the earth and an artificial satellite, or other spacecraft. The rotation
produced is a complicated function of the ionospheric electron density (which varies
from day to night), the wave frequency, and the orientation and strength of the earth’s
magnetic field at the particular latitude and longitude involved. Thus the polarization that
will arrive at a receiver from a linearly polarized transmitting antenna may be practically
unpredictable, even though it can be calculated if the electron density and earth’s field
are known. If it happens that the polarization is rotated just enough to put it at 90 degrees
with respect to a linearly polarized receiving antenna, no signal will be received. This is
the extreme case; for other angles of rotation, the signal strength (power) will be reduced
to a fraction equal to the square of the cosine of the angle between the wave polarization
and the receiving antenna polarization. These effects occur in the frequency range of,
roughly, 10 to 1,000 MHz, although ordinarily not above about 500 MHz in daytime and
perhaps 100 MHz or less at night.
If either the transmitting or the receiving antenna is circularly polarized (i.e., transmits
or receives circularly polarized waves), signal reception is possible regardless of polar-
ization rotation by the ionosphere. Circular polarization is also desirable in certain radar
and communications applications at higher frequencies.
A number of methods exist for producing circularly polarized radiation or reception
by an antenna. Some of them have already been mentioned in connection with the
description of specific antenna types, for example, the helix (sec. 4.6) and the spirals
(either conical or planar).
An important method of achieving circular polarization is the use of two dipoles
arranged to form a cross, that is, with their centers at the same point and their axes at
right angles to each other. If a transmission line connects directly to the center of one of
them, and if the two dipole-center feed points are joined by a quarter-wave line section,
there will be a 90-degree phase delay between the currents in the two dipoles in accor-
dance with equation (2–23), chapter 2, provided that the length of each dipole is adjusted
to make its input impedance nonreactive. The pair of dipoles will thus radiate two linearly
polarized waves at right angles to each other with 90-degree phase difference. As stated
in sec. 1.1, this combination produces an elliptically polarized wave, or if the two linear-
polarization amplitudes are equal it results in a circularly polarized wave. The two
amplitudes will be equal in the two directions perpendicular to the dipole axes. Thus in
these directions the radiation will be circularly polarized and also will have maximum
power density. In the plane of the two dipoles, the radiation will be linearly polarized
and of somewhat reduced power density. In this plane the radiation will be almost (but
not quite) uniform in all directions, that is, throughout 360 degrees. In all other directions
Antennas for Multiple Polarizations 303
¼-wavelength loop
of line (from AA to BB)
Dipoles
B
A
Two-wire
transmission
line
Line
input
FIGURE 7–17.
Arrangement of crossed dipoles for radiation of circular polarization (turnstile antenna).
the radiation will have varying degree of ellipticity, going from circular on the perpen-
dicular axis to linear polarization in the plane of the dipoles.
This arrangement of dipoles and the feed system is shown in Fig. 7–17. It is an espe-
cially important arrangement for two reasons. First, it is a basic method of producing
circular polarization. The pattern in this polarization is bidirectional but can readily be
made unidirectional by placing a reflector behind the dipoles, or combining several pairs
in an endfire array (with the cross-dipole planes perpendicular to the line of the array).
Second, it is useful when approximately uniform radiation is desired in a horizontal plane
with horizontal polarization.
This configuration is known as a turnstile antenna. To provide increased gain in the
horizontal plane and to reduce or eliminate the circularly polarized upward and downward
radiation, several turnstiles can be stacked vertically and fed in broadside-array fashion.
This arrangement is much used for television and FM broadcasting, using broad-band
dipoles. When a particular form of slotted-sheet dipole is used, the arrangement is known
as the superturnstile antenna, as was mentioned in the discussion of broad-band dipoles
(see Fig. 7–9). The quarter-wave phase-delay line section, if this phasing method is used,
should be arranged to produce as little obstruction as possible of the radiated fields and
also to have as little capacitive or inductive coupling to the dipoles as possible.
It is also possible to produce the necessary 90-degree phase difference by separately
feeding each dipole from a common source with parallel lines whose lengths differ by
a quarter wavelength, provided that both lines are matched by the dipole impedances
(no standing waves). The necessary phase shift between the dipoles can also be obtained
with a coaxial- or waveguide-type of phase shifter, and at low enough frequencies even
lumped-circuit elements (coils and condensers) may be used. Care must be taken, in
designing the feed system, to assure that the impedance relationships are such as to feed
equal power to each dipole and to provide a proper impedance match to the main feed
line.
Circular polarization can also be obtained with a horn radiator, either with a square
waveguide and a square-aperture pyramidal horn, or a circular waveguide and a conical
horn. In either case, the requirement is to excite a circularly polarized mode of propaga-
tion in the waveguide. This can be done in a variety of ways. A common way is to
introduce both polarizations into a rectangular waveguide, by means of in-phase probes
(Fig. 2–15, chapter 2) in two adjacent walls. Because of the rectangularity of the guide,
the two mutually perpendicular polarizations will have different phase velocities (this
assumes the waveguide is not square, but its height and width must be large enough to
permit propagation of perpendicular polarizations). The length of the rectangular guide
(measured from the probe position) is made just long enough to result in a 90-degree
phase difference between the two polarizations at the guide output.
A horn or other radiator with circularly polarized radiation can be used to illuminate
a paraboloidal reflector to produce a high-gain circularly polarized beam. Pyramidal
horns that are not square can be used to illuminate noncircular apertures, but special
methods must be used to compensate for the differences in phase velocities of the two
polarizations in the horn flare (since the flares are different for the different polarizations).
Lenses can also be used with such horns if the refracting medium is capable of passing
both polarizations with the same phase velocities.
A linearly polarized wave can be directly converted to circular polarization by means
of special transmission or reflection devices. An example of such a transmission medium
is a set of straight parallel vanes or plates with their planes parallel to the direction of
propagation and their direction inclined at 45 degrees to the linear polarization, so that
the wave in the regions between the plates can be considered to have two mutually per-
pendicular components, one parallel to the plates and one perpendicular to them. The
latter component will travel through the region at free-space velocity, but the former will
have a phase velocity that is determined by the plate spacing, according to the principle
of the metal-plate waveguide lens (except that the boundary surfaces of the parallel-plate
region are flat, or plane, instead of curved). If the thickness of this “lens” is properly
chosen, the two component waves can be caused to have a 90-degree phase difference
and the wave that emerges will be circularly polarized. This device is known as a quarter
wave plate, since it produces a quarter-wavelength effective path difference for the two
polarizations. A reflecting device that accomplishes the same result can be constructed
by using a similar set of vanes of half the depth, backed by a plane reflector.
Antennas for Multiple Polarizations 305
Circularly polarized high-gain arrays can be constructed by using crossed dipoles as

the radiating elements, fed as previously described, or by using helixes, pairs of slots, or
circularly polarized horns. Cross-polarized Yagi-Uda arrays, fed 90 degrees out of phase,
may also be used. Similarly, pairs of unidirectional log-periodic antennas can be used,
such as the wire-trapezoidal-tooth antenna of Fig. 7–13b, with a “staggered” arrangement
of the teeth so that the two polarizations are radiated with 90-degree phase difference at
all frequencies within the bandwidth range.
Many other forms of circularly polarized antennas are possible. The types that have
been described serve to indicate the general principles.
As mentioned in sec. 1.1, circular polarization is characterized as “right-hand” or
“left-hand” according to whether the electric vector is rotating clockwise or counterclock-
wise, respectively, as viewed from the source. Which of these senses of circular polariza-
tion will be radiated by a given antenna depends on whether the phases of the orthogonal
dipoles or fields lead or lag each other by 90 degrees. Moreover, if a given circularly
polarized antenna is connected to receive a given sense of circular polarization, it will
reject signals of the opposite sense. Therefore, to be able to receive signals of both
senses, an antenna should be connected to two receivers with different phasings, one
appropriate for one sense and one for the other. The above-mentioned four-port hybrid
ring coupler is especially convenient for this purpose; the desired result is accomplished
by connecting the antenna elements to two of the ports and the two receivers to the other
two ports.
A radar that transmits circularly polarized waves will receive from symmetrically
reflecting targets, like a sphere or a flat plate, echo signals that require the receiving
connection to the antenna to be phased in the sense opposite to the transmitting connec-
tion. For many types of complex targets, however, the reflected waves will contain both
circular-polarization senses. This fact can be used to allow receiving echo signals from
a desired unsymmetrical target (e.g., an airplane) while rejecting those from areas of
rainfall, since the raindrops reflect in the symmetrical manner.
7.3.2. Polarizers and Dual-Mode Transducers

Polarizers change polarization from one to another, and dual-mode transducers permit
transmitting or receiving more than one polarization. The output waveguide of a polarizer
or a dual-mode transducer is usually round or square.
Figure 7–18 shows a polarizer in square waveguide, with wave propagation from left
to right. To understand its operation, one can think of a vertically polarized wave being
resolved into two equal magnitude, equal phase waves oriented at ±45° to vertical. Now
assume the following:
• waveguide dimensions are small enough so that only one waveguide mode
(the dominant mode) is propagated
• a thin dielectric slab, as depicted in Fig. 7–18, is aligned at +45° to vertical
• the length of the slab is long enough that the phase delay of the +45° polarized
wave is 90° greater than that of the −45° wave
Then, looking from left to right (the direc-

ERC tion of propagation), the resultant electric
field is rotating clockwise. Thus, the
dielectric slab converts the vertically
polarized (V-POL) wave to a right circu-
larly (RC) polarized wave. Similarly, a
like slab oriented at −45° would generate
a left-circularly (LC) polarized wave.
EV For the polarizer of Fig. 7–18, a change
from right to left circular polarization can
occur by changing the input polarization
FIGURE 7–18. from vertical to horizon. Thus, the +45°
Circular polarizer with thin dielectric slab slab polarizer of Fig. 1–18 in combination
oriented at 45° to vertical. with a V-POL input can be used to trans-
mit or receive one circular polarization
EH (CP), and alternatively the other CP is
available with an H-POL input. If instead,
the slab causes 180° relative phase delay,
the V-POL input is transformed to H-POL
output. The operating principle of a polar-
izer in circular waveguide is similar to
that of the square waveguide example of
Fig. 7–18. For either the square or circular
waveguide case, tapering or notching both
ends of the dielectric slab reduces wave
EV reflections.
A polarization transducer permits the
FIGURE 7–19. adding of waves of different polarizations
Dual-mode transducer in square waveguide to create a wave of another polarization.
feeding a horizontally and vertically polarized Most transducers function reciprocally,
horn. so that they can separate waves into spe-
cific polarizations. Figure 7–19 shows a
dual-mode (i.e., dual-polarization) transducer that can provide separately a horizontally
or a vertically polarized wave into a square waveguide. Note that separate rectangular
waveguides couple V- and H-polarized waves independently to the square waveguide.
Although not shown, internal impedance matching is required if there are abrupt changes
in waveguide dimensions. Levine and Sichak (1954) provide input to a square waveguide
with separately polarized waves by using dipoles fed by coaxial cables, and Johnson
(1967) inputs a square waveguide with rectangular waveguides. Typically, the length of
transition (taper) between waveguide sizes is short, and often abrupt for compactness.
Bandwidths are ordinarily 5 to 10 percent for VSWR of 1.2 or less, and isolation between
V- and H-polarization inputs are at least 40 dB.
Figure 7–20 shows photographs of two commercially available transducers with rect-
angular waveguide inputs, one having a square and the other a circular waveguide output.
Note the abrupt changes in waveguide size that permit compact designs.
Omnidirectional Antennas 307
7.3.3. Corrugated Horn Feeds

It is sometimes desired that dual-polariza-
tion reflector antennas and their feeds
have equal azimuth and elevation beam-
widths. From sec. 6.4 it may be seen that,
for a square horn, the side lobes are larger
and the pattern is narrower in the E plane FIGURE 7–20.
than in the H-plane. Particularly for small Transducers having square and circular
horns, the lack of amplitude taper in the waveguide outputs. (Courtesy of Microwave
E plane may cause even greater diffrac- Development Laboratory, Inc.)
tion from the E-plane waveguide walls.
Some of these problems can be minimized
through use of corrugated horns. However,
they are heavier, larger, and more expen-
sive than smooth-walled horns. Universal
patterns for corrugated horns are given by
Love (1984, pp. 15–19). qf
Figure 7–21 illustrates a commonly
used corrugated horn, which is tapered
outward from a circular waveguide and
thus has a circular aperture. It has axially
symmetric beams, low side lobes, and
low cross-polarization. The concept of
corrugations is as follows: where the horn FIGURE 7–21.
taper begins, the slot depth is roughly l/2 A cone-shape tapered corrugated horn with
(giving an approximate short circuit like circular aperture. After Thomas (1978), © 1978
a shorted l/2 length transmission line), to IEEE.
keep the wave bound as though next to a
conducting wall. In addition, as the mouth of the horn is approached, the slot depth is
tapered so that near the throat the slot depths are roughly l/4. The l/4 slots approximate
the open circuit effect of a l/4 line) to detach the wave from the walls and thereby reduce
E-plane diffraction from the walls. The performance of sectoral and pyramidal horns can
also benefit from having corrugations, which cause the E-plane beamwidth to be broad-
ened and the side lobes to be greatly suppressed (Love 1984, pp. 15–43).
7.4. Omnidirectional Antennas

Much of the emphasis in antenna theory and practice is on increased directivity, which
improves the system performance in any point-to-point radio application. Narrow beam-
width allows discriminating (on reception) against interfering signals from directions
other than that of the desired signal, and on transmission it minimizes interference with
receiving stations in other than the desired direction. The resulting increased antenna
gain improves the signal strength in the desired direction.
Nevertheless in some applications a narrow beam is undesirable. In entertainment

broadcasting and in certain forms of communication an omnidirectional horizontal pattern
is required. In unstabilized spacecraft applications, isotropicity may be desired. Omnidi-
rectional refers to a constant pattern in one plane whereas isotropic refers to a constant
pattern over a sphere.
Omnidirectionality is fairly simple to obtain. For example, a vertical dipole or monopole
has an omnidirectional pattern in the horizontal plane, with vertical polarization. If horizon-
tal polarization is required, the turnstile antenna that has been described will give a fairly
uniform omnidirectional pattern; these can be arrayed vertically to give a narrow vertical
beamwidth and corresponding gain, as is commonly done for TV and FM broadcasting. At
still higher frequencies the biconical horn can be used for either vertical, horizontal, or even
circular polarization with proper excitation. At the lowest frequencies, of course, only verti-
cal polarization is practical for horizontal-plane applications above a conducting earth.
An isotropic antenna is not possible to achieve, even theoretically, as was mentioned
in the discussion of the concept of an isotropic radiator, in sec. 3.3. Such an antenna is
usually desired for installation on an unstabilized spacecraft, so that even though it may
spin or tumble it will always radiate toward the earth.
On artificial satellites or spacecraft that are not large in comparison with the wave-
length, the simple turnstile antenna (Fig. 7–17) works quite well. Its nearly omnidirec-
tional properties in the plane of the dipoles have already been mentioned in sec. 7.3.1,
as has the fact that in directions other than in this plane it radiates elliptically polarized
waves, becoming circularly polarized on the axis perpendicular to the antenna plane. The
polarization is anything but uniform over the entire pattern, but if the antenna on the
earth end of the circuit can receive circular polarization of both senses, there will always
be some signal of proper polarization for reception. This antenna has been used success-
fully on artificial earth satellites of diameters up to about one meter at frequencies up to
about 150 MHz (wavelength 2 meters). Each dipole is mounted so that each half projects
radially from the surface of the satellite, that is, the body of the satellite occupies the
space between the two halves of the dipoles. This fact does not seriously modify the
turnstile performance as long as the satellite is not too large and if its surface is fairly
“clean”—uncluttered by sizable projections.
At very much higher frequencies isotropicity is more difficult, because of the larger
size of the satellite in relation to the antenna size. One practical solution is to do away
with the need for omnidirectional radiation by stabilizing the satellite or spacecraft. (On
some spacecraft this has been done so that a directional antenna, aimed at the earth, can
be used.) Otherwise, it is necessary to place individual radiators at a number of points
around the surface of the vehicle so that at least one of them will be on the side toward
the earth at all times. At sufficiently high frequencies these radiators may take the form
of slots in the metal skin of the vehicle.
7.5. Electrically Small Antennas

In some circumstances it is impossible to accommodate an antenna of sufficient size
to have appreciable gain and high efficiency, owing to limitations of space or cost.
Electrically Small Antennas 309
This situation occurs most often at low frequencies, where an antenna of wavelength
dimensions is extremely large. Then it becomes important to consider antennas that are
physically and electrically small. The emphasis of the discussion is on their electrical
smallness, since mere physical smallness does not by itself make an antenna unusual. An
electrically small antenna is one that is small compared to the wavelength.
Some basic forms of electrically small antennas have already been discussed in secs.
4.1 and 4.5—the short dipole and the small loop. Their inefficiency was noted, particu-
larly in connection with the use of vertical grounded antennas for transmitting at LF and
VLF. These are actually monopoles or half dipoles operated in conjunction with their
ground-reflection images. They may sometimes be one hundred or more meters high, yet
they are electrically small because the wavelength may be several kilometers. Their inef-
ficiency is due to a combination of factors. The radiation resistance is very low, so that
a high current is required. At the same time the input impedance includes a high capaci-
tive reactance. If the required high current is to be obtained with a reasonable generator
(transmitter) rf voltage, this capacitive reactance must be tuned out with a large series
inductance (coil). The coil, in turn, has some resistance, in which appreciable loss can
occur due to the heavy current. The heavy current also flows into the ground so that
further losses will occur unless the ground conduction is excellent.
There are various means of reducing these losses. Capacitive top loading is used to
increase the radiation resistance, thus lowering the current required for a given radiated
power. Extensive systems of buried wires are used to increase the effective ground con-
ductivity. Loading coils (tuning inductors) are made of large conductors to reduce resis-
tance. These measures are necessary when the electrically short, grounded antenna is
used for transmitting. Consequently, at higher frequencies where efficient antennas may
be of reasonable size, electrically small antennas are not generally used in transmitting
applications.
There are times, at the low and medium frequencies especially, where electrically
small antennas may be necessary for receiving. Moreover, it is fortuitous that in receiving
applications at these frequencies the antenna efficiency may be unimportant, so that the
electrically small antenna may function very effectively. This may seem to be a contra-
diction of the “reciprocity” that has been ascribed to antennas in their transmitting and
receiving behavior, but it really is not. The reciprocity principle is correct and does hold
for electrically small antennas as long as only the radiated and received signal is consid-
ered. In reception, however, the important consideration is the signal-to-noise ratio.
The electrically small antenna of low efficiency may be effective when the alternative
is a larger antenna of about the same directivity, and when the predominant noise is
external, that is, enters the receiver via the antenna. These circumstances occur primarily
at frequencies below about 10 MHz. The predominant noise is atmospheric static, and
high-directivity antennas are usually out of the question, especially below 2 or 3 MHz.
Thus the alternative antennas may be a tall tower, possibly top-loaded, or a very much
shorter vertical monopole. The patterns, and hence the directivities, will be practically
the same in either case. The difference being considered between antennas is efficiency
(the factor k defined in sec. 3.3.5) and thus gain, and it can be shown that, within wide
limits, this will not affect the signal-to-noise ratio under the stated conditions.
Very common examples of electrically small antennas are those used for reception in
the broadcast band—535 to 1,605 kHz (approximately 200 to 600 meters wavelength).
As is well known, a short random length of wire serves very well as an antenna for local
reception, and at night even distant reception is possible. Automobile radios typically use
a short vertical monopole, about one meter in length, or about 0.01 wavelength or less.
Small “table radios” and also some larger sets often have built-in small-loop antennas
that are resonated by means of a parallel variable condenser; the combination then serves
as the tuned input circuit of the receiver rf amplifier. The ultimate in smallness is the tiny
“ferrite rod” antenna used in transistor pocket portable radios. This antenna is actually a
small-diameter elongated coil wound on a rod of ferrite material up to perhaps 10 cm in
length, connected in parallel with the rf amplifier tuning capacitor; it thus serves as a
combination antenna and tuned circuit as does the somewhat larger loop antenna of the
table radio. The ferrite core increases the rf flux in the coil. It also allows the necessary
inductance to be obtained with relatively few turns of wire so that the resistance is kept
low; hence the coil Q is high. Consequently both the induced voltage and the selectivity
of the circuit are greater than with an air-wound coil of the same size.
Electrically small loop antennas are also very commonly used for receiving at MF,
LF, and VLF, again because efficiency is not important. Also, since the wave polarization
is always vertical at the low frequencies (below 2 MHz), advantage may be taken of the
nulls that occur in the directions perpendicular to the plane of the loop, to eliminate
interfering signals from a particular direction, if the loop is made rotatable.
Sometimes an electrically small antenna may be desirable for transmitting at frequen-
cies where a quarter-wavelength vertical antenna would be possible but fairly expensive
and perhaps inconveniently large. A “short monopole” antenna will, in this case, have
just as good directivity, but the radiation resistance will be very low so that the efficiency
may be poor. A method of improving the efficiency, which has been incorporated in a
number of commercial designs, is the use of a resonant top-loading structure. In effect,
the antenna consists of a short vertical section that radiates (with vertical polarization)
and a horizontal section that is nonradiating (because of cancellation due to ground
reflection) but is of sufficient length, inductance, and capacitance to make the entire
structure resonant. That is, the input impedance is nonreactive and the radiation resistance
is much higher than it would be for a monopole without top loading. A metallic ground
plane is also employed. Therefore such antennas may be reasonably efficient even though
electrically small.
7.6. Direction-Finding Antennas

The term direction finding has a special and a general sense. In its special sense it refers
to the process of determining the direction of arrival of a signal from a distant source
purely by means of receiving equipment connected to a directional antenna system. In
the more general sense it may include systems that transmit as well as receive (e.g.,
radars). The term is used here in the general sense. The subject is very large, but all that
Direction-Finding Antennas 311
will be attempted here is a brief sketch of the antenna types and techniques, since the
basic operation of the antennas has already been covered in chapters 4, 5, and 6.
7.6.1. Low-Frequency Null Systems

The term passive is applied to systems or devices that function only when acted upon
by an external signal, as opposed to active systems that generate their own signals. Direc-
tion finders at low frequencies (LF and VLF) are generally passive. They are used as
aids to navigation and for locating the sources of unidentified signals, for various pur-
poses. They almost always utilize a null indication, that is, the direction of minimum
signal (ideally, zero) is determined, rather than maximum.
Multiturn small-loop antennas are customarily used. Since the wave polarization at
LF and VLF is always vertical, the plane of the loop is vertical, and provision is made
for rotating it about a vertical axis. An electrically small loop pattern in a horizontal plane
is like that of the short horizontal dipole (Fig. 4–4, chapter 4), with nulls in the directions
perpendicular to the plane of the loop. Therefore there is a 180-degree ambiguity of the
measured direction. This ambiguity may be resolved by employing in conjunction with
the loop a short vertical monopole located nearby (a negligible distance away in terms
of the wavelength). The null direction is first found with the monopole disconnected.
Then its output is connected to the receiver so that it adds to the loop voltage, but with
a 90-degree phase delay introduced by inductive reactance. The effective pattern of the
loop-monopole combination is an asymmetrical cardioid rather than the symmetrical
pattern of the loop alone, and by rotating the loop for maximum signal with the monopole
connected it can be decided which of the two null directions, observed with it discon-
nected, is the true direction of the incoming signal. This type of direction finding is
subject to errors from many causes, and numerous techniques have been devised for
minimizing them.
At somewhat higher frequencies, the ionosphere can produce appreciable horizontally
polarized downcoming signal components that tend to “fill in” the pattern nulls of a loop
and destroy the direction-finding accuracy. The Adcock antenna can overcome this effect.
It consists of two vertical dipoles or monopoles spaced a quarter wave or less apart and
connected to the receiver input with opposite polarity so that nulls occur in the directions
perpendicular to the plane of the dipoles (as in the patterns of the right-hand column of
the top row of Fig. 5–3, chapter 5). The pattern is essentially the same as that of a loop,
but the antenna is insensitive to horizontal polarization. The null ambiguity can be
resolved by the use of an additional (third) dipole or monopole, just as for the loop.
A goniometer is an electrical combining device used with multiple fixed antennas,
having a stator connected to the fixed antennas and a rotor connected to a receiver or
transmitter, allowing the pattern of the fixed antenna system to be rotated as the rotator
is rotated. By using two fixed vertical loops at right angles to each other (crossed loops)
in conjunction with a goniometer, mechanical rotation of the rotor may be made equiva-
lent electrically to rotating a loop antenna (Terman 1943, p. 879). Thus, this combination
of loop antennas with a goniometer permits the use of large fixed loops and also permits
locating them at a distance from the receiver, with transmission-line connections. Crossed
Adcock antennas can also be used with a goniometer.
7.6.2. Beam-Maximum Systems

At VHF and above it is more practical to employ an antenna with a very narrow direc-
tional beam, and to determine the signal direction by rotating the antenna until a signal
maximum is observed. This is the technique used for ordinary radar direction finding.
Radar antennas are usually rotatable, or the beam may be pointed in some manner.
Beamwidths as small as one degree or even less can be used, although at the VHF radar
frequencies beamwidths of 10 to 20 degrees are more common. The antennas may be
arrays, parabolic reflectors, or lenses. In special cases other antenna forms may be used,
such as surface-wave antennas. They may be mechanically steerable in both elevation
angle and azimuth angle. Beam steering by varying the phasing of an array is commonly
used for direction finding.
7.6.3. Lobing Antennas

When a narrow beam is “scanned” through the direction of an incoming signal, the loca-
tion of the maximum direction can be determined to within some fraction of the beam-
width. By employing optimum techniques this fraction can be made fairly small. It is
about one-tenth of the beamwidth for a radar system employing simple visual observation
of the signals on a cathode-ray-tube indicator. Even with optimum processing, however,
the accuracy is limited by the fact that the “nose” of a beam is blunt, rather than sharp.
A method of measurement that circumvents this difficulty has been devised for use
when extreme accuracy is required, as in weapon-pointing radars—the method of lobe
comparison, or lobing. In this method, the beam is shifted, without moving the antenna
as a whole, by equal fractions of a beam-
B A
width in both directions from the nominal
beam direction. This beam shifting can
be done either in rapid time sequence
Left lobe Right lobe (sequential lobing) or by generating two
pattern Crossover pattern
beams at the same time (simultaneous
Error signal lobing). The patterns of the two beams
amplitude for
directio B then overlap, as shown in Fig. 7–22. The
theory of operation can be described by
considering signals arriving first from
point A, and then from point B.
The antenna output in the simultane-
ous-lobing case is connected to two
receivers, one of which receives the dif-
Zero-error axis ference of the signals from the two lobes,
FIGURE 7–22. whereas the other receives the sum. If the
Patterns of a lobe-comparison antenna. signal is coming from A, there will be no
Direction-Finding Antennas 313
difference signal, since the amplitudes of the two lobes are equal. If the signal is coming
from B (an “off-axis” direction), there will be a difference, or error, signal, as shown by
the fact that the dashed line crosses the two lobe patterns at different levels. The sense
(direction) of the error is found by comparison of the phases of the sum and difference
signals in special receiver circuits that convert the error signal into a d-c voltage; the
polarity of this voltage will reverse as the signal direction changes from right-of-center
to left-of-center. The sum signal is also used, in radar systems, for signal detection and
range measurement. In the sequential-lobing scheme a single receiver is used, and the
received signals from the two beams are switched so that their amplitudes can be
compared.
It is apparent that the “difference pattern” of the two lobes (the difference of the pattern
amplitudes plotted as a function of angle) will be very sharp, rather than blunt like the
nose of a lobe. Because of this fact the accuracy possible by lobing is much greater than
by simply observing the maximum direction of a beam.
The two displaced beams can be generated in a number of ways. With an array antenna,
changing the phasing of the elements will produce the desired result. With a paraboloidal
reflector or lens antenna, the beam may be shifted in the desired manner by moving the
feed slightly off the optical axis, as shown in Fig. 6–3b, chapter 6. In a sequential lobing
system this shift may be made by mechanically moving the feed off-axis first in one
direction, then in the opposite direction, at a rapid rate. In either a sequential or a simul-
taneous lobing system, two feeds may be used, each displaced a small amount from the
optical axis, in opposite directions. In the simultaneous lobing system both feeds are
connected to the receiver through a hybrid junction (coupler) (sec. 2.6) so that a differ-
ence (error) signal is received. For sequential lobing, the receiver is switched between
the lobes.
Thus far, lobing only in one plane has been discussed. It may be desired, however, to
measure the incoming signal direction in two mutually perpendicular planes, for example,
vertically and horizontally. This can be accomplished by generating four lobes—left,
right, up, and down with respect to the nominal beam axis. With a paraboloidal reflector
or a rotational lens this requires four fixed feeds. A typical arrangement of four horns is
shown in Fig. 7–23. Their waveguide outputs are combined by means of hybrid junctions
in such a way that one receiver is fed
the difference between the signals of the
upper and lower pairs, and another the Four-horn
difference between the left and right pairs. feed cluster
Paraboloidal
A third receiver is fed the sum of all four Zero axis of
four-lobe
Paraboloid reflector
axis
horns. pattern
The sequential-lobing equivalent of the
four-lobe simultaneous system is accom- Four waveguides
plished by moving a feed (of a reflector To hybrid
or lens) in a circle around the zero axis. junctions
The beam direction then sweeps out a FIGURE 7–23.

cone about this axis. This method is called Four-horn simultaneous-lobing (monopulse)
conical scanning. It is most feasible feed for paraboloidal reflector.
with small microwave antennas, where the required mechanical motion is not very
great.
The simultaneous-lobe-comparison method, when used in a pulse radar system, is
known as monopulse lobing, since the signal amplitudes can be compared on a single-
pulse basis; the sequential-lobing system requires at least two successive pulses for its
operation. The method that has been described is also called amplitude monopulse, since
it is based on comparing the signal amplitudes in two angularly displaced lobes of a
single antenna. Another method, phase monopulse, uses two or four antennas placed side
by side, with their beams pointing in the same direction but with their phase centers
separated. If a signal arrives from an off-axis direction it will be received with equal
amplitude in each beam but with a phase difference, which may be used to generate an
error signal. When this phase method is used with nonpulsed systems it is referred to as
phase-comparison lobing or as an interferometer system. It is capable of greater accuracy
with a given antenna size because the sharpness of the error response is determined by
the separation of the antennas rather than by their individual beamwidths. The amplitude-
comparison lobing system, on the other hand, is more compact and more readily steerable
as a unit, for the same antenna gain and beamwidth.
In the phase-comparison or interferometer system, the interference pattern between
the two beams becomes multilobed as the separation of the antennas is increased. The
principle is identical to that of the interference between an antenna and its image in a
reflecting surface, as discussed in sec. 1.4 (see also Figs. 1–14 and 1–15). Therefore
ambiguity can exist if the individual beamwidths are appreciably greater than the lobe
widths, but the ambiguity can usually be resolved by one of a number of methods. This
technique is used in VHF satellite tracking systems. Interferometers are also commonly
used in radio astronomy applications.
Aircraft landing systems use a technique similar to the amplitude lobing principle,
where antenna patterns are transmitted from fixed ground stations and received by aircraft
receivers. For lateral guidance, ILS (Instrument Low-Approach System) radiates, from
the far end of a runway, two identical beams in the high-VHF region that are separated
in azimuth, but modulated at different frequencies. When the signal strengths of the dif-
ferently modulated VHF signals are equal, the aircraft is on lateral course. For vertical
guidance, ILS radiates, from the near end of a runway, two identical beams in the low
UHF region that are separated in elevation, but modulated at different frequencies. When
the UHF received signals of the differently moduated signals are equal, the aircraft is on
the vertical (i.e., proper glide) course.
Lobing is also used for passive direction finding at high frequencies, where it is supe-
rior to the loop or Adcock antenna methods.
7.7. Mechanical Scan Antennas

7.7.1. Introduction
In some applications, especially radar, it is necessary to “scan” a directional beam through
some angular sector—for example, around the horizon (360°) or through the vertical
Mechanical Scan Antennas 315
angle sector from the horizon to the zenith (90°). Sometimes a more limited sector can
be scanned in repetitive fashion, an extreme example being the conical scan lobing
system, which was described in sec. 7.6. Scanning in the horizontal plane is known as
azimuth scanning; vertical-plane scanning is called elevation scanning. The two motions
may sometimes be combined when it is desired to scan a large solid angle with a narrow
beam; typically, the azimuth scan is performed at a relatively slow rate and the vertical
scan is a rapid saw-tooth motion, with at least one vertical sawtooth scan occurring during
the time it takes for the azimuth scan to progress one beamwidth. (This insures that there
will be no gaps in the scan coverage.)
Relatively slow azimuth scanning is commonly accomplished by simply rotating the
entire antenna about a vertical axis at a constant rate. Typical speeds for radar antennas
are 1 to 30 rpm, depending on the antenna size. Limited sectors can be scanned with the
antenna as a whole remaining stationary, either by moving the feed of a reflector or lens,
or by varying the phasing of an array. These beam steering methods permit higher scan-
ning speeds. In fact, for many applications inertial-less scanning by electronic phasing
of an array is the preferred method for rapid beam scanning and/or tracking. Conse-
quently, chapter 8 is devoted entirely to the concepts of electronically scanned antennas
(ESA).
7.7.2. Rapid Mechanical Scan Antennas

High-speed repetitive scanning of limited sectors is sometimes accomplished at micro-
wave frequencies through the use of devices in which a reciprocating or sawtooth beam
motion is produced by purely rotational mechanical motion of a feed or mirror. These
devices often involve combinations of specially curved reflectors and parallel-plate
regions; typical examples are the Foster, Lewis, and Robinson scanners. The organ-pipe
scanner is another mechanism that includes a rotating waveguide horn that consecutively
energizes waveguides that have outputs arranged in a line, thereby transforming a con-
tinuous rotating motion into a reciprocating motion. The above-mentioned scanners are
described by Kelleher (1993).
Several different types of high-speed, mechanical scanning antennas were developed
at the Engineering Experiment Station (EES), Georgia Institute of Technology (now
Georgia Tech Research Institute). These antennas scanned, typically, 20 times per second
over an azimuth sector, and the active sector itself could be rotated in azimuth. Design
frequencies included 9, 16, 24, and 70 GHz. Two general configurations were used: (1)
multiple rotating, back-to-back reflectors and (2) multiple rotating horns feeding one or
two geodesic Luneberg lens that illuminated one reflector. Some of the back-to-back
scanners were used to transmit or receive either horizontal, vertical, or left or right cir-
cular polarization, and simultaneously receive two polarizations: horizontal and vertical,
or left and right circular.
Back-to-back reflector antennas were developed at EES during the late 1940s, 1950s,
and 1960s for radar tracking of projectiles down to the point of impact. Examples include
naval ship gun shells, bombs, and air-dropped mines. Attributes of the fast scan rate
include improved tracking of high-speed targets and elimination of the radar display
RING SWITCH COVER

TOP HONEYCOMB PANEL
14 ft
OFFSET PARABOLIC-
CYLINDER REFLECTOR
9 ft
PRIMARY FEED
32˝ 30˝
PRIMARY FEED
MOUNTING ANGLES
BOTTOM HONEYCOMB PANEL RADOME

FIGURE 7–24.
Exploded view of major MILORD antenna components.
flicker (that occurs with slow scan rate radars). The back-to-back reflectors rotated about
a vertical axis. Advantages of multiple reflectors include increased target revisit rate for
a given antenna spindle rotation rate, and increased beam-time on target for a given target
revisit rate. An azimuth sector of about 110 or 80 degrees was viewed by a radar, depend-
ing whether the scanner had three or four reflectors.
Figure 7–24 shows the principal components of the MILORD antenna (Holliman and
Hollis, 1956). MILORD is an acronym formed from the words Mine Impact LOcating
Radar Device.
Key features of the MILORD antenna follow:

Frequency = 9.375 GHz
Gain ≈ 39 dB
Horizontal polarization
Azimuth beamwidth = 0.7°
Elevation beamwidth = 3.5°
Weight = 5600 pounds (2540 kg)
Rotation rate = 240 RPM
Linear velocity at rim = 120 mi/h (193 km/h)
Beam scans at 16/s over 85° azimuth sector
Waveguide switch activates one reflector feed at a time over a 85° azimuth sector
For reduced aerodynamic drag, the honeycomb panels and the radomes enclose the
antenna to form a cylinder.
Figure 7–25 shows an exaggerated cross section through a feed and a reflector. Each
feed is of the hog-horn type, which consists of a parabolic reflecting barrier between
closely spaced aluminum plates and an H-plane sectoral horn at the focal point. The
linear aperture of each feed is 61 cm. The offset angle of the feed is 2°, giving a reduc-
tion in power reflected back into a feed. VSWR at the feeds was approximately 1.05 over
a 2 percent band.
A key component for a rapid sector-scan antenna is the high-speed waveguide switch.
The first sector scanners used a turnstile type switch. A turnstile switch is, in essence, a
rotary joint having multiple outputs, with all but one being short-circuited at a given time
(Long 1951). Later, the scanners used a ring switch design having a wider bandwidth,
and higher peak power performance (Johnson, Holliman, and Hollis 1960). A ring switch
contains one input junction and multiple output junctions. The basic junction geometry
consists of a cavity that couples energy to two rectangular waveguides, one is split lon-
gitudinally in the E-plane and the other is an output junction. The junction has a simple
geometrical configuration, requires no additional matching devices, and has a useful
bandwidth that exceeds 25 percent.
Figure 7–26 shows four dual polarized back-to back paraboloidal antennas that rotate
about a spindle at 150 RPM. The paraboloids are separately activated over a 70° sector,
and the sector is positionable over 226°. The 3 dB azimuth and elevation beamwidths
are each 1.5°; and the beams scan a 70° sector at 10, 5, or 2.5 scans per second. Separate
waveguides supply horizontal and vertical polarizations, and orthogonal circular polar-
izations are obtained by appropriate addition and subtraction of the H- and V-polarized
waves. The relative phase shift through the separate ring switches for the H and V polar-
ization paths was necessarily small, to provide isolation greater than 30 dB between the
two circular polarizations.
Figure 7–27 is a photograph of a rapid-scan, geodesic Luneberg lens antenna mounted
on a test stand (Hollis and Long 1957). The antenna, developed for use with a mortar
locating radar, produces two beams from separate geodesic lenses that scan sequentially
across a 40° azimuth sector. A parabolic cylinder reflector focuses the beams in elevation.
Parallel extensions connect the lenses with the reflector focal line, and they are folded
INPUT WAVEGUIDE
PRIMARY
FEED
PRIMARY FEED
PHASE FRONT
LATUS
α
α RECTUM
REFLECTOR
FOCAL LINE
2α
VERTICAL
REFLECTOR PHASE FRONT
REFLECTOR
FIGURE 7–25.
Exaggerated cross section through MILORD offset parabolic cylinder reflector and primary
feed. Reflector tilt angle a is 2°. From Holliman and Hollis (1956).
upward to permit mounting the lenses beneath the reflector. The two beams are separated
in elevation by means of line source feed horns that are located on opposite sides of the
focal line of the reflector. A sandwich-type radome covers the line source feeds.
The scanning mechanism is located in the cavity formed between the two lenses. Four
sectoral lens feed horns are spaced at intervals of 90° on two feed wheels, one for the
upper- and the other for the lower-elevation beam. The wheels are mounted coaxially
and the lens feed horns of one wheel are spaced 45° in azimuth from those of the other.
The feed-switching mechanism includes a four-way switch at the center of each feed
wheel, and a two-way chopper switch that

rotates at four times the speed of the feed
wheels. When the feed wheels rotate, the
upper and lower beams alternately scan
the 40° azimuth sector, each at a rate of
seventeen scans per second.
Figure 7–28 illustrates the geodesic
lens-scanning antenna for the Combat
Surveillance Radar AN/ MPS-29 (XE-1),
which operates at 70 GHz (Long, Rivers,
and Butterworth 1960). Key operating
parameters are gain 54.7 dB (includes
2.8 dB loss); 0.2° and 0.3° azimuthal
and elevation beamwidths, respectively;
twenty scans per second over a 30° scan
sector; vertical polarization. Details of the FIGURE 7–26.
lens, which is of a helmet type configura-
A dual-polarization rapid-scanner having four
tion, are given in sec. 6.8.8. A ring switch, back-to-back paraboloids, shown within the
which serves as a commutator for the lens machine shop of Engineering Experiment
feeds, consists of a stationary input section Station at Georgia Tech, circa 1963.
and a rotary output section that has eleven
sectoral horns that energize the geodesic
lens. The feed horns, seen protruding
below the rotating ring, are activated
sequentially as each passes through the REFLECTOR
scan sector of the input lips. The lens feed
horns, shown protruding below the rotat-
ing ring, are activated as each passes
through the scan sector of the input UPPER LENS SCANNER
lips. LOWER LENS FRAME
The path of the microwave energy PEDESTAL
through the ring switch is shown in Fig. TEST STAND
7–28. For 70 GHz, where waveguide
losses are appreciable, the E-dimension of
the ring switch waveguide and junction is FIGURE 7–27.
1.375l, corresponding to the conducting A rapid-scan, geodesic Luneberg lens-scanning
surface spacing of the lens. This height antenna mounted on a test stand (from Hollis
was selected to minimize lens losses, and and Long 1957; © 1957, IEEE).
was a compromise between ohmic losses
and losses from mode conversion (of the propagating waves) caused by bending in the
E-plane. The other switch cross-sectional dimension is the H-plane width of RG-98/U
waveguide, which is 0.875l. A hydraulic brake system allows, on operator command,
the beam scan to be stopped and positioned in 2 seconds to within less than 0.5 millira-
dian, anywhere in the 30° scan sector.
TRANSMITTED
ENERGY
2 1
6 5
1 – RING SWITCH INPUT SECTION

2 – RING SWITCH ROTATING SECTION
3 – GEODESIC LENS INPUT-LIP SECTION
4 – GEODEIC LENS
5 – PARALLEL PLATE EXTENSION
6 – REFLECTOR FEED ASSEMBLY
7 – REFLECTOR
FIGURE 7–28.
Exploded view of 70 GHz antenna having a scanning geodesic Luneberg lens (from Long,
Rivers, and Butterworth 1960).
7.8. Low-Noise Receiving Antennas

The principal criterion of performance in many radio systems is the signal-to-noise ratio
in the receiver. Noise is always present at some level, and the signal level must usually
be greater than the noise level for successful reception, although by special averaging or
integrating techniques some types of signals may be detected or read even though their
amplitudes are below the noise level. In all cases there is a minimum useful signal level
Low-Noise Receiving Antennas 321
that is expressed in terms of its ratio to the noise level. Consequently, it is important to
keep the noise level as low as possible.
Noise in radio systems is of various types and origins in different parts of the spectrum.
In the frequency region below about 10 MHz the principal noise is usually “static,” which
is generated in the atmosphere by lightning discharges and propagates for great distances
by ionospheric reflection. It enters the receiver via the antenna along with the desired
signal, and little or nothing can be done to minimize it. (A high-gain antenna will help
by increasing the received signal while the received noise remains virtually unchanged.
But at the very lowest frequencies, directive antennas are not feasible.)
In some situations, especially in the HF through UHF regions, man-made noise
from electrical machinery, appliances, power lines, and automotive ignition is the prime
offender. Much of this noise tends to be vertically polarized and to be stronger closer to
the ground. For the first of these reasons, horizontal polarization may be advantageous,
and is therefore used for FM and TV broadcasting. It is also beneficial, in these frequency
regions, to employ an elevated antenna connected to the receiver via a balanced or coaxial
transmission line, so that all reception takes place well above the ground.
From about 30 to 300 MHz, in the absence of man-made noise, cosmic noise predomi-
nates. This noise originates in the galactic system and is heard in audio-output receivers
as a “hiss” when the receiver gain is sufficiently high; on a radar A-scope it is called
“grass,” and on an analog TV screen it is sometimes referred to as “snow.” These terms
describe its approximate appearance. It is not seen or heard when a strong signal is
received because the receiver gain is then reduced (usually automatically). Cosmic noise
is, like static, unavoidable; it is received via the antenna along with the signal. (A high-
gain antenna is beneficial, however, because it increases the signal strength without
increasing the noise. This is true because the signal comes from a single direction whereas
the cosmic noise comes from virtually all directions of the sky.)
The cosmic noise decreases with increasing frequency, so that above about 1 GHz it
is of negligible importance. Some noise is radiated by the earth’s own atmosphere, and
by the ground. The atmospheric noise does not become significant until the frequency is
about 3 GHz or higher, and is not severe below about 10 GHz. The ground noise, on the
other hand, is relatively constant at all frequencies. The result of all these effects is that
there is a region of minimum noise between about 1 and 10 GHz, with cosmic noise and
static predominating at lower frequencies and atmospheric noise predominating at higher
frequencies. (See also sec. 9.13.2).
The foregoing discussion refers solely to noise that enters the receiver via the antenna.
Additional noise is generated within the receiver itself. Techniques exist, however, for
reducing this internal noise to such a level that the system performance is largely deter-
mined by the external noise, even in the low-external-noise frequency region. (Masers,
parametric amplifiers, and tunnel diodes and low-noise transistors are the basis of espe-
cially low noise amplifiers.) When the predominant noise is the noise that enters from
external sources, via the antenna, it is desirable to consider antenna designs that reduce
the reception of noise relative to the received signal reception.
In the frequency region UHF and above where the noise from the ground radiation is
a significant portion of the total noise, certain aspects of the antenna design do affect the
system noise level. As has been mentioned in chapters 3, 5, and 6, directional antennas
always have some side lobes and back lobes, through which some reception of ground
noise can occur even when the main-beam is pointed upward. Therefore, the objective
in designing a low-noise antenna is primarily to reduce the level of these lobes as much
as possible. It is especially important to do this when the antenna will be used for such
applications as space communication, radar, or telemetry with a low-noise receiver in
the frequency range somewhat below 1 GHz to somewhat above 10 GHz.
The design techniques that minimize side and back lobes have been discussed in sec.
5.6 and 6.3. There is, however, one further factor to be considered; namely, the ohmic
losses between the waves in space and the input terminals of the first amplifier of the
receiver. Because noise will be generated by any loss that exists in this part of the system
the length of transmission line or waveguide must be kept to a minimum. For this reason,
arrays are not ordinarily used as low-noise antennas, even though their side-lobe and
back-lobe characteristics can be controlled more readily than those of reflectors. The
losses in a typical array feed system (especially if the array is large) may generate more
thermal noise than would be avoided by total elimination of the side and back lobes.
(Thermal noise generation is discussed in sec. 9.13.)
The effect of these array losses can be circumvented by placing a low-noise receiver-
preamplifier at each element of the array, ahead of the feed-system losses. The
signal phase must be preserved in the preamplifier, so that the signals from the various
elements can be combined in correct phase relationship in the usual way, to form a
beam in the desired direction. By definition (IEEE 1993), an active antenna array is an
array which all or some of the elements are equipped with their own transmitter or
receiver or both. Usually there is not a complete transmitter or receiver located at an
element. Instead, located at the elements must of course be at least those components
necessary for minimizing losses and for providing the proper array element phases.
Generally, the active array antenna technique is very expensive, and it is therefore used
only in special applications such as in electronically beam-steered arrays for radar
(chapter 8).
Lenses can be designed for low side lobes, and also they offer the possibility of locat-
ing the receiver near the feed. For very high gain applications, however, lenses become
unwieldy in the low-noise frequency region below 10 GHz. Therefore, except for elec-
tronic beam scanning radar, the greatest emphasis in low-noise antenna design is on the
use of paraboloidal reflectors.
It is well known that a highly tapered aperture illumination will reduce the side-lobe
level, and somewhat greater tapering is usual in low-noise designs than when maximum
gain is the paramount consideration. At least equally important for reflector antennas is
the minimization of illumination spillover, because spillover causes back lobes directed
toward the ground. For the same reason it is important to minimize leakage through the
openings of a mesh reflector surface; that is, the openings must be very small compared
to the wavelength.
The Cassegrain feed system (sec. 6.3) is frequently employed to minimize the trans-
mission line length, to reduce or eliminate noise due to transmission-line loss. The
Cassegrain system is also less likely to result in back lobes due to spillover. The spillover
Low-Noise Receiving Antennas 323
Collimated radiation
reflector
aperture
Horn aperture
Plane of rotation Paraboloid focus and
about a phase center of horn
horizontal axis
Typical ray paths
rf input Reflector (offset
paraboloidal
section)
Horn walls
Paraboloid
axis
Continuation of
parabolic curve
Paraboloid
vertex
FIGURE 7–29.
Sectional view of a conical-horn-reflector antenna. Metallic enclosure (heavy line) is complete
except for opening at reflector aperture.
that occurs at the subreflector causes side lobes that look at the sky rather than at the
ground; hence they result in little or no additional ground-noise. When a conventional
focal-point feed is used, if low noise is important it is desirable to locate a receiver pre-
amplifier at the apex of the feed support. Care should be taken, when this is done, to
minimize aperture blockage, which also causes higher side-lobe levels. Parenthetically,
offset-fed reflectors are commonly used to reduce feed blockage and side-lobe levels for
home-use direct-broadcast satellite TV reception.
An antenna design that is perhaps ultimate in low-noise performance is a very large,
conical-horn-reflector antenna used by the Telstar satellite-communication ground termi-
nal at Andover, Maine. Figure 7–29 is a diagram of this type of antenna. Its design dra-
matically reduces spillover and reflector leakage, and also permits the receiver and
transmitter to be located close to the feed point, at the apex of the horn. The axis of the
horn is horizontal, but the mouth of the antenna can be pointed upward. This is possible
because a section of a paraboloidal reflector is inserted in front of the mouth of the horn,
so that the wavefront is redirected at a 90-degree angle from the horn axis. The apex of
the horn is at the focus of the paraboloid. The paraboloidal aperture is circular, 20.7
meters in diameter. The antenna is used for reception at a frequency of 4.08 GHz and
provides high gain and a narrow beam in combination with extremely low noise. The
gain is approximately 58 dB, and the beamwidth is about 0.23 degree. The antenna is
also used for transmitting at 6.39 GHz, where the gain is 62 dB and the beamwidth is
0.15 degree. The entire antenna, enclosed in a 210-foot-diameter inflatable radome, is
mounted on a turntable about 50 meters in diameter. The antenna rotates about the horn
axis, and thus the aperture can be directed toward any part of the sky. A detailed descrip-
tion of this antenna has been published (Hines, Li and Turrin 1963). A modification of
the design that results in a more compact structure has been devised. It is called the triply
folded horn reflector antenna (Giger and Turrin 1965).
7.9. Synthetic-Aperture Antennas

The synthetic-aperture antenna is not so much an antenna development as it is a signal-
processing development, although it does heavily involve antenna array theory. It is a
method of synthesizing a very large array aperture by taking advantage of the linear
motion of a relatively small antenna carried by an aircraft or spacecraft. The large aper-
ture is said to be synthesized in the sense that it does not physically exist in its entirety
at a single instant of time. Thus the very high gain, narrow beamwidth, and resulting
high resolution of a large-aperture array is obtained without the need for physically con-
structing a very large antenna (and just as importantly, without the need for carrying the
resulting large structure on an aircraft or spacecraft).
This technique is applicable primarily to mapping radar, which is known as synthetic
aperture radar, or SAR (Cutrona 1990). The radar equipment and its actual side-looking
antenna are carried by the aircraft or spacecraft (e.g., an earth-orbiting satellite), which
travels a straight course at constant altitude. (An earth orbit sufficiently approximates a
straight course during the array-synthesis processing time.) The region that can be
mapped by this method is a swath parallel to the vehicle’s course; the minimum and
maximum distances of this swath from the vehicle are determined by the downward
angles of the lower and upper boundaries of the vertical-plane beam pattern, subject to
additional constraints imposed by the horizon, and by the radar sensitivity and resolution
considerations.
The synthetic aperture can be regarded as a linear array, of which the actual antenna
is a single element. The linear motion of the vehicle positions this single element for
successively radiated radar pulses, at successively displaced positions. Now consider n
such positions and let the n successive radar echoes received at these positions be stored
in some manner that both amplitudes and phases are preserved. Then by suitable phasor
addition of these stored (time-delayed) signals, a resulting signal can be obtained that is
as if it had been received by a linear array of n elements. The effective angular resolution
of this synthetic aperture is the same as that of an n-element linear array.
It is evident that the storage and processing of the successively received echo signals
is the key to the feasibility of this scheme. All of the beam-forming considerations appli-
cable to “real” arrays are applied in this processing, such as tapering of the effective
element amplitudes (processing weights) to result in a desired side-lobe level. The first
SARs used special optical processing methods, because of the extremely large amount
of processing required. However, by 1984 digital signal processing (DSP) was fast
enough that it was used on the space shuttle-imaging-radar-B (SIR-B).
Synthetic-Aperture Antennas 325
There are three SAR modes of operation: stripmap, spotlight, and scan. In stripmap
SAR, the beam remains at a constant squint angle, which is usually perpendicular to the
flight. Spotlight is used to obtain an improved angular resolution of a known location of
a target of interest. As the platform moves, the beam-pointing direction is changed so as
to keep pointing at the target. Lastly, the scan mode is seldom used; but by means of
beam direction changes, it does permit observing a straight path that is not parallel to
the flight path. Even though there are different SAR modes of operation, the technique
employs side-looking from an elevated platform, and the resultant improvement in
angular resolution is one-dimensional, in the direction parallel to the vehicular motion.
The required resolution in the orthogonal (range) direction is obtainable because the
radar employs pulses. By means of pulse compression, a relatively long-duration trans-
mitted pulse can be transformed into a very short received pulse, and the range-dimension
resolution is determined by this received pulse length. Pulse compression is accomplished
by transmitting a very wide-band coded pulse, which is decoded and compressed by an
appropriate receiver filter. The pulse at the filter output is much shorter in duration than
the transmitted pulse, and the received signal sensitivity is dependent only on the trans-
mitted pulse energy; and the range resolution obtained depends only on the compressed
pulse length.
This combination of angle resolution by means of the synthetic aperture technique and
range resolution by pulse compression allows very high-resolution mapping of the surface
of the earth and other planets, monitoring of the positions of ships at sea, and study of
changes in the configuration of the earth or sea surface over time (Elachi 1987).
An interesting aspect of the SAR technique from the antenna-theory viewpoint is that
the best performance is obtained by processing the data so that, in effect, the antenna
beam is focused on the earth surface at the point of reflection of the radar pulse. Ordinary
arrays, as well as reflectors and lenses, usually produce a collimated beam, in which the
rays are parallel; in effect they are “focused at infinity.” This is the appropriate strategy
for transmission distances that are in the far-field region of the antenna. The simplest
types of SAR also produce a collimated beam. But by suitable data processing the SAR
beam can in effect be focused so that the rays converge to a point at the earth’s surface
(and this can be done at all ranges, since the data are separately processed for different
target distances). The advantage of this type of processing is that it permits a much larger
synthetic aperture to be formed than would be possible if a parallel-ray beam were
used.
For this synthetic-aperture procedure, the ordinary concept of a beamwidth is not
applicable. The relevant consideration in mapping is the azimuthal resolution, which is
expressed as a distance rather than an angle. For ordinary radar, and for far-field targets,
the resolution is of the order of qbR, where qb is the azimuthal beamwidth in radians and
R is the target range. But, for sufficiently large synthetic apertures, targets are not neces-
sarily in the far-field. It is shown by Cutrona (1990, pp. 21.4–21.7) that for the focused
case, and for a sufficiently large synthetic aperture D along the flight path, the azimuthal
resolution is equal to D/2, independent of the range and also independent of the frequency
(or wavelength) of the radar. Cutrona also shows that for the unfocused synthetic aper-
ture, the resolution is λ R 2 where l is the wavelength.
7.10. Antennas with Extremely Low Side Lobes

In many applications, a low-antenna-pattern side-lobe level is of great importance. This
is especially true for radar antennas if the radar can be expected to operate in an environ-
ment that results in a significant level of “clutter” echoes; that is, a profusion of echoes
from such reflecting objects as sea waves, rain, or irregular terrain. It is also true for
radars for military use, which may be subject to interference from enemy jamming trans-
mitters located in side-lobe directions of the radar antenna pattern. Low side lobes are
also important for satellite-relay communications where it may be desired to utilize the
same frequency for two or more antenna beams that are simultaneously transmitting or
receiving different messages to or from separate locations on the earth surface. Obviously
these transmissions will mutually interfere if the side-lobe levels are too high. These are
major reasons for desiring low side lobes. There are lesser reasons as well, such as the
general desirability of noninterference with other services and considerations of low-
noise antenna design.
The basic principles of achieving low side-lobe radiation or reception have been
described for array antennas in chapter 5 and for reflectors and lenses in chapter 6. As
described there, side lobes are reduced by appropriate tapering of the array-element
power distribution or of the lens or reflector aperture illumination intensity and by con-
structing reflectors and lenses with accurate surfaces. The lowest side-lobe levels attain-
able are generally with array antennas, for which the element power distribution is more
accurately and flexibly controllable than is the illumination taper for a reflector or lens.
For arrays as well as reflectors and lenses, a −20 dB side-lobe level is sometimes ade-
quate, −30 dB is considered good, and −40 dB is excellent. A level of −50 dB is very
difficult to achieve, but side-lobe levels of −50 dB and better have been achieved with
both array and reflector antennas.
The pioneering antenna design that proved the feasibility of obtaining ultralow side
lobes, as described by Schrank, was the AWACS radar antenna—an airborne system
operating at S-band (vicinity of 3 GHz) designed by the Westinghouse Corporation for
the E-3A airplane. This slotted waveguide array, of size about 7.6 meters wide by 1.5
meter high, has over four thousand radiating slot elements. The antenna is housed in a
so-called rotodome mounted above the aircraft fuselage; this is an aerodynamically
shaped radome, of dimensions that allow horizontal-plane rotation of the antenna for
azimuth scanning. Vertical steering of the beam is accomplished by means of high-preci-
sion ferrite phase shifters.
The key to the AWACS antenna achievement has been primarily
(1) the sophisticated computer calculations of the exact phases and amplitudes
of the required element excitations, taking into account mutual coupling
effects, and
(2) the preservation of extremely close mechanical tolerances in the construction
of the array. An excellent review of this subject has been published by
Schrank (1983).
Other array antennas have subsequently been designed and built with side-lobe levels
comparable to those of the AWACS, including at least one with dipole radiators rather
Antennas with Extremely Low Side Lobes 327
than waveguide slots, thus demonstrating that the technology is not confined to slot
arrays.
Although it is generally considered that better control of side-lobe levels can be
achieved with array antennas than with reflectors, side-lobe levels in the 40 to 50 dB
range have been reported for reflector antennas. These results are accomplished using
computer-aided design and precision construction. In addition to his review of low-side-
lobe array technology, Schrank also published a review article on low-side-lobe reflector
antennas (Schrank, 1985).
The decibel side-lobe level is conventionally expressed as the ratio applicable to the
highest side-lobe level. In some contexts, the average side-lobe level may be of greater
importance than the level of the highest side lobe. The average side-lobe level is usually
expressed in terms of its ratio to the isotropic level, denoted in decibels by the notation
dBi. For high-gain (narrow-beam) antennas, this ratio is approximately equal to the frac-
tion of the total power radiated (in transmitting-antenna terms) in the side lobes. Thus,
−20 dBi means that approximately one percent of the total power radiated is side-lobe
radiation. Since the main-beam power gain is also referred to the isotropic level, this
means that if the main beam gain is 35 dBi, for example, and the average side-lobe level
is −5 dBi, then the average side-lobe level is 40 dB below the main beam radiation. (Note
that in the dBi notation, positive numbers refer to levels above isotropic and negative
numbers refer to dB below isotropic level.)
In addition to airborne radar applications, low side-lobe design is especially
important for reflector or lens antennas used on satellites for earth communication when
frequency reuse is an objective. A paper by Burdine and Wilkinson (1980) discusses this
matter and describes techniques for minimizing the side lobes of reflectors and lenses.
(Many of the principles discussed in that paper are also described in chapter 6 of this
book.).
As discussed in secs. 7.8 and 9.13.2, as major objective in designing a low noise
antenna is to reduce the side lobes and back lobes as much as possible. The well-known
horn-parabola, a successful technique for reducing wide-angle and rear lobes, is dis-
cussed in sec. 7.8. Interestingly, a conical horn-parabola design provided one of the
lowest noise temperatures ever achieved (see sec. 9.13.2). In addition, this low-noise
satellite-communication ground terminal type of antenna has been also commonly used
for commercial microwave ground links, where the extremely low, wide-angle side lobes
allow high back-to-back isolation between antennas. Even so, strictly speaking, the horn-
parabola antenna is not an extremely low side-lobe type, because, according to Schrank
(1985), the near-in side lobes are not very low (about −20 dB).
An important aspect of the low side-lobe problem is measurement. Great care must
be taken to ensure that the side-lobe level measured on an antenna range is the true far-
field side-lobe level. As discussed in chapter 3, sec. 3.2.5, the conventional criterion for
an acceptable far-field measurement distance is 2D2/l, where D is the largest aperture
dimension and l is the wavelength. However, it has been pointed out by Hacker and
Schrank (1982) and Hansen (1984) that this criterion is inadequate for some types of
measurement, in particular for the measurement of very low side-lobe levels. The paper
by Hansen (1984) shows that for a Taylor-type pattern having a 50 dB first side-lobe
level, accurate measurement of this level to within 0.5 dB requires separation of the
measured and measuring antennas by at least 12D2/l. (However, according to Hacker
and Schrank (1982), measurement at lesser distance initially affects the measurement
accuracy for only the side-lobes nearest to the main-beam.)
7.11. Geometrical Theory of Diffraction, Method of Moments,

and Fractals
In the last several decades, new electromagnetic analysis techniques have evolved in an
effort to meet growing antenna design requirements. Many of these stem from mathe-
matical approximation methods made possible by readily available, powerful computer
capabilities. Use of the fast Fourier transform (FFT) was, of course, a major step toward
increasing computational speeds needed for pattern calculations. Plane-wave-spectrum
(PWS) scattering is another example of a mathematical technique made practical by
application of the FFT computer algorithm (Booker and Clemmow 1950; Wu and Rudduck
1974; Ryan 1981). Although the subject of advanced mathematical techniques is beyond
the scope of this book, this section addresses the Method of Moments (MoM), geometri-
cal theory of diffraction (GTD), and fractals. These are modern, advanced theoretical
models that, through the use of computer programs, are contributing to the development
of entirely new antenna types.
Each of these models (MOM, GTD, fractals) is connected with diffraction (radiation)
from sharp edges (curvatures small compared to a wavelength), which currently cannot
be analyzed by exact theoretical analyses (Balanis 2005, p. 721). Even so, the amount
of use of these models for providing practical antennas solutions is, seemingly, ever
increasing. This increased use has resulted because the necessary complex mathematical
equations are expressed in computer algorithms. Although solutions for some problems
involve heavy computational loads, new problem solutions are possible (with various
degrees of approximation) because of rapid increases in computer speeds and memories.
A brief summary follows.
7.11.1. Geometric Theory of Diffraction

In this book, much of the discussion on pattern calculations of aperture antennas involves
integrating the effects of amplitude and phase over the aperture (sec. 6.10), sometimes
referred to as the Kirchhoff method. Aperture calculations of this type are accurate for
calculation in the forward direction, which includes the main-beam and much of the
side-lobe region. It does not, however, properly account for “edge effects,” and for that
reason it does not correctly predict the pattern of an antenna in the lateral and backward
directions where edge effects predominate.
Another method of calculation called the geometric theory of diffraction (GTD) has
been developed that is useful when surfaces are large compared to a wavelength. With
GTD, the propagation of electromagnetic waves can be analyzed through use of ray
optics, like the reflection of rays from a large surface is calculated by use of the Fresnel
Geometrical Theory of Diffraction, Method of Moments, and Fractals 329
equations (Appendix D). Although GTD uses the ray-optics representation of electro-
magnetic propagation, it also incorporates diffraction theory and surface waves to account
for the effects of edge and surface discontinuities and surface-wave propagation. Surface
waves were briefly introduced in sec. 1.4.2.
Effects of each of the different electromagnetic phenomena used in GTD analyses are
discussed in the literature and are computed as coefficients (Hansen 1981). Then, each
of the coefficients, after being computed appropriately with algorithms, is used as a step
in the computational process for a specific antenna analysis. Depending on the geometry,
in GTD analyses there may be higher order effects like multiple reflections and diffrac-
tions. These occur in a complex environment, like an antenna on an automobile, near
ship structure, or near aircraft wings and engine nacelles. Burnside and Marhefka (1988)
published an extensive, one-hundred page treatment of GTD theory. Burnside, Rudduck,
and Marhefka (1980) published a summary of GTD computer codes developed at the
Ohio State University, and Marhefka (2000) is an example of more recently developed
GTD code. Examples of successful uses of GTD in calculating patterns of antennas
mounted on modeled aircraft and investigating the effects of the aircraft on the patterns
are addressed by Oortman and Ryan (2007).
An extensive set of antenna pattern calculations versus beam pointing direction for a
large, low-side-lobe beam-steered L-band (approximately 1.3 GHz) array, that includes
effects of aircraft interactions, is given by Allen (1992; 2004). The objective of the study
was to determine the optimum antenna location, when antenna performance and flight
dynamics are included. The required computations were very time intensive when made
with a major computer system at Lockheed Martin. Even so, the GTD method of locating
and moving an array on an aircraft was determined to be more cost-effective than a
program of measurements using a full-scale antenna and airframe or using a scale model
of antenna and aircraft.
7.11.2. Method of Moments

GTD applies to predicting antenna radiation from large aperture antennas, which are
usually very high frequency devices (VHF and above); and it can also be applied to large
planar arrays whose element excitations over the array surface can be considered equiva-
lent to a field distribution over an aperture. The method of moments (MoM), on the other
hand, is applicable primarily to wire antennas (e.g., dipoles and longer wire radiators),
but it has also been used for patch antennas (Harrington 1968). A class of user-oriented
digital computer programs called NEC (Numerical Electromagnetics Code) implements
MoM. For wire antennas see Djordjevic, Bazdar, Vitosevic, Sarkar, and Harrington
(1990), and for surface patches see Rao, Wilton, and Glisson (1982). MoM programs
compute currents, from which radiation patterns can be calculated. Feed-point imped-
ance, radiation efficiency, and gain can also be calculated. Most of the early work was
done on large-scale and/or microcomputers, but there are now programs available that
can be used with personal computers (Campbell 1984; Djordjevic et al. 1990; Miller and
Burke 1983).
7.11.3. Fractals and Fractal Antennas
A fractal is a design, shape, or figure that is generated by a methodology that repeatedly

uses an identical shape but of same and different sizes. A fractal antenna is constructed
to approximate fractal geometry. The mathematics of fractals is well suited for computer
computations because of their repetitive nature. Fractals can be used to generate pro-
foundly intricate and complex shapes, and they can be deterministic or random. Random
fractals contain a degree of randomness that permits the simulation of natural phenomena,
like trees or mountains, in great detail. A comprehensive review of fractal antennas with
an extensive list of references is included in Werner, Petko, and Spence (2007).
The fineness of detail of an object described with fractal geometry can be controlled
mathematically. For example, when a picture of a bush is generated, one only needs to
have a resolution small enough to be discernible to the eye. Similarly, in the case of
constructing an antenna, intricacies that are much, much smaller than the smallest appli-
cable wavelength need not be included. The procedure of dropping unnecessary details
of a fractral-generated configuration permits desired fractal antenna structures to be
manufacturable. Fractal antennas may be constructed through use of wire elements, and
these of course have multiple edges. As a consequence, user-oriented MoM computer
programs are used to analyze the edge diffraction effects of fractal wire antennas.
The use of a wire in short, more or less zig-zag, sections provides a greater overall
wire length in a smaller space than if the wire were straight. This fact, when generalized,
is known as the “space-filling” feature of fractal geometry. Figure 7–30 compares an
ordinary circular loop antenna with a fractal “Koch” loop antenna. Both loops have the
same outer radius, but it is obvious that the total wire length of the Koch loop is consid-
erably greater. In fact, the perimeter lengths of the circular and the Koch loops are 0.26l
and 0.68l and their calculated input impedances are 1.17 ohm and 26.7 ohms, respec-
tively. Thus, although each loop has the
Circle Loop Koch Loop same outer radius, it is much easier to
couple power to the fractal loop. It is
important to recognize that the impedance
of the fractal antenna is increased because
its effective length in increased, and con-
sequently fractals provide a means of
Equal Radius miniaturizing.
Fractal dipoles and patches are also
possible. Gianvitttorio and Rahmat-Samii
(2002) compare the performance of a
conventional dipole with the fractal dipole
models of Fig. 7–31. In each case, the
FIGURE 7–30. overall dipole lengths are equal. The wire
elements are contained within a plane for
Two wire loop antennas having the same outer
radius: a Koch fractal loop and a conventional the Koch and quasi-fractal tree dipole, but
circular loop (from Gianvitttorio and Rahmat- not so for the three-dimensional quasi-
Samii, 2002; © 2002, IEEE). fractal tree dipole. These fractal “trees”
Geometrical Theory of Diffraction, Method of Moments, and Fractals 331
are similar to a real tree in that the top of Iteration 0 1 2 3 4 5

every branch splits into more branches.
As with the Koch loop, the fractal
dipoles possess features of miniaturiza- Koch Dipole
tion. In other words, for increases in the
iterations (beginning with straight wires)
shown in Fig. 7–31, the resonant frequen-
cies are reduced. However, beyond the Quasi-Fractal
second iteration, the benefits of frequency Tree Dipole
reduction appear small, especially for the
Koch and fractal tree dipoles. The reso-
nant frequency for the conventional dipole 3D Quasi-
(iteration 0) is given as 1.90 GHz, and it Fractal Tree
is about 1.15 GHz for the second iteration Dipole
of the three-dimensional fractal tree dipole
(see Fig. 14 of Gianvitttorio and Rahmat- FIGURE 7–31.
Samii, 2002). Three types of fractal dipole antennas with
Patch antennas are described in sec. different iterations: Koch, fractal tree, and
4.9. A fractal patch can be made by three-dimensional fractal tree. In each case,
increasing the electrical length along the iteration 0 is the ordinary linear dipole (from
sides of a patch, by using fractal-shaped Gianvitttorio and Rahmat-Samii, 2002; © 2002,
IEEE).
(serrated) sides, instead of straight ones.
The characteristics of a 30 mm by
19.46 mm rectangular patch and a 30 mm
by 12 mm “torn-square” fractal patch,
shown in Fig. 7–32, are compared by
Gianvitttorio and Rahmat-Samii (2002).
Reportedly, fractal patches offer the
space-filling properties of fractal loops
and dipoles. Namely, when compared
with a conventional patch of the same FIGURE 7–32.
resonant frequency, the dimensions are A fractal patch antenna, where the edges along
smaller and the radiation patterns are the resonant dimension are a “torn-square”
approximately the same. Both of the fractal shape. The dot is a feed point (from
patches described above were resonant at Gianvitttorio and Rahmat-Samii, 2002; © 2002,
5.2 GHz, but the bandwidth of the rectan- IEEE).
gular patch was 1.8 percent and that of the
fractal patch was only 0.4 percent. For some applications, reduced size can be a worth-
while trade for reduced bandwidth.
Gianvitttorio and Rahmat-Samii (2002) also discuss the effects of reduced element
size on arrays. They compared array pattern measurements with calculations using MoM,
and reported similar results. Even though the size reduction of the elements is small in
terms of miniaturizing, patch-size reduction can significantly improve array performance
in two ways. First, with fixed center-to-center spacing of the elements, the reduced
element size increases the distance between element edges, and this reduces the mutual
coupling between elements. Second, reduced element size permits smaller center-to-
center spacing and thus more elements for a given array width, thereby permitting a wider
off-axis scan angle without creating grating lobes. Either type of use of reduced element
size can be effective in improving antenna patterns for off-normal beam directions.
Finally, the electromagnetics phenomena common to GTD, MoM, and fractal antennas
is radiation from sharp edges. Exact mathematical electromagnetic solutions cannot be
obtained for antenna configurations containing sharp edges, but it has been demonstrated
that accurate approximations are possible by using GTD and MoM. GTD or MoM is
used when dimensions are large or small, respectively, compared to a wavelength. The
mathematics of fractals are relatively easy to computer-generate, but fractal shapes have
sharp edges. However, experience shows that the electromagnetic characteristics of
fractal wire and patch antennas can be calculated with MoM techniques. Thus, it is appar-
ent that the antenna technology that has evolved from GTD, MoM, and fractals could
not have been developed without digital computers, because reduction to practice would
have been a mathematically impracticable procedure.
References
Allen, W. P., “Aircraft Interference Effects on AEW Antenna Patterns,” pp. 326–44 of
M. W. Long, Airborne Early Warning System Concepts, Artech House, 1992 and
SciTech, 2004.
Balanis, C. A., Antenna Theory, 3rd ed., John Wiley & Sons, 2005.
Booker, H. G., “Slot Aerials and Their Relation to Complementary Wire Aerials,” Journal
of the Institution of Electrical Engineers (London), 93, pt. IIIA, no. 4, 1946.
Booker, H. G., and P. C. Clemmow, “The Concept of an Angular Spectrum of Plane
Waves, and Its Relation to That of a Polar Diagram and Amplitude Distribution,”
Proeedings of the Institution of Electrical Engineers, vol. 97, pt. III, January 1950,
pp. 11–17.
Burdine, B. H., and E. J. Wilkinson, “A Low-Sidelobe Earth-Station Antenna for the
4/6 GHz B and,” Microwave Journal, vol. 23, no. II, November 1980, pp. 53–68.
Burnside, W. D., and R. J. Marhefka, “Antennas on Aircraft, Ships, or Any Large Complex
Environment,” Ch. 20 in Antenna Handbook—Theory, Applications, and Design, Lo,
Y. T. and S. W. Lee, eds., Van Nostrand, Reinhold Co., 1988.
Burnside, W. D., R. C. Rudduck, and R. J. Marhefka, “Summary of GTD Codes
Developed at the Ohio State University,” IEEE Transactions on Electromagnetic
Compatibility,” EMC-22, November 1980, pp. 238–243.
Campbell, D. V., “Personal Computer Applications of MININEC”, IEEE Antennas and
Propagation Society Newsletter, February 1984.
Cutrona, L. J., “Synthetic Aperture Radar,” ch. 21 in Radar Handbook, M. I. Skolnik (ed.),
McGraw-Hill, New York, 1990.
Deschamps, G. A., and R. H. DuHamel, ch. 18 in Antenna Engineering Handbook, H.
Jasik (ed.), McGraw-Hill, New York, 1961.
References 333
Djordjevic, A. R., M. B. Bazdar, G. M. Vitosevic, T. K. Sarkar, and R. F. Harrington,

Analysis of Wire Antennas and Scatterers: Software and User’s Manual, Artech House,
1990.
DuHamel, R. H., and D. E. Isbell, “Broadband Logarithmically Periodic Antenna
Structures,” IRE National Convention Record, 1957, Part I, pp. 119–28.
Dyson, J. D., “The Unidirectional Equiangular Spiral Antenna,” IRE Transactions on
Antennas and Propagation, AP-7, October 1959, pp. 329–34.
Dyson, J. D., “The Characteristics and Design of the Conical Log-Spiral Antenna,” IEEE
Transactions on Antennas and Propagation, AP-13, July 1965, pp. 488–99.
Elachi, C., Spaceborne Radar Remote Sensing: Applications and Techniques, IEEE Press,
1987, pp. 11–50.
Filipovic, D. S., and T. Cencich, “Frequency Independent Antennas,” ch. 13 in Antenna
Gianvitttorio, J. P., and Y. Rahmat-Samii, “Fractal Antennas: A Novel Antenna
Miniturization Technique, and Applications,” IEEE Antennas and Propagation
Magazine, Feb. 2002, pp. 20–35.
Giger, A. J., and R. H. Turrin, “The Triply-Folded Horn Reflector: A Compact Ground
Station Antenna for Satellite Communication,” Bell System Technical Journal, Sept.
1965, pp. 1229–53.
Hacker, P. S., and H. E. Schrank, “Range Distance Requirements for Measuring Low and
Ultralow Sidelobe Antenna Patterns,” IEEE Transactions on Antennas and Propagation,
AP-30, September 1982, pp. 956–66.
Hansen, R. C., Geometric Theory of Diffraction, IEEE Press, Institute of Electrical and
Electronics Engineers, 1981.
Hansen, R. C., “Measured Distance Effects on Low Sidelobe Patterns,” IEEE Transactions
on Antennas and Propagation, AP-32, no. 6, 1984, pp. 591–94.
Harrington, R. F., Field Computation by Moment Methods, The Macmillan Company,
1968.
Hines, J. N., T. Li, and R. H. Turrin, “The Electrical Characteristics of the Conical-Horn
Reflector Antenna,” Bell System Technical Journal., July 1963, pp. 1187–1211.
Hollis, J. S., and M. W. Long, “Luneberg Lens Scanning System,” IRE Transactions on
Antennas and Propagation, vol. AP-5, January 1957, pp. 21–25.
Holliman, A. L., and J. S. Hollis, “An X-Band Wide-Angle Scanner,” Record of the
Georgia Tech-SCEL Symposium on Scanning Antennas, Georgia Institute of Technology,
December 18–19, 1956, pp. 75–85.
Terms, 1993.
Isbell, D. E., “Log Periodic Dipole Arrays,” IRE Transactions on Antennas and
Propagation, AP-8, May 1960, 260–67.
Johnson, R. C., “Dual-Mode Coupler,” IEEE Transactions on Microwave Theory and
Techniques, November 1967, pp. 651–52.
Johnson, R. C., A. L. Holliman, and J. S. Hollis, “A Waveguide Switch Employing the
Offset Ring-Switch Junction,” IRE Transactions on Microwave Theory and Techniques,
September, 1960, pp. 532–37.
Jordan, E. C., G. A. Deschamps, J. D. Dyson, and P. E. Mayes, “Developments in

Broadband Antennas,” IEEE Spectrum, 1, no. 4, April 1964, pp. 58–71.
Kelleher, K. S., “Electromechanical Scanning Antennas,” ch. 18, pp. 18-2 to 18-23, in
Johnson, R. C., Antenna Engineering Handbook, 3nd ed., McGraw-Hill, 1993.
Kennedy, P. A., and R. W. P. King, Cruft Laboratory, Harvard University, Technical Report
155, April 1, 1953, “Experimental and Theoretical Impedances and Admittances of
Center-Driven Antennas” (work done under an Office of Naval Research contract).
Also reported in The Theory of Linear Antennas, by R. W. P. King, Harvard University
Press, Cambridge, Mass., 1956.
LeVine, D. J., and W. Sichak, “Dual-Mode Horn Feed for Microwave Duplexing,”
Electronics, September 1954, pp. 162–64.
Long, M. W., “A High-Speed K-Band Switch,” Proceedings of the IRE, December 1951,
pp. 1566–67.
Long, M. W., W. K. Rivers, Jr., and J. C. Butterworth, “Combat Surveillance Radar
AN/MPS-29 (XE-1),” Record of the Sixth Annual Radar Symposium, University of
Michigan, 1960. Reprinted in S. L. Johnston, Millimeter Wave Radar, Artech House,
pp. 461–78, 1980.
Love, A. W., “Horn Antennas,” ch. 15 in R. C. Johnson and H. Jasik, Antenna Engineering
Handbook, McGraw-Hill, 1984.
Mayes, P. E., “Frequency-Independent Antennas,” IEEE Antennas and Propagation Society
Newsletter, August 1982.
Marhefka, R. J., “Numerical Electromagnetic Code—Basic Scattering Code , NEC–BSC
(version 4.2)”, ElectroScience Laboratory Technical Report, The Ohio State University,
October 2000.
Miller, E. K., and G. J. Burke, “Personal Computer Applications in Electromagnetics,”
IEEE Antennas and Propagation Society Newsletter, August 1983.
Oortman, G. J., and C. E. Ryan, Jr., “Aircraft Antennas,” ch. 40, pp. 40-25 to 40-28, in
Rao, S. M., D. R. Wilton, and A. W. Glisson, “Electromagnetic Scattering by a Surface of
Arbitrary Shape,” IEEE Transactions on Antennas and Propagation, May 1982, pp.
1396–1423.
Rumsey, V. H., “Frequency Independent Antennas,” IRE National Convention Record,
1957, Part I, pp. 114–18.
Rumsey, V. H., Frequency Independent Antennas, Academic Press, 1966.
Ryan, C. E., “PWS Scattering Computation,” IEEE Antennas and Propagation Society
Newsletter, June 1981.
Schelkunoff, S. A., Electromagnetic Waves, Van Nostrand, Princeton, NJ, 1943. Also,
Schelkunoff and Friis, Antennas: Theory and Practice, Wiley, New York, 1952; and
Schelkunoff, Advanced Antenna Theory, Wiley, New York, 1952.
Schrank, H. E., “Low Sidelobe Phased Array Antennas,” IEEE Antennas and Propagation
Society Newsletter, April 1983.
Schrank, H. E., “Low Sidelobe Reflector Antennas,” IEEE Antennas and Propagation
Society Newsletter, April 1985.
Terman, F. E., Radio Engineers’ Handbook, McGraw-Hill, 1943.
Thomas, B. M., “Design of Corrugated Conical Horns,” IEEE Transactions on Antennas

and Propagation, vol. AP-26, March 1978, pp. 367–72.
Werner, D. H., J. S. Petko, and T. G. Spence, “Fractal Antennas,” ch. 33, pp. 33-1 through
33-28 in Antenna Engineering Handbook, 4th ed., J. L. Volakis (ed.), McGraw-Hill,
2007.
Wu, D. C. F., and R. C. Rudduck, “Plane Wave Spectrum-Surface Integration Technique
for Radome Analysis,” IEEE Transactions on Antennas and Propagation, May 1974,
pp. 497–500.

1. (a) A center-fed cylindrical dipole has a length-to-diameter ratio (L /D) of approxi-
mately 122. It is 12 wavelength long (L = l /2) at a frequency f0 = 500 MHz. At
approximately what two frequencies between 500 and 1,000 MHz will the reac-
tive component of its input impedance be approximately +100 ohms (neglecting
effects due to the feed-point connection)? (b) Another dipole of the same length
has an L /D ratio of approximately 17. At about what two frequencies in the same
range will the resistive component of its input impedance be 100 ohms? (Use Figs.
7–2 and 7–3.)
2. (a) A cylindrical dipole, at a particular frequency, has an input impedance
Z = 75 + j30 ohms. What are the values of input resistance and reactance of a
vertical monopole whose height (H of Fig. 7–5) above the ground plane is
exactly half the length of the dipole, and whose diameter is the same as the
dipole? Assume that the monopole ground plane is of infinite extent and perfectly
conducting. (b) If the dipole has a free-space directivity (D) of 1.5 relative to an
isotropic radiator, what is the directivity of the monopole?
3. A log-periodic antenna has a design ratio t = 0.8. At a frequency of 60 MHz, its
input impedance is purely resistive and 82 ohms in value. The upper and lower
cutoff frequencies are 300 MHz and 30 MHz. Assuming an ideal log-periodic
behavior between the cutoff frequencies, at what frequencies (in addition to
60 MHz) will the input impedance be exactly 82 ohms and purely resistive?
(Disregard fractions of a MHz.)
4. In each part, check the one of the four choices that you think is correct:
(a) An equiangular log-periodic antenna is desirable because:
(i) It is simpler to construct ⵧ
(ii) It has noncritical length dimensions ⵧ
(iii) It has very small impedance variation over its range of operation ⵧ
(iv) Its phase center does not vary with frequency ⵧ
(b) A self-complementary antenna is desirable because:
(i) Its phase center does not vary with frequency ⵧ
(ii) Its impedance does not vary with frequency ⵧ
(iii) It is light in weight ⵧ

(iv) The pattern does not have sidelobes ⵧ
(c) Some types of log-periodic antennas used as feeds for paraboloidal reflectors
are deficient for broad-band operation because:
(i) The beamwidth cannot be made narrow enough for proper illumination
taper ⵧ
(ii) The phase center varies with frequency ⵧ
(iii) The beam direction varies with frequency ⵧ
(iv) The beam cannot be made wide enough for proper illumination ⵧ
5. Name three basic types of antennas that can provide circularly polarized
radiation.
6. Name one type of antenna that will provide each of the following types of radia-
tion (or reception) characteristics: (a) Omnidirectionality (a theoretically perfect
circular pattern) in the horizontal plane, with vertical polarization. (b) Same
pattern as in (a) but with horizontal polarization. (c) An approximately uniform
pattern in the horizontal plane with horizontal polarization in that plane, and that
uses multiple dipoles.
7. State two conditions under which an electrically small antenna may be used for
reception with results virtually as good as would be obtained with a much larger
antenna.
8. With various types of direction-finding antennas, the direction of a received
signal is determined by finding the null direction rather than the direction of
maximum signal. In a single brief sentence, state why the null is used rather than
the maximum.
9. An antenna has a one-degree beamwidth in the vertical plane and a three-degree
beamwidth horizontally. The beam is scanned in both elevation and azimuth.
The elevation scan is a sawtooth motion, and each sawtooth scan takes 0.1
seconds. The azimuth scan is a much slower motion with constant angular speed.
What must this azimuth angular speed be in order for the azimuth scan to prog-
ress one azimuth beamwidth during the time of one vertical scan? Express the
result in terms of revolutions per minute.
10. Name two antenna design factors to which special attention must be paid to
achieve low-noise operation in the frequency range of 1 to 10 GHz. (One of
these factors relates to external noise, and one to internal noise.)
CHAPTER 8
Electronically Steered
Arrays
This chapter discusses electronically steered arrays (ESAs), also commonly called phased
arrays. These antennas have beams that can be electronically steered in pointing direction
or beam shape, without mechanical motion, and accomplished in microseconds. Although
the operation of all arrays depends on the proper phasing of the individual array elements,
the terms ESA and phased array have by usage come to mean an array in which the
beam is steered in direction and shape by varying the phasing of the elements. Present-
day ESA (or phased array) technology includes adaptive beam forming (ABF), which
merges rapid beam pointing and shaping with digital signal processing separately. With
ABF arrays, the phasing and time delaying of the signals from individual elements are
jointly adaptively adjusted and then summed to rapidly beam point, beam shape, and
Doppler filter desired received signals. In this way, the received signal output, from the
summed individual array signals, may be optimized, simultaneously, for both the strength
of the desired signal relative to thermal noise and relative to interference from undesired
signals.
The advancements in ESA (i.e., phase array) technology has resulted from intensive
and continuing research and development in the following areas:
(i) Improvement of the devices and circuitry used for producing the required
phase shifts and power division among the array elements;
(ii) Use of microprocessors or special-purpose computers to determine the
optimum phasing and power division to carry out a particular beam-steering
command;
(iii) Better understanding of factors affecting side-lobe levels, mutual coupling
effects on array-element patterns (and methods of minimizing the deleteri-
ous effects of such coupling), and effects of phase errors on side lobes and
main-beam gain;
(iv) Development of adaptive receiving arrays that are capable of automatically
adjusting their overall patterns to optimize signal-to-interference ratio by
placing pattern nulls in the directions of interfering signals;
(v) Development of solid-state amplifiers suitable for use at transmit-receive
array elements, making the use of amplifiers distributed among the array
elements more economically feasible and the arrays more efficient; and
337
338 Electronically Steered Arrays
(vi) Development of fabrication techniques such as the microstrip technology,

which further contributes to the economic feasibility of large arrays, espe-
cially when combined with solid-state technology. It also makes possible
conformal arrays—i.e., arrays that can conform to the surface contour of a
vehicle or to other nonplanar shape requirements.
8.1. Phased Array Principles

With all the elements of a linear array or a planar array in phase, the beam is directed
broadside, that is, perpendicular to the line or plane of the elements. On the other hand,
if the phases of the elements of a linear array progressively change by an amount per
element that is equal, in radians, to 2pd/l, where d is the element spacing and l is the
wavelength, the beam direction will be parallel to the line joining the elements (endfire).
These facts suggest that a progressive phase change per element along the array inter-
mediate between the broadside and endfire values might result in a beam directed at some
intermediate angle, as is the case.
An electronically steered array (ESA), or simply a phased array, is an antenna that
uses variable phase, frequency, or time delay at the array elements to scan the beam to
a designated angle in space.
Advantages of electronic steering include
rapid beam pointing
inertialess beam movement
fast beam shaping by aperture control
Disadvantages
high cost
high complexity
long, expensive design cycle
Because of its relative simplicity, frequency-scan arrays were the first widely used
electronically agile arrays for radar. However, because of the potential advantages of
electronic steering with phase shifters, momentum increased during the 1950s for the
development of practical, rapid phase changers that operate over a wide range of micro-
wave frequencies. Motivation for rapid, inertialess scan, narrow beam arrays continued
for years, but various technological limitations and costs continued as prohibiting factors.
Even today, although costs are greatly reduced, complexity and cost are still limitations
for the production of large (hundreds or more elements) phased arrays. Of course, phased
arrays for wireless, which generally require fewer array elements, are now widely used.
A major new thrust for array development began with the invention of adaptive digital
signal processing concepts for “smart” antennas, which require array elements to be
rapidly and independently controlled from instructions generated by digital computa-
tions. Today, phased array research, development, and production are presently major
technological activities.
Phased Array Principles 339
8.1.1. Time Delay and Phase Shift Requirements

Figure 8–1 illustrates a linear array with the main-beam electronically steered to a scan
angle qL, measured from broadside. The beam can be steered from one angle to another
by changing the relative time delay between elements. Note that, for a focused beam,
the wave phases from the elements on arriving perpendicular to the wavefront are equal.
This is true if, for each element, the overall time delay from element input to wavefront
is equal, and this can be done by adding a time delayer at the input of each element.
Then focusing is accomplished because, for each element, the delay time of the time
delayer plus the propagation time from element is equal. To illustrate, the path length
difference between the bottom and top elements to the wavefront is L sin qL. Thus, the
propagation time for the radiation from the bottom element to wavefront exceeds that
from the top element by L sin qL/c, where c is 3 × 108 m/s. Therefore, to retain focus at
all frequencies and at scan angle qL, a time delay increase of L sin qL /c is needed for the
Lc
os
θL
W
av
ef
dn
ro
nt
(lin
e
of
n
eq
δ
θL
ua
lp
ha
se
L
θL
Distance to wavefront: δn = dnsinθL
Element phase (radians): αn = (2π/λ) δn
inθ
L
Ls
FIGURE 8–1.
Effective aperture (line of equal phase, LcosqL) versus scan angle qL.
top element relative to the bottom element. Appropriate, decreasingly less time
delay is required for the lower elements for scan angle of qL. Clearly, for the beam to be
scanned in the −qL direction, the larger time delays will be needed for the lower
elements.
Consider now the following example, with qL = 30° and L/l = 20. Then focusing will
occur if a time delay is added to the top element that corresponds to the time it takes for
a wave in free space to traverse a distance of 10l. This delay can in principle be obtained
by adding a transmission line. However, for each scan angle a different length of trans-
mission line would be required—and the lengths would be different for each array
element. Now imagine an array with several hundred (or perhaps thousands of) elements;
and the capability to change to a different length of transmission line at each element
and each scan angle. Although time delayers permit an array to be focused over a wide
bandwidth, time delayers are bulky, heavy, and expensive. Therefore, time delayers are
used only when very broad-band performance is required. Instead, phase shifters (often
called phasers) are commonly used to provide agile beam-pointing capabilities.
As already noted, for a focused array the radiation from each element at a distant
observation point is in-phase. Now suppose there is a phase shifter (phaser) located at
each array element. Then, for narrow bandwidth operation, the required range of phase
shift is between zero and 2p, that is, between zero and 360°. However, each phaser must
change for each change in either scan angle or operating frequency. For focusing with
an element spacing d, there must a progressive decrease, from top to bottom in Fig. 8–1,
in phase delay a between elements as follows:
2π d 
α =  sin θ L
 λ 
(8–1)
Now assume a of (8–1) equals Df + 2pI, with I being any integer. Then, the phase
delay needed by a phaser is only Df. This mathematical unwinding of phase is called
modulo 2p. Thus, Df is expressed as mod (2p) in radians or as mod (360°) in degrees.
Recall that for perfect phasing of an array, the required phase shift for each element is
different, and all of the phases must change with a change in either scan angle qL or fre-
quency. Even so, to retain perfect array phasing, the needed phase delay of the phasers
never exceeds 360°.
8.1.2. Effective Aperture Versus Scan Angle

For steering angles qL other than zero, the effective aperture (the line of equal phase in
Fig. 8–1) is foreshortened in accordance with L cosqL. Then, for a planar array of width
Lx and height Ly that scans in azimuth AZ and elevation EL, the beamwidths will be
increased by the factors 1/cos qAZ and 1/cos qEL, in accordance with beamwidth being
inversely proportional to aperture width. Perhaps more detrimental, the gain will be
reduced by the product cos qAZ ⋅ cos qEL. The practical effect of gain loss due to aperture
foreshortening generally limits the use of off-normal scan angles to 60° and less.
Radiating Elements
Input Termination
S
Serpentine Feed
FIGURE 8–2.
Frequency scan linear array.
8.1.3. Frequency-Scan Antennas

In earlier times, ESA arrays that functioned by frequency scan were widely employed
for radar. This was because, for radar applications where transmit frequency can be con-
trolled, frequency-scan is a low-cost method for inertialess antenna beam scan (Ajioka
1993). Frequency scan antennas offer simplicity, low cost, and high reliability. Conse-
quently, frequency scan has been widely used for inertialess beam scanning. There are,
however, limitations because frequency beam-scan prevents other uses of frequency
change, such as for pulse compression or pulse-to-pulse frequency agility. Another
disadvantage is caused by the wide frequency spectrum needed, and the likelihood of
interference with other electronic systems. In addition, the transmitting and receiving
equipment required to accommodate changing frequency adds complexity and
expense.
With frequency scan, the desired change in array phasing is caused by changing fre-
quency. This changes the wavelength, so that if the phase change per element depends
on the phase delay in a transmission line or waveguide, the phase will vary as the
frequency varies. For a linear array, a basic series feed system is shown in Fig. 8–2,
where the length of line s is greater than the element-to-element spacing d. This effect
is accomplished by “folding” or “snaking” the waveguide or line between elements to
increase its length. The resulting feed is called a serpentine or snake feed. This type of
feed results in an increased phase change between adjacent elements for a given
frequency change.
Following (8–1), the angle of the beam with respect to the broadside direction can be
expressed in terms of the progressive phase change per element a and the frequency.
Then, for the beam focused in direction qL and spacing d between elements, the required
phase difference in radians between elements is
2π d 
α =  sin θ L + 2π I
 λ 
(8–2)
where l is free space wavelength and I is an integer. Letting ls denote the wavelength
within the transmission line of length s between elements, the interelement phase delay
is 2ps/ls. This phase delay must equal a of (8–2) if the array is focused. Thus,
 2π d  2π s
  sin θ L + 2π I = (8–3)
λ λs
Usually the connecting transmission line is air-filled waveguide operated in the TE01
mode. Its waveguide wavelength is given by (2–69) of chapter 2, and is repeated here
as follows:
c
λg = (8–4)
f 2 − fc2
In (8–4), c is the free space velocity of propagation and fc is the waveguide cutoff fre-
quency. If waveguide is used for the transmission line, ls = lg. Then from (8–3), if qL
is zero
s
I= (8–5)
λg0
Here lg0 is the wavelength within each connecting line of length s at the frequency f0
for which the beam is pointed in the broadside direction, that is, qL = 0. In other words,
I is the integer number of wavelengths along the transmission line at the wavelength
lg0.
Using (8–3), (8–4), and (8–5), it can be shown that
Iλ   f 2 − fc2   cI   f 2 − fc2  1 2 
12
sin θ L = 1 − = 1 −    (8–6)
d   f02 − fc2   fd   f02 − fc2  

In (8–6), only values of f within the interval for which |sin qL| ≤ 1 are permitted.
8.1.4. Array Bandwidth

Usually, modern phased arrays use phase shifters for rapid angle scanning and hopping,
but time delayers (sec. 8.2.4) are needed when wide-band performance is required. This
section discusses the principles that control the bandwidth of arrays that employ phase
shifters and time delayers.
Two categories of bandwidth are discussed in chapter 3. Category 1 in which an
antenna is required to handle frequencies over a wide band simultaneously; and category
2 in which the range of frequencies is covered over a period of time, but at any one time
the bandwidth needed is only a small fraction of the long-term requirement. The simul-
taneous case is called instantaneous bandwidth. It is the one needed to pass the signal’s
spectrum, and therefore the shape of a short pulse. Category 2, called tunable bandwidth,
is one where there is enough time available for setting antenna parameters prior to a fre-
quency change. Instantaneous bandwidth (category 1) presents the more difficult problem,
because the antenna must function properly, without adjustments, at any frequency within
the band.
As already discussed, arrays having broad instantaneous bandwidth must use time
delayers. Time delayers are heavy and expensive. For these reasons, most scanning arrays
use phase shifters for scanning, and consequently they have only limited instantaneous
bandwidths. Array elements and phase shifters (phasers) can limit overall bandwidth.
However, array elements usually can perform over bandwidths of 10% without major
changes in impedance or beamwidth. Recall that elements of a focused array require
equal time delay to a point in space, and phase delay for a given time delay is directly
proportional to frequency. Therefore, bandwidth is limited by phase shifters, because
their phase delay is nearly independent of frequency.
The beam direction of an array that is steered by phase shifters will automatically shift
as the frequency is changed. This beam shifting limits the array bandwidth. Most modern
arrays can, however, be “tuned” (by resetting the phase shifters) over a bandwidth of five
to perhaps ten percent. Frank (1972) gives an estimated first guess at bandwidth, without
tuning, as
bandwidth (%) = 3dB beamwidth (degrees) (8–7)
In other words, a phased array will have roughly a one percent instantaneous bandwidth
if it has a broadside −3 dB beamwidth of one degree. The actual bandwidth will depend
on the feed architecture, transmitted waveform, and scan angle. According to Frank, (8–7)
above will provide a bandwidth estimate that is rarely off by a factor of more than
two.
Determining a more accurate estimate requires a thorough examination of how the
array will be used. For example, the bandwidth decreases with increases in scan angle.
In addition, signals that are modulated, for example, pulsed signals, have a widened
spectrum, and this must be allowed in estimating needed bandwidth. In other words,
because of the different frequencies within a modulated signal, an array beam is spread
in angle. This spreading reduces the effective gain.
For a scanned beam to remain perfectly focused when frequency is changed, the
element phases must be changed. Without resetting, phase shifters are essentially constant
phase devices, with changes of only a few degrees over their operating bandwidths.
For (8–8), (8–9), and (8–10) that follow, it is assumed that the transmission lines to the
array elements are identical and phases of the phase shifters are independent of
frequency.
From (8–1), the beam is steered to a direction qL with a phase of
2π x 2π x
φ= sin θ L = f1 sin θ L (8–8)
λ1 c
at an element located a distance x from the array center. With the same phase setting, a
frequency f2 steers the beam to a new position qL + DqL. Then,
2π x 2π x
φ= f1 sin θ L = f2 sin (θ L + Δθ L ) (8–9)
c c
It can be shown that, for small changes in Df, where Df = f2 − f1, the change in direc-
tion of the beam DqL is small, and then (8–9) can be approximated as
Δf
Δθ L = tan θ L (8–10)
f
The relationship (8–10) indicates that the amount of beam shift DqL increases in
proportion to the product of fractional change in frequency and tangent of scan angle.
However, (8–10) is of limited usefulness, because it is valid only for small changes in
frequency.
To make practical estimates of beam shift, one must resort to approximations using
(8–9). Furthermore, the loss in signal strength for a given beam squint will necessarily
depend on beamwidth, and hence on aperture size. Frank gives graphs of loss in gain as
a function of beamwidth, scan angle, and spectral width. Equation (8–7), a commonly
used estimate for bandwidth, gives a gain loss of approximately 0.8 dB (due to the beam
shift) for an unmodulated signal with the beam scanned 60° from broadside. For a signal
with a broad spectrum, such as a short pulse, effective bandwidth may be less than indi-
cated by (8–7). This is because an array must be focused at all frequencies contained
within the spectral bandwidth of the signal being either transmitted or received.
Interestingly, antenna performance with short pulses can also be analyzed in terms of
the time required for a received pulsed wave to reach all of the array elements (Frank,
1972). This time is called the array fill time, which is (L sin qL)/c. Thus, fill time is the
time it takes for a wave traveling at speed c to traverse the distance L sin qL. The role of
a time delayer is, of course, to add delay that equals the fill time that will, in principle,
result in focusing that is independent of frequency.
8.1.5. Phased Array Patterns

Assume a planar array is located in the x-y plane (see, e.g., Fig. 5–14). Then, in accor-
dance with sec. 5.5.1, far-zone patterns can be calculated by summing the excitation of
the elements over their x and y values as follows:
g (θ , φ ) = ∑ ∑ a ( x, y ) exp [ jΨ ( x, y )] (8–11)
where
a(x, y) = the amplitude excitation of each element
Y(x, y) = a − (2p/l)d
with
a = phase excitation of each element and
d = pathlength difference of each element (radiator) to the observation point relative
to pathlength from the array center, with the observation point in the far zone (see
Appendix H),
d = −sin q (x cos f + y sin f)
Usually, the elements of rectangular arrays are energized so that their amplitudes and
phases along the x-axis aperture coordinates are independent of the y-axis coordinates,
and conversely. In that case, the element amplitudes and phases can be expressed as
products of terms that depend only on one coordinate, either x or y, as follow:
a ( x, y ) = a ( x ) a ( y )
α ( x, y ) = α ( x ) α ( y )
Then, the integrand of (8–11) is separable as two products, one that is a function of the
x coordinates and the other a function of the y coordinates. In addition, electronically
steered arrays are usually designed so the separations between elements are evenly
spaced, but not necessarily the same along the x and y axes. Then, the x and y coordinates
can be expressed as xm = mdx and yn = ndy, where dx and dy are the element spacings in
the x and y directions. Similarly, a(x), a(y), a(x), and a(y) can be expressed as am, an,
am, and an, respectively. Then (8–11) can be written as the following product of two
summations
g (θ , φ ) = ∑ am exp [ j (α m + (2π λ ) md x sin θ cos φ )]

(8–12)
⋅ ∑ an exp [ j (α n + (2π λ ) nd y sin θ cos φ )]
where far-zone amplitude patterns are calculated by summing the excitation of the ele-
ments over all m and n (x and y) values. The reader will recognize that each of the two
product terms of (8–12) is an array factor for a linear array, and thus each can be con-
sidered separately as if describing the radiation from a linear array.
Effects of beam steering are now considered, recognizing that scanning may be accom-
plished separately in each plane. An array beam is focused if the phases of the fields
from its elements are equal at a great distance. In other words, a focused antenna provides
a field of equal phase along a line normal to the beam-pointing direction, as in Fig. 8–1.
Consider now the pattern in the x-z plane for a line source along the axis. Then, since
there are no elements along the y axis an and an are zero. In addition, since the pattern
in the x-z plane is being considered, f = 0 and cos f = 1. Therefore, from (8–12), the
pattern in the x-z plane for a line source along the x-axis versus observation angle q is
M −1
2π
g (θ ) = ∑ am exp  j  α m + λ
md x sin θ  

(8–13)
m=0
where M is the number of elements and there are equally spaced elements located at
coordinates mdx.
The pattern amplitude |g(q)| is maximized when in (8–13) the term
 α + 2π md 
 m x sin θ  = 0 . Thus, for the beam to be steered for its peak amplitude in direc-
λ 
tion q = qL, each element phase am must be as follows:
2π
αm = − md x sin θ L (8–14)
λ
Then, when the phases are set in accordance with (8–14) to maximize the pattern in
direction qL, the pattern is as follows:
M −1
2π
g (θ ) = ∑ am exp  j λ

md x (sin θ − sin θ L )

(8–15)
m=0
The reader should recall that qL in (8–15) is the pointing direction of the major lobe and
q is the observation angle in an arbitrary direction.
8.1.6. Grating Lobes

A grating lobe is any lobe other than the major lobe that occurs when the radiation from
all elements add in-phase. Grating lobes for a nonscanning broadside array were dis-
cussed in sec. 5.5.4, where it is seen that additional major lobes will occur when the
element spacing is greater than l. Grating lobes can occur for a beam pointed off of
broadside if the element spacing is less than l. This occurs because the term g(q) of
(8–15) can have maximum values if
2π
md x (sin θ − sin θ L ) = ± 2πΙ (8–16)
λ
where I is an integer. When (8–16) is satisfied, the maxima of g(q) are grating lobes. As
a consequence of (8–16), element spacing must be less than l for an array that is scanned.
In particular, to prohibit grating lobes when the major lobe is pointed in directions other
than broadside, the maximum acceptable element spacing dmax is
λ
dmax = (8–17)
1 + sin ∆θ
where Dq is the largest angle qL to be scanned away from broadside. Parenthetically,

the derivation is not oblivious and to derive it the reader is referred to Ajioka (1993,
pp. 19–6 and 19–7).
Beam-Steering Technology 347
From (8–17), for Dq equaling 90°, dmax is l/2. Because of loss in effective aperture
due to pointing off of broadside, ESA designs typically limit the scan to a maximum of
60° off-broadside. Then, sin Dq = 0.866 and dmax = 0.54 l. Consequently, the element
spacings for ESAs are usually about l/2, or slightly less, to avoid the risk of creating
part or all of a full grating lobe.
Thus far it has been assumed that the array elements are arranged in a square grid with
separation in height and width of dmax between elements. Now consider a lattice configu-
ration of equilateral triangles, with each triangle leg having length dmax. Such a configu-
ration, when considering only the avoidance of grating lobes, will allow (depending on
maximum scan angle off-broadside) a 10 to 15 percent reduction in the overall number
of required elements (Sharp 1961). However, the triangular configuration creates a some-
what more stringent requirement on the fineness of phase quantization for controlling
side-lobe levels (Nelson 1969).
8.2. Beam-Steering Technology

8.2.1. Phase Shifters
An essential ingredient of a beam-steered phased-array is a device that can produce a
variable and controllable phase shift. Phase shifters are sometimes call phasers. (They
should not, however, be referred to as phasors, because that term denotes a mathematical
complex variable characterized by an amplitude and a phase angle.) Three basic methods
of producing phase shifts are: (i) variation of the velocity of propagation of waves in a
segment or segments of the antenna feed system; (ii) variation of the lengths of segments
of the feed path; and (iii) variation of the frequency, which results in a change in the
phase at the output of a fixed-length transmission-line. The frequency-variation method
(frequency scanning) already has been discussed (sec. 8.1.3).
Control of propagation-velocities or line lengths can be either analog (continuous) or
digital (step-wise variable). Both methods are used. The digital method can approximate
the continuous variation of phase to any desired degree of accuracy if a sufficient number
of digital bits is used. The most popular electronic phase shifters employ either ferrites
or diodes, but other technologies have matured in the last ten or so years (Romanofsky
2007). Phase shifter cost is a major factor in overall cost of electronically steered arrays
(ESAs), because it must be multiplied by the number of phasers used. For some arrays,
the number of phasers exceeds ten thousand; and an estimate (Romanofsky 2007,
p. 21–2) includes total phaser cost that approach 40 percent of the overall cost of steer-
able receive arrays.
The principal method of controlling the propagation velocity in waveguides and trans-
mission lines is through the use of ferrite material inserted into the space occupied by
the propagating waves. Ferrite is a ceramic-like insulator material that possesses mag-
netic properties, and has high resistivity which permits wave propagation with low loss.
The wave propagation velocity in ferrites can be varied by applying varying amounts of
a constant magnetic field to the material (superimposed on the propagating electromag-
netic fields). The strength of the constant field can be controlled by varying the magnitude
of a magnetizing direct current.
Some ferrite phase shifters require a continuous magnetizing current, while others are
of the “latching” type, in which a relatively short pulse establishes a magnetization level;
this level is maintained until it is reset by another dc pulse. However, when the magne-
tization level is to be changed for either the continuous-current or the latching type of
ferrite phase shifter, it is necessary first to set the magnetization momentarily at the satu-
ration level and then to apply the current required to set the desired level. This avoids
errors that would otherwise occur due to the hysteresis effect.
Ferrite phase shifters can be reciprocal or nonreciprocal. In the former case, the same
amount and direction of phase shift occurs for either direction of propagation. In the
latter case, the sign of the phase shift is reversed when the direction of propagation is
reversed. When a nonreciprocal phase shifter is used for both transmission and reception
with the same antenna (as is the case for many radar systems), the phase shifter must be
switched rapidly between the transmitting and receiving modes of operation. Despite this
requirement, nonreciprocal phase shifters are sometimes used in radar applications
because of other desirable characteristics. Switching times of a few microseconds are
achievable.
Ferrite phase shifters can be either analog or digital, and even when an intrinsically
analog shifter is used, it may be digitally controlled. These phase shifters become increas-
ingly inefficient and bulky as the frequency is decreased. They are not generally used
below about 1 GHz, although they have been used at frequencies as low as 200 MHz.
On the other hand, ferrite phase shifters are usable at frequencies of the order of 10 GHz
or higher.
The primary electronic method employed for varying the length of a transmission line
is diode switching. Diode phase shifters, unlike ferrite phase shifters, become increas-
ingly efficient as the frequency decreases, although they also can be used up to 10 GHz.
For purposes of basic understanding of its phase-shifting capability, the diode can be
regarded as a switch, although in one type of phase-shifter it behaves as a variable capaci-
tive susceptance. The simplest type of diode phase shifter is the switched line phase
shifter, consisting of elements in which diode switches select either of two transmission
line lengths. This is a digital device, as are most diode phase shifters. As an example of
its action, if the two transmission-line segments through which the signal can be alter-
nately switched differ in length by a quarter wavelength, the resulting phase shift will
be 90 degrees (since a 360-degree phase change occurs for each wavelength of a propa-
gation path). Other well-known types of diode phase shifters are the hybrid-coupled
phase shifter and the periodically-loaded-line phase shifter (Tang and Burns, 1993).
A digital phase shifter is constructed by cascading several units, each of which can be
switched between two fixed phase-shift values. The incremental phase shift of each suc-
cessive unit is twice as large as that of the next-smaller unit, and the largest increment
is 180 degrees. This permits all possible phase shifts to be achieved, in increments of the
smallest unit, by switching the appropriate elements from one state to the other. For
example, a 4-bit phase shifter has incremental phase-shift elements of 22.5, 45, 90, and
180 degrees (and zero), as shown in Fig. 8–3. This permits phase delays of 22.5, 45,
67.5, 90, 112.5, 135, 157.5, 180, 202.5, . . . , and so on. (For an n-bit phase shifter, the
number of possible delays is 2n, including zero phase shift which is equivalent to 360
Beam-Steering Technology 349
degrees.) This description of digital phase

shifting is applicable to digital ferrite
phase shifters as well as to diode Input 22.5° 45° 90° 180°
output
shifters.
The choice between ferrite and diode FIGURE 8–3.
phase shifters is made on the basis of Schematic of a 4-bit digital phase shifter.
many factors, of which the principal ones
other than cost include the frequency of operation, the rf power handling needs, the per-
missible insertion loss, permissible switching power, available space, and allowable
weight. The relative importance of these factors varies for different applications, so it is
difficult to make any general statements of preferences. Ferrite phase shifters are capable
of handling higher power levels and of operating efficiently at higher frequencies (although
both types are operable to at least X-band), and their insertion loss is usually lower. Diode
phase shifters become increasingly efficient at lower frequencies, and are somewhat less
critical in adjustment.
It is to be noted that, where phasers are used with either a receive module or a
transmit-receive (T/R) module (see Fig. 8–7), phaser losses are less critical. This is
because the phasers follow low-noise receive amplifiers and precede the transmit ampli-
fiers. In addition, where T/R modules are used, the phaser power handling requirements
are reduced because total transmit power is distributed between modules. Furthermore,
other advantages of diode phasers over ferrite phasers include lower cost, smaller size
and less weight. Therefore, diode phasers are used with receive or with T/R modules at
higher frequencies than might otherwise be possible.
8.2.2. Effects of Phase Quantization

It is evident that a digital phase shifter cannot provide the exact phase shift theoretically
required for the specified beam direction. However, it has been shown (Miller 1964)
that for a linear array of many elements the effect of phase errors is not great if the
average phase error over the array is algebraically near zero and if the errors are
random from element to element (i.e., not periodic). This conclusion is very similar in
its principles to the analysis of Ruze (sec. 6.6.2) concerning the effect on side-lobe levels
of deviations of a reflector surface from a true parabolic shape. When these conditions
are fulfilled (Cheston and Frank 1970, pp. 11–37), a 4-bit phase shifter will add less than
an additional 1.4 percent of the total radiated power to the average side-lobe power level
in a linear array of 100 elements, with a correspondingly negligible reduction of the
main-beam gain (0.06 dB). There may also be a slight but insignificant shift of the main-
beam direction. However, if the phase errors have any periodicity, from element to
element, much more significant effects will occur, including the formation of phase-
quantization lobes, which are analogous to grating lobes. There will also be some shift
of the main-beam direction. At some angles the phase-quantization lobes can be signifi-
cant. However, a method of reducing the peak phase-quantization lobes, attributed to
Miller (1964), consists of inserting fixed randomized phase shifts in the feed paths to
each array element. Quantization lobes can also arise due to amplitude quantization
effects when subarrays are used. For details of these matters, see Cheston and Frank
(1970, pp. 11–37).
8.2.3. Effects of Element Mutual Coupling

The design of large-scale electronically beam-steered arrays should include analyses on
effects of mutual impedance between array elements and on possible radiation blind
directions that can occur. In the early days of array development, unexpected “blind
spots” were observed for some arrays at certain scan angles (see, e.g., Allen 1962;
Cheston and Frank 1970; Hansen 1973, 2007; Stark 1974; Tang and Burns 1984, pp.
20–25/20–28; Mailloux 2005, ch. 6). It was eventually determined that mutual coupling
between array elements was a major cause. As discussed in sec. 5.5.1, the overall array
pattern is the product of an array factor (the pattern that would result if an equivalent
array of isotropic elements were used) and the actual pattern of a single element of the
array. However, as stated, the element pattern to be used is that of an element in the array
environment. That is, the element pattern to be used is not the pattern that is measured
for an isolated element.
Ideally, an antenna should provide a good impedance match between the transmitter
and free space. If there are impedance mismatches, a loss of input and radiated power
occurs. Thus a mismatch, even at one element, changes the array aperture illumination
and may cause an appreciable pattern change from that which would occur based on the
element’s assumed radiated power. A mismatch causes standing waves which produces
undesirable, higher voltage levels than does a matched condition. Furthermore, internal
reflections may cause the generation of spurious lobes in the overall array radiation
pattern. In fact, there are conditions, sometimes observed, where a matched condition
exists at the broadside beam direction, but at some other angle most of the power is
reflected and thus a scan “blindness” results. In spite of the difficulty of dealing with
mutual coupling, especially with beam scanning arrays, fixed-beam and electronically
steered arrays have been developed and satisfactorily operating for a number of decades
(Oliner and Knittel, 1972; Hansen, 2007).
It will be recalled, from secs. 5.1.4 through 5.1.7, that mutual coupling between ele-
ments changes the input impedance, radiated power, and the pattern of an array element.
Consequently, when phase changes are introduced to change the beam pointing direction
or operating frequency, element input impedances and radiated power change, and, in
general, the changes are different for each element. In addition, it has been found that
the major effect of beam scanning is individual element impedance change (thus change
in element current and element radiated power) and not element pattern change. Conse-
quently, a major design issue is to address the input impedances of the elements. This
means that mutual coupling analysis is needed to solve for the input voltages (V1, V2,
etc.) of Eq. 5–5, sec. 5.1.5, that produce the required currents.
There have been numerous publications written on the design of arrays (see, e.g., Allen
1962; Cheston and Frank 1970; Hansen 1973, 2007; Stark 1974; Tang and Burns 1984;
Mailloux 2005). Hansen (2007) is a recommended starting place before plans are set in
place for a major new steered-array development. There he outlines important steps for
Phased Array Feed System Technology 351
avoiding major flaws in design, which include detailed preliminary design, simulation,
computer analyses, and measurements.
8.2.4. Time Delayers

An array beam can be steered in any direction (subject to some constraints imposed by
the element pattern and by the array geometry) by varying the element phases over a
maximum range of zero to 360 degrees or 2p radians. In other words the maximum phase
shift required per element is modulo 2p radians. However, if a wide band of frequencies
is to be radiated or received, especially with large arrays (high gain, narrow beam), an
additional requirement is imposed because the desired phase shift cannot then be pre-
served over the full frequency range. That is, the modulo 2p principle does not then
apply. For very wide band operation, “true time delay” is required, which means that the
propagation times for the paths from a common point in the feed system to any point on
the radiated or received plane wavefront must be equal in the absolute sense, rather than
in the modulo 2p phase sense.
This means that instead of a phase delay of up to 360 degrees maximum, the delay to
some elements must have additional increments of full-wavelength path delays (at the
nominal frequency), so that this propagation time criterion will be satisfied. Devices to
accomplish this, though identical in nature to a phase shifter, must provide effective path
lengths equal to or greater than a wavelength at the nominal frequency.
Because of the much longer lengths of line involved, the insertion loss of time delayers
can be unacceptable in a passive array. This difficulty can be overcome, however, by the
use of transmitting and receiving amplifiers at the array elements; then the time-delayer
losses occur at low transmitter power levels so that the total power wasted is not large;
and in the receiving mode, the additional noise due to the time-delayer loss occurs
after sufficient amplification of the received signal so that the signal-to-noise ratio is
not appreciably degraded. An array using this principle is defined as an “active array,”
and the concept was introduced in sec. 7.8 in connection with low-noise receiving
antennas.
8.3. Phased Array Feed System Technology

The general principles of array feed systems are discussed in chapter 5 in terms of trans-
mitting. In summary, the total power radiated must be divided among the array elements
with some constant uniform phase gradient along any straight line through the array. The
greatest array gain and narrowest beamwidth is achieved with uniform distribution of
power among the elements, while an improved (reduced) side-lobe level results if the
power fed to the elements is reduced (tapered) toward each end element.
In the present chapter an additional requirement has been stated for broad-band
arrays; namely, the propagation times from the common feed source point (e.g., transmit-
ter) to all points on the radiated wavefront must be equal. For a broadside array this can
be achieved by either using equal-length paths from the source point to each element
or by use of a corporate feed structure. The difficulty with the first scheme is power
division among the elements; it is difficult to accomplish this for a many-element array
by branching from a single transmission line or waveguide into many lines all at one
division point. The corporate feed scheme, shown in chapter 2, Fig. 2–8(a), avoids
this problem while still preserving equal line lengths by the branching method.
At each branch, equal power division or any desired tapering can be obtained by suitable
impedance relationships at the junctions. Alternatively, hybrid couplers can be used to
minimize the coupling between element transmission lines, caused by impedance
mismatches.
If a beam at some direction other than broadside is wanted, then the lengths of the
lines to the various elements can be made unequal to provide the desired phase progres-
sion, or alternatively, phase shifters and/or time delayers can be inserted in the lines.
8.3.1. Feed Systems

ESA feed systems take many forms. Figure 8–4 shows multiple waveguide array ele-
ments (at left side) being fed by a corporate feed (at right side). Corporate feeds provide
equal path lengths and equal power to their outputs. This particular corporate feed sup-
plies four of the waveguide array elements of a multielement array that has apertures
oriented for vertical polarization. Moving to the left and down from the corporate feed
input, one can see a four-port hybrid junction, known as a hybrid tee. Its fourth port is
a “stubby” load (termination) at the top of the tee. The magic tee divides input power
equally between its two symmetrical ports, and each of these ports deliver power to two
additional magic tees. The load dissipates power reflected (by impedance mismatches)
from the symmetrical ports. There are, of course, many magic tees required for providing
equal power and equal path lengths to each of the multiple waveguide array elements.
In addition, there is a ferrite phase shifter in each waveguide path (between corporate
feed and array element) to provide separate phase control for each array element. The
phase shifter within the waveguide on the far left side is shown by means of a cutaway
view.
Figure 8–5 is a sketch of an eight-
CORPORATE FEED element corporate-fed linear array with
FERRITE eight phase shifters and two time delay-
PHASE SHIFTERS ers. The time delayers may be unneces-
sary, but they will increase the array
bandwidth. Let us assume the elements
are in a horizontal plane, so that the beam
is steered in azimuth. More complex
arrays can be formed by adding arrays
like the one in Fig. 8–5. For beam steer-
ELEMENT ARRAY ing in both azimuth and elevation, verti-
FIGURE 8–4. cal array elements are required. These
Phased array network with magic tee dividers might be on arrays each like Fig. 8–5,
and ferrite phase shifters. From Stark 1974; with each array replacing an array element
© 1974, IEEE. of Fig. 8–5. Then the antenna would be
ELEMENTS
PHASE SHIFTERS
HYBRID JUNCTIONS
RF LOADS
TIME DELAYERS
CORPORATE FEED
FIGURE 8–5.
Corporate feed for an eight-element array, consisting of eight phase shifters and two time
delayers.
an 8 by 8 planar array having 64 elements. It should be apparent that large, multielement

arrays may be devised by adding numerous sets of corporate fed arrays. When there are
a large number of array elements, tapering for side-lobe reduction may be implemented
by gradually reducing the number of elements toward the ends of the array.
Thus far the feed systems for arrays have been assumed to consist of transmission
lines or waveguides. It is also possible to feed an array optically—that is, in the same
manner that a reflector or lens would be fed (as illustrated in chapter 6, Figs. 6–1, 6–4,
6–10, and 6–14). An antenna having elements fed in this manner is called a space-fed
array. The “array” in this case is an assemblage of delay lines and/or phase shifters with
input and output ports (e.g., horns, dipoles, etc.) arranged as array elements. In this case,
the “array” and not a reflector or lens, does the focusing.
Examples of this type of antenna include the Rotman lens (Rotman and Turner 1963)
and the Bootlace lens (Volakis 2007, pp. 20–33). In the Rotman lens, the “array,” that
is, an assemblage of delay lines with input and output ports are located at the output of
a 2-D wave-guiding medium (see, e.g., Fig. 6–8). Figure 8–6 is an example of a Bootlace
lens in which a space-fed array of waveguides are energized by an E-plane feed. In this
example, the antenna contains twenty-four waveguides (8 columns by 3 rows), each
having a phase shifter or a time delayer so that the beam may be steered in both azimuth
and elevation. The reader will note the similarity with the metal waveguide lens of Fig.
6–23, except in that case the beam cannot be electronically steered.
8.3.2. Active Arrays and T/R Modules

Arrays that have a transmitter at each array element are called active arrays. For an
active radar array, a transmit/receive (T/R) module is connected to each array element.
Figure 8–7 illustrates a T/R module, where it can be seen to contain both a power ampli-
fier and low noise amplifier (LNA), for providing transmit and receive capabilities.
Additionally, T/R modules include the
element phase shifter and transmit/
receive switches.
It is to be noted that essentially a com-
plete radar can be conceptionalized by
E replacing each phase shifter at each
antenna element with a T/R module.
Then, each element would be energized
with the same power, and therefore there
F e e d would not be amplitude taper on trans-
ane
E-Pl mit. This is commonly done to maximize
FIGURE 8–6. transmit power. Side-lobe reduction on
A space-fed array that includes twenty-four reception is customarily done by taper-
waveguide-fed elements, each waveguide having ing on receive by control of the T/R
a phase shifter. receive gains or use of attenuation.
Pwr
ampl
Transmit
Transmit/ Transmit/
Phase
receive receive Antenna
shifter
switch switch
Receive
LNA
FIGURE 8–7.
Simplified diagram of a T/R module that serves as both transmitter and receiver at each
element of an active array. From Merrill I. Skolnik, Introduction to Radar Systems, 3rd ed.,
© 2001, p. 600, with permission of The McGraw-Hill Companies.
An additional advantage of using multiple active (transmit or receive) components is

reliability, because the loss of one or a few elements in a large array may not cause a
system outage.
8.3.3. Alternative Configurations, Including Subarrays

As may be imagined, there are many possible array network configurations. Return now
to Fig. 8–5, and assume each phase shifter is replaced by a T/R module. Each module
has a phase shifter, a low-noise receive amplifier (LNA), and possibly a low-power
transmit amplifier plus a transmit/receive switch. If the network were used for radar, an
initial transmitter signal at the network input provides an excitation for each low-power
transmit amplifiers via transmission through the hybrid junctions, and the low-transmit
power of each element is summed upon radiation. Amplitude taper is often not used for
transmitting, but it can be implemented by gain adjustments of the transmit amplifiers
within the T/R modules.
On receive, the signals from the array elements are summed coherently (phase and
amplitude) at each hybrid junction to provide a summed signal at the network output. It
is to be noted that, for a lossless network of eight elements, the signal to noise ratio is
substantially increased. For example, for an in-phase summation of eight equal amplitude
signals, the voltage would be increased by a factor of 8, that is, power increased by the
factor of 64. However, the summed noise power of eight equal-power independent noise
sources is increased by eight. Thus, the signal-to-noise improvement for this assumed
case is eight. For side-lobe reduction, gains of the LNA may be adjusted and some attenu-
ation may be desired for refined adjustments.
Modern solid-state amplifiers have considerably reduced the cost of large phased
arrays that use amplifiers located at the array elements, although this is still a fairly
expensive type of antenna. The cost can be reduced, however, by dividing the array into
subarrays, each of which contains several array elements, with only one amplifier per
subarray. It will still be necessary (if wide-angle two-dimensional scanning is required)
to provide a separate phase shifter for each element. However it then becomes possible
to provide only one time delayer per subarray. Each subarray is then regarded as an
element, and the steering of the actual individual elements is regarded as “steering” the
subarray-element patterns (Cheston and Frank 1970, Fig. 31, pp. 11–45). The time-
delayers then provide the array steering control.
8.3.4. Butler Matrix

Several matrix methods of feeding an array have been devised that result in the simulta-
neous generation of a number of beams at different directions from the array normal, in
such a way that receivers connected to different ports of the feed matrix will each receive
signals from only one of the beams, and a transmitter connected to one port will generate
only the radiated beam corresponding to that port. Or, if the power from a single trans-
mitter is divided equally and in-phase among the ports, all of the beams will be radiated
simultaneously. The resulting total pattern will then be the far-field vector-phasor sum
of the individual beam patterns. Finally,

if the transmitter power is divided
unequally among the ports, or with some
phase variation, the far-field superposi-
tion of the individual beams will be cor-
2 2 2 2 respondingly modified. This allows
generation of a “shaped beam” tailored to
3 1 1 3 any desired spatial coverage requirement.
Various modifications of this beam-
shaping principle are also used (Shelton
and Kelleher 1961).
An example of this matrix type of
1R 4L 3R 2L 2R 3L 4R 1L
feed—possibly the one that is best
……3-DB HYBRID COUPLERS
known—is the Butler matrix (Butler and
……PHASE SHIFTER, N × 22.5°
Lowe 1961). This and other matrix feeds,
FIGURE 8–8(a). including the Blass matrix, have been dis-
The Butler matrix for an eight-element, eight- cussed by Butler (Butler 1966). A sche-
beam array (after Butler 1966). matic of the Butler matrix, applied to an
eight-element, eight-beam array, is shown
1L 1R
in Fig. 8–8(a). In its basic form, this
2L 2R matrix is applicable to any number of ele-
3L 3R ments equal to an integral power of 2, that
4L 4R is, 2n, and there will then be m beams and
m ports, where m is an integer equal to or
less than 2n. The m beams are orthogonal,
which means (in this context) that the
peak of each beam corresponds to nulls
of the other beams. The matrix contains
fixed phase shifters and hybrid couplers,
denoted respectively in Fig. 8–8(a) by
FIGURE 8–8(b).
circles and rectangles. For this particular
matrix the couplers are 3 dB couplers
Beam pattern of the eight-beam Butler matrix
(equal power in the two coupled arms),
antenna (after Butler 1966).
and their two output arms are phased 0
and 90 degrees, or 90 degrees and 0,
depending on which arm is the input arm. The phase shift of each phase shifter is in units
of 22.5 degrees, or p/8 radians; the integer in each of the phase shifter rectangles indicates
the number of p/8 radian units that are shifted by that phase shifter. The output ports in
Fig. 8–8(a) are labeled numerically and by letter—L for left, R for right—and the corre-
sponding beams are similarly labeled in Fig. 8–8(b).
Numerous variations of this scheme, as described by Butler (1966), are possible,
including methods that produce a tapered power distribution for side-lobe reduction, “true
time delay” (wideband) matrices, and matrices that do the beam forming at the intermedi-
ate frequency of a receiving system.
8.3.5. Array Sum and Difference Patterns

The radar monopulse method of direction finding is described in sec. 7.6 of chapter 7.
It uses sum and difference patterns of an antenna for greater accuracy in determining
direction than possible if only the ordinary (sum) beam is used. This method was
described there in terms of a reflector type antenna with a multihorn feed. The desired
sum and difference patterns, in that case, are separately obtained at different receiver
ports by combining the separate feed-horn outputs in a system of hybrid couplers.
The same result can be obtained in an array (including electronically beam-steered
arrays) by feeding two symmetrical halves of the array separately and then combining
the signals from the two halves in a hybrid coupler to produce the desired sum and
difference signals. This permits monopulse operation in the plane perpendicular to the
line dividing the two halves. If the array is divided into quadrants, simultaneous
monopulse in both perpendicular planes (e.g., azimuth and elevation) can be obtained.
However, many special considerations arise in the application of monopulse techniques
to phased arrays for optimum generation of sum and difference patterns (Stark 1974,
p. 1694).
8.3.6. Microstrip and Printed-Circuited Technology

The geometry of the strip line is well adapted to construction by printed-circuit tech-
niques, in which the strip conductor and adjoining ground planes can be deposited on
low-loss substrate material by mass-production methods. An unbalanced stripline geom-
etry is illustrated in chapter 2, Fig. 2–9 (lower right), and a patch or microstrip antenna
is shown in chapter 4, Fig. 4–25.
The balanced form has the strip enclosed both below and above by a metallic ground-
plane. This transmission-line technology can be combined with ordinary printed-circuit
techniques to permit construction of monolithic phased-array modules containing not
only the transmission line components but also phase shifters, couplers, control circuitry,
radiating elements, and solid-state transmitting and receiving amplifiers. This fabrication
technique greatly reduces the cost as well as the bulk of large phased arrays. Moreover,
the “sheet” construction need not be completely planar; it can be made to conform to the
surface of a vehicle such as the fuselage of an airplane.
8.3.7. Conformal Arrays

An array thus constructed in nonplanar form is known as a conformal array. It is apparent
that the phasing of the elements must be calculated to produce a beam by taking the
nonplanar shape into account, but this can readily be done by a digital computer or, for
beam-steered arrays, by a dedicated microprocessor. The radiating elements can be con-
ventional dipoles mounted outside the ground plane, or suitably excited slots in the
ground plane, or so-called patch elements that can be regarded as a special form of slot
radiator (Mailloux 1982. For a general discussion of slot radiators see chapter 3,
pp. 196–199). Incidentally, the problem of controlling the side-lobe level is more difficult
in conformal arrays than in planar arrays because the element patterns in different parts
of the conformal array are oriented differently.
8.4. Adaptive Array Antennas

Adaptive array antennas have been developed from the technology of rapid beam point-
ing and shaping, along with being jointly merged with digital signal processing. The term
adaptive array refers to an array-type antenna that has the capability of adjusting (adapt-
ing) its pattern and the frequency filtering of the signals received from each array element
in some prescribed way that depends on phases and amplitudes, as well as the time delay
of signals received from external radiating sources. The full capabilities of adaptive array
antennas stems from the technology advancements made for beam scanning and beam
shaping of electronic steered arrays (ESAs), that is, phased arrays that were outlined as
items (i) through (vi) at the introduction to this chapter.
8.4.1. Background System Developments

An excellent review of the early state of development of adaptive array antennas was
published by Gabriel (1983), and the discussion in this section is based largely on that
paper.
In the early 1980s, the most common example of an adaptive array is the so-called
side-lobe canceller antenna, which adjusts its side-lobe pattern so that pattern nulls occur
in the side lobes that are in the directions of the sources of jamming (jammers), thus in
principle reducing the total received jamming signal to zero. In practice, of course, the
nulls are not absolute, yet a high degree of cancellation of the jamming signals can be
obtained, subject to certain restrictions. The restrictions are that there is a maximum
number of separate jamming directions that can be nulled, and the nulling performance
may be degraded for some angular distributions of the individual jammers. Also, the
adjustment (adaptation) of the pattern is not instantaneous; a finite time is required, and
this may place a limit on the radar scan rate.
The typical side-lobe canceller (SLC) consisted of a high-gain main antenna, which
was any type (e.g., a reflector, a lens, or an array), and an auxiliary array consisting of
several low-gain elements. These auxiliary elements are usually omnidirectional or nearly
so in the horizontal plane, with polarization identical to that of the main antenna. The
number of these auxiliary elements determines how many adaptive side lobe nulls can
be positioned arbitrarily to cancel jamming signals. This type of antenna is sometimes
referred to as a coherent side-lobe canceller (CSLC) to differentiate it from the noncoher-
ent device called a side-lobe blanker (SLB). The SLB utilizes a single low-gain horizon-
tally omnidirectional element, and when the signal received by this element is equal to
or greater than the signal received by the high-gain main antenna, the receiver output is
switched off, or “blanked.” This prevents side lobe jamming signals from being received,
but it does not permit detection of a desired main-antenna (higher gain) signal if it is
weaker than a jamming signal received by the low-gain antenna.
Adaptive Array Antennas 359
In the side-lobe canceller, the signals received by the auxiliary elements are processed
with the main-antenna signals in such a way that they create nulls of the main antenna
pattern in the jammer directions. The theory of this processing involves the concept of
the “degrees of freedom” available, which are equal to the number of low-gain elements
in the auxiliary array. This number in turn determines (generally is equal to) the number
of separate jamming-signal directions that can be nulled.
The mathematical algorithms of this signal processing and control tend to be very
complicated because they are partly based on multivariate estimation theory. Implemen-
tation of some of these algorithms can be either analog or digital, but in modern systems
digital technology is predominant. The adaptive process usually involves a “convergence
time,” and some of the simpler algorithms may converge relatively slowly, especially for
certain angular distributions of the interfering signals. However, rapid convergence times
are obtainable with the more sophisticated algorithms and techniques.
The side-lobe canceller, using a relatively few auxiliary array elements, is a “partial”
adaptive array. A fully adaptive array is one in which the entire system is an array and
all of the array elements contribute equally to the operation of the adaptive processor.
This allows much greater flexibility and control of the total pattern (i.e., there are more
degrees of freedom). However, the cost of the processor is much higher for arrays having
a very large number of elements, such as those used in many radar systems.
The principle of a fully adaptive array was demonstrated with computer computations,
however, in the above-referenced paper by Gabriel. In that 1983 paper, Gabriel’s array
had sixteen elements and was assumed to operate at S-band (approximately 3,000 MHz).
Further assumptions included a separate receiver for each array element and each of the
received signals, after conversion to an intermediate frequency (IF), was split into
two baseband I and Q signals (see Fig. 8–11). As is customary, I and Q denote the
in-phase and phase-quadrature components of an IF signal, and they preserve the phase
information in the “zero frequency” (baseband) channels that is necessary for digital
processing.
Figures 8–9(a) and 8–9(b) show results
of Gabriel’s computer analyses. In Fig. 0
8–9(a), the “quiescent” pattern of the
POWER IN DECIDELS
array is shown—that is, the pattern in the –10

absence of any jamming. Figure 8–9(b)
then shows the adaptive patterns at suc- –20
cessive time intervals in a continuous
adaptive process. As shown, a null region –30
exists at the angles of the three jamming
transmitters, which are indicated by the –40
three vertical arrows. Note that the average –90 –60 –30 0 30 60 90
side-lobe levels of the adaptive patterns SPATIAL ANGLE IN DEGREES
in the unjammed regions are somewhat FIGURE 8–9(a).
higher than those of the quiescent pattern; Adaptive array pattern with no interfering
this is the price paid for creation of (jamming) signal present (from Gabriel 1983;
the nulls in the jammer directions. It is © 1983, IEEE).
possible to retain the quiescent side-lobe levels in the adaptive patterns, but this requires
the application of additional computations in the adaptive algorithm (Frost 1972).
The technology of fully adaptive arrays is still in a state of intensive development,
and this was also the case as pointed out by Gabriel (1983). Some of the other early
classic papers include Gabriel (1976), Childers (1978), Gabriel (1980), Brennan and Reed
(1973), Brennan, Reed, and Swerling (1974), and Monzingo and Miller (1980). As
already indicated, development of adaptive or “smart” arrays continues to be an active
field, having its origins in the suppression of side lobes of a main antenna by use of
several smaller auxiliary antennas (Gabriel 1976). Today pattern optimization is done
with real-time weighting of received signals in such a way that the array on reception
can adapt to the interference environment. In principle it can also be done by weighting
the transmit antenna pattern, but this gen-
0 erally is not done. An extensive list of
up-to-date references is given by Gross
POWER IN DECIDELS
–10 (2007).
–20 8.4.2. Adaptive Beam-Forming

Arrays
–30
Figure 8–10 is a simplified example of an
adaptive beam-forming (ABF) system.
–40
–90 –60 –30 0 30 60 90 The phase and amplitude of signals from
SPATIAL ANGLE IN DEGREES each individual array element are ampli-
FIGURE 8–9(b). fied and filtered by a low-noise receiver
Patterns of the adaptive array at successive (Rx), converted to baseband, and digitized
time intervals after reception of a jamming with analog-to-digital (ADC) converters.
signal (from Gabriel 1983; © 1983, IEEE). Next, the digital signals are processed by
Rx ADC
ELEMENTS
Rx ADC
ARRAY
ADAPTED
+ OUTPUT
Rx ADC
WEIGHTS
ALGORITHMS
CONTROLLER
FIGURE 8–10.
Basic components of an adaptive array.
IF
MIXER
LOW
RF PASS A/D Q
MIXER FILTER
BAND
IF PASS
AMP FILTER IF
MIXER
LOW
PASS A/D I
RF LO FILTER
90°
IF LO
FIGURE 8–11.
Typical low-noise coherent (phase and amplitude) I and Q detection receiver.
computer algorithms for establishing phase and amplitude weights that modify the indi-
vidual received signals. Then, the modified signals are summed for providing interfer-
ence-suppressed outputs. The outputs can be from multiple signals that may have
simultaneously arrived at the array elements. For some algorithms, a sample of the
summed signals is fed back to the weight controller, as suggested by the dashed line in
Fig. 8–10.
Figure 8–11 illustrates the principal features of a receiver that supplies, in digital
format, the I and Q signals that contain phase and amplitude information, where ampli-
tude “a” is
a = I 2 + Q2
and phase j is
ϕ = arc tan ⎛⎜ ⎞⎟
Q
⎝I⎠
The mathematics of pattern optimization uses complex matrix theory. There is cur-
rently much activity in the development in adaptive algorithms, in computational capa-
bilities, and in improved array hardware (see, e.g., Balanis 2005). Presently used
algorithms are generally based on the early work of (1) the Howells-Applebaum method
(Applebaum 1976) and (2) a procedure due to Widrow and McCool (1976) that mini-
mizes the least square mean (LMS) between the array output signal and a reference
signal. The Howells-Applebaum method is commonly used with radar systems to affect
signal-to-noise optimization, subject to an array pattern formed in the absence of interfer-
ence. This requires sampling over a large number of range and Doppler cells to obtain
data to which the array system is optimized. The LMS method is used in communications
systems, for which a reference signal can use a replica of the received signal format.
Sometimes the LMS method is used with radar, and then the waveform of the transmit
signal can be used as the reference.
8.4.3. Space-Time Adaptive Processing

The antenna pattern optimization process of Fig. 8–10 is known as spatial adaptation.
Another processing method is temporal adaptation, which uses the separation and sup-
pression of signals on the basis on time delay or frequency. An adaptive system that
combines, cooperatively, both the spatial and temporal features of signals is said to use
space-time adaptive processing (STAP). The use of STAP, with its combined beam
shaping and Doppler processing features, has proved to be especially effective for
improving signal detection and clutter suppression for radar and mobile communications
on moving platforms.
Figure 8–12 is an oversimplified schematic for a generalized steerable array and signal
processor to provide both space and time adaptivity for the output voltage v(t). There are
N array elements and M time delays. The processing of time-delayed signals offers a
Antenna 1 Antenna 2 Antenna N
U1 (t) U2 (t) UN (t)
T w11 T T wN1
T w12 T wN2
w13 wN3
v (t)
FIGURE 8–12.
Generalized Space-Time Processor having N element channels and M delay taps per channel,
for cooperative space and time adaptive processing. From Barile, Fante, and Torres (1992);
© 1992, IEEE).
variety of frequency filtering options. For example, the availability of complex (ampli-
tude and phase) time-delayed signals permits fast Fourier transform (FFT) processing
that provides, in essence, signals from multiple band-pass filters. Finally, it is to be noted
that, in principle, the Fig. 8–12 configuration permits optimized frequency filtering and
beam shaping to be accomplished independently, or cooperatively (i.e., optimized
collectively).
As may be imagined, there are numerous design options. For example, the Fig. 8–12
configuration permits each antenna element to have a phase response that varies with
frequency. A phase response can compensate for the fact that a given path length l gives
a phase delay (2pl)/l = (2pl)f/c that depends on frequency. In other words, the tapped
delay-line may perform as an equalizer, to make the array have approximately the same
pattern over a range of frequencies. Wireless arrays using this phase feature are known
as wideband arrays and they are said to employ space-time, spatio-temporal, or two-
dimensional processors.
Thus, the wireless “two-dimensional” adaptive array provides the structure for sup-
pressing interfering signals based on frequency differences and on spatial separations
cooperatively. This configuration is said to be particularly valuable in heterogeneous
communications environments where signals of different bandwidths and different carrier
frequencies may be incident on the array (Liberti and Rappaport 1999, p. 102).
Features of a wideband wireless array are also of advantage for narrow band signals.
Consider, for example, multipath signal components that may or may not be separated
in angle or by significant time delays. The two-dimensional adaptive array can capture
multipath components separated in either time or angle, and thereby combine features of
both a spatial processor and a temporal equalizer (Liberti and Rappaport 1999, p. 102).
Computer throughput requirements and ultimate system performance can be altered
considerably, depending on how the element signals are processed. Processing may be
done on the signals from each array element, and then the separately processed element
signals can be jointly optimized. For example, one might first frequency-filter the M
time-delayed signals from each of the elements separately and then optimize the N
frequency-filtered element signals by adaptive beam-shaping and pointing. Alternatively,
the processing might first optimize beam shape and pointing direction for each of the M
different time delayed signals of the N elements separately, and then frequency-filter the
N signals that had previously been optimized by beam shaping and pointing.
Some readers may recognize that tapped delay-line filtering is commonly used in pulse
Doppler radar for suppressing clutter and noise. This filtering when combined with the
spatial adaptive features of Fig. 8–12 offers improved clutter and noise suppression
capabilities that are especially beneficial for radars on moving platforms (Skolnik 2001,
pp. 168–71).
Finally, fully adaptive arrays are costly and complex, but they offer many advantages
including
• Simultaneous transmission of independent messages in different directions

at the same frequency for increasing the capacity of band-limited cellular
telephone networks
• Fast adaptive pattern nulling for minimizing interference

• Corrections from array component imperfections and mutual coupling between
array elements
• Closely spaced multiple beams for increased network capacity
• Flexible time management in beam positioning and shaping the beams for
increasing network capacity and minimizing interference
• Improved detection and interference suppression for radar and communica-
tions on moving platforms.
8.5. Examples of Electronically Steered Arrays

Enormous progress has been made in the technology of large electronically steered
arrays, nevertheless, they are still very expensive, and they are used only in very special
applications that justify the cost. The principal application is for military radars. The
most complex and expensive antennas of this type utilize phase shifters for scanning
simultaneously in two orthogonal directions, for example, in azimuth and elevation.
These are sometimes referred to as phase-phase steered scanning arrays. Somewhat less
expensive are arrays that use phase scanning (either of the “true” phase-shifter type or
of the frequency-scan type which was described in sec. 8.1.3) in one direction (usually
elevation), and mechanical motion of the
entire antenna in the orthogonal direction
(usually azimuth).
An example of this arrangement is the
Marine Corps AN/TPS-32, an S-band
radar that scans by mechanical rotation of
the antenna in azimuth, with frequency
scanning from the horizon to about 20
degrees in elevation (Brookner 1977).
The pattern is a pencil beam of about 2
degrees azimuth beamwidth and 1 degree
elevation beamwidth. The antenna gain is
41 dB, and the side-lobe level is −25 dB
relative to the peak of the main beam. The
azimuth rotation rate is 6 rpm, and during
each 10-second azimuth scan the number
of elevation scans is sufficient to provide
azimuthal overlap of adjacent beams.
Figure 8–13 shows the Marine Corps
AN/TPS-59 radar, operating at L-band,
FIGURE 8–13. utilizes an array that combines mechani-
cal rotation in azimuth with true phase
The array antenna of the AN/TPS-59 radar
operates at L-band, and combines mechanical scanning in elevation (Johnson and Jasik
rotation in azimuth with phase scanning in 1984, p. 32–12). The array is rectangular,
elevation. (Courtesy of Lockheed Martin.) 30 feet in height and 15 feet wide, with
beamwidths of 1.6 and 3.2 degrees in elevation and azimuth, respectively. The elevation
coverage is from 0 to 19 degrees, with azimuth scan rates of either 6 or 12 rpm (option-
ally). The gain is 38.9 dB. Radiating elements are arranged in rows and columns, and
each of the fifty-four rows is fed by solid-state transmitter amplifiers; the needed rela-
tively large average power is made possible by the use of many low-power transmitter
amplifiers. Advantages of this technology are that failure of one or even a few of the
amplifiers does not totally disable the radar, and no portion of the system is subjected to
extremely high power levels with attendant problems of heating and voltage breakdown.
Receiver solid-state amplifiers and low-power-level phase shifters are similarly distrib-
uted. Both the TPS-32 and the TPS-59, incidentally, are transportable radars which can
be set up for temporary operation at field sites.
The Navy AN/SPY-1 shipboard multifunction S-band radar, a part of the Aegis weapon
system, has a phased-array antenna with true phase scanning in both azimuth and eleva-
tion (Bookner 1977; Scudder and Sheppard 1974). Four separate arrays are mounted on
four faces of the ship superstructure so that each covers an azimuth quadrant. The vertical
coverage of each array is sufficient so that the four arrays provide virtually hemispheric
coverage for the radar. Each array is of approximately hexagonal shape, about 12 feet in
width and 12 feet 7 inches in height. The vertical and horizontal beamwidths are of the
order of 2 degrees.
The radiating elements are rectangular horns (chapter 4, sec. 4.7), grouped into sub-
arrays as described in the preceding section of this chapter. The subarray arrangement is
somewhat different for transmitting and receiving. There are 32 transmit subarrays, each
containing 128 horn radiators. For receiving, the same horns are grouped into 64 sub-
arrays of 64 horns each. However, on reception four more subarrays are added to the
complete array, that is, these four subarrays are utilized for reception only. Thus there is
a total of 4,096 horns in the transmitting array and 4,352 horns in the receiving array.
The purpose of the additional four subarrays used on reception is to improve the sym-
metry of the array monopulse difference patterns, which are generated simultaneously
with the normal “sum” pattern, as described in Sec 8.3.3. The additional reception sub-
arrays also contribute to an improved side-lobe level on reception.
The phasing of the individual horns of the array is accomplished by ferrite (garnet)
nonreciprocal digitally-controlled latching phase shifters. The phasing is done on an
individual-element basis; the steering commands are based on the so-called row-column
method. The subarrays are not involved in this aspect of the antenna operation.
This description applies to the first operational model of the antenna, which differs
from the earlier experimental model described by Scudder and Sheppard (1974). As with
most major systems, newer antenna designs have been developed since the first opera-
tional model.
Each transmit subarray is fed by a separate high-power amplifier, whose output power
is then divided equally among the 32 subarray horns. On reception, the individual sub-
arrays do not have corresponding rf amplifiers; but there is a low-noise rf amplifier
mounted on the back of the array for each of the three monopulse channels (sum, azimuth
difference, and elevation difference). This does not eliminate the signal-to-noise reduc-
tion caused by losses in the on-array transmission-line components and phase shifters,
but it does avoid the loss that would other-

wise occur in the transmission line from the
array to the receiver components that are
located in the deckhouse.
The transmit array is fed with uniform
power distribution for maximum gain. The
receive array, however, has a modified
Taylor distribution for the sum monopulse
beam, to reduce the side-lobe level. The
difference monopulse beams utilize a
Bayliss distribution, a special distribution
for providing difference patterns with low
FIGURE 8–14.
side lobes (Hansen 2007, p. 20–19). The
Photograph of the Cobra Dane (AN/FPS-108) approximately hexagonal shape of the array
phased array antenna at Shemya, Alaska. also contributes to side-lobe reduction, for
(Courtesy Raytheon Company.)
both transmitting and receiving, in accor-
dance with the principle discussed for geo-
metrically tapered apertures in sec. 6.3.1, Fig. 6–6. In fact, the general configuration of
the AN/SPY-1 aperture is similar to that illustrated in Fig. 6–6 except that the height-to-
width ratio of the AN/SPY-1 antenna is closer to unity. This antenna represents near-
ultimate sophistication of phased-array design, exclusive of adaptive arrays.
An example of an extremely large array of the phase-phase scanning type, also repre-
sentative of the most advanced phased-array technology, is the Air Force AN/FPS-108
“Cobra Dane” system which operates at L-band. The Cobra Dane is a one-of-a-kind,
ground-based multifunction radar used for intercontinental ballistic missile (ICBM)
detection and analysis. It is located at Shemya, Alaska. Figure 8–14 is a photograph of
the antenna, which has a diameter of 95 feet (Johnson and Jasik 1984, pp. 32–14 to 34–17;
Brookner 1977, p. 34). It consists of a total of 15,360 active and 19,408 passive elements
in a “thinned” array. The passive elements, which are connected to dummy loads, are
placed in the otherwise empty positions where active elements would be if the array were
filled. Thinning can reduce the side-lobe levels in a manner similar to that achieved by
tapering of element excitations. From Willey (1962), a thinned array of a given aperture
size has a narrower beamwidth (and lower near-in side lobes) than would a filled array
with the same number of active elements (but smaller aperture size).
This antenna has a subarray structure with “true time delay” phasing of the subarrays.
This is required because of the large size of the antenna and the large bandwidth of the
transmitted and received signals. There are 95 subarrays, each containing 160 elements.
The phasing of the individual elements is accomplished by 3-bit digital diode phase
shifters. In the transmit mode, each subarray is individually fed by a traveling-wave-tube
power amplifier, and in the receive mode each subarray feeds a low-noise amplifier.
There are many array antennas employed in airborne and space applications. The
U.S. Air Force E-3 Sentry airborne warning and control system (AWACS), which became
available in the 1970s, is an example of an early operational airborne ESA radar that
employs an electronically scanned array (ESA). It scans mechanically in azimuth and
electronically in elevation to obtain height

measurement and stabilize the beam to
compensate for aircraft roll and pitch.
Today there are several recently devel-
oped military radars that have active ESA
antennas (AESAs), where transmit and
receive (T/R) modules are placed directly
at the array elements (Hendrix 2008). An
advantage of these antennas over mechan-
ical rotating antennas is the ability to
point its beam essentially instantaneously
in any direction within its field of view.
This flexibility enables an operator to use
the radar in almost unlimited combina-
tions of surveillance and tracking. Addi-
tionally, having separate receivers at each
AESA element permits adaptive nulling
for the suppression of interference.
Figure 8–15 shows an ultralow side
lobe UHF (approximately 440 MHz)
radar array antenna that scans electroni-
cally (and adaptively) in elevation and
FIGURE 8–15.
mechanically rotates in azimuth. It was
developed and tested by Westinghouse Ultra side lobe array that scans electronically
(and adaptively) in elevation and mechanically
Electric Corporation (now Northrop in azimuth. (Reprinted with permission of MIT
Grumman) to be a central part of a Lincoln Lincoln Laboratory, Lexington, Massachusetts.)
Laboratory experimental radar system for
the U. S. Navy. The antenna consists of
14 horizontal line arrays, each with 24 elements having a 65-dB Chebyshev amplitude
taper (Carlson, Goodman, Austin, Ganz, and Upton, 1990). Array aperture dimensions
are 5 meters height by 10 meters width, with gain of 29 dB. Fabrication includes a
dielectric stripline construction with equal path lengths to each element for improved
bandwidth, and extreme accuracy was required to achieve its ultra-low azimuth sidelobe
levels, reported to be more than 50 dB below the peak of the main beam.
As already noted, beam pointing in azimuth is accomplished mechanically, but in
elevation the transmit and receive patterns are controlled electronically. On transmit, the
fourteen rows of line arrays are separately energized and phased for pointing the transmit
beam in elevation. On reception, signals from the fourteen rows are separately amplified,
converted to digital format with phase and amplitude preserved, and adaptively processed
with phase and amplitude weighting for maximizing signal to interference levels in the
elevation dimension. Therefore, interference is suppressed in the azimuth dimension by
the ultra-low azimuth sidelobe patterns of the fourteen rows of line arrays, and interfer-
ence in the elevation dimension is suppressed by adaptively processing the signals
received from the fourteen horizontal rows of line elements.
References
Ajioka, J. S., “Frequency-Scan Antennas,” ch.19 in R. C. Johnson, Antenna Engineering
Handbook, 3rd Edition, McGraw-Hill, 1993.
Allen, J. L., “Gain and Impedance Variation in Scanned Dipole Arrays,” IEEE
Transactions on Antennas and Propagation, vol. AP-10, September 1962, no. 5,
567–69.
Applebaum, S. J., “Adaptive Arrays,” IEEE Transactions on Antennas and Propagation,
Sept. 1976, pp. 585–98.
Archer, D., “Lens-Fed Multiple-Beam Arrays,” Microwave Journal, vol. 18, no. 10,
October 1975.
Balanis, C. A., Antenna Theory, Analysis and Design, 3rd ed., Wiley-Interscience, 2005,
ch. 16.
Barile, E. C., R. L. Fante, and J. A. Torres, “Some Limitations on the Effectiveness of
Airborne Adaptive Systems,” IEEE Transactions on Aerospace and Electronic Systems,
vol. 28, no. 4, October 1992, pp. 1015–32.
Brennan, L. E., and I. S. Reed, “Theory of Adaptive Radar,” IEEE Transactions on
Aerospace and Electronic Systems, AES-19, no. 2, March 1973, pp. 237–52.
Brennan, L. E., I. S. Reed, and P. Swerling, “Adaptive Arrays,” Microwave Journal,
vol. 17, no. 5, May 1974, pp. 43 ff.
Brookner, E., Radar Technology, Artech House, 1977.
Butler, J. L., “Digital, Matrix, and Intermediate Frequency Scanning,” in Microwave
Scanning Antennas, vol. III, R. C. Hansen, ed., Ch. 3, Academic Press, 1966.
Butler, J., and R. Lowe, “Beamforming Matrix Simplifies Design of Electronically
Scanned Antennas,” Electronics Design, April 1961, pp. 170–73.
Carlson, B. D., L. M. Goodman, J. Austin, M. W. Ganz,and L. O. Upton, “ An
Ultralow-Sidelobe Adaptive Array Antenna,” The Lincoln Laboratory Journal, Vol. 3,
No. 2, 1990, pp. 291–310.
Cheston, T. C., and J. Frank, “Array Antennas,” ch. 11 in Radar Handbook, M. I. Skolnik
(ed.), McGraw-Hill, 1970.
Childers, D. G., Modern Spectrum Analysis, IEEE Press, 1978.
Frank, J., “Bandwidth Criteria for Phased Array Antennas,” in A. A Oliner and G. H.
Knittel, Phased Array Antennas, Artech House, 1972, pp. 243–53.
Frost III, O.L., “An Algorithm for Linearly Constrained Adaptive Array Processing,”
Proceedings of the IEEE, vol. 60, no. 8, August 1972, pp. 926–35.
Gabriel, W. F., “Adaptive Arrays: An Introduction,” Proceedings of the IEEE, Feb. 1976,
pp. 239–72.
Gabriel, W. F., “Spectral Analysis and Adaptive Arrays Superresolution Techniques,”
Proceedings of the IEEE, vol. 68, June 1980, pp. 654–66.
Gabriel, W. F., “Adaptive Processing Antenna Systems,” IEEE Antennas and Propagation
Newsletter, October 1983.
Gross, F. B., “Smart/Antennas,” ch. 25 in Antenna Engineering Handbook, 4th ed., J. L.
Hansen, R. C. (ed.), Significant Phased Array Papers, Artech House, 1973.
References 369
Hansen, R. C., “Phased Arrays,” ch. 20 in Antenna Engineering Handbook, 4th ed., J. L.
Hendrix, R., “Aerospace System Improvements Enabled by Modern Phased Array Radar—
2008,” Proceedings of the 2008 Radar Conference, Rome, Italy, May 26–30, 2008,
pp. 275–280.
Johnson, R. C., Antenna Engineering Handbook, 3rd ed., McGraw-Hill, 1993.
Johnson, R.C., and H. Jasik, Antenna Engineering Handbook, 2nd ed., McGraw-Hill, 1984.
Liberti, J. C., Jr., and T. S. Rappaport, Smart Antennas for Wireless Communications,
Prentice Hall PTR, 1999, p. 102.
Mailloux, R. J., “Phased Array Theory and Technology,” Proceedings of the IEEE, vol. 70,
no. 3, March 1982, pp. 263–64.
Mailloux, R. J., Phased Array Antenna Handbook, Artech House, 2005.
Miller, C. J., “Minimizing the Effects of Phase Quantization Errors in an Electronically
Scanned Array,” Proceedings of the 1964 Symposium, Rome Air Development Center
Document RADC-TDR-64–225, vol. 1, pp. 17–38.
Monzingo, R. A., and T. W. Miller, Introduction to Adaptive Arrays, John Wiley and Sons,
1980.
Nelson, E. A., “Quanization Sidelobes of a Phased Array with a Triangular Element
Arrangement,” IEEE Transactions on Antennas and Propagation, vol. 17 (May 1969),
pp. 363–65.
Oliner, A. A., and G. H. Knittel, Phased Array Antennas, Artech House, 1972.
Romanofsky, R. R., “Array Phase Shifters: Theory and Technology,” ch. 21 in J. L.
Volakis, Antenna Engineering Handbook, 4th ed., McGraw-Hill, 2007.
Rotman, W., and R. F. Turner, “Wide-Angle Microwave Lens for Line-Source
Applications,” IEEE Transactions on Antennas and Propagation, AP-11, no. 11,
November 1963, pp. 623–32.
Scudder, R. M., and W. H. Sheppard, “AN/SPY-1 Phased-Array Antenna,” Microwave
Journal, vol. 17, no. 5, May 1974.
Sharp, E. D., “A Triangular Arrangement of Planar-Array Elements that Reduces the
Number Needed,” IEEE Transactions on Antennas and Propagation, AP-9, March 1961,
pp. 126–29.
Shelton, J. P., and K. S. Kelleher, “Multiple Beams from Linear Arrays,” IEEE
Transactions on Antennas and Propagation, March 1961, pp. 154–61.
Skolnik, M. I., Introduction to Radar Systems, 3rd ed., McGraw-Hill, 2001.
Stark, L., “Microwave Theory of Phased Array Antennas: A Review,” Proceedings of the
IEEE, December 1974, pp. 1661–1701.
Steyskal, H., “Digital Beamforming at Rome Laboratory,” Microwave Journal, Feb. 1996,
pp. 100–124.
Tang, R., and R. W. Burns, “Phased Arrays,” pp. 20–31 to 20–48, in R. C. Johnson,
Antenna Engineering Handbook, 3rd ed., McGraw-Hill, 1993.
Thomas, D. T., “Multiple Beam Synthesis of Low Sidelobe Patterns in Lens Fed Arrays,”
IEEE Transactions on Antennas and Propagation, AP-26, no. 6, November 1978,
pp. 883–86.
Volakis, J. L., Antenna Engineering Handbook, 4th ed., McGraw-Hill, 2007.
Widrow, B., and J. M. McCool, “Comparison of Adaptive Algorithms Based On The

Method Of Steepest Descent And Random Search,” IEEE Transactions on Antennas and
Propagation, Sept. 1976, pp. 615–37.
Willey, R. E., “Space Tapering of Linear and Planar Arrays,” IEEE Transactions on
Antennas and Propagation, July 1962, pp. 369–77.

1. An array of 1 m width is to be designed for l = 0.1 m and for a maximum scan
angle 30° from broadside. If time delayers are used for scanning, what is the
maximum time delay needed?
2. Assume that a planar ESA array has gain of 100 in broadside direction. When
steered in azimuth 60° off broadside, what is the estimated gain in that off-
broadside direction?
3. If instead of time delayers, assume that phase shifters are used in Problem 1.
What phase delay in degrees expressed less than 360° (mod 360°) is required
for an end element when the beam is scanned 30° from broadside direction?
Express this phase delay relative to the phase at the opposite end of the array.
4. To avoid grating lobes from a line array having equal spacing between elements,
what is the recommended maximum spacing between elements for the following:
(a) a nonscanning broadside array, (b) an electronically steered array?
5. Assume an ESA designed for 2.0 GHz with the following features: 16 elements
spaced 0.4l along a horizontal line at 2 GHz. (a) What is the end-to-end length
between the first and last elements? (b) If delay-only phasing is used, what is
the maximum delay time required if steered in azimuth 60° from broadside?
6. Using the ESA of problem 5, prepare a graph of phase (relative to an end
element) versus element number, with phase expressed less than 360° (mod
360°), for the beam steered 30° from broadside.
7. An ESA has a 3 dB beamwidth of 4.0 degrees in azimuth when steered in broad-
side direction. What is its estimated azimuthal beamwidth when steered in
azimuth 60° from broadside?
8. Assume the ESA of problem 5 has the phases calculated in problem 6. In addi-
tion, assume the element amplitudes are equal. Show calculated normalized
patterns for 2.0 and 2.1 GHz on one graph, for azimuth angles of −90° to 90°
with amplitude graphed from 0 to −40 dB.
9. (a) How many phase-shift elements are needed with a 5-bit phase shifter?
(b) What are the minimum and maximum phase shifts required?
10. A linear array is to be used as a scanning antenna by varying the frequency to
change the phasing. The array element spacing is d = l0/2 = c/2f0, in the notation
of equation (8–6). The array is fed with a serpentine feed for which I = 5, using
coaxial line for which fc = 0. When the frequency of operation is f = 0.98f0,
by what amount will the angle of the beam be shifted from the broadside
direction?
CHAPTER 9
Antenna Measurements
Every experienced antenna engineer knows that successful antenna design requires a
knowledge of antenna theory, and except for the very simplest antennas, a certain amount
of “cut and try” is necessary before an initial design will perform satisfactorily. An
essential part of this cut-and-try process is measurement of the antenna’s performance.
The majority of antenna measurements lie within two basic categories: impedance
measurements and pattern measurements. The first category deals with one of the most
important antenna parameters—the input impedance. The second category is a very broad
and equally important one, with many subcategories, such as measurements of beam-
width, minor lobe level, gain, and polarization characteristics. Measurements of effi-
ciency and noise may also be desired in some instances.
Not all these possible measurements need be made in every situation. It is seldom that
the complete antenna pattern is measured, including side lobes and polarization charac-
teristics in all directions. Often, at the higher frequencies, it can be assumed that antenna
ohmic losses are negligible, and therefore the radiation efficiency need not be measured.
The input impedance is practically always important, however. The beamwidth, gain,
and side-lobe levels are also usually important, especially at the higher frequencies where
directional antennas are often used. Detailed polarization measurements are important in
special cases. For example, dual transmit and receive polarizations may permit doubling
the number of available communications channels. Measurement of antenna bandwidth
is not actually a separate measurement category; it consists of measuring impedance,
pattern characteristics, and other critical parameters over a band of frequencies.
As discussed in sec. 7.8, the antenna may affect the noise level of a receiving system
in special cases. Consequently, the technique of antenna noise measurements is especially
important in the lower atmospheric noise region of about 1 to 10 GHz. Therefore, the
basic technique of antenna noise measurement and the effect of antenna noise on overall
receiving system noise are described in the present chapter.
9.1. Antenna Patterns, General

The pattern (transmission or reception) of an antenna is a description (in the transmitting
case) of the field strength, and sometimes phase, at a fixed far-field distance from the
371
372 Antenna Measurements
q = 0°
Z
q
Ef
ANTENNA
POSITION
f)
N
TIO
(q,
q = 90°
EC
f = 270° Eq
DIR
EOU q = 90°
ATO Y
R f f = 90°
X
q = 90°
f = 0°
q = 180°
FIGURE 9–1.
Standard Spherical Coordinate System Used in Antenna Measurements. Source (IEEE 1979,
p. 4); © 1979, IEEE.
antenna, as a function of direction. In accordance with Fig. 9–1, direction is convention-

ally expressed in terms of the two angles, q and f, of a spherical coordinate system whose
origin is at the antenna. In the fullest sense the pattern also includes a description of the
polarization, that is, the spatial direction of the electric field within a plane tangent to
the far-field sphere.
The unqualified term “antenna pattern” implies that the pattern was that of the radiat-
ing far-field. In practice, this may be accomplished by placing the test antenna in a radi-
ating field having essentially constant amplitude and phase. The traditional method of
accomplishing this is to satisfy the 2D2/l, 20l, and 20D rules of thumb of sec. 3.2.5,
where D is the maximum antenna dimension as viewed from the measurement point and
l the wavelength.
The 2D2/l criterion ordinarily dictates large distances and out-of-doors environments
for large antennas (in terms of wavelength) at higher frequencies. Measurements
Antenna Patterns, General 373
satisfying this criterion, referred to as far-zone measurements, are discussed in sec. 9.2.
In addition, well-developed measurement methods that permit the use of much shorter
separation distances are described in later sections and include compact ranges (sec. 9.3),
near-field ranges (sec. 9.4), and scale models (sec. 9.6).
A complete pattern measurement consists of measuring field strength and electric field
direction (polarization) for many different values of the angles q and f for all values, in
principle. In practice, the number of q and f directions in which measurements must be
made depends on the complexity of the pattern and the need for detailed information in
the particular application.
Since complete three-dimensional patterns are difficult to plot on a plain sheet of paper,
and since the patterns in particular planes usually provide adequate information, patterns
are usually measured and plotted in planes. Sometimes the pattern in only one plane
conveys the information of AUT interest for a particular antenna. For example, in most
earth-to-earth communication or broadcasting systems, the horizontal-plane pattern is the
major consideration. Vertical-plane patterns in these cases usually have significance only
to the extent that they affect the gain of the antenna in the horizontal plane (i.e., the less
power that is radiated uselessly upward, the more there is available for radiation in the
horizontal plane). Exceptions to this statement include earth-to-earth systems that involve
propagation via an elevated reflection point or medium.
The vertical pattern is often of interest more directly, in applications involving an
earth-based antenna and some elevated receiving point (or vice versa), as in radio
astronomy, space communications, and in radar intended for air or space target detection
and tracking. The patterns in these two planes—horizontal and vertical (azimuth and
elevation)—suffice for practically all applications. The main-lobe pattern in other planes
can usually be adequately estimated from these principal-plane patterns. However, if the
detailed side-lobe patterns are of concern, as they may be in some radar applications and
in other special cases, oblique-plane patterns will be of interest, for the side lobes in these
planes cannot be inferred from the principal-plane patterns.
9.1.1. Radiating Near- and Far-Field Patterns

Figure 9–2 is a far-zone pattern of a paraboloidal reflector antenna, taken from a classic
measurements paper (Cutler, King, and Kock 1947). There one can see the wide-angle
lobe structure, which was attainable through careful measurement. Factors that affect,
generally, the lobe structure for a reflector antenna include:
• limitations of feed pattern amplitude and phase
• effects of aperture blocking by the feed
• feed pattern spillover from the reflector edges and effects of edge diffraction
• reflector surface inaccuracies
As discussed in chapters 5 and 6, strengths of close-in minor lobes can be controlled
by an antenna’s aperture amplitude and phase distribution. The vestigial (or shoulder)
lobe in Fig. 9–2 is a visible part of a first minor lobe (adjacent to major lobe) that remains
after the first null is destroyed by aperture phase error. Typically, with reflector antennas,
0
RELATIVE RADIATION INTENSITY IN DB
MAJOR LOBE
–10
–20
VESTIGIAL LOBE WIDE ANGLE
RADIATION
MINOR LOBES REARWARD
–30
RADIATION
–40
–50
20 0 20 40 60 80 100 120 140 160 180
DEGREES OFF AXIS
FIGURE 9–2.
Measured wide-angle far-zone amplitude (in dB) pattern of a reflector antenna. From Cutler,
King, and Kock (1947); © 1947, IEEE.
vestigial lobes occur because of a misaligned feed. Although patterns that include the
major lobe and close-in minor lobes may be calculated, ordinarily it is impossible to
precisely calculate the effects of wide-angle and rearward radiation caused by such
factors as aperture blockage, spillover, and edge diffraction. Obviously, effects of random
mechanical errors may be computed (accounted for) only in a statistical sense.
The radiating near- and far-field patterns of the close-in minor lobes, including the
vestigial lobe, may be calculated following the discussions of Appendix G. In addition,
the accompanying website includes example pattern calculations using Mathcad soft-
ware. These calculations include the more easily defined effects of aperture amplitude
and phase on the major lobe and close-in side lobes. However, measurements are required
in order to determine effects of aperture blocking, feed pattern spillover, edge diffraction,
and reflector surface inaccuracies. In other words, calculations cannot determine all of
the factors that affect side lobes, and thus the final appraisal of antenna performance
usually must be made based on measurements.
Although far zone patterns are nearly always needed, sometimes it is important to
measure and understand field behavior in the radiating near-field. Results of calculations
for the field distributions near an aperture are now discussed, where the phase and ampli-
tude of the field are strongly influenced by distance to the observation point. As a con-
sequence, Figs. 9–3 and 9–4 that follow show major effects of radiation from an aperture,
per se; that is, effects of edge radiation, spillover, and random phase errors are neglected.
Thus, the figures show only how reduced separation may affect amplitudes near the center
of radiation patterns. The calculations were made by following the methodology used in
the supplementary materials SM 4.4 and SM 4.5.
Antenna Patterns, General 375
Figure 9–3 shows how the constant K, 0
Amplitude in Decibels
where the observation point is separated –10
a distance K(D2/l) from the aperture –20
center, affects pattern shape. The reader –30
will recall that, from the rule of thumb of –40
sec. 3.2.5, the far-radiating field for elec-
–50
trically large antennas exists where K is
–60
two or greater. The effect of increased –10 –5 0 5 10
phase error, with reduced K (i.e., dis- Angle from Broadside in Degrees
tance), on filling the first pattern null may K approaches infinity
be seen. This null-filling effect due to K=1
K = 0.2
phase error is closely related to the cre-
FIGURE 9–3.
ation of the vestigial lobe of Fig. 9–2.
Figure 9–4 shows an entirely different Calculated radiation for three separation
distances K(2D2/l) versus angle from
aspect of separation distance than does
broadside.
Fig. 9–3. It shows how the amplitude of
the radiated field, near an aperture, varies
1
as a function of position along the
Field Strength (dB)
0.5
aperture. It depicts the near-field region
where radiation approximates a nearly 0
uniform plane wave, as conceptualized by −0.5

Fig. G–1 of Appendix G. Existence of this −1
approximation to a plane-wave field −1.5
permits, for antennas within this radiating −2
near-field, the measurement of far-field −2 −1.5 −1 −0.5 0 0.5 1 1.5 2
patterns within a compact range (see Aperture Coordinate (meters)
sec. 9.3). FIGURE 9–4.

Figures 9–3 and 9–4 were made from Relative field strength in dB along the aperture
calculations for a linear array of 66 iso- and separated by 6 meters from the array. The
tropic elements, with element spacing of irregular and smooth field strength variations
l /2 and with l = 0.1 m (3 GHz). The are caused by the equal and cosine aperture
amplitude distributions, respectively.
calculation for Fig. 9–3 assumed a cosine
amplitude aperture distribution, and for
Fig. 9–4 both a constant and a cosine amplitude distribution were assumed.
Figure 9–3 includes field distributions for three different separation distances. It can
be seen that the pattern nulls are deepest if the observation distance approaches infinity.
By reducing the distance to 2(D2/l), the usual rule of thumb for approximating far-field
conditions, there is no apparent change in the amplitude peaks, but the nulls near the
major lobe are less deep. For the 0.2(D2/l) distance, close-in nulls are no longer appar-
ent, and the rapid amplitude change versus q of the major lobe has been destroyed and
the beamwidth is increased. Then, from this example, it is seen that although the 2(D2/l)
distance permits retention of most of the pattern detail, some is lost. Thus, it should be
apparent that if accurate measurements of deep nulls and very low side lobes are needed,
observation distances greater than 2(D2/l) may be required.
Figure 9-4 shows calculated radiating near fields along a line parallel to and separated
from the aperture by 6 meters. The aperture width and wavelength are 3.25m and 0.1m,
giving about 211m for 2D2/λ. Clearly the 6-meter separation is electromagnetically very
near the aperture. The close-in field within a compact range, called the quiet zone, is the
region that is used to place a test antenna for measuring its radiating far-field patterns
(see sec. 9.3). There are two field strength curves in Fig. 9-4; one for which the element
amplitudes are equal, and the other for a cosine amplitude distribution. Near the aperture
center, the variation in field strength for the cosine tapered aperture is small. There,
however, for the array with equal amplitude elements, the variation is about 2 dB. These
differences in field strength variations near the aperture center underscore the fact that
aperture amplitude tapering of compact range antennas will improve the uniformity of
amplitude within the quiet zone.
It is appropriate to mention that the above-mentioned compact range quiet zone,
although being close to its radiating source, is at a distance great enough so that the
magnitude of the reactive near-field is negligible. In sec. 3.2.5, it is stated that l /2p is
the generally accepted boundary between the reactive and the radiating near-fields for
antennas that are short compared to l, yet there is not a commonly accepted guideline
for larger antennas. According to Yaghjian (1986, p. 33), from experience with near-field
probe measurements on aperture antennas, a distance of “l or so” appears reasonable as
an outer boundary for the reactive near-field. More specifically, according to E. B. Joy,
at a distance of 4 or 5l from a typical antenna under test (AUT), the reactive near-field
is less than 100 dB below the radiating near-field. This guide to relative strengths of
reactive and radiating near-fields was attained through personal communications with E.
B. Joy, and it is reportedly based on studies by him that follow equations (10) and (11)
of Joy and Paris (1972).
The near-field measurement technique (sec. 9.4) is entirely different from a compact
range measurement. It uses a sampling probe to measure the near-field radiation on a
planar, cylindrical, or spherical surface enclosing the antenna-under-test (AUT), and the
results are used to accurately determine far-field patterns of the AUT. To accomplish this,
detailed computer calculations are made based on carefully measured point-to-point
amplitude and phase measurements over the near-field measurement surface. The mea-
surements are made near the aperture, but ordinarily at a great enough distance so that
the reactive near-field is negligible, typically four to five wavelengths from the AUT.
Preferably, the measurements are made with a small sampling probe, to minimize probe-
aperture multiple reflections. In addition, the computer calculations commonly include
corrections to compensate for the probe pattern, gain, and polarization.
9.1.2. Pattern Statistics: Near- and Far-Field Statistical Gain

Accurate and detailed antenna patterns of the type acquired under controlled measure-
ment conditions are not always required. An example case is where interference from an
antenna is to be predicted. Recall from sec. 9.1.1 that the shapes of major lobes and the
side lobes depend on observation distance, when it is less than approximately 2D2/l. In
other words, the details of patterns are range dependent within the radiating near-field
Far-Zone Pattern Measurements 377
region. Yet, interference levels are gen- 30
GAIN (RELATIVE TO ISOTROPIC) db

AN/SPS-10
erally strongest between closely located 20
systems, for which separation distance
and observation angles are often unknown. 10
In addition, the impracticability of using
0
detailed patterns to predict susceptibility
to interference is compounded because of –10
minor-lobe structure changes with fre-
–20
quency and site configuration.
It is now known, from measurements, –30
that the statistics of patterns within either
–40
the radiating-near- or far-field regions can
°
4°
8°
°
°
0
°
°
8°
4°
0°
0
72
36
36
72
18
14
10
10
14
18
be useful for predicting radio-frequency
interference. This occurs because the FIGURE 9–5.
statistical distributions of antenna pat- 360 degree pattern through azimuth plane of
terns taken over wide angles are nearly AN/SPS-10 antenna. From Cain and Byers
(1968); © 1968, IEEE.
invariant to observation distance, whether
within either the near- or far-fields. In
other words, for a given antenna, the probability of its gain at any angle exceeding a
given value is dominated principally by the radiation patterns over a very wide range of
observation angles, almost exclusive of the major lobe or close-in minor (Johnson 1963;
Long 1960).
From the discussion above, the concept of statistical gain has meaning, if there is a
wide range of unknown but likely observation angles. In other words, the likelihood of
gain exceeding a given level is a useful means for estimating the susceptibility to inter-
ference, especially for narrow-beamwidth antennas. In fact, the statistics of radiation
patterns are useful for predicting near- and far-field interference from antennas, at in-band
or out of-band frequencies, and from antennas in the presence of interfering objects (Cain
and Byers 1969; Cain, Cofer, and Ecker 1970; Cain, Ryan, and Cown 1972; Cain,
Weaver, and Duffy 1974).
Figures 9–5 and 9–6 show results of gain measurements on the AN/SPS-10 radar
antenna (C-band). From Fig. 9–5, the gain versus angle of the antenna is above isotropic
(0 dB) for only about 5.3 degrees. Thus, the probability that the gain is less than isotropic
is 0.98, as illustrated in Fig. 9–6. Similarly, one can see from Fig. 9–5 that the gain is
less than −10 dB for more than half the angles. This is illustrated in Fig. 9–6, where the
probability is 0.74 that the gain is less than −10 dB. Although the peak gain of the antenna
is almost 30 dB, the median gain (the gain at which the cumulative probability is 0.5) is
only −13 dB. Thus, there is a significant difference between median gain and average
gain. (Average gain for a lossless antenna is unity, i.e., 0 dB).
9.2. Far-Zone Pattern Measurements

The measurement of a pattern involves two antennas: the one whose pattern is being
measured, and another some distance away. One antenna transmits (radiates) and the
1.0
PROBABILITY THAT THE GAIN IS LESS
0.8
THAN THE ABSCISSA
0.6
0.4
0.2
0.0
–50 –40 –30 –20 –10 0 +10 +20 +30 +40
GAIN (dB)
FIGURE 9–6.
Cumulative gain distribution in azimuth plane for AN/SPS-10 antenna. From Cain and Byers
(1968); © 1968, IEEE.
other receives. Because of the reciprocity principle, the antenna whose pattern is being
measured can be either the transmitting or the receiving member of the pair. The mea-
sured pattern will be the same in either case. In the following discussion the antenna
whose pattern is being measured will be called the antenna under test (AUT), and the
one used as the other terminal of the transmit-receive path will be called the range
antenna, regardless of which one transmits and which receives.
It is essential that the range antenna have the same polarization as the AUT antenna.
Ideally, the polarization of the range antenna should be controllably variable, so that the
pattern of the AUT antenna can be investigated for all polarizations (see Appendix B).
This is especially important in applications where very low side lobes are required,
because side lobes sometime have polarization different from that of the major lobe.
Typically, antennas transmit or receive electromagnetic waves over long distances.
Thus, knowledge of antenna characteristics with antennas in the radiating far-field is
needed, which is accomplished by placing the antennas under test in a plane wave of
uniform amplitude, phase, and polarization. The traditional method employs a free space
range, that is, one for which the separation between the AUT and range antennas meets
or exceeds the far-field requirements previously outlined.
The needed separation distances are generally large and often require the measure-
ments be made in out-of-doors environments. Thus, a major concern is the need to sup-
press unwanted reflections from the ground and nearby surrounding objects, even for
narrow-beam antennas. Typical free-space range geometries minimize effects of unwanted
reflections by using elevated transmit and receive sites, slanting the propagation path, or
by housing the range in an anechoic chamber lined with absorbing material.
Hollis, Lyon, and Clayton (1970) and IEEE (1979) include comprehensive discussions
regarding far-zone ranges. Methods that permit smaller separation distances between
antennas and thereby permit indoor pattern measurements include compact (sec. 9.3),
near-field (sec. 9.4), and scale model (sec. 9.6) ranges.
9.2.1. Low-Frequency Techniques

At the very lowest frequencies, where an electrically short vertical antenna is almost
universally used for transmitting (as in AM radio broadcasting), the pattern in the hori-
zontal plane is virtually uniform, and ordinarily it need not to be measured. Occasionally
two or more towers may be used in an array to create a null in a certain direction or
directions, and broad maxima in other directions. To measure such a low-frequency array,
its beam is held fixed in pointing direction, while the range antenna is transported around
it in a circular path at a constant distance. The range antenna, if its beam is directional,
is kept aimed at the AUT antenna so that only the AUT antenna pattern will affect the
result. The mobile antenna may be mounted on a panel truck, a helicopter, or an airplane.
Or, if the AUT antenna is on a ship or an island, the range antenna may be located on a
ship or boat.
If an absolute field-strength pattern is desired, it may be obtained by measuring the
voltage output of the range antenna. This voltage is then compared with the output
voltage of a calibrated signal generator to introduce a comparison signal into the receiver,
and adjusting its output to give the same voltage as obtained from the range antenna. A
similar procedure is used for relative-pattern measurement, except that it is unnecessary
to measure the absolute value of the range antenna output voltage. The receiver used in
such measurements must be very gain-stable. When the measurements are made over
land, the ground conductivity and the presence of nearby reflecting objects will affect
the results. Thus, it is customary to make measurements at more positions than would
ordinarily be required, to “smooth” the resulting pattern for obtaining an average of the
measurements made in approximately the same observation direction.
An important consideration in pattern measurement is to provide adequate separation
of the AUT and range antennas, to insure measurement of the far-field pattern. As dis-
cussed previously, a rule-of-thumb formula is that the separation must be at least
2D2/l where D is the maximum antenna dimension as viewed from the measurement
point and l is the wavelength. If, however, D is less than 0.707 l, a minimum permissible
separation is the distance of one wavelength. This second requirement is more likely
to be applicable to very low-frequency antennas, because their wavelengths are
very long.
9.2.2. High-Frequency Techniques

At frequencies above about 100 MHz, an antenna pattern is customarily obtained by
rotating the AUT antenna. This may occasionally be done at lower frequencies, and the
“low-frequency” procedure may sometimes be used up to considerably higher frequen-
cies. It is then especially important to make the field-strength measurements at points
Large reflecting object

(e.g., water tower)
n
tio
ec
dir
am
Be
Antenna
under
test Range
a b
antenna
Sidelobe Beam direction

being measured
FIGURE 9–7.
Effect of nearby reflecting object on pattern measurement.
that are in the clear—not too close to large buildings or power and telephone lines, for
instance.
When the pattern is to be measured by rotating the AUT antenna, both antennas should
be located so that they have an unobstructed view of each other, and also have the
required separation to insure a far-field measurement. A further requirement is that the
area between the antennas be clear of sizable reflecting objects, not only in the direct
line between them but for an appreciable distance on both sides. This requirement is
important if an accurate measurement of low-amplitude side lobes is to be made. Figure
9–7 illustrates the reason.
The range antenna is indicated to be the transmitting antenna, which is the customary
arrangement because it permits all the measurements (direction and signal strength) to
be made at a single location, that is, at the AUT antenna. If a large reflecting object, as
shown, is illuminated by the range antenna (whose beamwidth is assumed to be greater
than 2b ), some signal will be reflected toward the AUT antenna, arriving at the angle a
off the in-line direction between the two antennas. (The AUT antenna beamwidth is
assumed to be less than 2a.) This signal will be considerably less than the in-line signal
and will likely not seriously affect the measurement in the main lobe of the AUT antenna,
when the reflected signal will be received in the side-lobe portion of the AUT antenna
pattern. When the AUT antenna is rotated to allow measurement of its side-lobe pattern
at the angle a off axis, however, its main lobe will point directly at the reflecting object,
as shown in the diagram. Then the reflected signal received in this way may be compa-
rable to or even in excess of the in-line signal received in the side lobe, so that a consid-
erably erroneous measurement will result.
This effect is minimized by using a range antenna that is fairly directional, with high
gain in the direction of the AUT antenna and considerably reduced gain—or perhaps
even a null—in the direction of the reflecting object. However, this requires that the range
antenna be quite sizable and expensive. If there is just one major reflecting object in a
troublesome position, a null of the range antenna may be directed toward the object, if
this is possible without too greatly reducing the radiation in the desired direction or giving
it an incorrect polarization. As already noted, ideally, the polarization of the range antenna
should be controllably variable, so that the pattern of the AUT antenna can be investigated
for all polarizations. This is because the side lobes may have different polarization
properties than those of the main lobe.
Other possible remedies for the reflecting-object problem exist. One is to interpose
absorbing material (available commercially) between the range antenna and the object,
or between the object and the AUT antenna. Another is to erect a reflecting barrier that
will intercept the radiation going to or from the object and reflect it in some harmless
direction. This barrier is usually a flat sheet of solid metal or mesh material set at an
angle that will direct the reflected waves away from the AUT antenna.
9.2.3. Far-Field Ranges

When antenna measurements at far-field distances or greater are required on a regular
basis, it is customary to set up a far-field range for this purpose, consisting of a large
clear area with facilities at opposite ends for operating the AUT and range antennas. The
separation of the terminals of the range must of course be sufficient to satisfy the far-field
criteria, which depend on the aperture sizes and frequencies of the antennas to be
measured.
Potential problems and solutions relating to undesirable reflections from objects near
the pattern range have already been discussed above. A further problem in pattern mea-
surement is the reflection that occurs from the surface of the ground between the two
antennas. Specifically, waves propagate between the range and AUT antennas by two or
more paths: directly through the atmosphere and by reflections from various objects,
including the earth. Field strength at a point is affected by the interference (complex-
vector sum) of the direct and reflected waves. This interference phenomenon is called
multipath propagation. Appendix D discusses forward reflections from land and water
surfaces and multipath propagation. The discussion there pertains to reflections from
surfaces with various rms surface roughnesses, but for which the mean surface at the
reflection area is flat. For a perfectly smooth and flat surface, the beamwidths after reflec-
tion in both elevation and azimuth are the same as prior to reflection. However, the
surface roughness reduces the effective reflection coefficient for forward reflections and
it broadens the reflected beamwidths.
The multipath phenomenon results in an interference lobe pattern of the range antenna
radiation at the AUT antenna (see Fig. D–5, Appendix D). If this interference pattern is
such that there is appreciable variation of the incident field across the vertical aperture
of the AUT antenna, its pattern will not be correctly measured. On the other hand, if the
resulting interference pattern is such that the variation is slight within the AUT antenna
aperture, the resulting error will be insignificant. Effects of multipath may be calculated
by the methods described in Appendix D.
A special type of far-zone reflection range is one for which the site geometry permits
reflected waves from the ground to add constructively with waves that are propagated
directly between antennas. For this case (Hemming and Heaton 1973), the use of hori-
zontal polarization is preferred to vertical because ground reflections are less sensitive
to the reflection angle at the air-ground interface.
In general, the mean surface of terrain undulates and consequently there may also be
increased beamwidth broadening caused by terrain slope changes at areas of surface
reflections. Then, a simple multipath calculation may not be feasible. When adequate
terrain flatness is not present, a possible solution to the problem is to set up reflecting
“fences” at appropriate points to intercept the waves that would otherwise result in an
interference pattern, and reflect them in harmless directions. Or the terrain may be delib-
erately roughened to such an extent that no appreciable, specular reflection will occur.
Another option is to select a site so that a “valley” exists between the antenna locations;
then, if the range antenna has a fairly narrow vertical-plane beamwidth, there will be no
appreciable reflection from the “floor of the valley.” In an industrial or built-up area this
effect may be achieved by locating the AUT and range antennas on the roofs of tall
buildings separated by relatively clear areas, or on tall towers. In rural areas, two hilltops
may be used.
The foregoing discussion has been concerned with horizontal (i.e., azimuth) plane
pattern measurement. Elevation-angle patterns are virtually impossible to measure in the
presence of a reflection-interference lobe structure. One method of circumventing this
difficulty is to turn the AUT antenna on its side so that its “vertical plane” pattern can
actually be measured in a horizontal plane. The vertical patterns of antennas that can be
tilted upward can be measured from a helicopter or an airplane, if some method of mea-
suring the aircraft direction angle and range can be provided. Use of a precision tracking
radar or a global position system (GPS) may be ideal for this purpose.
9.3. Compact Ranges

The compact range provides a method of acquiring far-field patterns without a far-field
pattern range. In a compact range, patented by Johnson (1967), the test antenna is placed
near the aperture of a larger reflector antenna that creates a plane wave (Hess and Johnson
1982; Hickman and Johnson 1972; Johnson, Ecker, and Moore 1969). In this technique,
AUT is located in a collimated beam of the radiating near-field (called the quiet zone),
thus simulating far-field conditions and permitting a direct measurement of the far-field
pattern.
The compact range technique is useful for measurements of small- to medium-size
antennas. (In contradiction, Fig. 9–11 shows a huge compact range used for measuring
patterns from large and heavy structures). Johnson, Ecker, and Hollis (1973) include in-
depth discussions of the advantages and disadvantages. A disadvantage is that the range
reflector must be large compared to the antenna under test, and thus the technique is
usually impractical for measurements on very large antennas. Moreover, the surface-
accuracy required of the range reflector may be critical at frequencies above about
40 GHz. Obviously, basic advantages are that a measurement of the far-field pattern can
be made at short distances and indoors, with no more data processing required than for
far-field measurements.
Compact Ranges 383
Typically, compact ranges are located ROLLED

ABSORBING
EDGE
indoors, where they are not subjected to PANEL
adverse weather. Traditionally, they are in TEST
shielded, absorber-lined rooms that reduce ANTENNA
CUT
interference from nearby walls and other PARABOLOIDAL
objects. The shielded chamber also makes ABSORBER REFLECTOR
the compact range free from outside inter- FEED
fering signals and secure from outside
monitoring. Applications include antenna FIGURE 9–8.
pattern measurements, antenna imped- Sketch of a point-source compact range. From
ance measurements, gain and polarization Johnson, Ecker, and Hollis (1973); © 1973,
comparisons, boresight measurements, IEEE.
radar reflectivity measurements, and
illumination of animals and humans for
investigations of the biological effects of
microwave radiation.
Figure 9–8 is a sketch of a compact
range. One can see that the feed for the
paraboloidal reflector is below the turn-
table on which the test antenna is mounted.
The compact range reflector edges were
rolled to reduce the intensity of edge dif-
fracted waves toward the test antenna, in
an effort to minimize amplitude varia-
tions within the quiet zone. Note that the
test antenna is necessarily substantially
smaller than the compact range reflector.
Absorbing material used for reducing FIGURE 9–9.
interface reflections from the turntable An experimental compact range used to
and a nearby wall are also shown. investigate diffraction effects of reflector edges
The reflectors of some compact ranges (courtesy of Gerald Hickman).
have been offset parabolic cylinders and
others offset paraboloids. Figure 9–9 is a photograph of one of the experimental compact
ranges developed by Scientific Atlanta in the 1970s. In this figure one can see the absorb-
ing material along the walls and at the reflector base used to reduce stray reflections near
the facility. Also visible is a paraboloidal reflector antenna, which is mounted on a posi-
tioner and is being tested. According to Gerald Hickman (now with Electromagnetic
Sciences), experience showed that the serrations on the sides controlled edge diffraction;
however, the small rolled top and the straight bottom edge did not. Accurate surface tol-
erances were obtained by reflector shaping, accomplished with a special milling machine
developed for that purpose. Consequently, the compact range reflectors developed later
and produced for sale included serrations along all edges.
Figure 9–10 shows a compact range facility, a model currently available through MI
Technologies, a company that acquired the antenna instrumentation activities of Scientific
Atlanta. There one can see a large reflector

having serrated edges, the surrounding
walls and floor covered with absorbing
material, and a microwave parabolic
antenna with horn feed mounted on an
azimuth-over-elevation positioner. Some
of the features advertised include fre-
quency coverage from 1 to 110 GHz,
unique serrated reflector edges that mini-
mize edge diffraction effects, a precision-
machined monolithic reflector surface,
FIGURE 9–10.
and temperature-stable structures.
A compact range that includes a reflector with Figure 9–11 shows a 75-foot (22.9 m)
serrated edges, and a microwave parabolic
antenna on an azimuth-over-elevation outdoor compact range, located at the
positioner (courtesy MI Technologies). U.S. Army Electronic Proving Ground,
Ft. Huachuca, Arizona, that was designed
and developed by the Georgia Tech
Research Institute (Cotton 1988). It was
built to measure the patterns of antennas
on vehicles, including tanks and aircraft,
with dimensions up to 50 feet (15.2 m).
This compact range was developed for
initial operations from 6 to 40 GHz, but
the reflector offers the possibility for
extending to even higher frequencies.
The reflector is a section of the upper
half of a 150-foot focal-length parabola,
and it is positioned so that the parabolic
axis is horizontal. The parabola’s vertex
and focal point (and thus the feed posi-
tion) are at ground level, which is 2.5 feet
below the bottom edge of the reflector.
Only five especially designed wide-
bandwidth, low-cross-polarization, con-
stant-phase center feed horns are needed
FIGURE 9–11. to cover the 6- to 40-GHz range of fre-
Patterns of antennas mounted on an aircraft quencies (Bodnar 1986).
being measured in a compact range having a Reflector surface accuracy and edge
75-foot (approximately 23 m) diameter treatment are two of the principal con-
(courtesy of Henry P. Cotton). cerns in compact range design. The reflec-
tor surface consists of aluminum panels
assembled onto a rigid back structure, and the design specifications include panel surface
accuracy of 0.006 inch (about 1.5 × 10−6 m) rms. The reflector faces north, with the
backside shielded to minimize thermal distortions caused by the sun. The edge treatment
Near-Field Antenna Measurements 385
is used to minimize diffraction from the reflector edge for improving the uniformity of
the field across the aperture. Shaped serrated reflector edges are used that assisted in
providing quiet zone performance suitable for measuring antenna/vehicle radiation with
vehicle dimensions up to fifty feet (Joy and Wilson 1987).
Another major design problem for large compact ranges, when the test items are heavy,
is the AUT positioner, which is located approximately above the focal point. With a
compact range, the item being measured must be within the quiet zone, which is centered
roughly 40 feet (12.2 m) above ground. A positioner had to be designed and built, because
a suitable one was not available commercially. The positioner, with its base on the
ground, is designed to handle weights up to 140,000 pounds (69,300 kg), at heights
exceeding 40 feet, and move the test item from −1 to +91 degrees in elevation and con-
tinuously in azimuth.
A relatively new electronic timing means appears to have much promise for improving
quiet zone performance, by reducing the reflections from the compact range reflector
edges and from surrounding objects. The technique is, in essence, a short-pulse-length
radar range gating technique (Chang, Liao, and Wu 2004). Although additional electronic
instrumentation is required, radar range gating (also called impulse time-domain mea-
surements) may offer improved quiet zone performance as well as simplifications and
cost reductions in the traditionally used methods of reflector edge treatment and anechoic
chamber construction.
9.4. Near-Field Antenna Measurements*

Conventional far-field antenna measurements are made when the source antenna and the
antenna-under-test (AUT) are separated by the far-field criteria, which depend on D, the
diameter of the AUT and l, the wavelength of operation. The required separation can
be very large when testing large antennas at short wavelengths. In contrast, near-field
antenna measurements can be made with a separation of only a few wavelengths resulting
in a significantly smaller measurement facility (Johnson, Ecker, and Hollis 1973; Yaghjian
1986). The reduction in the size of the measurement facility, the high accuracy achieved,
and the ability to perform the measurements indoors has led to the popularity of near-field
antenna measurements. Near-field measurements require:
a. Use of a probe antenna to measure the phase and amplitude of two vector
components of the near-field of the AUT over a surface surrounding the
AUT
b. Characterization of the far-field pattern of the probe in phase, amplitude and
polarization
c. Application of well defined mathematical techniques to:
i. Remove the effects of the probe pattern, gain, and polarization from the
measured data (called probe correction)
ii. Transform the corrected near-field data and obtain the far-field of the
AUT
* This section contributed by Donald G. Bodnar, MI Technologies, Inc., Atlanta, Georgia.
Y The near-field to far-field transforma-

AN tion can be accomplished mathematically
SC ACE
R F in three coordinate systems, namely
SU
AUT PROBE planar, cylindrical, and spherical. Hence,
measurement systems for these three
Z geometries have been developed and are
discussed next.
X
9.4.1. Planar Near-Field

FIGURE 9–12.
Measurements
Typical planar near-field measured geometry
(courtesy MI Technologies). The theory of near-field antenna measure-
ments was first developed by Kerns (1981)
for a planar near-field (PNF) geometry
that is shown in Figure 9–12. The probe is moved to equally spaced positions in x and
in y on a planar surface located in front of the antenna. The movement of the probe is
accomplished using a mechanical positioner under computer control (Joy and Paris
1972). The phase and amplitude of the probe voltage as measured by a microwave
receiver is recorded by the computer at each x-y point. The process is typically repeated
for polarization of the probe oriented vertically and again for probe polarization oriented
horizontally. In this manner, the tangential component of the electric field on the mea-
surement plane is obtained. However, this measured near-field is distorted by the pattern
of the probe and this distortion must be removed to obtain the true near-field of the AUT.
The process of removing the probe distortion is called probe correction or probe com-
pensation and involves either mathematically calculating the complete far-field of the
probe or measuring its far-field pattern in amplitude, phase, and polarization at the fre-
quency of operation using a separate measurement. The true AUT near-field is obtained
from the measured AUT near-field and the probe far-field using the equations of Kerns.
Finally, a two-dimensional Fourier transform is performed to obtain the far-field of the
AUT from the corrected near-field data.
Planar near-field measurements are especially applicable to narrow-beam antennas
since the scan surface must be large enough to collect the majority of the energy radiated
by the antenna. Thus, horn antennas and reflector antennas are often measured using a
planar near-field scanner. In contrast, a dipole antenna cannot be measured with a planar
near-field measurement system since a great deal of energy is radiated in directions other
than through the scan surface.
A scanner for moving the probe over a planar surface and sampling the near-field
in equal increments, Dx, in x and equal increments, Dy, in y is shown in Figure 9–13.
MI Technologies has developed accurate and high-speed planar near-field scanners
and the associated high-speed microwave receivers that are needed for measuring
phased array antennas, which have many electronically switched beam-pointing
directions and many frequencies of operation. High-speed data acquisition is very
important in these measurements due to the massive amount of data that must be
collected.
Near-Field Antenna Measurements 387
9.4.2. Cylindrical Near-Field

Measurements
The measurement geometry for cylindri-
cal near-field measurements is shown in
Figure 9–14 and the technique is dis-
cussed by Leach and Paris (1973). The
mathematical formulation employs cylin-
drical wave modes instead of plane waves
as is required in planar near-field mea-
surements. However, the formulation of
Kerns (1981) also applies to the cylindri-
cal geometry. Measurements are made in
equal increments, Dy, in y, and in equal
increments, Dq, in q measured in the x-z
plane. A typical scanner for this geometry
is shown in Figure 9–15. Cylindrical mea-
surements systems are well suited for fan FIGURE 9–13.
beam antennas such as cellular telephone Planar near-field (PNF) scanner consisting of a
base station antennas as well as pencil vertical traveling tower over a horizontal slide
(courtesy MI Technologies).
beam antennas. The technique is not
applicable to antennas that have a great
deal of radiation straight up and down Y
since the scanner does not capture this SCAN

energy. Economical and flexible scanner SURFACE
designs have been developed by MI
Technologies, which are required for
price sensitive wireless antenna applica- PROBE
tions (Pearson, Wang, Bodnar, and Tian
2007).
Z
9.4.3. Spherical Near-Field
Measurements
Spherical near-field measurements -X
(Hansen 1988; Leach and Paris 1973)
measure the near-field on the surface of a FIGURE 9–14.
sphere that surrounds the antenna as illus- Typical cylindrical near-field measured
trated in Figure 9–16. Data is collected at geometry (courtesy MI Technologies).
equal increments in q and j (the standard
spherical coordinate angles). This data is
then probe corrected and transformed to the far-field to obtain the far-field of the antenna
under test. Any antenna can be measured with a spherical system (e.g., a low-gain, broad-
beam antenna as well as fan-beam and pencil-beam antennas) since all of the energy
radiated by the antenna is captured by the

measurements (Hansen 1988). A scanner
implementation for spherical measure-
ments is shown in Figure 9–17. The
azimuth scanner rotates the AUT in
azimuth while the arch moves the probe
along a circular arc in elevation. This
combination of positioners allows near-
field measurements over a sphere. Posi-
tioner accuracy and positioner alignment
are critical issues in spherical near-field
measurements especially at microwave
and millimeter wavelengths and MI Tech-
nologies has developed techniques for
handling both of these issues.
9.5. Polarization Measurement

Polarization theory is discussed in Appen-
dix B. The polarization of the radiated
field from an antenna may be measured
by measuring the received signal voltage
with a linearly polarized receiving-range
antenna as its polarization is rotated
through 360 degrees. If two maxima and
two nulls are observed, the field is lin-
early polarized in the direction corre-
sponding to the maxima. (The maxima
will be 180 degrees apart, and the nulls
will be 90 degrees from the maxima. The
direction of the nulls can be measured
FIGURE 9–15. more accurately than that of the
maxima.)
Cylindrical near-field system employing a
vertical scanner and an azimuth turntable If maxima and minima (rather than
(courtesy MI Technologies). nulls, or zeros) are observed, the field is
elliptically polarized, and the ratio of the
maximum to the minimum field intensity
is called the polarization ratio or ellipticity. When this quantity is measured along the
axis of the main beam of the antenna under test, it is called the axial ratio.
When the field intensity is constant as the range antenna polarization is rotated, the
field is circularly polarized. (Circular polarization is of course just the special case of
elliptical polarization for which the polarization ratio is one.)
Commonly used linearly polarized range antennas include half-wave dipoles and
waveguide horns. In fact, any linearly polarized antenna will serve. Polarization measure-
ments may also be made using two fixed receiving antennas linearly polarized at right
Scale-Model Measurements 389
angles to each other, and measuring the Y

ratio of the intensities received by each.
If the field is elliptically polarized, the PROBE
phase difference of the signals in the two
antennas must be measured. For an illus-
tration, see Kraus (1988, p. 838).
9.6. Scale-Model Z
Measurements
Because of the practical difficulties of
-X
making measurements on large antennas,
and the expense of constructing experi-
mental designs of large size, the scale-
model technique is very useful in antenna FIGURE 9–16.
development and research work. The Typical spherical near-field measured geometry
principle of the scale model is that if two (courtesy MI Technologies).
antennas are of exactly the same shape
but differ in size by a scale factor S, their
electromagnetic behavior will be the same
if the smaller antenna is operated at a
frequency S times as great as that at which
the larger antenna is operated.
This principle is self-evident for simple
antennas such as a half-wave dipole, but
in fact it holds for all types of antennas.
For example, if two paraboloidal reflec-
tors are of exactly the same form (same
f/D ratio and same aperture shape) but one
is ten times the size of the other, and if
they have identically shaped feeds that
also differ in size by a 10 : 1 ratio, then if
the larger one is operated at 1 GHz and
the smaller one at 10 GHz, their operating
parameters will be identical. The gains,
beamwidths, side-lobe levels, and input
impedances will be the same; in fact the
patterns will be exactly alike. Also, the
percentage bandwidths will be the same.
The same would be true of a third antenna
five times as large as the larger of the first
two, operated at 200 MHz. Thus a large FIGURE 9–17.
antenna can be scaled down, or modeled, Spherical near-field antenna measurement
by any desired factor by operating it at a system employing an arch and azimuth
sufficiently high frequency. turntable (courtesy MI Technologies).
Frequencies in the region of 10 GHz are very popular for scale-model work, because
the scaled antennas in this region (known as X-band) are of very manageable size,
usually, and also the necessary apparatus is readily available at this frequency: wave-
guides and fittings, signal sources, receivers, and special measurement equipment.
However, some antennas if scaled to so high a frequency would be too tiny and delicate
for practical handling. For example, a half-wave dipole is only about 0.6 inch long.
Another popular frequency region, when an X-band model would be too small, is S-band
(the region of frequency around 3 GHz).
The scale-model principle is exact if the scaling is exact. Not only the size, but also
the conductivity of the model must be scaled, for exact results. The conductivity of the
model must be greater than that of the full-sized antenna by the factor S. However, if
the conductivities of both antennas are quite high, as is usually the case, and if the con-
ductors of the model do not become too minute, the conductivity scaling is not of much
importance. Models are sometimes silver-plated to increase their conductivity. If parts
of the antenna involve dielectric or magnetic materials, the sizes of these parts are scaled,
but the dielectric constants and the permeabilities are not; they should be of the same
values in both the model and the full-sized antenna; however, their losses, which often
depend on conductivity, must be scaled.
The exactness of scaling of the linear dimensions of the antenna is the most important
scaling factor. The scaling requirement applies to all dimensions, such as the thickness of
the walls of a waveguide horn, the openings of the mesh of a reflector surface, and even the
deviations from a true paraboloidal surface—all of these must be smaller in the model by
the factor 1/S, if the results are to be completely accurate. On the other hand, if only approxi-
mate results are needed, the scaling need not be applied to some of the details of the antenna,
if they are very small compared with the wavelength. Inexact scaling affects the accuracy
of impedance measurements more than it does pattern measurements and gain.
An advantage of the use of scale models is that it reduces not only the size of the
antenna but also the required size of a pattern range, or separation of AUT and range
antennas for any far-field measurements. This can be seen by considering the separation
criterion 2D2/l. Since D is decreased by the scale factor 1/S, and l is reduced by the
same factor, the net effect is to reduce the quantity 2D2/l by the factor 1/S.
Because of this reduction scale-model pattern and gain measurements are sometimes
made in an anechoic chamber, which is an enclosure whose walls, floors, and ceilings
are lined with a special material that absorbs radio waves almost completely. Thus no
reflections occur, and the room behaves as if it were free space. To make an enclosure
of this type of suitable dimensions for use with full-scale antennas in the VHF and UHF
regions would be prohibitively expensive. But for model antennas at S-band or X-band
or small full-sized antennas it becomes quite practical. The absorbent material is com-
mercially available.
9.7. Antenna Pattern Measurement Equipment

The basic subsystems of an automatic pattern measuring facility include a signal source,
receiver, AUT positioner, and pattern recorder. Figure 9–18 is a block diagram of a
Antenna Pattern Measurement Equipment 391
SIGNAL
TEST SOURCE SOURCE
ANTENNA ANTENNA
TEST
POLARIZATION
POSITIONER
POSITIONER
SOURCE
TOWER
POSITIONER TEST SOURCE SOURCE SOURCE

RECEIVER
CONTROL POSITIONER TOWER TOWER CONTROL
INDICATORS INDICATORS CONTROL
PATTERN
RECORDER
FIGURE 9–18.
Block Diagram of a Typical Antenna Measurement System. Source (IEEE 1979, p. 19);
© 1979, IEEE.
simplified antenna measurement system. At major pattern measurement facilities, the

subsystems function interoperatively, whereby phase and/or amplitude can be accurately
measured and automatically recorded as a function of the coordinate system of the
antenna under test (AUT). Usually, a central control system console is employed to allow
operator selection of such variables as transmit and receive polarizations, transmit fre-
quency, and range of measured angular coordinates.
For most operating frequencies of interest, most if not all the needed subsystems are
available commercially. For example, multiband network analyzers permit coverage over
the present practical limits of radio frequencies (tens of kHz to 100 GHz and even
higher). A network analyzer (sec. 9.11.2) may be used to provide any or all of the fol-
lowing: signal sources with rapid frequency change over wide ranges of frequencies,
receivers for tracking signals from signal sources, rapid phase and amplitude measure-
ments, automatic measurements and computer calculations using the measured results;
and multiple visual displays of operating parameters, measured, and calculated results.
Additionally, signal sources, receivers, and RF components may be procured separately;
and computers with their accessories can provide computations, visual displays, and
f ROTATIONAL AXIS
q ROTATIONAL
A AXIS
q ROTATIONAL Z
AXIS X f ROTATIONAL
AXIS
q f
O O
Y Z
A R S q
f
S
X
Y
Z A
f ROTATIONAL q ROTATIONAL
q ROTATIONAL q AXIS AXIS
Y O f ROTATIONAL
AXIS O S AXIS
X
A Z
S
(a) (b)
FIGURE 9–19.
Antenna Positioners: (a) azimuth-over-elevation positioner (b) elevation-over-azimuth
positioner. Source (IEEE 1979, p. 25); © 1979, IEEE.
printed recordings. Even complete, or near complete, measurement systems that include
a central control console may be purchased. Usually, personal computers are used to
control the automated measurement system and perform needed calculations.
The AUT positioner, its control system, and its position indicators constitute a key
subsystem, which must be selected based on the sizes and weights of the antennas to be
measured and on the types of tests to be made. The positioner must of course have the
needed movements and yet to be highly stable, handle the requisite weights, and with-
stand prevalent environmental conditions. Additionally, the position controls and indica-
tors must be accurate and reliable. A wide variety of positioner types, along with their
position controls and position indicators, are commercially available. Perhaps the sim-
plest positioners are those that rotate continuously in either azimuth or in elevation.
Figure 9–19 includes sketches of two often-used movements, namely azimuth-over-
elevation and elevation-over-azimuth.
Positioners are commercially available in various sizes and weight-handling capaci-
ties. Figure 9–20 shows a large positioner that includes a bearing having a diameter of
85 inches (about 2.2 m). According to Gerald Hickman, formerly of Scientific Atlanta,
this model was used for positioning an F-111 aircraft. Smaller, conceptually similar
positioners having platform diameters as small as about 6 inches (15 cm) are also
commercially available.
Directivity and Gain Measurements 393
9.8. Directivity and Gain

Measurements
Directivity and gain are discussed in
Section 3.3. Directivity D is a measure of
an antenna’s ability to concentrate radi-
ated power per unit solid angle in a certain
direction, and gain takes into account the
antenna efficiency as well as its direc-
tional properties. In principle, directivity
D can also be determined from the num-
erical integration of a complete far-field
pattern, obtained from detailed relative
strength measurements. In that case, abso-
lute calibrations are not needed, but gain
cannot be determined without knowledge
of the antenna losses. The integration
method is useful for antennas whose pat-
terns are simple, but it may be of limited
use for high-gain antennas with compli-
cated side-lobe or multilobed patterns, FIGURE 9–20.
because detailed measurement and inte- A large antenna positioner of the type used for
gration are required over the entire pattern positioning an F-111 aircraft (courtesy of
Gerald Hickman).
versus q and f of the antenna. It is inter-
esting to note that these measurements
must be performed with high relative accuracy, and directivity measurement does not
require an absolute calibration. However, gain that includes antenna losses is generally
of greater practical importance.
Gain is the ratio of the radiation intensity, in a given direction, to the radiation intensity
that would be obtained if the power accepted by the antenna were radiated by an isotrope.
Note that the definition of gain does not include losses resulting from impedance and
polarization mismatches. The lossless isotrope is assumed to radiate all of its input power,
but some of the power delivered to the actual antenna may be dissipated in ohmic resis-
tance (i.e., converted to heat). Thus, gain takes into account the antenna efficiency as
well as its directional properties. Consequently, the efficiency factor k is the ratio of the
power radiated by the antenna to the total input power. Therefore, the relationship
between gain G and directivity D is
G = kD (9–1)
where k ≤ 1 to account for dissipative (I2R) losses.
If the power input to (i.e., accepted by) the antenna is Pin,
• the power density pisotrope from the isotrope is Pin/4pR2 and
• the power density from the antenna pantenna is E2/377, where E is the rms value
of the electric field strength from the antenna.
Therefore,
E2
4π R 2 E 2 4π R 2 pantenna
G = 377 = = (9–2)
Pin 377 Pin Pin
4π R 2
Efficiency k is Pout/Pin, and thus from (9–2) the directivity can be expressed as
D= = (9–3)
377 Pout Pout
9.8.1. Absolute-Field-Strength Method

A direct method of gain measurement is based on (9–3) above, and it requires an absolute
measurement of the antenna input power Pin, and the electric field intensity E or power
density at distance R from the antenna. The measurement is, of course, made in the direc-
tion of maximum radiation. However, if this method is to give the gain of the antenna
itself, using (9–3), the measurement must be made under free-space propagation condi-
tions. In other words, the prevailing conditions must not include multipath interference
by the earth, or any other factor that may modify the free-space inverse-square law. On
the other hand, if the measurement is made using (9–2) with the antenna in its operating
location, the gain measured is the effective gain of the antenna in combination with its
environment. When earth reflection is involved, this measured gain will depend on the
elevation angle of the observation point (as indicated by Fig. 1–15, chapter 1), as well
as on the antenna height and the earth’s reflection coefficient.
If these factors are known or can be measured, the gain of the antenna by itself (i.e.,
as would be measured in free space) can be deduced. For example, sometimes there may
be enough information available to determine the reflection coefficient of the earth and
the propagation factor F, as defined by equation (1–50), chapter 1. Then if a value of
field intensity E is actually measured, one could calculate the gain for conditions of free-
space propagation. In terms of the factor F, (9–2) can be rewritten as (9–4) so that it
expresses the free-space gain of the antenna even if the field intensity E or the power
density is measured under non-free-space conditions:
G= = (9–4)
377 Pin F 2 Pin F 2
The absolute field intensity E can be measured at low frequencies by means of a

multiturn loop-receiving antenna, as described for small loops in sec. 4.5.1. At higher
frequencies at which a “small loop” is impractical, it is more convenient to make the
measurement in terms of the received power density pr. This quantity is related to the
receiving-antenna capture cross section (effective receiving area) Ar by the formula
Pr 4π Pr 4π Pr
pr = = = (9–5)
Ar kDr λ 2
Gr λ 2
where Pr, Dr, and Gr refer to the received power, directivity and gain of the receiving
antenna. (For a review of receiving cross section area, see chapter 3, sec. 3.4).
9.8.2. Standard Gain Antennas

A standard gain antenna is one whose gain is accurately known so that it can be used in
measurement of other antennas. Certain simple forms of antennas can be constructed to
have gain of known amount, based on previous calibrations or calculations. Pyramidal
horns (sec. 4.7) are commonly used at microwave frequencies as gain standards.
Alternatively, a standard antenna can be obtained by a gain measurement that
does not require an antenna of known gain. This method requires two identical
antennas. One is used as a transmitting antenna and the other for receiving. They are
separated by a distance R, and aimed at each other. The transmit antenna input power
Pin and the received power Pr are both measured. Gain in this case can be found easily
from the Friis transmission equation of sec. 3.4.2, which, with the two gains equal,
becomes
G2λ 2
Pr = Pin F2
(4π R ) 2
Then, solving for gain one obtains
4π R Pr
G= (9–6)
λF Pin
This procedure is likely to be most accurate when the propagation factor F is unity,
that is, under effectively free-space conditions with minimal earth-reflection interference
effects. Equation (9–6) can also be applied successfully under conditions that permit
accurate calculation of F, for example, when reflection occurs from a smooth water
surface located between the two antennas.
The gain calibration procedure described above, that uses (9–6), is known as the two-
antenna method. This method and the three-antenna method, to now be described, are
usually performed at a site where the ground and other reflections are negligible, so that
F is assumed to be unity. For the three-antenna method, three sets of two-antenna mea-
surements are performed with three antennas that may have different gains, designated
here as GA, GB, and GC. Then, by expressing the Friis equation in logarithmic form and
setting F = 1, three simultaneous linear equations are obtained that follow:
4π R ⎞ ⎛P ⎞
10 log G A + 10 log GB = 20 log ⎛⎜ ⎟ + 10 log ⎜ r ⎟
⎝ λ ⎠ ⎝ Pin ⎠ AB
4π R ⎞ ⎛P ⎞
10 log G A + 10 log GC = 20 log ⎛⎜ + 10 log ⎜ r ⎟
⎝ λ ⎟⎠ ⎝ Pin ⎠ AC
4π R ⎞ ⎛P ⎞
10 log GB + 10 log GC = 20 log ⎛⎜ ⎟ + 10 log ⎜ r ⎟
⎝ λ ⎠ ⎝ Pin ⎠ BC
Thus, the three gains may be determined from the above simultaneous equations.
9.8.3. Power-Measurement Methods

Transmitted and received powers can be measured by various methods. A common
method of transmitted power measurement is to extract a small fraction of the power in
the transmission line by means of a directional coupler and apply it to a power meter.
Typically, a power meter consists of a bolometer connected to one arm of a bridge circuit
for converting bolometer-resistance changes to indications of power. To obtain the value
of the high transmitted power levels, Pt, the power thus measured is corrected for the
directional-coupler attenuation and for any additional attenuation that may have been
inserted between the coupler and the bridge to reduce the power to the level for which
the bridge is calibrated.
Received power is usually measured by a comparison method, using a calibrated signal
generator and a receiver with some type of output indicator. The general procedure is
to adjust the receiver gain until a useful output indication is obtained; and it should be
ascertained that the signal level is high enough so that practically all the indication is
due to the signal and not to the background noise. Then the transmitter is turned off, and
the calibrated signal generator is turned on. It may either be connected to the receiver in
place of the antenna, or the antenna may be left connected and the test signal introduced
into the receiver input by way of a directional coupler in the antenna-to-receiver trans-
mission line. The signal generator output is then adjusted until it gives an indication equal
to that obtained with the actual received signal. From the signal generator calibration,
corrected for attenuation of the directional coupler, the value of the received power,
Pr, can be calculated.
9.8.4. Gain Measurement by Comparison

At high frequencies the most common method of gain measurement is by comparison
of the signal strengths transmitted or received with the unknown-gain antenna and a
standard-gain antenna. This comparison is most conveniently made on a pattern range,
with the same general setup of equipment used in pattern measurement and with the range
antenna transmitting. The gain of this antenna need not be known, nor does the propaga-
tion factor F affect the result as long as F does not vary appreciably over the apertures of
the AUT antenna and the gain standard. All that is required of the range antenna and its
associated transmitter is that they do not vary the amount or frequency of the radiated
power in the direction of the AUT antenna throughout the measurement procedure.
Since the gain of the unknown antenna is ordinarily higher than the gain of the gain
standard, the standard antenna is first connected to the receiver (in the manner of an AUT
antenna in pattern measurement), and aimed at the range antenna. The receiver gain is
adjusted to give a convenient output meter indication (after it is determined that the level
of the signal is well above noise level). Then the antenna whose gain is to be measured
is connected in place of the standard-gain antenna, and attenuation is introduced into the
transmission line between the antenna and receiver until the output indication is the same
as it was with the gain-standard antenna. If the attenuation factor, expressed as a power
ratio greater than one, is L, the gain of the unknown antenna, Ga, is
Ga = LGs (9–7)
where Gs is the standard-antenna gain. Inasmuch as antenna gains and attenuator calibra-
tions are often expressed in decibels, it is frequently convenient to make the calculation
in decibels, in which multiplication is replaced by addition:
Ga(dB) = Gs(dB) + LdB (9–8)
In the unlikely event that the unknown antenna has a smaller gain than the standard, L
in (9–7) is expressed as a number less than one, and the decibel value of L in (9–8) is
negative.) The basic set-up for gain measurement by the comparison method is dia-
grammed in Fig. 9–21. It is essential in this method of gain measurement that both the
AUT and the standard antenna are equivalently impedance-matched to the load presented
to them by the transmission line. The best way to insure this is to make VSWR measure-
ments with each of them connected in turn, and adjust the matching of each for a flat
line (VSWR = 1). This method basically compares antenna power gains, but directivities
may also be determined if the antenna radiation-efficiency factors are known.
Antenna being Distance 2D2/λ

measured (primary or greater
antenna)
Secondary
antenna
D
Standard-gain horn
Signal
Output source
Receiver
meter
Attenuator
FIGURE 9–21.
Set-up for gain measurement by comparison method.
9.9. Antenna Efficiency

The term efficiency has two different connotations in its application to antennas. One is
related to the dissipative losses and the other to the ratio of the directivity to the aperture
area.
9.9.1. Radiation Efficiency

The radiation efficiency kR, a number between zero and one, expresses the ratio of the
power radiated PR to the total power accepted Pin by the AUT. The difference of these
two quantities is the power dissipated PD in ohmic or dielectric losses. The radiation
efficiency is also the factor applied to the directivity D to obtain the gain G. There are
in principle several ways of measuring kR, indicated by the following equations:
PR
kR = (9–9)
Pin
Pin − PD
kR = (9–10)
Pin
G
kR = (9–11)
D
The first equation requires direct measurement of the total radiated power, which is pos-
sible only in special cases. Measurement of the total input power to the antenna Pin is
not difficult, since this power flows in the transmission line connecting the transmitter
to the antenna.
The second equation requires measurement of the dissipated power PD. This can
sometimes be done, especially at low frequencies, by measuring the resistance of conduc-
tors in which current flows, and multiplying these resistances by the square of the current.
At LF and VLF, the components of the antenna system whose rf resistances must be
measured are the ground system, the loading coil, and the conductors of the antenna
itself.
The third equation is not directly useful, but it may be combined with (9–4) to
obtain
4π R 2 E 2
kR = (9–12)
377 Pin F 2 D
This equation is especially useful at VLF with short vertical grounded radiators (mono-
poles). For these antennas, D may be taken as twice the directivity of the corresponding
short dipole; hence, ordinarily, D is approximately 3. Within line-of-sight distances and
over water or moist ground, it can be assumed that F = 1. Therefore, if measurements
Radiation Resistance 399
are available of both the total input power to the antenna Pin and field strength E at dis-
tance R from the antenna, the radiation efficiency can be determined. Since the definition
of kR requires E to be the far-field strength, R must be a distance satisfying the far-field
criteria, as discussed in sec. 3.2.5.
9.9.2. Aperture Efficiency

The other connotation of the term “efficiency” relates to the equation for the directivity
of a large-aperture type antenna—a horn, a large unidirectional planar array, a parabolic
reflector, or a lens. From equation (6–13), chapter 6, directivity is expressed in
(9–13) as:
4π A
D = k A 2  (9–13)
 λ 
where A is the geometric area of the aperture (for antennas that have a physical area such
as those listed immediately above) and l is the wavelength (both expressed in the same
system of units). In this context, kA is called the aperture efficiency. If the field intensity
over the aperture of an antenna is uniform (no “taper”), with no spillover or leakage,
kA = 1. This is the largest value of kA practically attainable, ruling out “supergain” anten-
nas, which are not feasible except to a limited degree in a few cases. Typical values of
kA range from somewhat less than 0.5 to nearly 1.0. The measurement of kA is virtually
synonymous with measurement of directivity D, since from (9–13):
Dλ 2
kA = (9–14)
4π A
9.10. Radiation Resistance

The radiation resistance of an antenna is defined by equation (3–28), chapter 3; it is the
ratio of the power radiated to the square of the antenna current. As mentioned, it is in a
sense a fictitious quantity, since it is referred to an arbitrary point in the antenna and has
different values for different reference points. It is conventional to refer it to the current-
maximum point, although it may also be referred to the feed point. In many cases the
two points are one and the same; an example is the case of a center-fed half-wavelength
dipole. When referred to the current-maximum point, it is sometimes known as the loop
radiation resistance, since a current maximum is also called a current loop.
If there is no ohmic loss in the antenna—that is, if all the input power is radiated—then
the radiation resistance referred to the feed point is equal to the resistive component of
the antenna input impedance. In this case, measurement of the antenna input impedance
constitutes a measurement of its radiation resistance. If the feed point is not a current-
maximum point, the loop radiation resistance may be calculated from the feed-point
radiation resistance from the formula:
2
 I feedpoint 
Rr (loop) = Rr (feedpoint )  (9–15)
 I maximum 
where I denotes the currents at the points indicated by the subscript notation.
If there is appreciable ohmic loss, so that the antenna radiation efficiency factor kR is
less than one, the radiation resistance is found from equations (3–26) and (3–29) of
chapter 3 to be
Rr = k R Ri (9–16)
where Ri is the input resistance, that is, the resistive component of the input
impedance.
It is apparent that the radiation resistance is sometimes a rather nebulous concept and
not always easily measured. In general, it is a useful concept only when it is readily
measurable. It has no meaning for antennas in which there is no clearly defined current
value to which it can be referred.
9.11. Impedance Measurements

Impedance is defined in terms of the current I that flows if a voltage E is applied between
a pair of electrical terminals. According to Ohm’s law, impedance Z is Z = E/I. Imped-
ance is a complex quantity, having a resistive real part R and a reactive imaginary part
X, such that Z = R + jX. The algebra of complex numbers is reviewed in Appendix C.
Impedance Z can be determined by measuring the voltage and current at the pair of
terminals, and the phase angle between them. At low radio frequencies, voltmeters and
ammeters are practical instruments. The direct measurement of the phase angle, however,
is not a simple matter, although it can be done. A simple method at low radio frequencies
is to apply a small voltage derived from the current I to one deflection axis of an oscil-
loscope and a sample of the voltage E to the other axis. The resulting pattern will be an
ellipse; and the value of the phase angle can be determined from the dimensions of the
ellipse. This direct method of impedance measurement may sometimes be useful.
However, comparison methods are more commonly employed. The most basic method
is the bridge method, which is essentially a sensitive method for measuring an unknown
impedance by comparing it with a known impedance.
Prior to the 1970s, before the commercial development of network analyzers, the
preferred impedance measurements methods were the impedance bridge and the standing
wave method. For measurement of low-frequency antenna impedances, bridges were
virtually always employed, and up to frequencies of about 30 MHz they were the method
of choice. From 30 MHz to perhaps as high as 1 GHz, the bridge method was used,
although the bridge impedance arms might then consist partly of transmission-line ele-
ments rather than purely “lumped” capacitances and inductances.
The standing wave method was used for frequencies above a few hundred MHz,
whereby the field strength along the transmission line was measured. The so-called
Impedance Measurements 401
slotted line was commercially available for measuring standing wave ratios within coaxial
transmission lines or waveguides, up to frequencies of about 100 GHz. As already noted,
today the preferred impedance measurement method is the network analyzer, for which
multiband designs are commercially available for almost all radio frequencies, from
10 kHz to 100 GHz. The basics of network analyzers are given in sec. 9.11.2. However,
the standing wave method is first discussed in sec. 9.11.1, because its principles are basic
to the understanding of power transfer to antennas and their related components.
9.11.1. Standing-Wave Method

At frequencies above 30 MHz, a transmission line of sufficient length to exhibit a stand-
ing wave, as discussed in secs. 2.1 and 2.2 (see, for example, Fig. 2–5), is not unrea-
sonably long. It then becomes practical to measure the antenna input impedance by
determining the VSWR (voltage standing wave ratio) on the feed line, and the distance
from the antenna terminals to the first voltage minimum on the line. Calculation of the
antenna impedance by this method is based on equation (2–46) of chapter 2, in conjunc-
tion with equations (2–37), (2–47), and (2–48). These equations show how the complex
reflection coefficient of the load can be found from measurement of the VSWR and dis-
tance d from the load to the voltage minimum, and that the load impedance, ZL, can be
calculated when r is known.
The Smith Chart is described in sec. 9.12. It can be employed to determine ZL directly
from the values of VSWR and d, without going through any calculations, and without
the intermediate determination of the reflection coefficient, r. Therefore, the method
consists of measuring the quantities VSWR and d, and then making use of the Smith
Chart.
The quantities to be measured are depicted graphically in Fig. 9–22. (This diagram
shows only one of many possible patterns that might be observed.) The VSWR is the ratio
Vmax/Vmin. The wavelength, l, is twice the separation of the voltage minima. The ratio
d/l, rather than d itself, is the quantity actually used in the calculation.
Voltage
Vmax V pattern
Vmin
0
λ d
Source of Transmission
voltage Line conductors ZL line
FIGURE 9–22.
Diagram showing quantities to be measured in standing-wave method of impedance
determination.
The basic procedure for measuring the position of the voltage minimum, and the
VSWR, is to move a short sensing probe along a transmission line and find the positions
of maximum and minimum field strengths. To meet these requirements, it is possible to
employ a special slotted line—a horizontal section of coaxial line or waveguide that has
a long narrow slot cut into the top of its outer wall, along its length. A short metal pin,
the sensing probe, projects through the center of the slot and into the coaxial line or
waveguide (without touching the walls). The upper end of the probe connects to a detec-
tor (usually a crystal diode). Because the probe is of such small dimensions, it does
not disturb the field appreciably, and the detector does not constitute an appreciable
load.
Prior to about 1975, slotted lines were commercially available over a wide range of
frequencies, but because of the development of highly successful network analyzers, they
seem to have become almost extinct. Even so, the slotted section remains a low-cost
instrument for making VSWR and impedance measurements. For frequencies of several
hundred megahertz up to perhaps 3 GHz, coaxial slotted lines of 50-ohm characteristic
impedance were standard. At higher frequencies and up to about 100 GHz, slotted wave-
guides were used. Another method of obtaining VSWR is to use a directional coupler,
such as the one represented in Fig. 2–19, chapter 2. If the voltage outputs of both ports
2 and 4 of this coupler are separately measured, one measurement will represent the
amplitude of the forward-traveling wave in the line, and the other will represent the
reflected-wave amplitude. The ratio of the sum of these amplitudes to their difference
gives the VSWR, in accordance with equation (2–43), chapter 2. A measurement setup
that uses a directional coupler in this way is called a reflectometer. If also the phase dif-
ference of the two voltages is measured and the distance from the coupler to the load is
known, the load impedance can be computed.
9.11.2. Network Analyzers

Traditional network analyzers are used to measure the amplitude and phase characteris-
tics of the transmission and reflection properties of a device under test (DUT). From
these measurements, properties such as input impedance, reflection coefficient, and
transmission loss can be determined. Figure 9–23 is a schematic of a traditional transmis-
sion/reflection network analyzer, where a frequency source, a DUT, a three-input receiver,
a processor, and a display are depicted. The three receiver inputs are the incident (input),
reflected, and transmitted signals. These three signals are obtained from the couplers
designated A, R, and B and, for brevity, the signal voltages are also designated A, R,
and B.
Most of the measurement data from a network analyzer is derived from ratios, either
magnitude only or complex ratios (amplitude and phase) of the signal voltages A, R, and
B. For example, the magnitude ratios R/A and B/A provide the reflection coefficient
magnitude and the transmission loss through the DUT. In addition, the complex ratios
provide information needed for determining impedance and phase change on transmis-
sion through the DUT. From the ratios, the processor also determines such quantities as
VSWR and input impedance. The display provides the various calculated data, and some
Impedance Measurements 403
A R B
DEVICE
FREQUENCY
UNDER LOAD
SOURCE
TEST
INCIDENT REFLECTED TRANSMITTED

A R B
A R B
RECEIVER
PROCESSOR/
DISPLAY
FIGURE 9–23.
Schematic of a basic network analyzer.
commercially available network analyzers even provide such complete information as a

Smith Chart (see next section) plot versus frequency.
There are many types of network analyzers that are available commercially, some of
advanced design for special applications. As already indicated, frequency sources usually
provide a wide range of frequencies, and others have variable output power so that mea-
surements of component linearity may be accomplished. Some analyzers determine fea-
tures known as the S-parameters of a network, which are needed for applications where
networks are cascaded into complex configurations. S-parameters are determined by two
sets of complex (amplitude and phase) ratios: the transmission and reflection ratios (1)
at a network input with the output terminated with a matched load and (2) at the network
output with the network input terminated with a matched load.
Network analyzers may measure magnitudes only or magnitude and phase, operate
manually or automatically, and cover a frequency range continuously or in steps. Modern,
commercially available network analyzers permit rapid (virtually simultaneous) phase
and amplitude measurements over wide ranges of frequency. Multiband designs are used
for very wide frequency coverage, permitting coverage over the present practical limits
of radio frequencies (tens of kHz to 100 GHz). Analyzers often allow measurements at
two or more ports, thus permitting isolation or cross-polarization measurements versus
frequency.
As already described, early UHF, microwave, and millimeter wave impedance mea-
surements used a slotted transmission line, commonly called a slotted line. Next came
the simple single-frequency network analyzer that compared incident and reflected waves
within the transmission line and “automatically” provided calculated impedance results.
This automation began in the late 1960s, and rapidly evolved with the subsequent devel-
opment of tunable, wideband, solid-state sources. The technology of network analyzers
expanded from the early, single low-radio-frequency input and transfer impedance mea-
surements to the present state-of-art whereby amplitude and phase determinations are
made from the lower radio frequencies up to the microwave and millimeter wavelength
regions.
Features of modern network analyzers include
• rapid tunable, wideband solid-state signal sources
• rapid-phase and amplitude measurements
• automatic measurements and computer calculations using the measured
results
• multiple visual displays of operating parameters and measured and calculated
results.
9.12. The Smith Chart

When the VSWR and the position of a voltage minimum have been measured, calculation
of the antenna input impedance can be made from the basic transmission-line equations
of chapter 2. The considerable labor of using these equations is usually avoided by using
an impedance chart, which is a graphical representation of the impedance relationships
expressed by the equations.
A common feature of impedance charts is that they deal with dimensionless ratios,
rather than directly with physical quantities. The ratios involved are primarily ratios of
impedance, length, and voltage: specifically, the ratio of the load impedance of the line
to its characteristic impedance (ZL /Z0), the ratio of the distance from the load to a voltage
minimum to the wavelength (d/l), and the ratio of maximum to minimum standing-wave
voltages (VSWR). Conversions from two of these ratios to the physical quantities ZL and
d, and vice versa, are readily made, since Z0 and l are presumed to be known quantities.
Because they deal with ratios, the same charts can be used for all characteristic imped-
ances, frequencies, and absolute voltage levels.
The Smith Chart is the most widely used impedance chart, and it became popular
shortly after the concept was published by its inventor (Smith 1944). Because of its
common usage, it is commercially available in printed form and is in effect a special
graph paper made for plotting impedances. The basic plan of the Smith Chart is shown
in Fig. 9–24. Within the circular boundary there are two orthogonal families or sets of
circles. Orthogonal means, roughly, perpendicular, in the sense that the circles of one
family intersect those of the other family perpendicularly, that is, at right angles. (Actu-
ally it is the tangent lines to the circles at the point of intersection that form right
angles.)
There is one point on the chart through which every circle of both families passes;
this is the point at the exact bottom of the chart. The circles of one family pass through
The Smith Chart 405
this point horizontally; those of the other R/Z0 = 0
family go through it vertically. The first X/Z0 = –0.5 X/Z = 0 X/Z0 = +0.5
0 R/Z0 = 0.33
of these families of circles represent con-
stant values of the ratio RL/Z0, and will Center of R/Z = 1
0
be referred to as “R circles.” The second X/Z0 = –1 chart
X/Z0 = +1
family of circles corresponds to constant
values of XL/Z0 and will be referred to as R/Z0 = 3
the “X circles.” RL and XL are of course X/Z0 = –2 X/Z0 = +2
the resistive and reactive components
of the load impedance, ZL. The X circles
to the left of the center line are negative
FIGURE 9–24.
values of XL/Z0, representing capacitive
reactance, and those on the right are posi- Basic construction of the Smith Chart.
tive, representing inductive reactance.
The vertical center line is the XL = 0 line. The R circle that passes through the exact
center of the chart represents RL/Z0 = 1. Therefore the exact center point of the chart
corresponds to a load impedance that is a pure resistance of value equal to the charac-
teristic impedance; that is, at this point RL = Z0 and XL = 0. (This is of course the load
value that results in a unity value of VSWR.)
These two families of circles are in effect a system of coordinates, one coordinate set
representing the resistive component and the other set the reactive component of the load
impedance. Any particular point on the chart corresponds to a load impedance, ZL, whose
components are given by the two orthogonal R and X circles that intersect at that
point.
In addition to these R and X coordinates, there is another set of coordinates for the
measured quantities, VSWR and d/l. These coordinates are not printed on the chart, since
they would result in a hodgepodge of lines and make it difficult to read the chart. Instead,
they are to be plotted in by the user for the specific measured values in a particular case.
The VSWR coordinates are circles whose centers are at the center of the Smith Chart—
that is, at the XL/Z0 = 0, RL/Z0 = 1 point. This point corresponds to VSWR = 1, and the
circles of increasing size correspond to increasing values of VSWR. The largest of these
circles forms the outer boundary of the chart and represents an infinite VSWR (∞).
The d/l coordinates are radial lines emanating from the center of the chart. A circular
scale of values of d/l is provided on the outer periphery of the chart. The full circle spans
the range from d/l = 0, at the top of the chart, through d/l = 0.25 at the bottom of the
chart, to d/l = 0.5 again at the top; thus the complete circle of values corresponds to
values of d going from zero to 12 wavelength. The values increase counterclockwise,
when d is the distance from the antenna terminals to the first voltage minimum. This
direction on the scale is usually marked “wavelengths toward the load,” which refers to
the location of the null when the load is short-circuited, with respect to the voltage
minimum with the short removed. A complementary scale is also usually provided,
marked “wavelengths toward the generator.” This scale increases in the opposite direction
and corresponds to the distance from the voltage minimum (short removed) to the nearest
null (load shorted) in the direction of the generator (signal source).
FIGURE 9–25.
Example of impedance and admittance calculation, using the Smith Chart.
An example of the use of the Smith Chart is shown in Fig. 9–25. In this example it
is supposed that the VSWR has been measured and found to be 5.3. Therefore a circle
has been drawn, with its center at the center of the Smith Chart, passing through the
VSWR = 5.3 point on the vertical center line. The d/l value has been found to be 0.16,
and a radial line has been drawn from the center of the chart to this point on the d/l scale
(the one labeled “wavelengths toward load”). At the intersection of this circle and radial
line, a pair of the R and X circles (shown dashed) also intersect. These are seen to be the
ones for RL/Z0 = 0.6, and XL/Z0 = −1.4. Therefore, the antenna input impedance in this
case consists of a resistive component RL = 0.6Z0, and a negative (capacitive) reactance
component XL = −1.4Z0. If (for example) the characteristic impedance of the line is
Antenna Noise Measurement 407
50 ohms, then RL = 30 ohms, and XL = −70 ohms. (Note that XL here designates “load
reactance” and not, as in some cases, “inductive reactance.”)
This example illustrates the basic use of the Smith Chart. It can also be used to solve
impedance-matching problems. For this purpose it is often convenient to work with
admittance, rather than impedance. As discussed in chapter 2 using equations (2–26)
through (2–34), admittance Y is the reciprocal of impedance. Its real and imaginary
components are called, respectively, conductance G and susceptance B. To find these
admittance components, in terms of their ratio to the characteristic admittance of the line
Y0 (= 1/Z0), is extremely simple using the Smith Chart. The radial line for the applicable
value of d/l is merely extended (as shown dashed in Fig. 9–25) until it intercepts the
opposite side of the VSWR circle. The values of GL/Y0 and BL/Y0 are then read at this
point, from the same set of orthogonal circles that represent RL/Z0 and XL/Z0 at the inter-
section of the solid radial d/l line with the VSWR circle. In this particular example it is
found that GL/Y0 = 0.26 and BL/Y0 = +0.6. Since Y0 = 1/Z0, if Z0 = 50 ohms then Y0 =
0.02 mho. Consequently GL = 0.26 × 0.02 = 0.0052 mho, and BL = +0.6 × 0.02 =
+0.012 mho, in this example. (Note that whereas capacitive reactance is negative, capaci-
tive susceptance is positive; inductive susceptance is negative.)
9.13. Antenna Noise Measurement

As pointed out in sec. 7.8, the lowest antenna noise is generally present in the frequency
region extending from approximately 1 to 10 GHz. Antennas designed for low noise in
this region will generally be at least moderately directive, with a well-defined beam, and
they will usually be steerable. Low-noise antennas have moderate or narrow beamwidths,
low-side-lobe levels, high ohmic efficiency, and impedances are well matched. An example
of a low-noise antenna is the earth-based terminal of a space communications system.
9.13.1. Elementary Noise Theory

The noise performance of an antenna is expressed in terms of a noise temperature rating.
This rating actually expresses the available noise power at the antenna output terminals.
“Available” in this context means “deliverable to a matched-impedance load.” The use
of the somewhat fictitious temperature rating stems from the fact that a basic source of
noise in electric circuits is the thermal agitation of electrons in resistive circuit compo-
nents. From Nyquist (1928), it is known that the open-circuit rms noise voltage Vn gener-
ated in a resistance of value R ohms at an absolute temperature T is
Vn = 4 kTRB volts (9–17)
−23
where k is Boltzmann’s constant (1.38 × 10 watt-second per Kelvin*), and B is the
bandwidth in Hertz within which the noise voltage is measured. (The formula is actually
* Parenthetically, absolute temperatures are correctly expressed in Kelvins and not in degrees Kelvin.
Following IEEE (1992, p. 320), the Kelvin or absolute temperature scale is related to the Celsuis and
Fahrenheit scales by the formula:
TKelvin = 273.15 + TCelsuis = 255.37 + ( 5 9)TFahrenheit

an approximation that holds for most conditions at frequencies in the radio spectrum. A
more exact formula is required at extremely high frequencies and at extremely low
temperatures.)
It is well known that the maximum power that can be delivered by a generator or other
source having a given internal electromotive force (open-circuit voltage) is obtained
when the external load resistance is equal to the internal resistance of the source. (More
generally, the external impedance must be equal to the complex conjugate of the source
impedance.) From this it readily follows that the thermal noise power deliverable to this
optimum load by a thermal noise source of resistance R at temperature T is
Vn2
Pn = = kTB watts (9–18)
4R
Hence the quantity kTB is the available noise power. It does not depend on the particular
value of resistance, but only on the absolute temperature T (for a given bandwidth B).
Therefore it has become customary to rate a noise source whose available noise power
is Pn in terms of its equivalent noise temperature calculated from (9–18)—that is,
T = Pn/kB. This rating practice is followed even where the noise is not actually of thermal
origin, provided it “resembles” thermal noise in statistical and frequency spectrum
characteristics.
A thermal-noise voltage is a randomly fluctuating voltage that conforms to a particular
statistical law and contains a uniform mixture of all frequencies within the radio spec-
trum—or, in particular, within any selected band, such as the intermediate frequency (i–f)
passband of a receiver. The general appearance of a noise-voltage waveform, as viewed
on an oscilloscope connected to the output of a receiver (after detection), is shown in
Fig. 9–26. This waveform is the half-wave-rectified “envelope” of the rf noise; the rf
cycles have been filtered out. (Such noise is sometimes called “grass.”)
Because of their statistically random nature, noise voltages from two different (inde-
pendent) sources are uncorrelated. Thus, when combined or superimposed in a circuit or
as electromagnetic waves in space, they add so that the resultant rms voltage (or field
intensity) is the square root of the sum of the squares (root-sum-square) of the super-
imposed voltages (or fields). Consequently, the resultant noise power is the sum of
the individual-component noise powers; it is unnecessary to consider whether phase
relationships cause addition, cancellation,
or some intermediate result. This deduc-
tion follows from the fact that the relative
Voltage
phases of uncorrelated noise voltages

fluctuate; they do not maintain a constant
100 µsec relationship over time. Therefore, if an
Time
antenna receives noise contributions from
FIGURE 9–26. a number of independent sources, the
Oscillogram of a typical time sample of total noise power (temperature) will be
detected thermal noise; 100-msec sample, simply the sum of the individual noise
amplifier bandwidth 500 kHz. powers (temperatures).
The mathematical analysis of noise, the interrelationships of signals and noise, and
the physical origins of noise have been intensively investigated, and an extensive litera-
ture exists. See, for example, Blake (1986, ch. 4) or Davenport and Root (1987).
9.13.2. Antenna Noise Temperature

As was pointed out in sec. 7.8, antenna noise comes from a number of sources, both
external and internal to the antenna. The internal sources are thermal in nature and are
due to ohmic losses. The sources of external noise are graphically depicted in Fig. 9–27.
The mechanism of generating electromagnetic thermal noise waves in dissipative propa-
gation media, such as the earth, and in the atmosphere at frequencies where it partially
absorbs radio waves, is similar to that of thermal noise generation in resistive circuit
elements. Noise radiated from some sources, such as the galactic system of which the
earth and the sun are members, is not truly thermal in nature, but within limited band-
widths it resembles thermal noise, so that its contribution to total antenna noise may be
described in terms of an equivalent noise temperature.
Because some of the external noise is unavoidable, when an antenna noise temperature
is measured it is necessary to evaluate the result in relation to the minimum possible
noise temperature that could be achieved at the operating frequency. From measurements
from many workers on the various noise sources depicted in Fig. 9–27, and by using the
principle of addition of noise power from independent sources, it is possible to plot curves
of approximate noise temperatures of an ideal antenna as a function of frequency. Figure
9–28 includes such curves.
Calculated noise temperatures for an ideal antenna in the frequency range from
100 MHz to 100 GHz, due to galactic and atmospheric noise are included in Fig. 9–28.
In the left-hand region, below 1 GHz, the maximum values will be observed if the
antenna is pointing toward the center of the galaxy (the “hottest” part of the sky, in
radio-astronomical parlance); whereas the minimum values are observed toward the
galactic poles (where the sky is “cold”). In the right-hand region, above 1 GHz, the
maximum values are observed with an antenna pointed toward the horizon (elevation
angle q = 0°), looking through the thickest possible section of the atmosphere; whereas
the lowest values are observed looking straight up, toward the zenith (q = 0°). Noise
temperature maxima caused by water vapor and oxygen resonances can be seen centered
at 22 GHz and 60 GHz, respectively. The values shown in Fig. 9–28 assume that there
is no noise due to ohmic losses in the antenna, and none due to ground noise received
via minor lobes of the pattern. That is, the antenna is assumed to be lossless and to
have no minor lobes. Techniques for minimizing losses and minor lobes are discussed
in sec. 7.8.
Experience shows that the noise temperature of a well-designed low-noise antenna in
the 1 to 10 GHz range should be less than 60 K, and it may be as low as 4.5 K. This
latter value was reported (DeGrasse et al. 1959) from use of a 5.65-GHz horn-reflector
antenna of the type described in sec. 7.8 (the same type used in the Telstar satellite com-
munication system). Of the 4.5 K, 2.5 K was attributed to tropospheric noise, and the
remaining 2 K to ground noise, via minor lobes.
Sun
Galaxy
Main beam
Side and
back lobes
lo
no
sp
Beam-elevation
he
re
angle
Antenna
Tr
op
os
ph
er
e
Earth
FIGURE 9–27.
External sources of antenna noise.
9.13.3. Measurement 10,000

NOISE TEMPERATURE (K)
Techniques q = 0°
q = 1°
1,000 q = 2°
As in antenna gain measurement, the q = 5°
q = 10°
most convenient technique of measure- q = 30° 0°
q = 90°
ment of noise temperature is comparison 100
90°
with a source of known (and variable)
noise temperature. Figure 9–29 shows the 10
measurement set-up.
The switch in this arrangement must 1
be a high-quality transmission-line 100 1,000 10,000 100,000
switch—a coaxial switch if the line is FREQUENCY (MHz)
coaxial, or a waveguide switch when the FIGURE 9–28.
line is waveguide. It is first set to connect
Noise temperature of an idealized antenna
the receiver input to the antenna, and the located on the earth’s surface, as a function of
receiver gain is adjusted to cause the frequency and beam elevation angle q (from
output meter to read a convenient value. Blake 1969).
This reading is noted. The antenna should Antenna

be pointing straight overhead (zenith
direction) when this reading is taken, or
at as high an angle as possible. The mea-
surement should be performed in clear Receiver Output
(low noise) meter
weather, with no precipitation or over-
cast, and at a time when the sun is low in Transmission-line switch
Standard
the sky so that it will not be in the main- noise
beam of the antenna pattern. At frequen- source
cies below 1 GHz, the approximate FIGURE 9–29.
galactic temperature should be estimated
Set-up for antenna noise temperature
in the overhead direction, from published measurement.
sky-noise maps (see, e.g., Ko 1958).
Next the switch is connected to the
Variable rf
calibrated variable noise source, and its Cooled load To
Ts ≅ 4 °K to attenuator transmission-line
effective temperature (noise power output) 80 °K Tt ≅ 290 °K switch
is adjusted until the output meter reading
is the same as that previously noted. The FIGURE 9–30.
noise temperature of the calibrated source Method of obtaining variable noise output from
is then the same as that of the antenna. It a fixed-temperature source.
is essential that the receiver gain remains
constant during the entire measurement procedure. Also, the antenna and the noise source
must be equivalently matched to the receiver.
A variable-temperature source may be obtained by using a constant-temperature source
and an attenuator. The arrangement is shown schematically in Fig. 9–30, where Ts is the
temperature of the cooled source and Tt is the temperature of the attenuator. This method
is described in a paper by Schuster, Stelzried, and Levy (1962). The attenuator affects
the output temperature of the source-attenuator combination in two ways. First, it attenu-
ates the noise from the source. Second, it generates thermal noise itself, since it is a lossy
device at a nonzero thermodynamic temperature. Not all the noise thus generated is
delivered to the receiver; part of it is coupled back to the source.
Characteristics of an attenuator that influence its output noise are its thermal tempera-
ture and its loss factor. The loss factor L is defined as the ratio power input to power
output. Therefore, L is the reciprocal of transmission efficiency. Stated differently,
L = 1/G, where G is power gain (which for an attenuator G ≤ 1). Thus, that part of the
attenuator output noise power Psource due exclusively to the input noise source of tem-
perature Ts is
Psource = kTs B ⎛⎜ ⎞⎟
1
(9–19)
⎝ L⎠
Now let Ti be the attenuator noise temperature, and thus kTiB is the attenuator noise
power at the attenuator input. Recall that the available power from separate noise sources
is the sum of the noise powers. Thus, the total noise power at the attenuator input is
kTsB + kTiB, and the total noise power Pn(total) at the attenuator output is
1
Pn(total ) = [ kTs B + kTi B ] (9–20)
L
To determine Pn(total) of (9–20) as a function of Ts, Ti, and L, an equation for Ti is

needed. First, consider the special case when the attenuator thermal temperature Tt and
the noise source temperature Ts are equal. Then, Pn(total) necessarily equals kTtB because
Tt is the only temperature present. By solving (9–20) when Ts = Tt and Pn(total) = kTtB,
it is seen that Ti, the equivalent noise temperature that reaches the attenuator input,
becomes
Ti = Tt ( L − 1) (9–21)
Note that (9–21) was found true regardless of the magnitudes of either Ts and Tt,
providing only that Ts = Tt. In fact, Dicke et al. (1946) report that noise power
created internally by an attenuator is always equal to (9–21). Consequently, by replacing
Ti in (9–20) with Tt(L − 1), one finds that attenuator output power versus Ts, Tt, and
L is
= kB ⎡⎢ s + Tt ⎛⎜ 1 − ⎞⎟ ⎤⎥
1 T 1
Pn(total ) = [ kTs B + kTt ( L − 1) B ] (9–22)
L ⎣L ⎝ L⎠⎦
Then, from the relationship Pn(total) = kBTn with Tn being the effective noise tempera-
ture at the attenuator output terminals, one finds
+ Tt ⎛⎜ 1 − ⎞⎟
Ts 1
Tn = (9–23)
L ⎝ L⎠
When L approaches infinity, it can be seen from (9–23) that Tn = Tt, whereas if
L = 1, Tn = Ts. Therefore, by varying L effective noise temperatures in the approximate
range Ts to Tt are available by the method of Fig. 9–29.
If the attenuator is calibrated in decibels, and the decibel attenuation expressed as a
positive number is AdB, the value of L is given by
L = anti log ⎛⎜ dB ⎞⎟
A
(9–24)
⎝ 10 ⎠
The standard reference temperature for noise measurements is 290 K (IEEE, 1992, p.
850), which is a chilly “room temperature” of 16.8° C (62.3° F). The method (Fig. 9–30)
of obtaining variable output noise temperature will yield values that range from Ts up
to or down to 290 K. This depends on whether Ts is less than or greater than 290 K.
Ordinarily Ts is less than 290 K, if it is presumed that the antenna noise temperature a
low-noise antenna is to be measured.
9.13.4. Low-Noise-Temperature Comparison Sources

To measure a low-antenna noise temperature, a comparison source of noise with a tem-
perature near that of the antenna is required. The ordinary “dummy load” or resistive
termination of a transmission line or guide will have a noise temperature equal to its
thermal temperature, in accordance with the Nyquist equations (9–17) and (9–18). Since
the thermal ambient temperature is ordinarily around 290 K, such a load must be artifi-
cially cooled to obtain a low-noise source. This is the method actually employed.
If the load is cooled in a cryostat with liquid nitrogen, a noise temperature of about
80 K will result (−193.2° C, −315.7° F). This is low enough for measurement of moder-
ately low antenna noise temperatures. But for very low-antenna temperatures a lower
comparison source is preferred. One may be obtained by using liquid helium in place of
the liquid nitrogen. The resulting noise temperature is about 4 K.
These very low values of noise temperature are effective only if the load is connected
to the receiver through a very low-loss transmission line. If appreciable loss is present,
the effective temperature will be increased in accordance with (9–23). This is the mecha-
nism used for providing increased effective noise temperatures when values higher than
those of the cooled load are required.
9.13.5. Measurements Using Ambient Temperature Noise Sources

According to Imbriale (2007, p. 45-21), a variety of noise temperature techniques and
measurements are used in connection with the National Aeronautical and Space Admin-
istration (NASA) Deep Space Network (DSN) program. The low noise ground-based
systems are typically calibrated by injecting a known amount of power into a receiver
input. However, it can be difficult to accurately determine the noise temperature of this
power when referenced to the receiver input, because of different temperatures and losses
of the components and transmission line between the noise source and the receiver.
Also according to Imbriale (2007, p. 45-22), the ambient temperature method described
by Stelzried (1971) eliminates the difficulty of calibrating the noise source temperature
and it is most often used for low noise (<<300 K) receiving systems in the DSN program.
Note that from (9–23) above, with Ts = Tt the output temperature Tn equals Ts and Tt. In
other words, with the noise source and its connecting components all at the same ambient
temperature, the noise source temperature available is the ambient temperature, indepen-
dent of the component losses.
The ambient temperature technique is used to determine the receiver operating noise
temperature Top, with Top = Ta + Te, Ta = antenna noise temperature, and Te = receiver
effective noise temperature at the receiver input. A measurement is made by alternately
connecting the receiver input with a waveguide switch between the antenna and a
matched load at ambient temperature load. Thus, the measurement is made by carefully
determining the ratio R, where
Tp + Te Tp + Te
R= =
Ta + Te Top
with
Tp = physical temperature K of ambient termination
Te = receiver effective noise temperature K at its input
Ta = antenna noise temperature K
The ratio R is determined by means of a precision attenuator located in the receiver IF
stage, which is positioned remotely enough that a change in attenuator settings does not
influence the effective receiver input temperature Te. Another readily met requirement is
that the receiver response be linear between its input and the attenuator.
The receiving operating noise temperature Top from the equation for R above is
Tp + Te
Top = (9–25)
R
Then, Top is calculated by using (9–25) with the measured a value of R, measured
ambient temperature Tp, and the effective receiver noise temperature Te that is either
measured or estimated. Notice that if Te << Tp, Te may not need to be known
accurately.
9.14. System Noise Calculations

As already discussed, noise temperature is a useful measure for the noise level created
by components. Often, however, the term “noise figure” is used to quantify the noise
levels created by mixers, amplifiers, and even complete receiving systems. Therefore,
the term is introduced in sec. 9.14.1, and its relationship with the concept of system noise
temperature is discussed in sec. 9.14.2.
9.14.1. Noise Figure

By definition, noise figure is the actual output noise power relative to the output power
if the input noise temperature T0 is 290 K (IEEE 1992). Specifically, noise figure NF
is the ratio of actual output power to the output power kT0BG, if T0 were 290 K.
Therefore,
Pno kT BG + kTi BG kBG (T0 + Ti ) T

NF = = 0 = =1+ i (9–26)
kT0 BG kT0 BG kT0 BG T0
Here, as used previously, B is the noise bandwidth, G is the power gain of the device
being evaluated, and Ti is the effective noise temperature of its internally generated noise
(referred to the device input). Parenthetically, with T0 being 290 K, the product kT0 is
System Noise Calculations 415
4 × 10−21 watt-second; which is easier to remember than Boltzmann’s constant (1.38 ×

10−23 watt-sec per Kelvin).
Then, from (9–26) and with T0 being 290 K, one finds
Ti = T0 ( NF − 1) = 290 ( NF − 1) (9–27)
Thus, (9–27) gives the internally generated noise temperature Ti of a device when referred
to its input in terms of its noise figure NF.
An example is now given on determining the output noise power Pout of a transistor
amplifier having NF of 2 dB (NF = 1.58) and an input source with noise temperature Ts
of 200 K. Then, from (9–27) Ti is 290(1.58 − 1) or 168.2 K. Thus, the noise temperature,
referred to the device input, is Ti + Ts or 368.2 K. Now assume the noise bandwidth
B and gain G are one MHz and 30 dB (1,000), respectively. Thus, the average noise
power output Pout of the transistor amplifier with input source temperature Ts of 200 K
becomes
Pout = k (Ti + Ts ) BG = 1.38 × 10 −23(368.2 )106 × 103 = 5.08 × 10 −12 watt
9.14.2. System Noise Temperature

Sources of noise within a receiving system include the antenna with its externally
received noise, the transmission line between the antenna and the receiver, and the
receiver itself. For a narrow-beam antenna and a low-noise receiver, effects of antenna
pointing may be observed because the sources of external noise include land and sea
surfaces, the sun, galactic radiation, and the atmosphere.
Noise may arise anywhere within the receiving system, so that the noise level is dif-
ferent from point to point within the system. Of great importance, of course, is the noise
power at the receiver detector (demodulator). However, for the purpose of a signal-to-
noise calculation for a complete system, it is convenient and customary to refer the
system output noise to the antenna output. Done in this way, system noise temperature
Tsys is defined such that the system output power N referenced to the antenna output is
N = kTsys B (9–28)
where
k = 1.38 × 10−23 watt-sec/Kelvin (Boltzmann’s constant)
Tsys = system noise temperature in Kelvins
B = receiver noise bandwidth
Then, if the net available power gain between the antenna output and the receiver output
is G, the actual noise-power of the receiver output is G kTsysB. Thus it is important to
state what reference point has been chosen when the concept of system noise temperature
is employed.
In general, the effect of a noise source at one point in the system may be referred to
another point by properly multiplying or dividing the source noise power by the net
power gain of the components between the two points. Then, when both signal and noise
powers are referred to the same point in a system, their ratio is the same as the output
signal-to-noise ratio, provided that the gain is the same for signal and noise.
Following Blake (1961), the system noise temperature Tsys of a system that includes
antenna, transmission line, and the receiver is
Tsys = Ta + Tr ( Lr − 1) + Lr Te (9–29)
where
Ta = antenna noise temperature, referenced to the antenna output, which includes

the externally received noise plus the thermal noise of the antenna itself
Tr = thermal temperature of the transmission line between the antenna and the
receiver
Lr = the transmission line loss factor
Tr(Lr − 1) = noise temperature of the transmission line referenced to its input (antenna
output)
Te = noise temperature of the receiver referenced to its input
LrTe = noise temperature of the receiver referenced to the antenna output
It is important to note that in (9–29), Ta, Tr(Lr − 1), and LrTe are each equivalent noise
temperatures referenced to the antenna output terminals. The reader will recognize that
a transmission line is a fixed attenuator. Therefore, the noise temperature created by the
transmission line and referenced to its input (antenna output) may be calculated using
(9–27) in sec. 9.14.1. The terms Te and LrTe denote the internally generated receiver noise
referenced to the receiver input and to the antenna output, respectively. The factor Lr
accounts for the transmission line attenuation between the receiver input and the antenna
output. In other words, since Lr ≥ 1, the noise temperature of the receiver referenced to
the antenna output LrTe exceeds the internally generated receiver noise temperature Te
referenced to the receiver input.
Alternatively, system noise temperature Tsys can be expressed in terms of receiver noise
figure NF, by substituting (9–27) into (9–29). Then, Tsys becomes
Tsys = Ta + Tr ( Lr − 1) + Lr [290 ( NF − 1)] (9–30)
Sometimes the term system noise figure NFsys is used to express the performance of
an overall receiving system. In that case, the system noise temperature and the system
noise figure are related as follows:
Tsys = 290 ( NFsys − 1) (9–31)

References 417
TABLE 9-1 Calculated System Noise Temperatures. (Assumptions include 300 K

transmission line temperature, low and moderate transmission line losses and receiver
noise figures.)
NF (receiver) Lr Ta (Lr-1)Tr Lr290(NF-1) Tsys

1.5 (1.8 dB) 1.047 (0.2 dB) 60 K 14.1 K 152 K 226 K
1.5 (1.8 dB) 1.047 (0.2 dB) 300 K 14.1 K 152 K 466 K
1.5 (1.8 dB) 2.0 (3 dB) 60 K 300 K 290 K 650 K
1.5 (1.8 dB) 2.0 (3 dB) 300 K 300 K 290 K 890 K
4.0 (6.0 dB) 1.047 (0.2 dB) 60 K 14.1 K 911 K 985 K
4.0 (6.0 dB) 1.047 (0.2 dB) 300 K 14.1 K 911 K 1225 K
4.0 (6.0 dB) 2.0 (3 dB) 60 K 300 K 1740 K 2100 K
4.0 (6.0 dB) 2.0 (3 dB) 300 K 300 K 1740 K 2340 K
Table 9-1 below gives calculated noise temperatures when using the assumptions
300 K transmission line thermal temperature, plus low and moderate values of transmis-
sion line losses Lr and receiver noise figures NF. Note that signal-to-noise ratio (SNR)
is proportional to system noise temperature Tsys. Therefore, significant changes in signal
detection performance require substantial changes in the ratios of Tsys, system noise
temperature. Note that changes in the antenna noise temperature Ta will result in signifi-
cant improvement in signal-to-noise only if both Lr and NF are very small. From Table
9-1, it is apparent that small values of Tsys require small values of both Lr and NF. Notice
that as Lr is enlarged, the noise contributions from both the transmission line and from
the receiver are increased. It is thus doubly important to place the receiver as close to
the antenna output as practicable, for minimizing Lr and to thereby improve the system’s
signal detection performance.
References
Blake, L. V., “Antenna and Receiving System Noise-Temperature Calculation,”
Proceedings of IEEE, vol. 49, October 1961, pp. 1568–69.
Blake, L. V., “A Guide to Basic Pulse-Radar Maximum-Range Calculation,” Naval
Research Laboratory 6930, December 23, 1969; also see L. V. Blake, ch. 2 in M. I.
Skolnik (ed.), Radar Handbook, 2nd ed., McGraw-Hill, 1990, p. 2.29.
Blake, L. V., ch. 4, Radar Range-Performance Analysis, Artech House, 1986; or ch. 4,
Munro Publishing, 1991.
Bodnar, D. G., “Feed Study for USAEPG Compact Range,” Final Technical Report,
Project A-3922, Georgia Tech Research Institute, August 1986.
Cain, F. L., and K. G. Byers, Jr., “Statistical Gain Characteristics of Radar Antennas at
Very Short Fresnel Zone Distances,” IEEE Transactions on Electromagnetic
Compatibility, EMC-11, February 1969; pp. 1–9.
Cain, F. L., and K. G. Byers, Jr., “Relations of Site Effects to Statistical Gain
Characteristics of Radar Antennas,” 1968 IEEE Electromagnetic Compatibility
Symposium Record, pp. 339–48.
Cain, F. L., J. W. Cofer, and H. A. Ecker, “Prediction Method to Statistically Determine

Potential Antenna Interference,” Proceedings of the IEEE International Symposium on
Electromagnetic Compatibility, July 1970, pp. 407–16.
Cain, F. L., C. E. Ryan, and B. J. Cown, “Prediction of Near-Field Antenna Coupling in
the Presence of Obstacles,” Proceedings of the IEEE International Symposium on
Cain, F. L., E. E. Weaver, and E. F. Duffy, “Prediction of Near-Field Coupling between
Misaligned Antennas,” Proceedings of the IEEE International Symposium on
Chang, D., C. Liao, and C. Wu, “Compact Antenna Test Range without Reflector Edge
Treatment and RF Anechoic Chamber,” IEEE Antennas and Propagation Magazine,
August 2004, pp. 27–37.
Cotton, H. P., “Development of an Outdoor Compact Antenna Range,” ITEA Journal of
Test and Evaluation, vol. 9, no. 1, 1988, pp. 28–31.
Cutler, C. C., A. P. King, and W. E. Kock, “Microwave Antenna Measurements,”
Proceedings of the IRE, December 1947, pp. 1462–71.
Davenport, W. B., and W. L. Root, An Introduction to the Theory of Random Signals and
Noise, Wiley-IEEE Press, 1987.
DeGrasse, R. W., D. C. Hogg, E. A. Ohm, and H. E. D. Scovil, “Ultra low noise
measurements using a horn reflector antenna and a traveling wave maser,” Journal of
Appied.Physics, vol. 30, December 1959, p. 2013.
Dicke, R. H., et al., “Atmospheric Absorption Measurements with a Microwave
Radiometer,” Physical Review 70, Sept. 1946, pp. 340–48.
Dyson, J. D., “Measurement of Near Fields of Antennas and Scatterers,” IEEE
Transactions on Antennas and Propagation, vol. AP-21, July 1973, pp. 446–60.
Hansen, J. E., Spherical Near-Field Antenna Measurements, Peter Peregrines Press,
1988.
Hemming, L. H., and R. A. Heaton, “Measurement of the Radiation Patterns of Full-Scale
HF and VHF Antennas,” IEEE Transactions on Antennas and Propagation, July 1973,
pp. 532–38.
Hess, D. W., and R. C. Johnson, “Compact Ranges Provide Accurate Measurement of
Radar Cross Section,” Microwave Systems News, September 1982, pp. 150–60.
Hickman, T. G., and R. C. Johnson, “Boresight Measurements Utilizing a Compact
Range,” Abstracts of the 1972 Spring USNC/URSI Meeting, Washington, D.C., April
13–15, 1972, p. 11.
Hollis, J. S., T. J. Lyon, and L. Clayton, Microwave Antenna Measurements, Scientific-
Atlanta, Inc., 1970. This publication may be available through MI Technologies, Inc.
(www.mi-technologies.com).
IEEE, The New IEEE Standard Dictionary of Electrical and Electronic Terms, 1992.
IEEE Standard Test Procedures for Antennas, The Institute of Electrical and Electronics
Engineers, Inc., IEEE Standard 149, 1979.
Imbriale, W. A., “Earth Station Antennas,” ch. 45 in Antenna Engineering Handbook, 4th
ed., J. L. Volakis (ed.), McGraw-Hill, 2007.
References 419
Johnson, R. C., “Mutual Gain of Radar Search Antennas,” Proceedings of the Ninth Tri-
Service Conference on Electromagnetic Compatibility, IIT Research Institute, Chicago,
Illinois, October 1963.
Johnson, R. C., “Radar Search Antennas and RFI,” IEEE Transactions on Electromagnetic
Compatibility, EMC-6, July 1964, pp. 1–8.
Johnson, R. C., “Antenna Range for Providing a Plane Wave for Antenna Measurements,”
U. S. Patent 3,302,205, January 31, 1967.
Johnson, R. C., H. A. Ecker, and J. S. Hollis, “Determination of Far-Field Antenna Patterns
from Near-Field Measurements,” Proceedings of the IEEE, December 1973;
pp. 1668–94.
Johnson, R. C., H. A. Ecker, and R. A. Moore, “Compact Range Techniques and
Measurements,” IEEE Transactions on Antennas and Propagation, September 1969,
pp. 568–76.
Joy, E. B., and D. T. Paris, “Spatial sampling and filtering in near-field Measurements,”
IEEE Transactions on Antennas and Propagation, vol. AP-20, May 1972,
pp. 253–61.
Joy, E. B., and R. E. Wilson, “Shaped Edge Serrations for Improved Compact Range
Performance,” Proceedings of the 1987 AMTA Symposium, Seattle, Washington.
Kerns, D. M., “Plane-Wave Scattering-Matrix Theory of Antennas and Antenna-Antenna
Interactions,” NBS Monograph 162, National Bureau of Standards, Boulder, CO, June
1981.
Kraus, J. D., Antennas, 2nd Edition, McGraw-Hill, 1988.
Ko, H. C., “The Distribution of Cosmic Background Radiation,” Proceedings of the IRE,
46, No. 1, January 1958, pp. 208–15.
Leach, W. M., Jr., and D. T. Paris, “Probe Compensated Near-Field Measurements on A
Cylinder,” IEEE Transactions on Antennas and Propagation, July 1973; pp. 435–45.
Lee, T. H., and W. D. Burnside, “Performance Trade-Off Between Serrated Edge and
Blended Rolled Edge for Compact Range Reflectors,” IEEE Transactions on Antennas
and Propagation, January 1996, pp. 87–96.
Long, M. W., “Wide Angle Radiation Measurements,” Proceedings of the 6th Conference
on Radio Interference Reduction and Electronic Compatibility, IIT Research Institute,
Chicago, Illinois, October 1960.
Mahmoud, M. S. A., T. H. Lee, and W. D. Burnside, “Enhanced Compact Range Concept
Using an R-Card Fence: Two-Dimensional Case,” IEEE Transactions on Antennas and
Propagation, March 2001, pp. 419–28.
Nyquist, H., “Thermal Agitation of Electric Charge in Conductors,” Physical Review, vol.
32, July 1928, pp. 110–13.
Pearson, L. W., X. Wang, D. G. Bodnar, and B. Tian, “Spherical Near-Field Scanning
Measurements of Base Station Antennas,” 2007 Antenna Systems / Short-Range
Wireless Conference, Denver, Co, Sept 26, 2007.
Pistorius, C. W., and W. D. Burnside, “An Improved Main Reflector Design for Compact
Range Reflectors,” IEEE Transactions on Antennas and Propagation, March 1987,
pp. 342–47.
Schuster, D., C. T. Stelzried, and G. S. Levy, “The Determination of Noise Temperatures

of Large Paraboloidal Antennas,” IRE Transactions on Antennas and Propagation, vol.
AP-10, no. 3, May 1962, 286–91.
Smith, P. H., “An Improved Transmission Line Calculator,” Electronics, vol. 17, January
1944, pp. 130–33.
Stelzried, C. T., “Operating Noise-Temperature Calibrations of Low-Noise Receiving
Systems,” Microwave Journal, vol. 14, June 1971, p. 41.
Volakis, J. L., Antenna Engineering Handbook, 4th ed., McGraw-Hill, 2007.
Yaghjian, A. D., “An Overview of Near-Field Antenna Measurements,” IEEE Transactions
on Antennas and Propagation, vol. AP-34, January 1986, pp. 30–45.

1. A paraboloidal-reflector antenna is designed for operation at 3 GHz. Its largest
aperture dimension is 20 feet. At what minimum distance should a far-field range
antenna be placed for use in measuring the radiation pattern?
2. An electrically short vertical radiator (monopole) at 100 kHz has an input power
of 5000 watts. At a distance of 5 km, the field strength is 0.082 volts rms
per meter. What is the radiation efficiency of this antenna? (Assume F = 1,
D = 3.)
3. The two terminals of an antenna pattern range are on opposite shores of a lake
that is 2 km wide. Two identical paraboloidal antennas are mounted on towers
at these terminals and are aimed at each other. One is used as a transmitting
antenna with a power input of 10 watts. The other antenna receives a signal
whose power level is measured and found to be 1.5 mw at the antenna output
terminals. The operating frequency is 100 MHz. It has been determined that the
propagation factor, due to the interference of direct waves and reflected waves
from the lake surface, is F = 1.9. What are the gains of these antennas?
4. A microwave lens has a circular aperture whose diameter is 3 meters. Its direc-
tivity at 10 GHz is measured and found to be D = 59,000. What is the aperture
efficiency?
5. A gain-standard horn is known to have a gain G = 12.5. It is being used to
measure the gain of a large directional antenna by the comparison method. When
the antenna being measured is connected to the receiver, it is found to be neces-
sary to insert an attenuator adjusted to attenuate by 23 dB, in order to have the
same receiver output that was observed with the horn connected. What is the
gain of the large antenna? Express the answer both as a power ratio and in
decibels.
6. It is desired to build a scale model of the antenna of Problem 1, with the largest
aperture dimension scaled to 18 inches. (a) At what frequency must this model
be operated in order to have the same pattern as the full-size antenna? (b) At
what minimum distance should a far-field range antenna now be placed for
pattern measurement?
7. A dummy antenna connected to a receiver consists of a resistance of proper value

to match the receiver input impedance. It is cooled with liquid nitrogen to a
temperature of 80 K. The receiver has a power gain factor of 1016 (160 dB). Its
bandwidth is 100 kHz. If the receiver were ideal, in the sense that it contained
no internal noise sources, what would its noise power output be, with this dummy
antenna connected to its input terminals?
8. A paraboloidal antenna used for satellite communication is pointed at the zenith
sky, and the receiver gain is advanced until a measurable noise output is observed
on an output meter. The reading is noted. Then the receiver input is switched to
a liquid-helium-cooled load at 4 K whose impedance is identical to that of the
paraboloidal antenna. An attenuator, which does not change the impedance pre-
sented to receiver, is connected between the load and the receiver. Its actual
(thermal) temperature is Tt = 300 K. When its attenuation is adjusted so that the
loss factor is L = 1.05, the noise output meter is observed to have the reading
that was noted with the antenna connected. What is the noise temperature of the
paraboloidal antenna? (Note: At high microwave frequencies, some error may
be incurred by using equations based on Nyquist’s approximation, equation
(9–17), at the very low temperature of liquid helium; but for the purposes of this
problem, assume that the equations given in this chapter are valid.)
9. A receiving system has the following properties: noise bandwidth 1.0 MHz,
antenna noise temperature 200 K, receiver noise figure 6 dB, transmission line
line thermal temperature 20° C, transmission loss 1.0 dB. What is the receiver
noise temperature referenced to the antenna output?
10. For the receiving system of Problem 9, show that the system noise temperature
is 1,365 K.
11. For the receiving system of Problem 9, find the signal-to-noise ratio expressed
in dB, for a received signal level of 10−10 watt at the antenna output.
Answers to Problems
Chapter 1
1. 2.59 v/m
3. (a) 1.93 × 10−8 w/m2. (b) 7.16 × 10−6 a/m
5. 15.52°
7. (a) 159 km. (b) 198 km.
9. 1.27 × 10−5 w/m2
Chapter 2
1. The plotted curve is a sinusoid of unit amplitude, with its negative maximum at
x approximately 0.75 m and its positive maximum at x approximately 2.25 m.
3. (a) Irms = −0.22 − j1.82 amp; |Irms| = 1.83 amp
(b) f = −96.9°
5. (a) 215.4 ohms. (b) L = 3.43 × 10−6 henry (3.43 µh)
7. (a) r = 0.52 + j 0.3; r = |r|ejf with |r| = 0.6 and f = 30°
(b) ZL = 1198 + j1125
9. 50 ohms
Chapter 3
1. (a) F; (b) T; (c) T; (d) F; (e) F; (f) F
4π
3. D = = 2.1 × 105 , or 53.2 dB
Ω
5. (a) 23.9 m2 (b) 5.5 m
7. (a) Rr = 75 − 8 = 67 ohms (b) kr = 67/75 = 0.89 (c) G = 1.46
9. N
Chapter 4
1. The radiation field is ten times the reactive field at distance r = 10 m.
3. (a) 12.5a. (b) 2.5a. (c) Zero
423
424 Answers to Problems
5. (a) n = 3 (qmax = 45°) (b) NE-SW or NW-SE

7. (a) qE = 56° and qH = 67° (b) 6.5 or 8.1 dB, using Eq. 4–24.
9. Approximately 33 ohms, from Sec. 4–3.
Chapter 5
1. (a) a = 120°.
(b) f 0°, 180° 30°, 150° 60°, 120° 90° 210°, 330° 240°, 300° 270°
E 0.50 0.71 0.83 0.87 0.26 0.07 0
sin (2π sin θ ⋅ sin φ )⋅ sin θ

3. E A(rel) =
π
4 sin  sin θ ⋅ sin φ 
2 
5. (a) 0.45p = 1.4 radians = 81°.
(b) E(90°) = 0.647; E(270°) = 0.159. Front-to-back electric-intensity ratio: 4.07
(12.2 dB).
7. (a) Area = 20 square meters; side length 4.5 meters approx.
(b) 361 dipoles.
9. The angle of the V should be 52 degrees (actually 51.7° is the result obtained by dou-
bling the angle calculated from equation (4–14) with n = 10).
Chapter 6
1. From Eq. 6–1, (D/2)2 = (4)(10)(6.4) = 256. (a) Therefore, Da = 32 feet. (b)
f/D = 10/32 = 0.3125.
3. cot b = 0.6 − 1/2.4 = 0.1833; b = 79.43°. Therefore (a) 2b = 158.9°. (b) t2 =
−12(79.43/100)2 = −7.58 dB. (c) t1 = 40 log cos b/2 = −4.56 dB. Therefore t1 + t2 =
−12.14 dB. Hence (from tabulation given on p. 238) sidelobe level is about −26 dB.
5. l = 2d = 1 meter = 3.28 feet = 39.36 inches. Da = 1.414 meters = 4.64 feet = 55.68
inches.
7. The reflector antenna is: (i) lighter in weight; (ii) less complicated; (iii) less
expensive.
9. φ = +/−5.74°
Chapter 7
1. (a) 535 MHz and 825 MHz ( f1 = 1.07f0 and f2 = 1.65f0).
(b) 500 MHz and 875 MHz ( f1 = 1.00f0 and f2 = 1.75f0).
(Answers within ± 5 MHz of these figures are acceptable.)
3. 31, 38, 48, 75, 94, 117, 146, 183, 229, and 286 Mz.
Answers to Problems 425
5. (i) crossed dipoles; (ii) horns; (iii) helixes.

7. (i) When the larger antenna would have virtually the same pattern (directivity) as the
electrically small antenna, and
(ii) the predominant noise in the receiving system is of external origin—that is, enters
by way of the antenna.
9. The required angular speed in azimuth is 3 degrees in 0.1 second, or 30 degrees per
second. This is equivalent to 1800 degrees per minute, or 5 rpm.
Chapter 8
1. 1.67 ns
3. Zero degrees
5. 2.6 ns
7. 8 degrees
9. (a) 5, (b) 11.25 degrees and 180 degrees
Chapter 9
1. 743 meters or 2438 feet
3. G = 54 or 17.3 dB
5. G = 2512 or 34 dB
7. 1.1 watt
9. 1088.9 K
11. 37.3 dB
APPENDIX A
Maxwell’s Equations
James Clerk Maxwell (1831–1879) is generally regarded the founder of electromagnetic

theory, which he conceived in 1864 (Herrera 1991). His theory began with equations that
described all previously known results on both electric and magnetic experiment and
theory. From those equations he removed a mathematical inconsistency by adding a term
to include a then unknown electrical phenomenon, called displacement current. Displace-
ment current flows even in a perfect dielectric. Consequently, his completed equations
contain both the previously known conduction current and the up-until-then unheard of
displacement current. Electromagnetic (E-M) wave propagation requires the presence of
displacement current, a concept invented by Maxwell.
A brief overview of Maxwell’s equations follows. These equations appear in numerous
books, yet the details of the information they contain are not expressible solely with a
few words. The mathematics of Maxwell’s equations is complex and thus, without
appropriate knowledge and experience, their overarching meanings are not fully appreci-
ated. These equations are of tremendous importance, because they summarize all known
laws of electromagnetics. In fact, the existence of E-M wave propagation was not
recognized until after the development of these equations.
E-M fields are described in terms of four quantities (E, H, B, D) written in bold type,
to indicate that they are vectors having both magnitude and direction. E and H are the
electric and magnetic intensities, respectively. D is the electric displacement and B is the
magnetic induction. The sources of the E-M fields are electric charge and current, which
are in general functions of both time and position. The symbol r denotes electric charge,
a scalar term having magnitude only; and current is described by current density J, a
vector.
Discourses on E-M theory often begin with a presentation of the four Maxwell equa-
tions, (A–1) through (A–4) below. The International System of Units (abbreviated SI), a
modern version of the metric system, is used. These equations employ two commonly
used partial differential functions, from vector calculus, called divergence (abbreviated
div) and curl, as follow:
427
428 Maxwell’s Equations
div D = ∇⋅ D = ρ (A–1)
div B = ∇⋅ B = 0 (A–2)
−∂B
curl E = ∇ × E = (A–3)
∂t
∂D
curl H = ∇ × H = J + (A–4)
∂t
Equations (A–5) and (A–6) that follow include the definitions of divergence and curl in
rectangular coordinates.
∂Ax ∂Ay ∂Az

div A = ∇⋅ A = + + (A–5)
∂x ∂y ∂z
 ∂A ∂Ay  ∂A ∂A  ∂Ay ∂Ax 

curl A = ∇ × A = i  z −  + j  x − z  + k  −  (A–6)
 ∂y ∂z   ∂z ∂x   ∂x ∂y 
No attempt in made here to provide the mathematics required for a comprehensive

understanding of Maxwell’s equations. Instead, the equations are included to provide a
glimpse into the nature of the Maxwell theory. Equations (A–1) through (A–4) are E-M
field equations from which the electric and magnetic fields, produced by charge density
r and current density J, can be determined. The fields also depend on the material
medium in which they exist; thus additional fundamental equations are applicable that
involve conductivity s (reciprocal of resistivity), electric permittivity e, and magnetic
permeability m. Other applicable equations are Ohm’s law, expressed as E = J/s, and the
electric and magnetic field relationships D = eE and B = mH.
A brief mention is now made of the physical meaning of Maxwell’s four equations.
Equation (A–1) can be interpreted as saying that the E-field vector is a direct measure
of the total enclosed electric charge r. From (A–2), it can be reasoned that no magnetic
flux line B has a beginning or end. Equation (A–3) is from Faraday’s law, which states
that electromagnetic force is induced in a circuit by time rate of change in B. In (A–1)
through (A–3) plus (A–7) below, Maxwell encapsulated mathematically all electrical
properties that had been experimentally verified. The development of these equations
was truly significant. However, without question, the invention of (A–4) was even more
significant, because it contained predictions that had never been observed.
Maxwell’s development of (A–4) began with an equation that can be expressed as
curl H = J, (A–7)
that describes previous experimental observations of the magnetic field H resulting from
a steady current density J. However, from vector calculus it is known that div curl A is
Maxwell’s Equations 429
zero. Therefore div J = 0, which means that “the current is always closed and there are
no sources or sinks” (Slater and Frank 1947, p. 84). Plates of a capacitor are examples
of both a source and a sink. In this regard, consider discharging a capacitor. Current starts
at the positive plate, its charge decreases as the current flows to the negative plate, and
there the charge is diminished and removed. Thus, one plate serves as a source and the
other a sink.
To mathematically include sinks and sources in electrical phenomena, Maxwell
invented displacement current density, which is ∂D/∂t. It is entirely different than conduc-
tion current density J that flows, for example, because of the conductivity of wire. Beliefs
pronounced by Maxwell include (a) displacement current only exists in the presence of
a varying field, and (b) displacement and conduction currents will produce, separately
or collectively, a magnetic field.
Thus, because matter has both conducting and dielectric properties, Maxwell
added ∂D/∂t to J in (A–7) to provide his revolutionary equation (A–4). After this
change, the divergence of curl H from (A–4) is zero and the contradiction between
mathematics and physical phenomena, as conjectured by Maxwell, is removed. Then,
the combination of (A–3) and (A–4) provide the theoretical mechanism whereby E-fields
create H-fields, and conversely; thereby creating electromagnetic waves that propagate
indefinitely.
A basis for displacement current density having amplitude ∂D/∂t follows. Let
v = voltage across two parallel capacitor plates
d = spacing of the plates
A= area of each plate
e = dielectric constant
C= eA/d, where C is capacitance
I = current input to a fixed capacitance
From basic circuit analysis, the current I that flows into a fixed capacitance is related to
the change in voltage across its capacitor plates as follows:
I = C ⋅ ( ∂v ∂t ) = ( ε A d ) ⋅ ( ∂v ∂t ) = ε A ⋅ ( ∂E ∂t ) = A ⋅ ( ∂D ∂t )
Then, the current density along the capacitance plates is I/A, that equals ∂D/∂t. Thus,
since a variable voltage applied to a capacitor forms a closed circuit, the displacement
“current” density between the plates equals the current density at the capacitor plates.
Therefore, total current density (J + ∂D/∂t) is continuous around the circuit.
Maxwell’s equations are simplified when considering plane waves in a uniform mate-
rial having permittivity e, magnetic permeability m, zero conductivity s, and zero electric
charge r. Even further simplification occurs if we assume the propagation medium is
lossless, so that s = 0. Then Maxwell’s equations, in differential form, become
div E = div ( D ε ) = 0 (A–8)
div H = div ( B µ ) = 0 (A–9)

430 Maxwell’s Equations
∂H
curl E = − µ (A–10)
∂t
∂E
curl H = ε (A–11)
∂t
Equations (A–8) through (A–11) are much simpler than (A–1) through (A–4). Equa-
tions (A–8) and (A–9) differ from (A–1) and (A–2) only because r is now zero. Equation
(A–10) says a changing magnetic field will produce an electric field, and (A–11) says
that a changing electric field will produce a magnetic field. Since conductivity is assumed
zero, (A–11) does not include the conduction current term J = s E of (A–4), that other-
wise also contributes to producing magnetic field. Equations (A–8) through (A–11) can
be solved to obtain the plane wave equations (1–2) and (1–3) of chapter 1 (see, e.g.,
Slater and Frank 1947, pp. 90–93). However, the necessary mathematical detail is not
included here. Instead, the effects of (A–10) and (A–11) on producing E-M waves are
now discussed.
As noted above, a changing electric field produces a magnetic field and that in turn
produces an electric field, and so on. Thus, there is a series of energy transfers started
when either an electric or magnetic disturbance occurs. Energy is transferred from the
electric to the magnetic field, and back to the electric field, and repeated indefinitely.
Therefore, in this process energy is being transferred from one form to the other as it
propagates through space, the result within a uniform material is that electromagnetic
waves propagates at constant velocity. Then, the effect of an electric or a magnetic dis-
turbance at one location creates an E-M wave that later reaches another location. The
resulting time delay depends on separation distance between the two locations, and on e
and m that control the propagation velocity.
Maxwell was able to calculate the velocity of electromagnetic waves, if they were to
exist, in terms of e and m. Then, based on his careful electrical measurements on e and
m in about 1865, he expected that E-M waves travel at the rate of about 3 × 108 m/s.
Furthermore, he collected all the best measurements on the speed of light, and he found
that the average of those available to him was about the same that he had predicted from
his measurements and wave calculations (Skilling 1942, p. 112). Thus, Maxwell hypoth-
esized that light is an electric wave and it appeared to him that he had substantiated his
hypothesis. However, many scientists did not accept Maxwell’s hypothesis until 1887.
This was when Heinrich R. Hertz proved that radio waves and light have similar proper-
ties, including the same velocity in a vacuum.
As a wave passes through an imaginary surface in space, there is a flow of power
through the surface area. The power density at the surface can be expressed as a vector
P, called the Poynting vector after a mathematician of the nineteenth century. In addition
to the magnitude of power density, the vector P gives the direction of power flow, that
is, direction of propagation. In terms of vector analysis,
P = E×H (A–12)
References 431
where E and H are the electric and magnetic field vectors. The magnitude of the vector
cross product A × B is A⋅B sin g, where g is the angle between A and B. Thus, since E
and H are perpendicular, the magnitude of P is simply E⋅H. With E and H denoting
effective (rms) amplitudes, the average of the Poynting vector, that is, power density, is
E2/377. The direction of the Poynting vector, as with other cross products, is obtained
from a right-hand rule. Specifically, if the fingers of the right hand curve from E to H,
the thumb shows the direction of propagation. Thus, from the relative pointing directions
of the E- and H-fields, one can determine the direction of propagation.
In conclusion, experimental optics and ray analyses for light were known before
Maxwell conceived and developed his equations, but without them a theoretical explana-
tion for the propagation of light did not exist. After Maxwell developed the equations in
1864, he used them to calculate the propagation velocity of light, and the calculations
agreed with previously measured values. Still later, in 1887, Heinrich Hertz demonstrated
that the velocities (and other properties) of radio waves and light were the same. Thus,
from the use of Maxwell’s equations and experimental observations, the theory of E-M
waves was established. Today it is known that the phenomenon of E-M radiation covers
the entire energy spectrum from very high frequency cosmic rays to low-frequency, long-
wavelength radio waves.
References
Herrera, J. C., “Electromagnetic Radiation,” in Encyclopedia of Physics, 2nd ed., by R. G.
Lerner, R. G. and G. L. Trigg, VCH Publishers, 1991, pp. 285–88.
Skilling, H. H., Fundamentals of Electric Waves, John Wiley and Sons, 1942, p. 112.
Slater, J. C., and N. H. Frank, Electromagnetism, McGraw-Hill, 1947.
APPENDIX B
Polarization Theory
Wave polarization is introduced in sec. 1.1.3 and further discussed in sec. 3.10. Polariza-
tion is a property of an electromagnetic wave, which describes the time-varying direction
and amplitude of its E-vector. Specifically, polarization describes the figure traced as
a function of time of the extremity of the E-vector at a fixed location in space, when
observed looking in the direction of propagation. Polarization of an antenna is defined
as the polarization of the far-field wave that it radiates. As discussed in secs. 1.1.3 and
3.10, linearly polarized waves are the most commonly used. When the amplitude of
E-vector varies and the E-vector points only up and down, the wave is vertically or V
polarized. Similarly, if the amplitude and direction of the E-vector varies and its direction
is only horizontal, the wave is horizontally or H polarized. Circular polarization (CP) is
the next most frequently used polarization. Its E-field has constant magnitude, but its
direction rotates (within a plane) continuously through 360° and it completes a cycle in
one wave period.
The most general form of a wave polarization is elliptical. The E-vector of an ellipti-
cally polarized wave can be regarded from three viewpoints. Namely, it can be considered
to be (1) a rotating vector, the end point of which traces out an elliptical helix whose
axis lies in the direction of propagation; (2) the resultant of the E-vectors of two linearly
polarized waves of the same frequency; or (3) the resultant of the oppositely rotating E-
vectors of two circularly polarized waves of the same frequency. Figure B–1 illustrates
the E-vector of an elliptically polarized wave at various positions in space, at a fixed
instant of time. An ellipse within an x-y plane is shown at the left end of this spiraling
wave. The ellipse is created by tracing the end-point (terminus) of the E-vector onto the
x-y plane, during the duration of one rf cycle.
B.1. Elliptically Polarized Waves Resolved into Linearly

Polarized Components
An elliptically polarized wave can be created by the vector sum of two linearly polarized
waves, polarized along the unit vectors 1x and 1y as follows:
E = 1x E x + 1y E y (B–1)
433
434 Polarization Theory
Transmitting where
Antenna
E x = E xm sin (ω t − γ z ) (B–2)
and
Y E y = E ym sin (ω t − γ z + δ ) (B–3)
Left Screw-Sense in Space
X In (B–2) and (B–3), Exm and Eym are the
Z Counterclockwise Rotation in Plane peak amplitudes of the x- and y-polarized
FIGURE B–1. electric fields.
Equations (B–1) through (B–3) describe
A left circularly polarized wave exhibits
counterclockwise rotation of the E-vector in a an elliptically polarized wave of radial
plane, when viewed in propagation direction Z. frequency w and with propagation con-
(Adapted from Huynen 1970, p. 10.) stant g traveling in the positive z direc-
tion. There is a phase difference d between
its x- and y-directed components, which
Eym
are in space quadrature and which have
maximum values Exm and Eym. If viewed
in the direction of propagation, the end-
E2 point of the E-vector traces an ellipse on
a fixed plane in space, as shown in
Fig. B–2. This ellipse is known as a
polarization ellipse.
Exm Readers will recognize the similarity
of a polarization ellipse and a Lissajous
E1
figure. A Lissajous figure is used to deter-
mine relative amplitude and relative phase
of two equal-frequency sinusoids with an
oscilloscope. In the polarization ellipse of
Fig. B–2, the E-field is obviously largest
Note: when pointed along the major axis and
+z is into page smallest when pointed along the minor
+x is to the left
+y is up axis. Also, notice that the field strengths
in Fig. B–2 along the x- and y-axes are
FIGURE B–2.
smaller than Exm and Eym, respectively.
Resolution of an elliptically polarized wave These properties, as well as major-axis tilt
into linearly polarized components along the +x
and +y axes. and E-vector rotation direction, depend
on the relative phase d and the ratio
Exm/Eym.
Now examine (B–1) through (B–3). If Eym = 0, the E-vector varies in magnitude and
direction, but it is always along the x-axis. Thus, the wave is linearly polarized in the x
direction. Similarly, if Exm = 0, the E-vector is always along the y-axis and the wave is
linearly polarized in the y direction. Now suppose Exm ≠ 0, Eym ≠ 0, and d = 0 or
Elliptically Polarized Waves Resolved into Circularly Polarized Components 435
d = 180°. Then the amplitudes Ex and Ey 2

are synchronized, being zero and at their Eym
1
maximum amplitudes (Exm and Eym) Exm
1
simultaneously. In this case, however, the 2
tilt angle of the linearly polarized E-field –180° –135° –90° –45° –0° +45° +90° +135° +180°
then depends on the ratio Exm/Eym. With δ
Exm = Eym and the special cases d = 0 or FIGURE B–3.
d = 180°, the linear polarization is ori- Polarization ellipses with waves propagating
ented at 45° to the X- and the Y-axes. into the page. The polarization is right circular
Other special cases are when d = ±90°. if d = −90° and left circular if d = +90°.
Then, the major axis will lie along the x-
axis if Exm > Eym and along the y-axis if
Eym > Exm. After horizontal and vertical
polarizations, circular polarization (CP) EL E
is the most frequently used (sec. 7.3). CP
is the special case of d = ±90° and Exm =
Eym. Using (B–2) and (B–3), let d = 90°
and t = z = 0; then Ex = 0 and Ey = Eym.
ER
Now consider a later time so that wt =
90°; then Ex = Exm and Ey = 0. Thus, E
rotates from the +y direction to the +x
direction during the time that causes wt to
change from zero to 90°. In other words,
E rotates counter-clockwise when looking
outward (the direction of propagation);
that is, the polarization is left circular. Note:
Correspondingly, if d = −90° the polariza- +z is into page
+x is to the left
tion is right circular. +y is up
It should now be apparent that knowl-
FIGURE B–4.
edge of the amplitudes of the two orthog-
onal (perpendicular) linear components Resolution of a polarization ellipse into a left-
and the phase difference between them is circular EL and a right-circular ER polarized
wave. Note that the relative phase D of EL and
sufficient to define a polarization ellipse. ER is determined by the orientation of the
Figure B–3 includes the polarization major axis.
ellipses for selected ratios Eym/Exm and
relative phases d.
B.2. Elliptically Polarized Waves Resolved into Circularly

The resolution of the rotating E-vector into two circularly polarized components is shown
in Fig. B–4, where EL and ER are the magnitudes of left circular and right circular waves
respectively. A circularly polarized wave which has a clockwise rotation of the E-vector
when one looks in the direction of propagation is defined as a right circular, or RC,
436 Polarization Theory
polarized wave. Similarly, the E-vector of a left circular, or LC, polarized wave rotates
counterclockwise if the observer looks in the direction of propagation.
An elliptically polarized wave can be created by the vector sum of an RC and an LC
wave. This is done below by using phasor notation to indicate the E-field direction versus
both time and position. In other words, the phasor serves the role of a unit vector having
variable direction. In (B–4) through (B–6) below, EL and ER are the constant amplitudes
of the left- and right-circular polarized waves, and bold type indicates a vector having
both amplitude and direction.
E = EL + ER (B–4)
where
EL = EL e j (w t −g z) (B–5)
and
E R = E R e - j (w t - g z + D ) (B–6)
In (B–6), D is the difference in the pointing directions of EL and EL when EL is pointed

along the reference axis. The positive exponent indicates counterclockwise rotation and
the negative exponent indicates clockwise rotation. If EL = ER, the polarization is linear
and its orientation depends on the relative phase D. Note that, at a point in space, EL and
EL rotate in opposite directions. Therefore, since their amplitudes are equal, the peak
amplitude for the resulting linearly polarized wave is 2EL = 2ER.
From Fig. B–4, it can be seen that the major and minor semi-axes of this ellipse are
the sum and difference, respectively, of ER and EL. The ratio of the major to the minor
axis is called the axial ratio, which can be expressed as
E R + EL
AR = (B–7)
E R − EL
From Fig. B–4 it also can be determined that the rotation direction of the E-vector is
the same as that of the larger of ER and EL. Thus the ratio of the magnitudes of the RC
and LC waves determines the direction of rotation, in addition to the axial ratio AR.
B.3. Relationships between Linearly and Circularly

A relationship between the linear and circular resolutions may be obtained by considering
the components of the circular waves along the x- and y-axes. The real parts of (B–5)
and (B–6) are along the x-axis, and the imaginary parts are along the y-axis. Then the
magnitudes of the linearly polarized components are as follows:
References 437
E x = EL cos (ω t − γ z ) + ER cos (ω t − γ z + ∆ ) (B–8)
E y = EL sin (ω t − γ z ) − ER sin (ω t − γ z + ∆ ) (B–9)
More detailed discussion of the theory of polarized waves can be found in Kraus
(1988, pp. 70–80), and in Stutzman (1993, pp. 1–65). Stutzman is also recommended
reading on multipolarization antennas, systems, and phenomena. Long (2001) is recom-
mended for reading on the polarization and depolarization properties of radar echoes
from land and sea.
The magnitudes of the four polarizations (H, V, RC, LC) may be received simultane-
ously with four separate antennas. Alternately, the separate signals from two orthogonal
linear or two orthogonal circular polarization antennas, when combined appropriately,
can be used to obtain simultaneously the magnitudes of the four separately polarized
signals. Further, a dual linear or a dual circular polarization feed (coupler) can be
employed, thereby allowing the amplitudes of each of the four polarizations to be simul-
taneously obtained with an appropriately designed antenna.
References
Huynen, J. R., Phenomenological Theory of Radar Targets, Drukkerij Bronder-Offset N.V.,
1970, p. 10.
Kraus, J, D., Antennas, 2nd ed., McGraw-Hill, 1988.
Long, M. W., Radar Reflectivity of Land and Sea, 3rd ed., Artech House, 2001.
Stutzman, W. L., Polarization in Electromagnetic Systems, Artech House, 1993.
APPENDIX C
Review of Complex-
Variable Algebra
Complex variables are (variable) numbers containing the factor j, which is a symbol
denoting −1. The notation −1 means “the number that, when squared (multiplied by
itself), equals −1.” But none of the “real numbers” fits this specification. There is no real
number that, when multiplied by itself, equals −1. When any real number, positive or
negative, is squared, the result is a positive number. In particular, (−1)2 = +1.
Therefore, the “number” that, when squared, equals −1, represents an extension of the
system of real numbers. It is called an “imaginary” number—possibly unfortunate ter-
minology, since it implies something that does not “really” exist. Imaginary numbers
exist just as much as do the so-called real numbers. Although j is “imaginary,” it need
not be mysterious. It is just another way of writing −1. It is a different kind of number
from the familiar real numbers, which are 0, ±1, ±2, ±3, etc. Yet obviously it has some-
thing in common with them, since the symbol “1,” which is part of the quantity −1, is
a real number.
When j (or as it is denoted in physics, i) is multiplied by any real number (for example,
j5 or −j6), the resulting product is an imaginary number. Thus the totality of imaginary
numbers is obtained by multiplying j by all the real numbers. Another way of writing j5
is 5 −1 or −25 . Another way of writing −j6 is −6 −1 or − −36 . (Note that − −36
is not the same thing as +36 , which is of course a real number equal to either +6 or
−6.)
If two imaginary numbers are multiplied together (for example j3 × j5), the ordi-
nary rules of algebraic multiplication apply. In other words, j3 × j5 = j215. But
j 2 = j × j = −1 × −1 = −1 . Therefore, j3 × j5 = −15, which is a real number. Thus the
product of two imaginary numbers is a real number (further evidence that imaginary
numbers are not wholly “unreal”).
By the same process of reasoning, all even powers of j are real numbers. Thus, j 2 = −1
(as was just shown); j 4 = j 2 × j 2 = −1 × −1 = +1; j 6 = −1; j 8 = +1; and so on. On the
other hand, odd powers of j are imaginary: j 3 = j 2 × j = −j; j 5 = j 4 × j = +j; and so forth.
The reciprocal of j is equal to −j, as can readily be seen by the following argument:
1/j = j/j 2 = j/(−1) = −j. Reciprocals of odd powers of j may thus be shown equal to minus
the value of the odd power; that is: 1/j 3 = −j 3 = +j, etc. Reciprocals of even powers are
equal to the even power: 1/j 2 = j 2; 1/j 4 = j 4; etc. Reciprocal quantities are of course also
439
440 Review of Complex-Variable Algebra
expressible with negative exponents; 1/j = j −1; 1/j 2 = j −2; and so on. In short, j may be
manipulated algebraically in exactly the same way as any symbol.
A complex number is the sum of a real number and an imaginary number. Thus if
z = x + jy, where both x and y are any real numbers, then z is a complex number. (Actu-
ally, since either x or y may have the value zero, in which case z will be either purely
imaginary or purely real, the purely real and purely imaginary numbers are both special
cases of complex numbers.) The real number x is called the real part of z, and the real
number y is called the imaginary part of z. This statement is sometimes written in the
abbreviated form: Re (z) = x; Im (z) = y. Notice that y is a real number, though it is called
the imaginary part of z. The imaginary number is jy.
Manipulation of complex numbers follows all the rules of ordinary algebra. The
symbol j is treated like any other algebraic symbol, although it may at times be conve-
nient to convert expressions like j2 and j3 into their equivalent forms, −1 and −j, and
so on.
As an example, if the two complex numbers z = x + jy and w = u + jv are to be added,
the result is
z + w = ( x + u) + j( y + v ) (C–1)
That is, the real part of the sum is the sum of the real parts of the added complex numbers,
and the imaginary part of the sum is the sum of the imaginary parts.
Multiplication of imaginary numbers is performed as follows:
zw = ( x + jy ) (u + jv ) = xu + jxv + jxu + j 2 yv = ( xu − yv ) + j ( xv + yu ) (C–2)
(since j2 = −1).
The real numbers are represented as points on a line, with one point designated zero;
numbers to the left are “mirror images” of numbers to the right except that those to the
left are negative and those on the right are positive. The complex numbers are represented
as points in a plane by a cartesian coordinate system. The horizontal coordinate axis
is the real-number line just described, and the vertical coordinate axis is an identical
imaginary-number line, its zero coinciding with the real-axis zero. This complex-plane
representation is illustrated in Fig. C–1 for the particular complex numbers x = 4, y = 3
(z = 4 + j3), and x = −2, y = 5 (z = −2 + j5).
In this representation the direct or straight-line distance from the origin to the
complex-number point is called its modulus or absolute value. The modulus is (by
definition) a positive real number, regardless of which quadrant of the coordinate plane
contains the number point. It is the length of a line. The modulus of z1 in Fig. C–1 is
designated r1. From the Pythagorean theorem for right triangles it is apparent that
r1 = x12 + y12 = 42 + 32 = 25 = 5 . That is, the modulus of a complex number is the
(positive) square root of the sum of the squares of its real and imaginary parts. The
modulus is denoted algebraically by the “absolute value” brackets ||; thus:
z = x + jy = x 2 + y2 (C–3)
Review of Complex-Variable Algebra 441
The angle (q1 in Fig. C–1) that the + jy

z2 = –2 + j5
modulus line forms with the real part of
z1 = 4 + j3
the complex number is called the phase
angle. Its tangent is the ratio of the imagi- r1
y1
nary part (y1) to the real part (x1). Certain q1
physical quantities characterized by phase –x x
0
x1
angles are conveniently represented by a
“directed line segment” called a phasor,
having a length equal to the modulus of a
– jy
complex number, and a direction given by
its phase angle. Accordingly, such a quan- FIGURE C–1.
tity is represented by a complex number. Complex-plane representation of real and
Phasors are very similar in their nature imaginary numbers (z-plane).
to two-dimensional vectors, but the latter
term is properly reserved for physical quantities with two rectangular components having
mutually perpendicular directions in space. Thus, an electric field intensity is a vector,
which may be represented by the vertically and horizontally polarized components. A
voltage developed by a current flowing through a resistance and a reactance in series, on
the other hand, may be represented as a phasor, of which the real part represents the
component of the voltage in phase with the current, and the imaginary part the component
90 degrees out of phase with the current. Similarly an impedance is a phasor whose real
part is the resistance and whose imaginary part is the reactance.
Complex numbers may also be represented by a complex-exponential notation based
on the relationship of the real and imaginary parts to the modulus and the phase angle.
In this notation the complex number z = x + jy is written
z = z e jθ = re jθ (C–4)
The equivalence of this representation and the original one is based on deMoivre’s
theorem, which defines the meaning of a complex exponent as follows:
e jθ = cos θ + j sin θ (C–5)
(The proof of this theorem is based on concepts of advanced calculus.) Rewriting (C–4)
in terms of deMoivre’s theorem gives
z = r cos θ + jr sin θ (C–6)
By inspection of the right triangle of Fig. C–1 (in which z = z1, r = r1, etc.) it is apparent
that
x = r cos θ 
 (C–7)
y = r sin θ 
442 Review of Complex-Variable Algebra
Substituting these results into (C–6) gives
z = x + jy (C–8)
as was to be shown. The complex exponential notation is extremely useful in mathemati-

cal analyses involving complex numbers.
If two complex numbers have the same real parts, and imaginary parts of the same
magnitude but opposite sign, they are called complex conjugates of each other. Thus the
complex conjugate of (x + jy) is (x − jy), and vice versa.
Multiplying a complex number by its conjugate yields the square of the modulus;
thus:
( x + jy ) ( x − jy ) = x 2 − j 2 y 2 = x 2 + y 2 (C–9)
In the complex exponential notation the conjugate of rejq is re−jq. This is shown from de
Moivre’s theorem in the following way:
e − jθ = e j( −θ ) = cos ( −θ ) + j sin ( −θ )
(C–10)
= cos θ − j sin θ
because cos(−q) = cos(+q) and sin(−q) = −sin(+q). That the product of the complex
conjugates yields the square of the modulus is also readily seen from this notation, as
follows:
re jθ ⋅ re − jθ = r 2 e( jθ − jθ )
(C–11)
= r 2 e0 = r 2
In algebraic equations involving products and quotients of complex numbers, it is

usually required to reduce each side of the equation to separated real and imaginary parts.
For products, this is accomplished simply by performing the indicated multiplication, as
exemplified by (C–2), and grouping together the real terms and the imaginary terms. For
a quotient, the procedure is slightly more complicated, but basically simple. It consists
of first multiplying both the numerator and the denominator of the fraction representing
the quotient by the complex conjugate of the denominator. This procedure eliminates
imaginary terms from the denominator. Then the numerator is separated into real and
imaginary parts. For example:
a + jb ( a + jb ) ( c − jd ) ac + jbc − jad − j 2 bd ac + bd  bc − ad 
= = = 2 + j 2 (C–12)
c + jd ( c + jd ) ( c − jd ) c −j d
2 2 2
c +d 2  c + d 2 
An important theorem of complex algebra is that if two complex numbers are equal
to each other, their real parts and their imaginary parts are separately equal. This means
that every complex equation is equivalent to two real equations. Thus, if
Review of Complex-Variable Algebra 443
x + jy = u + jv (C–13)
it necessarily follows that
x=u
y=v (C–14)
(two equations). This theorem has many applications. For example, the relationship
between impedance and admittance is Z = 1/Y, where in general Z = R + jX and Y =
G + jB. Therefore
1 R − jX R  X 
G + jB = = 2 = 2 − j 2 (C–15)
R + jX R + X 2
R +X 2  R + X 2 
Therefore, applying the theorem, it is found that
R
G= (C–16)
R + X2
2
−X
B= (C–17)
R + X2
2
as was stated without detailed proof in equations (2–30) and (2–31), chapter 2.
APPENDIX D
Complex Reflection Coefficients

and Multipath Effects
Chapter 1 (sec. 1.4.5) includes an introduction to the polarization dependencies of the

reflection coefficient for a smooth, flat surface and to the effects of multipath propaga-
tion. The present appendix includes equations and example calculations for reflection
coefficients of smooth and rough earth surfaces. Furthermore, equations are given for
calculating the pattern propagation factor F for both flat earth and spherical earth geom-
etries. In addition, the SM that accompanies this book provides examples using Mathcad
for calculating reflection coefficients and the pattern propagation factor F. Therefore, by
changing the initial assumptions, the user may readily attain computer calculated solu-
tions to a wide variety of problems involving reflections applicable to either flat earth or
spherical earth geometries.
D.1. Fresnel’s Equations

Fresnel’s equations are basic relationships that give reflection coefficient magnitude and
phase as a function of polarization, arrival angle of incoming e-m waves, and complex
permittivity. These equations include the cases with electric vector perpendicular to and
in the plane of incidence. The two cases are indicated by EH and EV in Fig. D–1, and
correspond to horizontal and vertical polarizations. Elsewhere, these components are
sometimes designated as Eparallel and Eperpendicular, respectively. Note that the direction
for vertical polarization is perpendicular to horizontal polarization and to the direction
of propagation. Otherwise, it is not precise because the direction of vertical polarization
depends on angle Y. For calculations involving reflections of an elliptically polarized
wave (Appendix B), the Fresnel equations are used to separately determine the reflection
coefficients for the horizontally and vertically polarized components.
Now assume that the electromagnetic wave incident on the surface is in free space
and the surface material is nonmagnetic, as it is ordinarily for the earth. Then, from
Fresnel’s equations, the complex (amplitude and phase) reflection coefficients for hori-
zontal and vertical polarizations (GH and GV) follow:
445
446 Complex Reflection Coefficients and Multipath Effects
sin ψ − (ε r − cos2 ψ )
12
ΓH = = ρ H exp ( − jφ H ) (D–1)
sin ψ + (ε r − cos2 ψ )
12
ε r sin ψ − (ε r − cos2 ψ )
12
ΓV = = ρV exp ( − jφV ) (D–2)

ε r sin ψ + (ε r − cos2 ψ )
12
The reader will notice that (D–1) and (D–2) contain −fH and −fv terms, meaning that
positive fH and fv denote phase delays. This use of phase delay follows Kerr (1947, pp.
396–97) and is consistent with the fact that the phase of reflected waves for land and sea
lags that of the incidence wave. This practice of defining fH and fv as phase lags differs
from sec. 1.2, where positive f denotes leading phase. As also discussed in sec. D.2,
there can be uncertainties regarding the correct phases when using (D–1) and (D–2),
because the square root of a complex (real and imagery parts) number has two answers
and they are at different angles. Consequently, because of this ambiguity in phase, it is
useful to remember that, for land and sea, reflected phase always lags.
As shown in Fig. D–1, the angle y is the angle between the incident wave (and the
reflected wave) and the tangent to the surface at the point of reflection. The term er is
the complex permittivity, relative to permittivity of vacuum, expressed as
ε r′ − jε r′′ = ε r′ − j60λσ (D–3)
where l is the free-space wavelength in meters and s is conductivity in mhos per meter.
Example values of e′r and e″r for soil and water are given in Table D-1. By definition,
er = 1 for a vacuum and it is essentially the same for the earth’s atmosphere. In addition,
Appendix H includes the electrical properties of some low-loss dielectric materials used
for radomes.
For normal incidence (y = 90°), where EH and EV are parallel to the reflecting plane,
the H and V reflection mechanisms are necessarily identical. However, from (D–1) and
(D–2) GH and GV become
Path of Path of
Incident Wave Reflected Wave 1 − ε r1 2
EV V EV
ΓH = (D–4)
1 + ε r1 2
EH H EH
ε r − ε 1r 2 ⎛ ε −1 2 ⎞ ⎛ ε r − ε 1r 2 ⎞
ΓV = = = −ΓH
ε r + ε 1r 2 ⎜⎝ ε −1 2 ⎟⎠ ⎜⎝ ε r + ε 1r 2 ⎟⎠
y y Horizontal (D–5)
FIGURE D–1. The cause for the sign difference between

Reflection from a smooth, flat surface. EH and GV and GH is based solely on geometry
EV are electric vectors for horizontal and and without regard to phase change, per
vertical polarizations, respectively. se. Note from Fig. D–1 that the direction
Reflection Coefficients for Smooth Land and Sea 447
TABLE D-1 Approximate Electrical Properties of Soil and Water Sources: Kerr (1947,
p. 398) and Dicaudo (1970, p. 14–30)
Material l s (mhos/m) e′r e″r

Vacuum (defined) 0 1 0
Ice 3 cm 3.2 25.6 × 10−4
Snow 3 cm 1.3 6.5 × 10−4
Seawater 3 m/20 cm 4.3 80 774/52
Freshwater lakes 1m 10−3/10−2 80 0.06/0.60
Moist ground 1m 10−2 30 0.6
Very dry ground 1m 10−4 4 0.006
Very wet sandy loam 9 cm 0.06 24 32.4
Very dry sandy loam 9 cm 0.03 2 1.62
Texas, very dry soil 3.2 cm 0.0074 2.8 0.014
Note: e″r = j 60 ls, and thus it is highly sensitive to wavelength. The wavelength dependency of e′r is much less, and its values given in
Table D-1 are typical of the kilohertz to gigahertz frequency range. Note that for ground, e′r and e″r each increases with an increase in
moisture content.
of the reflected EV at Y = 90° is opposite 1.2

Magnitude of Reflection Coefficient
to its incident direction at Y = 90°.

1
However, a change in E-field direction
does not occur for EV at Y = 0°. Further- 0.8
more, the change in E-direction does not
occur for EH at either Y = 90° or Y = 0°. 0.6
Thus at y = 90° and with equal phase
0.4
change upon reflection, GH and GV must
differ in sign. 0.2
0
0 10 20 30 40 50 60 70 80 90
Grazing Angle in Degrees
D.2. Reflection Coefficients for V-POL, 10 MHz
Smooth Land and Sea V POL, 10 GHz
H-POL, 10 MHz
Figures D–2 and D–3 show the general H-POL, 10 GHz
trends in reflection coefficient versus FIGURE D–2.
grazing angle y and frequency for a flat, Magnitude r of reflection coefficient for
smooth seawater surface. Figure D–2 smooth sea surface versus grazing angle Y.
includes reflection coefficient amplitude 10 MHz & 10 GHz; horizontal & vertical
r for both H- and V-polarizations at two polarizations.
frequencies, 10 MHz and 10 GHz, with
e′r = 80 and s = 4.3 mhos/m assumed.
Note that r is essentially unity at zero grazing angle for each polarization and frequency.
The magnitude rH is usually close to unity at all y angles, but it drops to as low as 0.8
at the higher frequencies and larger y values. Figure D–3 shows examples of phase delay
on reflection for H- and V-polarizations at 10 MHz and 10 GHz. For H-polarization, the
200
Phase Delay of Reflection Coefficient
phase delay is 180° (or slightly greater)

180 for all grazing angles and the two
160 frequencies.
140 From Figs. D–2 and D–3, it is seen that
120 rV and fV are more sensitive to frequency
100 and grazing angle than are rH and fH.
80 However, GV and GH are both −1 at zero
60 grazing angle (horizon direction) at all
40
frequencies. Note that rV has minimum
20
values at small angles Y, and that the par-
0
0 10 20 30 40 50 60 70 80 90 ticular Y for the minimum rV depends on
Grazing Angle in Degrees frequency. For a lossless dielectric, the
V-POL, 10 MHz
V-POL, 10 GHz minimum for rV is zero and that angle Y
H-POL, 10 MHz is known as Brewster’s angle. For imper-
H-POL, 10 GHz
fect dielectrics such as land and sea, the
FIGURE D–3. minimum is not zero and the angle is the
Phase delay f of reflection coefficient for pseudo-Brewster angle. From Fig. D–3,
smooth sea surface versus grazing angle y. it can be seen that, near the pseudo-
10 MHz & 10 GHz; horizontal & vertical Brewster angle, the phase delay fV
polarizations.
changes rapidly between 180° and 90°
with increases in grazing angle, and then
it slowly approaches zero phase delay with further increases in y. In addition, the pseudo-
Brewster angle decreases with decreases in frequency.
In sec. D.1, the reader is cautioned regarding the correct phases when using (D–1) and
(D–2), because the square root of a complex number has two answers with different
angles. Kerr (1947, pp. 400–401) discusses how the complex numerators and denomina-
tors of the Fresnel equations vary versus y; and he thereby resolves the ambiguity in
phase and establishes the general shapes of fH and fV versus y. Namely, in accordance
with Fig. D–3, fH is 180° at y = 0 (grazing) and increases slightly with increases in y;
and fV is also 180° at y = 0, but it decreases to 0° at y = 90°. Thus, as discussed in sec.
D.1, fH = −fV when V- and H-polarized waves arrive normal to a surface, because of the
defined pointing directions of the two polarizations.
A computer may supply an incorrect angle, because software often provides only one
answer for a multivalued function. In Mathcad, for example, the answer given for the
root of a complex term is the principal value, that is, the smallest positive angle relative
to the positive real axis in the complex plane. However, ambiguities regarding signs and
magnitudes of fH and fV are properly resolved when their general shapes versus y are
consistent with the paragraph above.
Sections SM 1.1 and 1.2, in the SM that accompanies this book, provide numerical
examples using Mathcad for calculating complex reflection coefficients for smooth flat
land and sea surfaces. There it can be seen that the shapes of rV, rH, fV, and fH versus
grazing angle for land are, generally, similar to those of seawater. However, land is a
poorer conductor than sea water, the pseudo-Brewster angles are somewhat larger, and
at most grazing angles rv and rh are smaller for land than for sea. Furthermore, as the
Reflection Coefficients for Rough Surfaces 449
grazing angles approach zero, the complex reflection coefficients for horizontal and verti-
cal polarizations approach −1 for both land and sea.
D.3. Reflection Coefficients for Rough Surfaces

The field scattered by a rough surface is the sum of two components: a specular compo-
nent and a diffuse component. For specular reflections, the angle of incidence and the
angle of reflection are equal. However, diffuse scattering can occur in any direction rela-
tive to the incident EM field. A specularly reflected component is a coherent wave, that
is, one that has deterministic amplitude and phase and can be described as a function of
position. For example, the sum of a specularly reflected field and a coherent “direct” field
will produce a periodically varying field, if sampled in the vertical dimension above the
flat surface. On the other hand, the amplitude and phase of a diffusely scattered field are
unpredictable and are random functions of position. Therefore, a diffusely scattered field
added to a coherent field will produce a total field that varies randomly as a function of
position.
The Rayleigh criterion is introduced in sec. 1.2, as being a rule-of-thumb for estimat-
ing if a semi-rough surface will reflect as if the surface were smooth (specular reflector).
There is not a specific, clearly defined specular/nonspecular boundary, because the
amount of specularity gradually changes with changes in surface roughness, incidence
angle, and wavelength. Note that in Appendix D and the accompanying SM, the grazing
angle y of Fig. D–1 is used, instead of qi of Fig. 1–6, because of its convenience in
multipath analyses. Thus, y is the complement of qi, and sin y replaces cos qi in equation
(1–25).
A useful characterization of surface roughness is sh, the standard deviation of the
surface height distribution. Mathematically, sh equals the rms of heights on the surface,
if the average height is the zero reference height. As a guideline for roughness, following
Rayleigh’s criterion (chapter 1), a surface is considered not smooth if
σ h sin ψ ≥ λ 8 (D–6)
Experience indicates that the “significant wave height,” the height of a sea wave that
most observers guess from a quick observation, exceeds sh of sea surfaces by a factor
of about four. Thus, the standard deviation is actually substantially smaller than peak-to-
trough surface height.
A further refinement in calculating the effects of roughness provides an estimate of
the effective value of the magnitude of G, as follows:
  2πσ h sin ψ  2 
Γ = ρo exp  −2    (D–7)
  λ 
Here ro is the magnitude of the reflection coefficient for a smooth, flat surface. From
(D–7), |G | is approximately ro if sh sin y << l and it equals 0.29ro if sh sin y = l/8.
Thus, as is already known, surface reflections become more specular-like with decreased
roughness, decreased grazing angle, and with increased wavelength (decreased

frequency).
D.4. Pattern-Propagation Factor F

Effects of the earth and its environment on received signal strength are conventionally
included by means of the pattern propagation factor F (sec 1.4.7). Pattern-propagation
factor is, according to the Institute of Electrical and Electronics Engineers (1997), the
ratio of the field strength amplitude that is actually present at a point in space to that
which would be present if free-space propagation had occurred with the antenna beam
directed toward the point in question. Consequently, the power at a point is F2 times its
value if the propagation path were in a vacuum. Therefore, as discussed in sec. 1.4.7,
received power Pr for one-way transmission can be expressed as
t t Gr λ
2
PG
Pr = F2 (D–8)
( 4π ) R
2 2
where Pt is transmit power, Gt and Gr are transmit and receive antenna gains, and R is
distance between transmit and receive antennas.
Figure D–4 shows the basic multipath process that consists of the phasor addition of
waves at point P. The shortest path is the direct path and it has length designated as Rd,
Rd
A R2
d
h2
R1
r
h1
y y
B
–h1
FIGURE D–4.
Geometry for a flat earth, with image of source A located at −h2.
Flat Earth Geometry 451
and the other path is the indirect path and has length R1 + R2. The phase difference of
the two waves is affected by three factors: (l) the path-length difference d which equals
(R1 + R2 − Rd), (2) the phase change of the reflected wave that occurs upon reflection,
and (3) the phase difference, if any, of the fields radiated by the antenna in the direct-ray
and reflected-ray directions (qd and qr, respectively).
The two waves may also have amplitude differences caused by the following factors:
(l) each wave is subject to the inverse-square-law power-density reduction, and they
travel different distances, (2) the reflected wave undergoes at least some loss of intensity
in the process of reflection, and (3) antenna patterns are, in general, not constant in
the vertical plane and may therefore have different field intensities in the qd and qr
directions.
In most multipath situations, the path difference d is small compared to the path lengths
of the direct and reflected paths. Therefore, the usual small difference in amplitudes
caused by these path length differences d can be neglected, and the primary effect of d
is on the phase difference between the waves that arrive from the direct and reflected
paths.
As already noted, the elevation pattern of the antenna affects the relative strength of
the rays in directions qd and qr, and therefore possible effects of the pattern must be
addressed. Often in multipath situations, however, the angles qd and qr are small enough
that the pattern’s relative strengths at those angles are insignificant. Then, if both the
direct and indirect paths are energized equally, the resultant electric field at point P is
E = Ed {1 + Γ e − j(2π λ )δ } = Ed {1 + ρe − j[(2π λ )δ +φ ] } (D–9)
where Ed is the field due to the direct wave, d is the difference in lengths between the
indirect and direct paths, and f is the phase delay created by the reflecting properties of
a surface. Therefore, in accordance with (D–9), the propagation factor F is
F = E Ed = 1 + ρe − j[(2π λ )δ +φ ] (D–10)
Note that F consists of two components, of amplitude one and r, separated by angle
(2p/l)d + f. Thus, from use of the law of cosines, it is seen that (D–10) becomes
F 2 = (1 + ρ 2 + 2 ρ cos ψ ) (D–11)
where Y = (2p/l)d + f.
D.5. Flat Earth Geometry

The flat earth approximation is useful for applications involving relatively short ranges
and small transmit and receive heights, so that the effects of the earth’s curvature are
negligible. This section includes the equations applicable to the flat earth approximation
and sec. D.7 discusses the spherical earth model. Section 1.4.3, chapter 1, describes a
simplified flat earth multipath model based on a number of assumptions, including
• a flat, smooth, and perfectly conducting surface

• direct- and reflected-wave paths that are effectively parallel at the point of
observation, and
• horizontal polarization.
Often the antenna height h1 above the earth is small compared to Rd, the distance between
the antenna and the observation point P. Then, it may be true that Rd >> h1 and thus
consequently R2 is effectively parallel to Rd. With these assumptions, there are several
simplifications regarding Fig. D–4; namely, y = qd, d = 2h1 sin qd, and knowledge of h2
is not required. In general, however, accurate multipath calculations require the use of
the more detailed geometrical equations, such as those that follow and those for the
spherical earth model included in sec. D.7.
Figure D–4 provides for a more general analysis of the flat earth model, and the
following equations are applicable:
2
h −h
Rd = ( h2 − h1 ) + G = G 1 +  2 1 
2 2
(D–12)
 G 
2
h +h
R1 + R2 = (h2 + h1 )2 + G 2 = G 1 +  2 1  (D–13)
 G 
When G2 > > (h1+ h2)2, useful approximations are
 1 h − h 2
Rd ≈ G 1 +  2 1   (D–14)
 2 G  
and
 1 h + h 2
R1 + R2 ≈ G 1 +  2 1  
 2 G  
By using the equations (D–14), it can be seen that the difference d, in the reflected
wave path-length R1 + R2 and the direct wave-path length Rd, may be expressed as
2h1h2
δ = ( R1 + R2 ) − Rd ≈ (D–15)
G
Equation (D–15) is often a valid approximation for ground communications systems,

because usually for those systems G2 > > (h1+ h2)2.
A number of exact equations can be obtained from the geometry of Fig. D–4, and they
include the following:
h2 − h1
sin θ d = (D–16)
Rd
Multipath Dependencies on Frequency, Polarization, and Surface Roughness 453
cos θ d = G Rd (D–17)
h1 + h2
sin ψ = (D–18)
G
θr = ψ (D–19)
Recall that the solution of multipath problems in exact form requires knowledge of y
versus qd. However, the relationship between y and qd is not available, mathematically,
in a simple closed-form. However, accurate multipath solutions can be acquired by com-
puter with step-by-step procedures. For example, to calculate an antenna pattern at dis-
tance Rd and with antenna height h1, let h1 and Rd be constants. Then, the following are
attained: qd versus h2 from (D–16), G versus qd from (D–17), y versus G from (D–18),
and qr from (D–19).
D.6. Multipath Dependencies on Frequency, Polarization, and

Surface Roughness
The phenomenon of interference between the direct and indirect waves produces polar-
ization sensitive lobes above the reflecting surface. Generally, the positions of the lobes
do not coincide for horizontal (H-POL) and vertical (V-POL) polarizations, because of
differences in the phases (fH and fV) and amplitudes (rH and rV) of the reflection coef-
ficients. Figures D–5 through D–8 that are now discussed were prepared by using the
flat earth model.
Examples of field strengths above a flat, smooth sea for H- and V-polarizations at 1
and 4 GHz are given in Fig. D–5. Notice that an advantage of V-polarization, especially
at lower frequencies, is that the relative field strength near the surface is larger. Inter-
ference between direct and reflected waves also may create maxima and minima in
field strength versus range, somewhat analogous to the variation in Fig. D–5 with height
at a fixed range.
Some differences in the propagation factor F for vertical and horizontal polarizations
versus grazing angle and frequency are shown in Fig. D–6. Here it is assumed that the
earth is perfectly smooth and the following are additional assumptions: a non-directive
antenna at height 3 m above the earth, frequencies of 1 MHz (l = 300 m) and 100 MHz
(3 m), and e′r = 16 and s = 10−2 mho/m. It may be seen that:
• at 1 MHz, because there is 180° phase change for H-POL reflection: (a) V-POL
signal always exceeds H-POL signal (i.e., FV > FH), and (b) at less than 10°,
FV exceeds FH by at least 30 dB.
• at 100 MHz, because pathlength differences between the direct and indirect
paths cause large changes in phase versus grazing angle: (a) H-POL signals
can be either smaller or larger than V-POL signals, and (b) H-POL peaks are
larger than the V-POL peaks and the H-POL minima are smaller than the
V-POL minima because rH > rV.
f = 1 GHZ
θ = 0.8° θ = 4° θ = 8°
6 6 6
5 5 5
HEIGHT ABOVE SEA (ft)
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
0 1 2 0 1 2 0 1 2
f = 1 GHZ RELATIVE FIELD STRENGTHS
θ = 0.8° θ = 4° θ = 8°
6 6 6
5 5 5
HEIGHT ABOVE SEA (ft)
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
0 1 2 0 1 2 0 1 2
RELATIVE FIELD STRENGTHS
SOLID LINES INDICATE HORIZONTAL POLARIZATION
DOTTED LINES INDICATE VERTICAL POLARIZATION
FIGURE D–5.
Electric field amplitude patterns above a smooth sea relative to free-space amplitudes. 1 and
4 GHz, with 0.8°, 4°, and 8° grazing angles. From Long, Wetherington, Edwards, and Abeling
(1965).
Figures D–7 and D–8 show the propagation factor F for H-POL with and without
surface roughness. The curves were prepared under the following assumptions: sh = 0
(flat, smooth earth) and sh = 0.3 m (representing a recently plowed agricultural field),
l = 3 m (100 MHz) and 0.3 m (1 GHz), and land with er = 16 − j60ls and s = 10−2 mho/
m. In Fig. D–7 for 100 MHz, the surface roughness causes only a slight filling of the
minimum at 30°, and there are little effects of surface roughness at smaller grazing angles.
Also in Fig. D–7, the effect of surface roughness at 100 MHz is not strong, even for
Spherical Earth Geometry 455
grazing angles up to 70°. In Fig. D–8, for 10

1 GHz, effects of surface roughness are
Propagation Factor F in dB
0
not apparent for angles less than 3°, and –10
for the rough surface essentially all effects
–20
of multipath have disappeared for angles
above 10°. Thus, in some applications, –30
the use of higher frequencies may be a –40
useful method of minimizing the effects –50
of multipath interference. There are prac-
–60
tical limits, of course, on the use of the
higher frequencies (shorter wavelengths) –70
0 10 20 30 40 50 60 70 80 90
imposed by the general increase in atmo- Grazing Angle in Degrees
spheric attenuation. In summary, the elec- V-POL, 1 MHz
H-POL, 1 MHz
tromagnetic effects of roughness depend V-POL, 100 MHz
on sh, q, and l through their relationships H-POL, 100 MHz
with the roughness parameter sh sin y/l FIGURE D–6.
and on its effect on the magnitude of Propagation factor F in dB versus grazing
reflection co-efficient in accordance with angle. V- and H-polarizations, 1 and 100 MHz,
(D–7) of sec. D–3. antenna 3 m above flat smooth land.
D.7. Spherical Earth Geometry 10
The earth’s overall shape is roughly

Propagation Factor F in dB
spherical and its curvature decreases the 0

path-length difference d between the
direct and reflected waves. Another effect
–10
of the earth’s curvature is to decrease the
amplitude of reflected waves, caused by
them being spread out or diverged by the –20
convex surface of the earth.
Electromagnetic waves propagating
within the earth’s atmosphere do not –30
0 30 60 90
travel in straight lines but bend slightly Grazing Angle in Degrees
because of refraction. One effect of refrac- 20 log F, 100 MHz, H-POL, Smooth Surface
20 log F, 100 MHz, H-POL, Rough Surface
tion is to extend the distance to the
horizon, thus increasing radio and radar FIGURE D–7.
coverage; another effect is to change the Propagation factor F versus grazing angle.
elevation pointing direction of the antenna 100 MHz, horizontal polarization, smooth
land and plowed land (standard deviation
pattern at long distances. In computations, s = 0.3 m).
h
a commonly used method is to (1) replace
the actual earth of radius a by an equiva-
lent earth of radius ae = ka and (2) replace the actual atmosphere by a vacuum (free
space) in which the waves propagate in straight lines. It is customary to use the value
k = 4/3 for approximating the effects of refraction. Although k = 4/3 is typical,
10 measurements may yield larger or smaller

values (U.S. Department of Commerce
1966).
0 Figure D–9 illustrates straight-line
Propagation Factor F
propagation in accordance with the spher-

ical earth model. Note that, unlike flat-
–10 earth geometry, the horizontals at the two
terminals (points A and P) are not paral-
lel. The horizon is located where path Rd
–20 is tangent to the earth’s surface. There-
fore, at the horizon, ray paths Rd and R1
are parallel, with angle y being zero. The
–30
range to the horizon Rh is the distance
0 5 10 15 from point A to the point that the tangent
Grazing Angle in Degrees ray (for which y = 0) strikes the earth.
20 log F, 1 GHz, H-POL, Smooth Surface Range to the total horizon Rmax is the
20 log F, 1 GHz, H-POL, Rough Surface range from point A to point P, when Rd
FIGURE D–8. exceeds Rh and point P is located along
Propagation factor F versus grazing angle. the extension of the tangent ray beyond
1 GHz, horizontal polarization, smooth and its tangent point at the earth. According
plowed land (standard deviation sh = 0.3 m). to the spherical earth model, Rmax is the
maximum practical range for detecting an
electromagnetic wave emitted from point
A, given that the heights at points A and P are h1 and h2.
The range to the horizon Rh is obtained from the right triangle relationship
(ae + h1 )2 = Rh2 + ae2

or
Rh = 2ae h1 + h12 (D–20)
Similarly, the distance from the tangent point to point P of height h2 is 2ae h2 + h22 .
Therefore, the maximum practical range for detecting an electromagnetic wave emitted
from point A, given that the heights at points A and P are h1 and h2, is
Rmax = Rh + 2h2 ae + h22 (D–21)
The equations that follow are from Blake (1986, pp. 249–58). They allow the param-
eters of Fig. D–9 to be computed if the heights h1 and h2, the total ground range G, and
the equivalent earth radius ae are given. The first step is to obtain the ground ranges
G1 and G2, where G is (G1 + G2). This requires solving for two parameters, p and x,
that follow:
Spherical Earth Geometry 457
P
Horizontal
at point P
Rd
Horizontal
at Point A R2
h2
d
Antenna
Effective
A R1 Earth
r
y Surface
y
h1 B G2
G1
ae ae
Earth Center
C
FIGURE D–9.
Geometry for spherical-earth model. (Effective earth has radius ae, compatible with assumption
of straight ray paths A-P and A-B-P). Adapted from Blake (1991, p. 255).
2
 h −h 
ae( h1 + h2 ) +  
2 G
p= ξ = sin −12aeG 1 3 2  (D–22)
3  2  p 
From p and x, G1 and G2 are calculated by the following:
G sin ξ
G1 = −p G2 = G − G1 (D–23)
2 3
Finally, one can attain R1, R2, and Rd by using the following:
G  G 
R1 = h12 + 4ae( ae + h1 ) sin 2 1  R2 = h22 + 4ae( ae + h2 ) sin 2 2 
 2ae   2ae 
(D–24)
2  G1 + G2 
Rd = ( h2 − h1 ) + 4 ( ae + h1 ) ( ae + h2 ) sin 
2
 
2ae 
In (D–24), the angles within the sin2 terms are each expressed in radians.
Now that equations for all linear dimensions of Fig. D–9 are available, the angles qd,
qr and y can be obtained as follow:
2ae( h2 − h1 ) + h22 − h12 − Rd2

θ d = sin −1 (D–25)
2 ( ae + h1 ) Rd
If h1 and h2 are both much smaller than ae, which is true for most applications,
h −h R 
θ d ≅ sin −1 2 1 − d  (D–26)
 Rd 2ae 
The exact equations for qr and y, along with approximations for the usual scenarios for
which h1 << ae, follow
2ae h1 + h12 + R12 h R 

θr = sin −1 ≅ sin −1 1 + 1  (D–27)
2 ( ae + h1 ) R1  R1 2ae 
and
2ae h1 + h12 − R12 h R 

ψ = sin −1 ≅ sin −1 1 − 1  (D–28)
2ae R1  R1 2ae 
From Fig. D–9 it can be seen that

qd is the elevation angle, relative to the horizontal at point A of the direct ray between
points A and P;
Divergence Factor 459
qr is the depression angle, relative to the horizontal at point A of the indirect path ray
between point A and the earth (point B); and
y is the angle measured above the horizontal at the earth (point B) of the indirect path
ray between point A and the earth.
For many applications where multpath interference is significant, the elevation (verti-
cal) antenna pattern is sufficiently broad that the pattern strengths at the beam center and
at angles qd and qr are approximately equal. Then, the approximation of (D–9) in sec.
D.4 is valid. Otherwise, multipath analyses must include effects of the shape of the eleva-
tion pattern.
The pathlength difference d is, of course, equal to (R1 + R2) − Rd. Consequently,
because this equation involves differences in numbers that are large compared to d, sig-
nificant computing errors can occur. Blake (1986, p. 257) uses a more accurate method
for calculating d, which follows:
4 R1R2 sin 2 ψ
δ= (D–29)
R1 + R2 + Rd
Thus, by starting with only the heights h1 and h2, the ground range G and an assumed
equivalent earth radius ae, one can make calculations for the various path lengths and
angles of Fig. D–9. The accompanying SM includes spherical earth calculations using
Mathcad, where the user can choose the initial assumptions in accordance to his or her
needs.
D.8. Divergence Factor

For a spherical wavefront incident on a spherical surface, the power density after reflec-
tion is reduced by a divergence or spreading of the ray paths. This geometric (not elec-
tromagnetic) description causes the calculated reflected field strength to be reduced with
decreased grazing angle. However, at grazing angles of about one degree or less, forward
propagation from a curved surface is not accurately described by the reflection of rays.
Then, instead, its description requires advanced electromagnetic diffraction techniques
beyond the scope of this book (Meeks 1982, Ch. 6).
The letter D denotes the divergence factor, a power reflection coefficient based entirely
on geometry. Specifically, D is the calculated fractional reduction in the field strength
resulting from geometry that describes rays reflected from a curved surface. Field strength
is proportional to the square root of the ray density. Thus, D is the square root of the
ratio of the beam’s solid angles before reflection from a smooth spherical surface divided
by the solid angle after reflection. To include effects of divergence, the magnitude of the
specular reflection coefficient is expressed as
 2πσ h sin ψ  
2
Γ = ρo exp  −2    D (D–30)
  λ 
where ro is the reflection coefficient magnitude if the surface is smooth and the expo-
nential function is the roughness factor that was introduced in Sec. D.3.
From detailed geometrical analysis, Kerr (1951, p. 406) in his equation (16) provides
a general equation for D. This equation, expressed in terms of the parameters of Fig. D–9,
was obtained by the present author by using Kerr’s equations (11) and (12). That equa-
tion that follows Kerr and various approximations for D are included in SM 3.6 of the
accompanying SM. To obtain another and yet simpler equation, Kerr uses assumptions
that are valid if the effects of divergence are appreciable. These assumptions include
small grazing angles, and terminal heights and path lengths small compared to the earth’s
radius. Kerr’s approximation follows:
−1 2
 2G1G2 
D = 1 + (D–31)
 aeG sin ψ 
A number of other approximations for D appear in the literature (Balanis 1984, 2005;
Beckmann and Spizzachino 1963). These approximations and (D–31) are compared in
SM 3.6 with both terminal heights (h1 and h2) being 20 km, and each gives almost identi-
cal calculated values. Consequently, each of the equations will provide acceptably cal-
culated geometrical values for D if neither platform height exceeds 20 km. However, an
accurate geometrical calculation of D does not necessarily provide a valid electromag-
netic result.
The risk of using the calculated divergence factor for very small grazing angles (about
one degree or less) is now illustrated. Recall that a small D value implies a small reflected
field, and then the total field would approximate the free space field. Contrarily, measure-
ments indicate that the field strength actually becomes smaller as y becomes smaller and
the horizon is approached. However, because of diffraction from the earth, the pattern
propagation factor F is very small, but not actually zero at ranges that exceed the calcu-
lated horizon Rmax (Blake 1991, pp. 271–74). Accordingly, incautious use of a D value
can cause the calculated pattern propagation factor F versus range to incorrectly approach
the free-space value of unity.
References
Balanis, C. A., R. Hartenstein, and D. DeCarlo, “Multipath Interference for In-Flight
Antenna Measurements,” IEEE Trans. on Antennas and Propagation, January 1984,
pp. 100–104.
Balanis, C. A., Antenna Theory, Analysis and Design, 3rd ed., Wiley-Interscience, 2005,
pp. 208–11.
Beckmann, P., and A. Spizzachino, The Scattering of Electromagnetic Waves from Rough
Surfaces,” The MacMillan Company, 1963, p. 224.
Blake, L.V., Radar Range-Performance Analysis, Munro Publishing, 1991; Artech House,
1986.
Dicaudo, V. J., “Radomes,” ch. 14 in M. I. Skolnik, Radar Handbook, McGraw-Hill, 1970.
Institute of Electrical and Electronics Engineers, IEEE Standard 686-1997, 1997.
References 461
Kerr, D. E., Propagation of Short Radio Waves, McGraw-Hill, 1951.

Long, M. W., R. D. Wetherington, J. L. Edwards, and A. B. Abeling, “Wavelength
Dependence of Sea Echo,” Engineering Experiment Station, Final Report, Contract
N-62269-3019, Georgia Institute of Technology, July 1965.
Meeks, M. L., Radar Propagation at Low Altitudes, Artech House, 1982.
Reed, H. R., and C. M. Russell, Ultra High Frequency Propagation, 2nd ed., Boston
Technical Publishers, 1964, p. 140–41.
U.S. Department of Commerce, A World Atlas of Atmospheric Radio Refractivity, ESSA
Monograph No. 1, 1966.
APPENDIX E
Radomes
A radome is a dielectric structure that is used to protect an antenna from its environment
(wind, rain, ice, salt spray, dust, insects), and it may be used to reduce aerodynamic drag.
Radomes serve as feed covers, covers attached to the antenna, or covers within which
an antenna moves. Therefore, they have a wide variety of shapes and wall structures
(Huddleston and Bassett 1993).
A radome can cause unwanted changes in beam-pointing direction, create energy loss
by reflection and absorption, cause gain loss and side-lobe degradation by defocusing,
create internal reflections that produce impedance mismatches, increase side-lobe levels,
and may also depolarize the radiation. Rulf (1985) discusses electrical design issues for
an airborne radome used with a large, very-low side-lobe antenna. The reader should be
aware that radome design is a highly specialized subject. For example, the design for a
radome with pointed nose for high-speed aircraft may be more complex than the antenna
design.
Schrank, Evans, and Davis (1990) discuss a variety of radome types, and descriptions
of some of the more common radome wall cross sections are given here and illustrated
in Fig. E–1.
Thin wall. A solid dielectric, with thickness usually less than l/10 measured in the
dielectric. The reflections from the air-radome and radome-air interfaces have oppo-
site phases. Thus, if the radome wall thickness is thin compared to the wavelength
within the radome material, the net effect is near-cancellation of the reflections. A
thin wall radome has good electrical properties, but it may be weak structurally
when designed for frequencies above 1 GHz.
Half wavelength. This is a solid dielectric surface having an electrical thickness near
one-half wavelength. A half-wavelength-thick surface is nonreflecting at its design
frequency. Because of the half-wavelength requirement, its bandwidth and its useful
range of incidence angles are limited.
A-sandwich. A commonly used three-layer wall configuration consisting of two thin
dielectric skins separated by a thicker but low-dielectric constant core. The core
might be a honeycomb or a foam material. Since the reflections from the two
463
464 Radomes
Solid, thin wall or l/2 thickness skins will be roughly equal in ampli-
A sandwich, low dielectric constant foam tude and phase, a quarter-wavelength
or honeycomb core, skins spaced l/4 spacing of the skins will minimize their
C sandwich, two back-to-back A sandwiches, combined reflections
for increased strength C-sandwich. Two back-to-back A-sand-
Multiple-layer sandwich, thin layers of fiberglass wiches. It can be used when the ordi-
with low density cores, for good electrical nary A-sandwich does not provide
properties and strength
sufficient strength.
FIGURE E–1. Multiple-layer sandwich. These include
Commonly used radome wall cross sections. many thin layers of fiberglass with low-
density cores for providing great strength
and good electrical performance.
Dicaudo (1970) includes numerous useful graphs that show electrical performance
versus incidence angle for single-layer, A-sandwich, and C-sandwich walls; and for
polarizations parallel and perpendicular to the plane of incidence formed by the direction
of propagation and the normal to the radome surface. These curves show transmission
efficiency and insertion phase delay versus the radome-wall thickness-to-wavelength
ratio. The data are included for several dielectric materials.
Relative permittivity is the permittivity of a medium relative to its free space value eo
of 8.85 × 10−12 farads/meter. To account for the ohmic loss of a dielectric material, a
complex relative permittivty er is used that follows:
ε r = ε r′ − jε r′′ = ε r′ − j 60λ0σ (E–1)
where l 0 is the free-space wavelength in meters and s is conductivity of the medium in

mhos per meter. The term e′r is the conventional relative dielectric constant ordinarily
quoted for materials, and it and the index of refraction n of a nonmagnetic material are
related as follows
(εr′ )1 2 = n = c v (E–2)
where v is wave velocity in the medium and c is wave velocity in free space. Then from
l = v/f and the wavelength l in an open, unbounded (not a waveguide) medium is
λ = v f = c (ε r′ )1 2 f  = λ0 (ε r′ )1 2 (E–3)
where l0 is the free space wavelength. Thus, for a dielectric constant of 4 (typical of
epoxy resin fiberglass radomes), the internal wavelength would be one-half l0. Then, for
this example, a layer of l0/10 thickness corresponds to l/20 internal thickness and would
be electrically acceptable as a thin radome or as a thin wall for a layered radome.
The term e″r accounts for a dielectric’s ohmic loss, and the ratio e″r/e′r is defined as the
loss tangent or tan d. Although not relevant to antennas and transmission lines, another
electrical term is dielectric power factor PF, where PF = sin d. Loss tangents for radome
Radomes 465
TABLE E-1 Approximate Electrical Properties of Selected

Materials at 10 GHz
Material e′r Loss tangent (e″r/e′r)

Vacuum (defined) 1 0
Plexiglas 2.6 1.5 × 10−2
Polystyrene 2.54 4 × 10−4
Teflon 2.08 4 × 10−4
Duroid 2.7 3 × 10−3
Aluminum oxide 9.2 3 × 10−4
Fused silica 3.6 2 × 10−4
Boron nitride 4.4 1 × 10−4
dielectrics are in the range of 10−4 to 10−2. Thus, for these materials loss tangent and
power factor are essentially equal.
Table E-1 includes typical electrical properties of some radome materials at 10 GHz.
For further details, see Bodnar (2007) and Burks (2007). Although e′r varies with fre-
quency, the values given are generally typical for kiloHertz and gigaHertz frequencies.
On the other hand, e″r = j 60 l s and thus loss tangent e″r is a strong function of fre-
quency. Electrical properties also depend on fabrication method, material density, and
temperature.
The loss per unit length of a dielectric is proportional to loss tangent and is derived
theoretically by Ragan (1948, pp. 28–29). The attenuation, expressed in nepers per meter,
within the dielectric is
ANp = (π λ ) tan δ in nepers per m (E–4)
or
ANp = π tan δ in nepers per λ within the dielectric (E–5)
By using the relationship between attenuation in nepers ANp and in decibels AdB, namely
ANp = 0.115 AdB, we obtain
AdB = 27.3 tan δ in dB per λ within the dielectric (E–6)
or by using the index of refraction n, that is, (e′r )1/2, of the dielectric medium,
AdB = 27.3n tan δ in dB per free space wavelength, λ 0 (E–7)
As examples of possible dielectric losses, we now consider radome wall thicknesses

of l/20 and l for a dielectric having a relatively high loss tangent of 10−2. Then from
(E–6), the attenuations in dB are
466 Radomes
AdB = 27.3 ⋅10 −2 ⋅ (1 20 ) = 1.4 × 10 −2 dB for λ 20 thickness and 10 −2 loss tangent
and
AdB = 27.3 ⋅10 −2 ⋅1 = 0.27 dB for λ thickness and 10 −2 loss tangent
Thus it is seen that ohmic loss, when using a relatively high loss material, is insignificant
for a thin wall radome. However, the loss can be appreciable if the thickness were one
wavelength. Another major concern is the likely negative effect of the radome on antenna
focusing. Generally, thin wall radomes do not create significant degradation to antenna
gain or side lobes. However, much caution is needed when using the thicker radomes,
often needed for structural strength.
References
Bodnar, D. G., “Materials and Design Data,” ch. 55, pp. 55–3 through 55–5 in J. L.
Volakis, Antenna Engineering Handbook, 4th ed., 2007.
Burks, D. G., “Radomes,” ch. 53, pp. 53–16 and 53–17 in J. L. Volakis, Antenna
Engineering Handbook, 4th ed., 2007.
Dicaudo, V. J., “Radomes,” ch. 14 in M. I. Skolnik, Radar Handbook, McGraw-Hill, 1970.
Huddleston, G. K., and H. L. Bassett, “Radomes,” ch. 44 in R. C. Johnson (ed.), Antenna
Engineering Handbook, 3rd ed., McGraw-Hill, 1993.
Ragan, G. L., Microwave Transmission Circuits, vol. 9, M.I.T. Laboratory Series,
McGraw-Hill, pp. 28–29, 1948.
Rulf, B., “Problems of Radome Design for Modern Airborne Radar, Part 1,” Microwave
Journal, Jan. 1985, pp. 145–48, 152–53; “Problems of Radome Design for Modern
Airborne Radar, Part 2,” Microwave Journal, May 1985, pp. 265–67, 271.
Schrank, H. E., G. E. Evans, and D. Davis, in ch. 6, “Reflector Antennas,” pp. 6.44–6.52
in M. I. Skolnik, Radar Handbook, 2nd ed., McGraw-Hill, 1990.
APPENDIX F
Far-Zone Range-Approximation
and Phase Error
The far-field region, that is, the far zone, is the region of an antenna field where the
antenna radiation pattern is essentially independent of the distance from the antenna.
In free space, with the antenna maximum dimension D being large compared to a
wavelength, the far-field region is commonly taken to be greater than rmin, where
rmin = 2 D 2 λ (F–1)
l is wavelength (IEEE Standard 100–1992, p. 482). Greater distances are desirable,

especially if accurate measurements are needed for antennas with very low side
lobes. The distance criterion (F–1) is based on the phase variation caused by the
path-length change d over a linear aperture of dimension D, resulting from spherically
spreading radiation from a point P, as illustrated by Fig. F–1. Sometimes, however,
additional criteria are necessary for assurance that a measured pattern replicates a far-field
measurement.*
From a right triangle of Fig. F–1 with sides R, D/2, and (d + R), one attains
R 2 + ( D 2 ) = (δ + R ) = δ 2 + 2δ R + R 2
2 2
or
( D 2 )2 = δ 2 + 2δ R (F–2)
Ordinarily d << R, and then d2 << 2 dR. Thus,
( D 2 )2 ≈ 2δ R
*While reviewing the manuscript, E. B. Joy added two additional criteria: (1) rmin = 20 l, that pertains to
electrically small antennas and assures that the reactive near-field is negligible at rmin and (2) rmin = 20 D,
that limits the relative down range far-field decrease across the antenna under test to an acceptable value.
As with the factor 2 in (F–1), choice of the factors 20 depends on the accuracy of measurement required.
According to J. S. Hollis, T. J. Lyon, and L. Clayton, Jr., (1970, sec. 14.2), the relevant effects are usually
considered negligible if rmin ≥ 10l and rmin ≥ 10D.
467
468 Far-Zone Range-Approximation and Phase Error
or
δ ≈ ( D 2 )2(1 2 R ) = D 2 8 R (F–3)
d
We can now find the phase error caused
by the path-length difference d if rmin =
R 2D2/l, with d << 2R in accordance with
(F–1). Then from (F–3),
R
D P δ = D 2 8 R = D 2 8 (2 D 2 λ ) = λ 16 = 22.5°
(F–4)
R
Therefore, when the separation distance
is 2D2/l between an aperture (of maximum
dimension D) and the source, the
d maximum phase error is 22.5° (l/16). In
addition, it is to be noted that the phase
error caused by the pathlength difference
d varies in proportion to D2, from zero at
FIGURE F–1.
the aperture center to a maximum at an
Path-length difference d for an aperture aperture edge.
dimension D and range R. A question now to be addressed is the
required minimum far-zone distance
2D2/l, when d is not much less than 2R, the assumption used to get the l/16 far-zone
criterion of (F–4). Returning to the exact equation (F–2) and letting l/16, it may be seen
that D = l/8 if R = 0. Thus, the pathlength difference d of Fig. F–1 is less than l/16 if
D < l/8. In other words, according to the l/16 far-zone criterion, the far zone always
exists if the antenna maximum aperture dimension D is less than l/8. However, it is to
be noted that both reactive and radiative fields surround an antenna (sec. 3.2). Thus,
although the d < l/16 far-field criterion may be satisfied at very short distances from
very small antenna aperture, the reactive near-field may exceed the far-field at separation
distances less than 20l (sec. 3.2.5).
References
Hollis, J. S., T. J. Lyon, and L. Clayton, Jr., sec. 14.2 of Microwave Antenna
Measurements, Scientific-Atlanta, 1970.
The New IEEE Standard Dictionary of Electrical and Electronic Terms, 1992.
APPENDIX G
Radiating Near and Far Fields,

and the Obliquity Factor
Appendix G discusses the radiating E- and H-fields that surround an energized antenna,
including radiation patterns versus separation distance from an antenna aperture. This
material is supplemented by the accompanying website that contains example calcula-
tions using Mathcad for radiating near- and far-field patterns, where the Huygens-Kirch-
hoff formulation is used for the near-aperture radiating field calculations (Good 1990).
G.1. The Radiating Fields

The E- and H-field regions that surround an energized antenna are of three types: the
reactive near-field, the radiating near-field, and the radiating far-field. The locations of
these fields depend on distance to the observation point, wavelength, and aperture size.
The reactive near-zone field decays the most rapidly with distance from the antenna so
that, except at relatively close distances, its magnitude is insignificant compared to the
magnitudes of the radiating fields. Historically, the closer-in and more distant radiating
fields have been called the Fresnel and Fraunhofer regions, but today the terms “radiating
near-field” and “radiating far-far field” are preferred (IEEE, 1992). Furthermore, these
fields are not located at clearly defined distances, but the “far field” region is where the
radiation pattern versus angle is for practical purposes independent of distance. As a rule
of thumb, the far-zone distance is said to be at and beyond 2D2/l, where D is the largest
dimension of the aperture. This and other requirements are given in Appendix F.
Figure G–1 is a depiction of the three regions as given by Johnson, Ecker, and Hollis
(1973). It illustrates the propagation paths within the radiating far-field region as if they
were emitted radially from the antenna, that is, as if the energy emanates from a specific
point (like a radiating point source). Notice that, within the radiating near-field region of
this high-directivity antenna in Fig. G–1, propagation is shown as being both along paral-
lel paths from the aperture and transitioning in the diverging directions of the far-field
region. Existence of the parallel paths indicates, at least theoretically, that the phase and
amplitude are both constant perpendicular to the propagation path. These features permit
469
470 Radiating Near and Far Fields, and the Obliquity Factor
REACTIVE the measurement of accurate far-zone

RADIATING
NEAR-FIELD
REGION
radiation patterns within a compact range
FAR-FIELD
REGION (see sec. 9.3), even though the antenna
under test is within the radiating near-
field of the compact range antenna.
ANTENNA
Radiation patterns for aperture anten-
nas are most often calculated based on
assumptions called high-frequency scalar
RADIATING
NEAR-FIELD
approximations. High frequency refers to
REGION the aperture being large enough that the
FIGURE G–1. effects of radiation from aperture edges
The three regions surrounding a high directivity
are negligible and the distance to the
antenna. (Johnson, Ecker, and Hollis, © IEEE observation point is at least several wave-
1973). lengths; and scalar refers to the field over
the aperture being almost completely
linearly polarized, with only a small fraction of the energy being cross-polarized.
Silver (1949, pp. 169–99) published classic equations for the close-in (Fresnel) and
the more distant (Fraunhofer) radiation fields, that are now described. Assume an aperture
is in the x-y plane, in accordance with Fig. G–2. Also assume the amplitude and phase
excitation on the surface (the aperture) is described by the term A(x, y), and let each
surface point be located a distance r1 from observation point P.
Equation (2) of Silver (1949), page 170, is a general diffraction equation developed
for continuous aperture dimensions that are large compared to a wavelength. For specific-
ity, let the aperture be rectangular with ±a/2 and ±b/2 being its edges along the x and y
directions. With two assumptions, described below, Silver’s equation (2) reduces to the
Huygens-Kirchhoff diffraction formulation (Lerner and Trigg 1990); for which pattern
field amplitudes can be expressed as
2π
−j r1
g(θ , φ ) =
a2 b2 e λ
 cos η ( x, y ) + 1  dxdy
∫− a 2 ∫− b 2 A ( x, y ) r1  2  (G–1)
where A(x, y) is the amplitude and phase distribution over the aperture, h(x, y) is the
angle between the outward direction of r1 and the Z axis at each x,y aperture position.
The assumptions include: (1) the r1 and R lengths are at least several wavelengths and
(2) the phase fluctuations across the aperture are small enough so that the phase distribu-
tion is essentially uniform.
Persons familiar with physical optics will recognize (G–1) as being the phasor sum-
mation of Huygens’ wavelets (Fig. 1–9, chapter 1) that emanate from the aperture x,y
positions, with their amplitudes modified based on propagation direction, in accordance
with the factor (1/2)[cos h(x, y) + 1]. This amplitude multiplier is the Fresnel obliquity
factor, which has been verified with mathematical rigor (see, e.g., Slater and Frank 1947).
Accordingly, forward propagating wavelets have unity amplitude if h(x, y) = 0°; and the
The Radiating Fields 471
wavelets have zero amplitude in the backward direction if h(x, y) = 180°. In general,
h(x, y) is a complicated function of the aperture coordinates and q. Fortunately, however,
the obliquity is not needed for calculating array patterns, and often it is not needed when
calculating radiation from continuous aperture antennas.
Another approximation, valid for the far-field and much of the near radiating field,
assumes the lengths r1 and R are large compared to the linear aperture dimensions a and
b. Now, with this, we neglect the variation of cos h over the aperture and we replace h
with q and r1 with R. Then, (G–1) is simplified and becomes (G–2) that follows
2π
( cos θ + 1) a2 b2 −j r1
g(θ , φ ) = ∫− a 2 ∫− b 2 A ( x, y ) e λ dxdy (G–2)
2R
In the radiating near-field region described by (G–2), in accordance with Appendix H,

the distance r1 varies with the observation angle q, the x and y aperture coordinates, and
the distance R between the aperture origin and the observation point P. Thus, the nor-
malized near-fields are functions of R and there are gradual changes versus R from the
radiating near-field out to and including the far-field.
It is to be noted that field amplitudes are highly sensitive to the pathlength difference
d between r1 and R, and a useful relationship follows:
2π 2π 2π 2π
− r1 − (δ + R ) − R − δ
e λ = e λ =e λ e λ
To obtain an equation for use only within the radiating far-field, distance r1 in (G–2) is
replaced by using (H–6) of Appendix H. Then, the integrand of (G–2) is simplified,
q P (xp, yp, zp) r1 P (xp, yp, zp)
R
R q
y z
y f
f
x
(a) (b)
FIGURE G–2.
Spherical coordinates with aperture in x-y plane: (a) shows observation point P at distance R
from origin and (b) shows point P at distance R from the origin and distance r1 from a specific
position on the x-y plane.
and the equation for the pattern amplitude in the radiating far-field becomes (G–3) that
follows:
2π
( cos θ + 1) a2 b2 j sin θ ( x cos φ + y sin φ )
g(θ , φ ) = ∫− a 2 ∫− b 2 A ( x, y )e λ dxdy (G–3)
2R
Note that, consistent with the definition of far zone, the integral of (G–3) is not a
function of R.
Recall that q is the observation direction and is measured from the normal to the
aperture (the z-axis). For q of 30°, 60°, and 90°, the obliquity factor reduces the calculated
E-field amplitude by 0.6, 2.5, and 6 dB, respectively. On the other hand, the beam point-
ing direction is determined by the phase term within the aperture excitation A(x, y). Often
the main beam (major lobe) is pointed along the z-axis (q = 0). Then, especially for
narrow-beam antennas, the q values of greatest interest are small enough that the obliq-
uity factor is assumed unity. Therefore, the term (1 + cos q)/2 in (G–3) is commonly
neglected. Furthermore, the equation for g(q, f) is usually simplified by neglecting the
multiplicity factor 1/R. Then the electric field pattern in the radiating far-field may be
expressed as
2π
a2 b2 j sin θ ( x cos φ + y sin φ )
g(θ , φ ) = ∫ ∫ A ( x, y )e λ dxdy (G–4)
−a 2 −b 2
If the limits of integration of (G–4) are replaced by infinity and minus infinity, the
integral becomes a Fourier integral of two variables. Then, for practical purposes (G–4)
is a Fourier integral, because the aperture illumination A(x, y) is zero outside the aperture.
Thus, the integral in (G–4) vanishes beyond the aperture dimensions ±a/2 and ±b/2. With
this in mind, the reader will recognize that the far-zone pattern of an aperture antenna is
the Fourier transform of its aperture illumination. This Fourier-transform relationship
was especially important for antenna engineers before the general availability of digital
computers. Then, calculations for patterns based on aperture distributions were made
through use of published Fourier transform tables. Nowadays, the fast Fourier transform
(FFT) is useful for pattern calculations that include a large number of aperture data
points.
The Huygens’ concept of summing wavelets is applicable to arrays, but the obliquity
factor is replaced by unity. For a continuous aperture, the summation is of the wavelets
within a plane wave that propagates in the +z direction, and the obliquity factor provides
for no propagation in the −z direction. For array factors (patterns without inclusion of
element patterns), the summation is of a finite number of wavelets that propagate isotro-
pically from each element, and thus the obliquity factor does not apply to arrays.
Therefore, to get array factor far-field patterns, one may use (G–1) through (G–4) by
replacing (a) the integrals with discrete summations and (b) the obliquity factors with
the numeral one.
Principal-Plane Patterns Versus Range 473
G.2. Principal-Plane Patterns Versus Range

This section discusses x-z plane pattern calculations versus range, for continuous aper-
tures and arrays located on the x-axis. Let us assume that there are N radiating elements
with equal phase and with different amplitudes a(n) that are functions of the positions
x(n) along the x-axis. Then, the integral can be replaced by discrete summation, where
for continuous apertures the elements are infinitesimal and their spacing approaches zero.
Then, (G–1) becomes
2π
−j r1
λ
 cos η ( n ) + 1 
N
e
g(θ ) = ∑ a (n) r1  2  (G–5)
n =1
Ordinarily, r1 is large compared to the dimensions of the antenna, and then the bracketed
term in (G–5) becomes (cos q + 1)/2. However as noted in sec. G–1, for array elements
the (cos q + 1)/2 term is inapplicable and is replaced by numeral one.
In general, pattern calculations are highly sensitive to r1 and the pathlength difference
d, where d is r1 − R. For r1, equation (H–4), Appendix H, is exact. Then, with radiating
elements located only at positions x(n) on the x-axis, r1 becomes
0.5
 x (n) x (n)  
2
r1 = R 1 − 2 sin θ +  
 R  
(G–6)
 R
Furthermore, where aperture dimensions are small compared to R, d is considerably

simplified, as determined from equation (H–6), Appendix H. Consequently, for far-field
patterns in the x-z plane (f = 0) and for an aperture with its coordinates x(n) on the
x-axis, one obtains
δ = r1 − R ≈ − x ( n ) sin θ (G–7)
Now, using (G–6) and (G–7) when R is much larger than r1, one obtains (G–8) that
follows.
2π
( cos θ + 1) N j x ( n ) sin θ
g(θ ) =
2R
∑ a( n ) e λ (G–8)
n =1
Equation (G–8) may be used for far-field calculations, provided R is 2D2/l or greater.
As in the previous equations, the obliquity factor (cos q + 1)/2 for arrays is replaced by
numeral one. In addition, this factor is often assumed to be unity for aperture antennas;
because major interest is usually near small q values, where the obliquity factor is essen-
tially one.
Example pattern calculations versus range are included in the accompanying
website.
References
Good, Myron L., “Diffraction,” in Encyclopedia of Physics, 2nd ed., R. Lerner and G. L.
Trigg (eds.), 1990, VCH Publishers, p. 252.
Hansen, R. C., “Aperture Theory,” in R. C. Hansen (ed.), Microwave Scanning Antennas,
Academic Press, Inc, New York, 1964.
Terms, 1992.
Johnson, R. C. (ed.), Antenna Handbook, 3rd ed., p. 1–10, McGraw-Hill, 1993.
Johnson, R. C., H. A. Ecker, and J. S. Hollis, “Determination of Far-Field Antenna Patterns
from Near-Field Measurements,” Proceedings of the IEEE, December 1973, pp.
1668–94.
Slater, J. C., and N. H. Frank, Electromagnetism, 1947, Mc-Graw-Hill, ch. 13.
APPENDIX H
Path Length Differences

from a Planar Aperture
Appendix H includes calculations on the differences in distances from points on a planar

surface to a distant point P. The results are used in Appendix G for determining antenna
patterns versus range.
Consider the spherical coordinate system of Fig. H–1(a) with point P located at
coordinates xp, yp, zp in space and where P is located a distance R from the origin
(x = y = z = 0). Notice that angle q is measured from the z-axis, but angle f is measured
from the x-axis and it is always in the x-y plane. From standard spherical coordinate
relationships,
x p = R sin θ cos φ ≡ Rα
y p = R sin θ sin φ ≡ Rβ (H–1)
z p = R cos θ
In (H–1) above, the abbreviations Ra and Rb are used to denote xp and yp to simplify
equations that follow. Notice that
12 12
R = ( x p ) + ( y p ) + ( z p )  = ( Rα ) + ( Rβ ) + ( R cos θ ) 
2 2 2 2 2 2
(H–2)
Next, referring to Fig. H–1(b), we find the distance r1 between points at xa and ya on the
aperture (za = 0) and point P is as follows
12 12
r1 = ( xa − x p ) + ( ya − y p ) + ( z p )  = ( Rα − xa ) + ( Rβ − ya ) + ( R cos θ ) 
2 2 2 2 2 2
(H–3)
After expanding (H–3) and noting from (H–2) that [(Ra)2 + (Ra)2 + (R cos q)2] = R2, we
find
12
 2 x 2 + y2 
r1 = R 1 − (α xa + β ya ) + a 2 a  (H–4)
 R R 
Now assume that the aperture dimensions are small compared to R, that is, xa << R
and ya << R. Then, r1 of (H–4) is simplified and becomes (H–5) that follows
475
476 Path Length Differences from a Planar Aperture
q P (xp, yp, zp) r1 P (xp, yp, zp)
R
R q
y z
y f
f
x
(a) (b)
FIGURE H–1.
Spherical coordinates with array elements in x-y plane. In (a), point P is at distance R from
origin; in (b), point P is at distance R from the origin and distance r1 from a position xa,ya on
the x-y plane.
12
r1 ≈ R ⎡⎢1 − (α xa + β ya )⎤⎥ ≈ R ⎡⎢1 − (α xa + β ya )⎤⎥
2 1
⎣ R ⎦ ⎣ R ⎦
= R − xa sin θ cos φ − ya sin θ sin φ (H–5)
Therefore, when R>> both xa and ya, r1 exceeds R by d, where
δ = r1 − R = − sin θ ( xa cos φ + ya sin φ ) (H–6)
Now consider effects on r1 and d with R fixed, and with P only on one of the principal
planes, x-z or y-z. Then, for the x-z and y-z planes: f = 0 and f = p/2 (90°), respectively.
Thus, from (H–6), with R fixed and P in the x-z plane
δ = r1 − R = − xa sin θ ( for φ = 0 ) (H–7)
and with P in the y-z plane
δ = r1 − R = − ya sin θ ( for φ = π 2 ) (H–8)
Now assume the x and y axes are vertical and horizontal, respectively, as in Fig.
H–1(b). Then from (H–7) and (H–8), which are applicable when xa << R and ya << R,
we see that to close approximations the following are true:
(1) for P on the vertical plane (x-z), r1 and d are functions of the xa (vertical)
location on the aperture—not on the ya (horizon) location; and similarly
(2) for P on the horizontal plane (y-z), r1 and d are functions of the ya (horizon-
tal) location on the aperture—not on the xa (vertical) location.
Path Length Differences from a Planar Aperture 477
Now continuing to refer to Fig. H–1(b), we see that

• the principal plane pattern in azimuth is the q dependent pattern when
f= 90°; and
• the principal plane pattern in elevation is the q dependent pattern when
f= 0.
It is to be underscored that all discussion following (H–4) is limited to situations
for which the antenna aperture dimensions are small compared to the distance R,
between antenna origin and the observation point P. Therefore, that discussion is appli-
cable only to the far zone. In general, however, radiating near-field pattern calculations
must include the exact pathlength difference d (i.e., r1 − R) attainable from (H–4). Pat-
terns versus distance R are discussed in Appendix G and relevant calculations are
included in the accompanying SM.
APPENDIX I
Effects of Random Aperture

Phase Errors
In arrays, aperture phase errors are caused by element misalignment and improper phase
excitation, and surface errors are a major source of aperture error for reflector antennas.
Section 6.5 includes a general discussion on reflector errors, and the present appendix
addresses relatively simple calculations on the effects of random phase errors.
In Fig. I–1, it may be seen that the difference DR in the pathlengths aa′ and bb′ is
∆ R = 2h sin α (I–1)
Thus the phase difference in radians DY caused by the path length difference DR between
paths aa′ and bb′ is
∆ψ = ( 2π λ ) ∆ R (I–2)
Then, if the height distribution for a random rough surface is designated by Dh, the
distribution of phase differences can be expressed as
∆ψ = ( 2π λ ) 2 ∆hsin α (I–3)
Experience shows that random surface roughness is well described by surface heights,
which can be expressed statistically as a normal (i.e., Gaussian) distribution. Then, in
Mathcad, the surface height distribution Dh becomes
∆h = rnorm ( n, µ, σ ) (I–4)
where
n = number of data points
m = average value
s = standard deviation
If m = 0, the standard deviation s equals the rms of the distribution in height.
479
480 Effects of Random Aperture Phase Errors
a′ When DY of (I–3) is sufficiently small,

b the relative phases of the rays from the
various reflecting points is small and their
b′ a phases tend to add constructively. Then,
a the reflections are specular. Otherwise, to
a account for the phase differences DY
h
caused by rougher surfaces, effects of the
FIGURE I–1. terms l, Dh and sin a in (I–3) must be
included.
The difference DR in the path lengths of aa′
and bb′ is 2 h sin a. Thus the phase difference If a surface is curved, as for most
DY is (2p/l) DR = (4p/l)hsin a from Long antenna reflectors, effects of curvature on
2001, courtesy Artech House. (I–3) must also be included, as is now
discussed. From Fig. I–2, which depicts a
ray reflected from a parabolic surface, it
can be seen that 2a + b = 180° or
α = (180° − β ) 2 (I–5)
Therefore, to include effects of random

errors on the parabolic surface, effects of
the angle b on DY must be included.
Then, for (I–3) to be used with Fig. I–2,
Dh = rnorm(n, m, s) and a = (180° − b )/2.
Thus, for a curved reflector surface having
a uniform surface roughness, the spread
in phase error (standard deviation)
depends on the angle b and therefore it
varies along the aperture x dimension.
A cautionary note: up to this point a in
Appendix I is used to denote the grazing
FIGURE I–2. angle at a reflecting surface. In (I–6) that
Geometry showing that a = (180° − b )/2. follows, ai is the phase of the ith element
Thus, for a uniform surface roughness, since a of an array and it is therefore unrelated to
depends on b, the range of random phases DY this previous use of a.
varies over the reflector surface. The following is an example of the
effects of random phase error on the
pattern of an array of closely spaced array elements, having a length of 4 meters and
located on the x-axis. The pattern calculated is a function of q and it is in the x-z plane,
with f = 0. It is assumed that l = 0.1 m and that a continuous aperture is simulated by
sampling at 401 points on the x-axis. Thus, the sample spacing is 0.1 l. In order for the
aperture to be centered at the x-y-z origin, we sum over the index i and define the aperture
coordinates as
xi = (i − 200 )10 −2
Effects of Random Aperture Phase Errors 481
Then, using the general relationship from array theory for the amplitude of the electric
field (sec. 5.5.1), for the far zone we have
{ 2π
}
401
E (θ ) = ∑ ai exp j α i +   xi sin θ  (I–6)
i=0   λ 
where
ai = the amplitude of the ith element
ai = phase of the ith element
We now make the following additional assumptions, namely, that (1) amplitude is con-
stant over the aperture with each ai = 1 and (2) the phase is uniformly and normally dis-
tributed over the xi.
Figure I–3 shows results of assuming rms phase errors of l /16 and l /8 (22.5° and
45°), and comparing those results with zero phase error (for the calculations, see the
accompanying SM). Note that the pattern peak levels are reduced with increases in
–5
–10
Field Strength in dB
–15
–20
–25
–30
–9 0 9
Off-Broadside Angle in Degrees
Zero Phase Error

RMS Phase = 22.5 degrees
RMS Phase = 45 degrees
FIGURE I–3.
Antenna patterns from apertures with rms phase errors of 0, l/16, and l/8.
482 Effects of Random Aperture Phase Errors
the rms errors. Additionally, the loss in gain caused by l /16 rms phase error is only a
fraction of a dB, but it is discernible. On the other hand, because of the randomness in
phase error and the resulting randomness in side-lobe levels, the patterns are not sym-
metrical between the negative and positive angles from the major lobe.
Reference
Long, M. W., Radar Reflectivity of Land and Sea, 3rd ed., Artech House, 2001, p. 52.
Index
absorption, attenuation due to, 13 AN/SPS-10 radar antenna (C-band), gain

absorption coefficient, 20 measurements on, 378
active arrays, 354 AN/SPY-1 shipboard multifunction S-band
active ESA antennas (AESAs), 367 radar, of U.S. Navy, 365–366
active element pattern, 177 AN/TPS-32, 364, 364 fig.
adaptive array antennas antenna design, 371
adaptive beam-forming arrays, antenna measurements
360–362, 360 fig., 362 fig. compact ranges, 382–385, 383 fig.,
advantages of, 365 384 fig.
background system developments for, directivity and gain, 393
358–360, 359 fig.–360 fig. absolute-field-strength method,
basic components of, 360, 360 fig. 394–395
space-time adaptive processing in, gain measurement by comparison,
362–364, 362 fig. 396–397, 397 fig.
wireless two-dimensional, 363 power-measurement methods, 396
adaptive beam-forming (ABF) system, standard gain antennas, 395–396
337, 360–362, 360 fig., 361 fig. of efficiency, 398–399
Adcock antenna, 311, 312 equipment for, 390–392, 391 fig.–
Aegis weapon system, 365 393 fig.
AESAs. see active ESA antennas of impedance, 400–404
airborne warning and control system near-field, 385–388, 386 fig.–389 fig.
(AWACS), 366 of noise temperature, 409–414, 411 fig.
aircraft landing systems, antenna patterns antenna patterns, 371–377, 372 fig.,
for, 314 374 fig., 375 fig.
Alford loop antenna, 147 fig. far-zone pattern measurements,
amateur (ham) radio, 4 377–382, 380 fig.
amplitude tapering, see tapering near- and far-field statistical gain,
AN/FPS-108 Cobra Dane system, 366, 376–377, 377 fig., 378 fig.
366 fig. positioner for, 392, 393 fig.
AN/MPS-29 (XE-1), geodesic lens- radiating near- and far-fields patterns,
scanning antenna for, 317, 318 fig. 373–376, 374 fig., 375 fig.
483
484 Index
antennas receiving, 30
array, 169 resonant, 139
broad-band, 285 standard gain, 397
current and voltage in longer, 132–136, structures of, 89–91
133 fig.–135 fig. surface-wave, 159
defined, 1, 87 synthetic-aperture, 285, 324–325
direction-finding, 310–314, 312 fig., transmitting vs. receiving, 30–31
313 fig. antennas, long-wire
electrically small, 308–310 classification of, 138–139
with extremely low side lobes, directivity of, 141
325–327 effect of ground on, 143
fractals, 328–330, 330 fig., 331 fig. polarization of, 144–145
frequency-independent, 293–301, 295 radiation patterns of, 138–141, 140 fig.,
fig., 296 fig., 298 fig.–301 fig. 142, 142 fig.
frequency-scan, 338, 341, 339 fig. radiation resistance of, 141, 141 fig.
helical, 148–149, 148 fig. uses of, 143
ideal, 87 antenna-under-test (AUT), 385
leaky-wave, 159 aperture, of antenna, 269
loop, 144–148, 144 fig., 147 fig. circular, 279, 279 fig.
low-noise receiving, 320–324, 323 fig. of parabolic aperture, 240
mechanical scan, 314–321, 316 fig., shape of, 233
318 fig.–320 fig. aperture analysis, 329
microstrip, 157–159 continuous line source in, 273–276,
for multiple polarization 273 fig., 277 fig., 278 table
for circular polarization, 302–305, general, 274
303 fig. historical note on, 270
corrugated horn feeds, 307, 307 fig. pattern calculations for continuous,
dual-mode transducers, 306, 306 fig., 269, 271, 273, 275, 277, 279
307 fig. pattern vs. Q and F in, 279, 279 fig.
polarizers, 305–307, 307 fig. principal plane patterns in, 279, 278 table
nonresonant, 142, 142 fig. separable x and y distributions in, 275,
omnidirectional, 307–308 277, 277 fig.
parameters for, 87 aperture dimensions
aperture dimensions, 113–115 and beamwidth, 119
bandwidth, 111–113 interdependent with gain, 113–114, 116
beamwidth, 105–107 aperture distribution, 151
directivity and gain, 98–103 calculating, 209
effective area, 103–105 defined, 236
input impedance, 110–111 Dolph-Chebyschev (D-C) distribution,
minor lobes, 107–108, 108 fig. 203
polarization, 112 planar and volume arrays with uniform,
radiation resistance and efficiency, 190–191, 193, 191 fig.
108–109 for reflector antenna, 251, 255, 255 fig.
patch, 157–159 Taylor distribution, 239
Index 485
aperture efficiency, 114, 420 radiation resistance in, 179

aperture phase errors, effects of random, two-isotropic-element, 170–171, 174,
480–482, 481 fig. 171 fig., 173 fig.
aperture tapers, 239 artificial-dielectric media, 260
Arecibo, Puerto Rico, Arecibo atmosphere, absorption attenuation in, 14
Ionospheric Observatory at, 252 atmospheric gases, and propagation on
array fill time, 344 radio, 32
array patterns attenuation
broadside linear, 183, 184 fig., due to absorption, 13, 14
186 fig. logarithmic expression of, 12
general equations for, 194, 209, 214 for point and line sources, 230
arrays of transmission lines, 65
electronically steered, 221 attenuation factor, 14
elements of, 169 AUT, 392. see antenna-under-test
horn, 220 AWACS. see airborne warning and
long-wire, 218–219, 219 fig. control system
microstrip, 220 AWACS radar antenna, 326
multielement uniform linear, 182–183, azimuth dimension, interference
185, 187 suppressed in, 367
and mutual coupling, 221–222 azimuth scanning, 315
parasitically excited endfire, 188–189,
189 fig. Babinet’s principle, 155
patch, 220–221 baluns, 70
planar, 209–211, 213–217 in broad-band design, 287
slot, 220, 220 fig. defined, 163
with uniform aperture distribution, examples of, 163, 163 fig.
190–191, 193, 191 fig. bandwidth
array tapering. see also tapering defined, 112
Dolph-Chebyschev patterns, 203, impedance, 111, 112
204 fig. instantaneous, 112
for side-lobe reduction, 201, 203 pattern, 111
Taylor distribution for, 206–208 in phased array, 345
array theory, 169 specification of, 286
element directivity and mutual tunable, 112
coupling, 181–182 beam-maximum systems, 312
input and mutual impedances in, beam shift, estimates of, 344
178–179 beam steering
mutual coupling in, 177–178, effects of, 345, 349
181–182 by feed offset, 269, 270 fig.
parallel-dipole two-element array, beam-steering technology
173–175, 176 fig. effects of element mutual coupling, 350
pattern vs. Q and F, 174 effects of phase quantization, 349
principle of pattern multiplication in, phase shifters, 349–351, 351 fig.
179–181, 184 time delayers, 351
486 Index
beamwidth coaxial lines, characteristic impedance

in aperture analysis, 274 table, 278 for, 72
table Cobra Dane (AN/FPS-108) phased array
and aperture shape, 234 antenna, 366, 366 fig.
definition for, 106–107 coherent side-lobe canceller (CSLC), 358
half-power, 106, 106 fig. collimation
for horn antennas, 245, 247 defined, 228
and line-source feeds, 239 ray, 281
measurements for, 372 collinear broadside array, 183, 183 fig.
for planar arrays, 213–216 compact range, 382–385, 384 fig.,
practical significance of, 104–105 385 fig.
principal-plane, 106 complementary antennas, 298
for short dipoles, 125, 123 fig. complex-variables, 439–443
for slot radiators, 155 conducting materials, of antennas, 90–91
Taylor distribution for, 206 conductors, 90–91
Beverage antenna, 143–144 conformal arrays, 377–378
biconical antennas, 289–290, 290 fig. conical scan lobing system, 313
biconical horn, 154–155, 155 fig. conical scanning, 311–314
Blass matrix, 356 conical spiral antenna, 297, 297 fig.
Booker’s principle, 300 coordinate systems, 92–95, 98 fig.
Bootlace lens, 353–354 corner-reflector antennas, 254–255, 254
Brewster angle effect, 40 fig., 255 fig.
broad-band antennas, 285 corona, considerations in, 91
basic principles of, 286–288 corrugated horns, 307, 307 fig.
biconical antennas, 289–290, 290 fig. cosmic-ray radiation, 3
dipoles, 288–289, 288 fig., 289 fig. coupling devices
folded dipoles, 291–292, 291 fig., 292 fig. directional couplers, 81–84,
miscellaneous dipoles, 292–293, 294 fig. 82 fig.–84 fig.
monopoles, 290–291, 290 fig. hybrid couplers, 81–84, 82 fig.–84 fig.
superturnstile antenna, 292, 292 fig., for transmission-line-to-waveguide
293 fig. coupling, 78–79, 80 fig.
broadside arrays, 174, 185, 185 fig., current
186 fig. displacement, 122
amplitude tapering of, 200–201, 200 and electromagnetic fields, 1
table, 202 fig. in long antennas, 131–135,
beamwidth and gain of, 213–216 132 fig.–133 fig.
and continuous aperture antenna, curtains, 188
196–197, 197 fig. Cutler, Maine, U.S. Navy installation in,
linear, 181–184, 182 fig., 184 fig. 131
broadside patterns, of planar arrays, 211 cutoff frequency, for waveguides, 75
Butler matrix, 355–356, 356 fig. cylindrical lens, 257
Cassegrain feed system, 245–246, 244 fig. decibels, space attenuation in, 13
and noise reduction, 323–324 delta-match feed, 159, 159 fig.
Index 487
design ratio, for frequency-independent element, 181–182

antennas, 295 of elemental dipole, 125–126
device under test (DUT), measurements and Hansen-Woodyard condition, 188,
of properties of, 401 216
dielectric constants, 17, 447 table, for helical antenna, 150
465 table IEEE definition of, 99
dielectric lenses, 260–262 of long-wire antenna, 139, 143
dielectric rods, as waveguides, 81 of long-wire arrays, 219
dielectrics, artificial, 258 and pattern multiplication, 181
diffraction, 15, 224 of planar array, 190, 209–213
defined, 25–26, 26 fig. and radiation intensity, 99
at edge of obstacle, 25–26, 27 fig. of reflector antenna, 249
geometric theory of, 326–327 for slot radiators, 155, 157
diffuse reflection, 20 and solid angle, 99, 100 fig.
digital signal processing (DSP), 322 discone antenna, 294 fig.
diode switching, 348–349 displacement current, 123
dipole antenna, 388 divergence, definition of, 428
dipole-array patterns, two-element, divergence factor, 459–460
172–175 Dolph-Chebyschev (D-C) distribution,
dipoles. see also short dipoles 203
biconical antennas as, 289 Dolph-Chebyschev (D-C) patterns
and broad-band antenna design, 287, simplified calculations for, 205
288–289, 288 fig., 289 fig. and Taylor distribution, 206, 208
circular polarization with, 302–303, dual-mode transducers, 305
303 fig. ducting, 35
folded, 291–292, 291 fig., 292 fig. DUT. see device under test
fractal, 329–330
half-wave, 135–137, 136 fig. E-3A airplane, 326
horizontal-plane patterns of two earth’s reflection, and measured gain, 394
vertical, 174 fig. earth’s reflection coefficient, variations in,
miscellaneous broad-band, 293–294, 39–40
395 fig. earth’s surface
directional couplers, 81–84 effects of, 31–32
direction-finding antennas reflection from, 35–39
beam-maximum systems, 311 E-2C aircraft, UHF array on, 216
lobing antennas, 311–313, 311 fig., Echo 1, 34–35
312 fig. Echo 2, 34–35
low-frequency null systems, 310–311 edge taper, defined, 236
directive antenna types, comparison of, effective area, 103–105
279 Effelsberg, Germany, fully steerable
directivity radio-telescope antenna at, 252
calculation of, 181, 209 efficiency, antenna
defined, 98 aperture efficiency, 399
and electrically small antenna, 310 radiation efficiency, 398–400
488 Index
“egg crate” construction, 261 microstrip and printed-circuited

EHF. see extremely high frequency technology, 357
electrically small antennas, 308–310 and T/R modules, 354–355, 354 fig.
electric field intensity, geometry of, electronic scanning array
123 design of, 169
electrodynamic phenomenon, radiation as, use for, 285
121 elemental dipole, 120, 151
electromagnetic (E-M) wave propagation, current and voltage in, 132
427 directivity of, 125–126
electromagnetic (E-M) waves, velocity of, pattern of, 123–125, 124 fig.
430 total power radiated by, 125
electromagnetic field, 1 elementary doublet, 120
energy transferred in, 123 element directivity, effects of, 181–182
electromagnetic spectrum, 3 ellipsoidal dipole, 294 fig.
electromagnetic waves, 159 endfire arrays, 175, 186–188, 187 fig.
characteristics of feed arrangement for, 185, 187 fig.
absorption, 13–15 parasitically excited, 188–189
decibel (logarithmic) expression of patterns calculations of, 199–201
attenuation, 12–13 unidirectional, 185
field intensity and power density, energy conservation, principle of, 8
10–12 Engineering Experiment Station (EES), at
frequency and wavelength, 3–4 Georgia Institute of Technology,
polarization, 6–7 271, 315, 319 fig.
rays and wavefronts, 7–8 environmental wave-propagation effects
space-time relationships, 4–5 beyond-the-horizon propagation,
spherical waves and inverse-square 32–35, 33 fig.
law, 8, 10 and earth environment, 31–32
wave velocity, 2 radio-transmission equation, 42–44
defined, 2 and reflection from earth’s surface,
environmental wave-propagation 35–39
effects, 31–44 transionospheric, 41–42
origins of, 21 variations in earth’s reflection
and radiation requirements, 28–30 coefficient, 39–40
electromagnetic wave theory, 1 wave nomenclature, 40–41, 41 fig.
electronically scanned antennas (ESA), equiangular log-periodic antennas,
315 299–300
electronically scanned array (ESA), in equivalent-circuit line representation,
AWACS, 366 of transmission line, 48–49,
electronically steered arrays (ESAs), 169, 49 fig.
337, 364, see also phased arrays extremely high frequency (EHF)
adaptive array antennas, 358–364 characteristics of, 3 table
advancements in, 337 frequency allocations for, 4
and beam-steering technology, 347–351 extremely low frequency (ELF),
Butler matrix, 355–356, 356 fig. characteristics of, 3 table
Index 489
Faraday rotation, of polarization of wave, matrix methods, 355–356, 356 fig.

41–42 for MILORD antenna, 316–317,
Faraday’s law, and Maxwell’s equations, 316 fig.
428 for parabolic reflectors, 228–230,
far field 230 fig.
and inverse-square law, 10 for paraboloid, 255
minimum permissible distance for, 95 for phased arrays, 357–358, 359 fig.
statistical gain, 376–377 primary, 249
far-field range, for antenna measurements, Fermat’s principle of least time, 265–266
381 ferrite phase shifters, 348–349
far zone, defined, 469 field-strength pattern, 36–40, 37 fig.
far-zone amplitude, measured wide-angle, flat earth geometry, 450–453, 450 fig.
373, 374 fig. Fourier integral computer, development
far-zone approximation, 467–468, of, 271
468 fig. Fourier transform, 472
far-zone measurements, 373, 377–383 fractal dipole antennas, types of,
far-field ranges, 381–382 330–331, 331 fig.
high-frequency techniques, 379–381, fractals, 285, 328, 330, 332
380 fig. Franklin antenna, 185, 185 fig.
low-frequency techniques, 379 Fraunhofer region, 96, 233, 469
fast Fourier transform (FFT), 328, 363, free space
472 electromagnetic waves in, 2
Federal Communications Commission plane electromagnetic wave in, 5–6,
(FCC), 4, 177 6 fig.
the feed, defined, 233 point-source wavefronts and rays in,
feed design, for reflector or lens, 9 fig.
235–236, 235 fig., 236 fig. and wave velocity, 3
feed lines, 89 frequency
feed methods defined, 3
basic, 159–164, 159 fig.–163 fig. in electromagnetic spectrum, 3
“delta match,” 161, 161 fig. frequency allocations, in U.S., 4
for Franklin antenna, 185, 185 fig. frequency-independent antennas
for horn radiator, 150–151, 153, 163, equiangular log-periodic antennas,
163 fig. 299–300, 299 fig.
for longer antennas, 132–135, 134 fig. and log-periodic principle, 294–297,
with quarter-wave transmission-line 295 fig., 296 fig.
transformer, 161, 163, 161 fig. practical log-periodic designs,
“Zepp,” 161–162, 162 fig. 297–300, 298 fig.
feed systems properties of, 295–296
Cassegrain, 244–245, 244 fig. self-complementary antennas, 300–301,
corporate, 351–353, 353 fig. 300 fig., 301 fig.
for endfire arrays, 186–188, 187 fig. spiral antennas, 301, 301 fig.
line-source, 239–240, 240 fig., frequency-scan antennas, 341–342,
241 fig. 341 fig.
490 Index
Fresnel equations, 328–329, 445–447 Georgia Institute of Technology,

Fresnel obliquity factor, 469–470 Engineering Experiment Station
Fresnel region, 96, 233, 469–470 at, 271, 315
Friis transmission equation, 103–104 global positioning satellite (GPS),
Ft. Huachuca, Arizona, U.S. Army frequency allocations for, 4
Electronic Proving Ground at, goniometer, 311–312
384 GPS. see global positioning satellite
GTD. see geometric theory of
gain, 42 diffraction
in aperture analysis, 274 table, grating lobes, 186, 198, 346–347, 349
278 table Green Bank, West Virginia, U.S. National
and aperture dimensions, 113–115 Radio Astronony Observatory at,
and beamwidth, 105 252
defined, 103, 399 grounding, 91
determination of, 107, 106 fig. ground reflection, and long-wire antennas,
of horn antennas, 242 143
IEEE definition of, 99 ground wave, 40–41
for long-wire arrays, 220, 221 fig. group velocity, 72–74
measurements for, 381 guy wires, 90, 91
and pattern multiplication, 181, 184
and radiation intensity, 99–101 half-wave dipole, 133, 134, 136–139,
and solid angle, 99–101, 100 fig. 141–144, 134 fig., 137 fig.,
specified, 98–99 138 fig.
gain calibration procedures, 395 Hamming function, 209
gain measurement Hansen-Woodyard, 186, 188, 200, 201
absolute-field-strength method, 394 helical antenna, 148–150, 148 fig.
by comparison, 410–413, 410 fig. directivity for, 150–152, 154, 155
power-measurement methods, 396 feed method for, 159, 162
gain standards, 395–397 normal mode, 149
geodesic lens, 266–268, 266–268 figs. use of, 153, 157
geometric optics, 22, 233 Hertz, Heinrich, 120, 122, 430, 431
collimation, 228–229 Hertzian dipole, 120
focusing, 228–229, 231, 235 high frequency (HF), characteristics of, 1,
geometric theory of diffraction (GTD), 2, 3 table
285, 328–329 horn antennas, 283
geometry aperture dimensions for, 113
flat earth, 451–453, 450 fig. feed patterns for, 246, 248, 249
fractal, 335 and shapes of major lobes, 246, 248,
planar near-field, 386–387, 387 fig. 252
spherical earth, 455–457, 459, 456 fig., wide-beamwidth patterns for, 247, 252
457 fig. horn array, 220
spherical near-field measured, 387, horn radiators, 150–153
389 fig. biconical horn for, 154, 155, 154 fig.
stripline, 367 circular polarization with, 308
Index 491
feed methods for, 159, 162, 161 fig. input impedance, 110, 111
radiation patterns for, 151, 152, 154, of basic folded dipole, 291
155, 157 calculation of, 414, 415
horns, corrugated, 307, 307 fig. for electrically small antennas,
Howells-Applebaum method, of pattern 308–311
optimization, 360–362 equations for, 59, 62
Huygens-Kirchhoff formulation, 469, 470 and feed methods, 159, 162
Huygens’ principle, 2, 25–26, 26 fig., measurement of, 389–391
151, 227 of short dipoles, 126–129, 127 fig.
Huygens’ wavelets, 470, 472 Institute of Electrical and Electronics
hybrid couplers, 80–82, 82 fig. Engineers (IEEE), 43, 450, 460
Instrument Low-Approach System (ILS),
IEEE. see Institute of Electrical and 314, 319
Electronics Engineers, Inc. insulators
illumination protection of, 90, 91
defined, 236 uses for, 83
reflectors, 227, 229–252, 228 fig., 230 intercontinental ballistic missile (ICBM)
fig., 235 fig., 243 fig., 244 fig. detection and analysis, 366
ILS. see Instrument Low-Approach interference, 15, 394
System and antenna applications, 20, 23, 25
images, principle of, 21, 21 fig., 22, effects of, 29
129 problem of, 53
impedance charts, 404 interferometer system, 314
impedance match International Telecommunications Union
in broad-band design, 285–290, 292, (ITU), 3–4
293, 296 inverse-square law, 10
between transmitter and free space, and attenuation, 12–14
221, 223 and power density, 8, 10–12
impedance measurements, 401–404 and radiation process, 28
network analyzers for, 391, 400–404, spherical waves and, 8
403 fig. ionosphere
standing-wave method, 401, 404, and long-distance radio propagation,
401 fig. 33, 34
impedances and radio waves, 2, 40–41
in array theory, 169, 171 ionospheric scatter, 34
characteristic, 51, 53, 55, 71, 72, 76 irises, reactance introduced into
mutual, 177 waveguides by, 77, 77 fig.
impedance transformation, analysis of, Isbell log-periodic dipole array, 299, 230,
65, 67, 69 305
impedance transformer isotrope, defined, 8
single-stub, 66, 66 fig. isotropic point source, 169–171,
transmission line as, 54, 55, 64, 65 171 fig.
induction field, 28 ITU. see International
infinitesimal dipole, 120 Telecommunications Union
492 Index
jamming, 7 linear arrays

“J” antenna, 162 broadside, 183–186, 186 fig., 190–194,
Jodrell Bank, radio telescope antenna at, 191 fig.
252 element spacing of, 215
pattern calculations for, 194–199
Kirchhoff method, 328 synthetic aperture as, 324
Koch loop antenna, 330, 331, 330 fig. linear superposition, principle of, 22
Kraus, John, 148 line-source feed, 239–241, 240 fig., 249
line sources, pattern calculations for, 271,
Law of Conservation of Energy, 8, 10 276, 277 fig.
leaky-wave antenna, 159 Litz wire, 90
lens load impedance, determination of, 63
beamwidth and, 233, 238, 239 loading coils, and radiation resistance,
Bootlace, 353 309
for low side lobes, 326 lobes
Rotman, 353 grating, 184, 186, 194, 198, 208, 218,
waveguide horns as feeds for, 245–247 332, 346–347, 349, 370
zoned, 259 for horn antennas, 245
lens antennas phase quantization, 347, 349
beam steering by feed offset with, 269 lobing antennas, 312–315, 312 fig.
compared with reflectors, 275, 281 Lockheed Martin, 329, 364 fig.
dielectric lenses, 264–266 log-periodic antennas
focusing with dielectric and metal plate equiangular, 299–301, 299 fig.
lenses, 262 problems and exercises, 335
geodesic Luneberg lenses, 265–268, log-periodic principle, 294–297, 295 fig.,
268 fig. 296 fig.
lens refracting media, 260, 262, 268 long-wire antennas, directional properties
low side-lobe design for, 327 of, 218
metal plate lenses, 262, 264, 265, loop antenna, 87
265 fig. Alford, 147 fig.
optical lenses, 257, 258 circular, 146
surface configurations for, 257–259, coaxially positioned, 147
259 fig. feed methods for, 145
use of, 257, 266 Koch, 330–331, 330 fig., 331 fig.
variable-index-of-refraction lenses, resistance of, 145
261 small, 144–145, 144 fig.
lens design, 262 lossless infinite line, 50–51
light low frequency (LF), characteristics of,
free-space velocity of, 73 3 table
Maxwell’s hypothesis for, 430 Luneberg lens, 261–262, 262 fig.
lightning-induced currents, protection beam steering by feed offset with,
against, 91 269
light waves, reflection of, 19 geodesic, 260, 265–269, 262 fig.,
Lincoln Laboratory, 367 266 fig., 268 fig.
Index 493
Luneberg lens antenna, rapid-scan, MILORD antenna, 316–317, 316 fig.,

geodesic, 317, 319, 319 fig. 318 fig.
Molnya orbits, 35
magnetic dipole, oscillating, 120 Molnya satellites, 35
magnetic field. see also electromagnetic monopoles, 290–291, 290 fig.
field conical skirt, 294 fig.
generation of, 28 folded, 294 fig.
of short dipoles, 120–123, 121 fig., horizontal-plane patterns of two
122 fig. vertical, 176 fig.
Marconi, Guglielmo, 32 uses for, 291
masts, 88 monopulse, 314
matched load condition, 55–56 monopulse lobing, 314
matching principle, in impedance monopulse operation, 357
relations, 65 multielement array, 178, 182, 184,
Mathcad calculations, 39, 97, 445, 459, 182 fig., 184 fig.
469, 479 multipath analyses, 459
Mathcad software, 250, 374 multipath effects
Maxwell, James Clerk, 1, 122, 427 and flat earth geometry, 451–453
Maxwell’s equations, 1, 427–431 of pattern propagation factor F, 450
for analysis of radiation, 151 multipath propagation, 381
and determination of gain, 105, 106 fig. mutual coupling
for electric currents and associated between array elements, 169, 172
fields, 27 in array pattern analysis, 221
for electromagnetic analysis of element in beam-steering technology,
coupling, 178 347–349
Huygens’ principle deductible from, 25 effects of, 177–178, 181–182
physical meaning of, 428 with planar arrays, 271
plane waves in uniform material in, 429 mutual impedance, 178–179
on reciprocity, 30
medium frequency (MF), characteristics narrow-beam antennas, 378, 386
of, 3 table National Aeronautical and Space
metallic-obstacle delay medium, 260 Administration (NASA)
metal-plate lens, 262–265, 264 fig., Deep Space Network program, 413
265 fig. and satellite communications, 34
metal-plate waveguide refracting medium, National Telecommunications and
260 Information Agency, 4
metals, as conductors, 89–90 near-field
meteor scatter, 34 radiating, 96, 470–472
Method of Moments (MoM), 285, reactive, 96
328–329 near-field measurements, 96, 382,
microstrip antennas, 157–158, 158 fig. 385–386
microstrip array, 220 cylindrical, 386–387, 387 fig., 388 fig.
microwave region, 4 planar, 386–387, 386 fig., 387 fig.
millimeter wavelength region, 4 spherical, 386–389, 389 fig.
494 Index
near-field statistical gain, 376–377 line-source feed for, 239

network analyzer, 391, 400–404, pattern evaluation for, 251–252, 251 fig.
403 fig. stacking, 256
noise parabolic reflector
atmospheric, 321 compared with lens antennas,
calculated temperatures for, 409, 257–258
410 fig. feed-positioning requirements for,
comparison source of, 413 241–244, 243 fig.
cosmic, 321 paraboloidal reflector
at high frequencies, 29 Cassegrain feed for, 244, 244 fig.
measurement techniques, 410–413, elliptical-aperture, 237–238
411 fig. feed-support for, 242, 243 fig.
in radio systems, 321 simultaneous-lobing (monopulse) feed
system noise temperature, 414–417, for, 312, 312 fig.
417 table paraboloidal reflector antenna, far-zone
theory, 407–408 pattern of, 373, 374 fig.
using ambient sources of, 413 paraboloids, dual polarized back-to back,
noise, antenna 317, 319 fig.
external sources of, 410 fig. paraboloid surface, 230, 230 fig., 232
temperature, 409, 410 fig. parasitic elements, 188
noise figure, 414–417, 417 table parasitic excitation, 188
noise measurements, 371, 412 Pascal’s triangle, 202
noise theory, 407 patch antennas, 157–159, 158 fig.
nomenclature, wave, 40, 41 fig. design and development, 158
nonresonant antennas, 139, 143 with fractal patch, 331, 331 fig.
normal mode helical antenna, 149 patch arrays, 220–221
null-filling effect, due to phase error, path length differences from aperture,
375, 375 fig. 476–477, 476 fig.
null systems, low-frequency, 311–312 path length differences, from planar
aperture, 475–477, 476 fig.
obliquity factor, 469–474 pattern measurements, 377
ohmic loss, for thin wall radome, 466 pattern multiplication, principle of,
omnidirectionality, 308 179–181, 184
optical principles, radio-wave, 15–26 pattern optimization, 361–362
optics pattern-propagation factor F, 43–44,
geometric, 22, 227–234 450–451, 450 fig.
organ-pipe feed, 267 patterns, endfire array, 199–201, 200 fig.
see also radiation patterns
parabola, geometry of, 228, 229 fig. pedestals, 88–89
parabolic antenna, steps for developing, phase-comparison lobing, 314
251–252 phased arrays
parabolic cylinder, 230, 230 fig. alternative configurations, 355
aperture of, 232 array bandwidth, 342–343
attenuation for, 230–231 array sum and difference patterns, 357
Index 495
conformal, 357–358 point source feed, 249

defined, 169 polarization
effective aperture vs. scan angle, 340 circular, 7, 42, 113, 150, 235, 300,
feed system technology for, 351–358, 302–305, 303 fig., 388, 433
352 fig.–354 fig., 356 fig. defined, 433
frequency-scan antennas, 341–342, effect of Faraday rotation on,
341 fig. 41–42
grating lobes, 346–347 of electromagnetic waves, 6–7
patterns calculated for, 344–346 elliptical, 388, 389
and phase shift requirements, 339 horizontal vs. vertical, 39
principles of, 338 linear, 7
for scanning, 221 measurement of, 371, 398
subarrays, 355 multipath dependencies on, 453–455,
time delay, 339–340, 339 fig. 454 fig.–456 fig.
phased array technology, 338 polarization pattern, 93, 113
phase distribution patterns, of current and polarization theory, 433
voltage, 132–133, 133 fig., and circularly polarized components,
134 fig. 435–436, 435 fig.
phase error and linearly polarized components,
effects of random aperture on, 433–435, 434 fig., 435 fig.
479–482, 480 fig. relationships between linearly and
far-zone range-approximation and, circularly polarized components,
467–468 436–437
phase pattern, 93 polarizers, 305–306, 306 fig.
phase-phase steered scanning arrays, 364 polyrod antenna, 159
phase quantization, effects of, 349–350 positioner, 390–391, 391 fig.
phasers, vs. phasors, 347 in compact range design, 384
phase shifters, 347–349, 349 fig. types of, 403
phase velocity, 72–73 power density
pillbox antenna, 240 fig. calculation of, 8, 10
planar aperture, path length differences defined, 8
from, 475–477, 476 fig. of electromagnetic fields, 10–11
planar arrays and reflection of radio waves, 19
broadside patterns of, 213 power-density ratio, 12
calculations for, 209–210 power dividers, for transmission lines,
elements of equal amplitude, 212–213 67–69, 67 fig., 69 fig.
element spacings for, 210, 210 fig. power transmission coefficient, 19
equations for patterns of, 213–215 Poynting vector, 430–431
pattern vs. Q and F in, 191–192 principal-plane patterns, defined, 93
separable aperture distributions, propagation
211–212 beyond-the-horizon, 33, 33 fig.
with uniform aperture distribution, 190 round-the-world, 34
plane wave, space-time relationships for, with spherical earth model, 456,
5–6, 6 fig. 457 fig.
496 Index
propagation medium, 2 radiation resistance

isotropic, 10 in array theory, 181
with nonlinear properties, 22 calculation of, 108–109
refractive index of, 16 concept of, 102, 104, 108
and wave impedance, 11 for electrically small antennas, 309
pseudo-Brewster angle, 40 input impedance and, 177
of long-wire resonant antenna, 141
quantization lobes, 349 measurement of, 401
quarter wave plate, 304 in resonant antennas, 139
of short dipoles, 125
radar band designations, 4, 5 table radiators, 293
radar monopulse method of direction adapted to feeding reflectors, 235
finding, 357 broad-band dipoles, 292, 303
radiating far-field, 96, 470–472 half-wave dipole as, 136
radiating near-field, 96, 470–472 helical antennas, 148–150
radiation horn, 152–155
and electric currents, 27 long-wire, 141
and electromagnetic waves, 26 loop antennas, 144–147
from electrons in galactic space, 29 primary, 239
major and minor lobes of, 107 short dipoles, 121
requirements for, 28, 29 fig. beamwidth for, 126, 129
from sharp edges, 332 dipole radiation, 120
radiation efficiency, 111, 400 directivity of, 125, 137
radiation field, 28, 124 input impedance of, 126, 132
radiation intensity, determination of, with nonuniform current distribution,
102 127, 130
radiation patterns, 95 pattern of, 123–124, 126
absolute and relative, 94 radiation resistance of, 130
from aperture analyses, 246 short vertical antenna with ground
of array, 171, 177–178 image, 129
array factors, 194, 197 top-loaded antenna, 130
of continuous aperture antennas, slot radiators, 155, 157
248 vertical-tower, 138
coordinate systems, 91, 92 fig. radio broadcasting, two-element array in,
definition of, 93 175
and earth-reflection interference, 98 radio frequency identification (RFID),
in free space, 97–98 96
of horn antennas, 151, 153, 247 Radio Frequency Spectrum, 3 table
of long-wire antennas, 139 radio-horizon effect, and earth’s shadow,
near-field and far-field, 95–97 31 fig.
of nonresonant antennas, 142 radio propagation, long-distance, 32–34
in a plane, 93–94 radio transmission, 1
of resonant antennas, 139 radio transmission equation, 42
for uniform linear arrays, 185 radio transmitter, power delivered by, 65
Index 497
radio-wave optical principles, 15, 17, 19, reciprocity

21, 23, 25 between conductance and resistance, 58
Huygens’ principle, 25–26 defined, 21
interference, 22–23, 25–26 for electrically small antennas, 309
principle of images, 21, 21 fig. principle, 31
reflection, 18–21 reflection, 15
refraction, 15–18 defined, 15–16
radio waves diffuse, 20
beyond-the-horizon propagation, 32, from earth’s surface, 35–36, 38, 40
33 fig., 35 laws of, 229
and earth environment, 31 at plane boundary, 19, 19 fig.
and earth’s reflection coefficient, 39 from semi-rough surface, 21, 21 fig.
and reflection from earth’s surface, specular, 20, 21
35–36, 38 and standing waves, 52, 76
velocity of, 5 reflection coefficients, 19
radomes divergence factor, 459–460
defined, 469 and Fresnel’s equations, 445
electrical properties of materials used pattern-propagation factor, 450
for, 465 for rough surfaces, 449
problems caused by, 479 for smooth and rough earth surfaces, 445
types of, 486 for smooth land and sea, 447
use for, 103 variation in earth’s, 39–40
random phase errors, effects of, 479 and VSWR, 61, 63
range reflector antenna
compact, 385, 417 advantages of, 284
far-zone approximation, 470, 472 and beamwidth, 237–238
principal-plane patterns vs., 473 Cassegrain, 245
range antenna, in far-zone pattern compact range, 384, 384 fig., 385
measurements, 388, 396 compared with lenses, 285
ray collimation, 281 construction of, 253–254
Rayleigh criterion, 21, 449 corner-reflector antennas, 255–256
rays design, 251–252
characteristics of, 9 directivity for, 234
defined, 5 far-zone amplitude pattern of, 373,
point-source, 9 fig. 374 fig.
in propagation medium, 7 feed-positioning requirements for, 241
ray-wave front analysis, 227 illumination of, 237–239
reactive field, and dipole polarity, largest, 252–253
121 low side-lobe design for, 329
reactive near-zone field, 469 offset, 245, 252
reception parabolic, 232–234, 230 fig., 235 fig.
defined, 30 parabolic cylinder, 235, 243 fig.
and polarization, 42 surface accuracy requirements for, 253
reception patterns, 91 waveguide horns as feeds for, 248
498 Index
refraction self-complementary antenna, 300

atmospheric, 32 example of, 306
effects of, 467 logarithmic-spiral, 301 fig.
with gradual variation of refractive problems and exercises, 335
index, 17–18, 18 fig. SEP see scan element pattern, 177
at plane boundary, 18 fig. SHF. see super high frequency
refractive index shielding, principle of, 29
of air, 18 ships, search radar antenna on, 106
defined, 16 short-circuited lines, 60
resonant antennas, 136 short dipoles
long-wire, 169 beamwidth for, 202, 205
patterns of, 139, 140 fig., 142 center-fed, 120
resonant circuits, 57–58 characteristics of, 111
RFID. see radio frequency identification circuits of center-fed, 127 fig.
rhombic antenna, 219, 219 fig. compared with half-wave dipole, 139,
rotational lens, 258 141
Rotman lens, 353 current and voltage in, 132–133
roughness, electromagnetic effects of, 455 current-voltage relationship for, 120
rough surfaces, reflection coefficients for, and dipole radiation, 120
459–460 effective length of, 129
round-the-world propagation, 34 input impedance of, 126
isolated, 119
SAR. see synthetic aperture radar with nonuniform current distribution,
satellite communications, early, 34 127, 130
satellites pattern of, 125–126
geostationary, 35 short vertical antenna with ground
low earth orbiting, 35 image, 129, 132
Molnya, 35 and spherical coordinate systems, 123
turnstile antenna on artificial, 308 short monopoles, total radiated power of,
scale-model measurements, 389 131
scan angle, 269 side-lobe blanker (SLB), 358
scan antennas, rapid mechanical, 341 side-lobe canceller (SLC) antenna, 358,
scan element pattern (SEP), 177 359
scan mode of operation, 325 side-lobe levels, measurements for, 389
scanning sidelobe ratio, 107
conical, 323 side lobes
phased arrays for, 221 in aperture analysis, 274
scanning arrays, phase-phase steered, 364 and aperture dimensions, 113
scatter methods, limitation of, 34 array tapering for reduction of, 201,
Scientific Atlanta, Inc., 271, 383 203
sea of Dolph-Chebyschev patterns, 207
field strengths above smooth, 453, for elliptical apertures, 237
454 fig. extremely low, 326–327
reflection coefficients for, 458–460 of reflector antenna, 251–252
Index 499
signal-to-noise ratio, 320 of impedance, 401 fig., 404, 406 fig.,

skin effect, phenomenon of, 50 407
sky wave, 40–41 standing-wave pattern, of current and
slot array, 220 voltage, 132
slot radiators standing-wave ratio, with attenuation,
bandwidth properties of, 293 62–64
impedance properties of, 155 standing-wave voltage patterns, 62
use for, 203 STAP. see space-time adaptive processing
waveguide slots for, 157 static-charge buildup, protection against,
slotted lines, 402–403 91
Smith Chart, 63, 66–67, 289, 401 static field, and dipole polarity, 121
basic plan of, 404 steady-state lossless line equations, 54
example of use of, 415 steering angle, 269
soil, electrical properties of, 447 table stripline geometry, 357
solid angle, 101, 102 strip lines, 69, 71
space attenuation, 13, 14 stripmap mode of operation, 331
for paraboloidal reflector, 240 stub transformers, 66
for point and line sources, 230 submillimeter, characteristics of, 3 table
space-fed array, 353–354 super high frequency (SHF)
space shuttle-imaging-radar-B (SIR-B), characteristics of, 3 table
324 frequency allocations for, 4
space-tapering, 208. see also tapering superposition, principle of, 22–25
space telemetry applications, for helical superrefraction, 35
antenna, 150 superturnstile, 292, 292 fig., 293 fig.
space-time adaptive processing (STAP), superturnstile antenna, 303
362 surface roughness
space-time relationships multipath dependencies on, 453
for electromagnetic waves, 10, 22 reflection coefficients for, 445, 447
and interference phenomenon, 25 table, 448, 449
specular reflection, 20, 21 surface-wave antennas, 159, 188
“speed of light,” in vacuum, 2 surface wave propagation, 33
spherical coordinate system, standard, synthetic-aperture antennas, 285, 324–325
372 fig. synthetic-aperture radar (SAR), 324, 332
spherical earth geometry, 455, 457 system noise temperature, 415–417, 417
spherical earth model, to estimate wave- table
propagation path, 32
spiral antennas, 301 taper, see tapersing
spotlight mode of operation, 325 tapering
stacking, for superturnstile TV amplitude, 191, 191 fig., 192, 194–197,
transmitting antenna, 293, 302 197 fig., 198–203, 204 fig.,
standing-wave, 62–64, 67 205–206, 206 fig., 207–209
standing-wave measurement and Dolph-Chebyschev synthesis, 205,
determination of load impedance by, 207
63 in low-noise designs, 322
500 Index
tapering (cont’d) lossless infinite line, 50

with low-side-lobe array distributions, for multiple loads, 71 fig.
208 parallel branching, 67
and mutual coupling, 181 power dividers for, 67
for elliptical apertures, 237 reflection and standing waves, 52
for side-lobe reduction, 353, 355–356 reflection coefficient and VSWR, 60
Taylor distribution resonant and nonresonant, 143
aperture, 203, 205–206, 206 fig., standing-wave patterns, 62
207–209, 211, 213–214, strip, 69, 70 fig., 71
216–218 two-conductor, 69, 76
optimum illumination, 238 two-wire, 47, 48 fig.
tables, 274, 278 waveguides as, 47–48, 71–90
TE. see transverse-electric transmit and receive (T/R) modules, 362–
television, frequency allocations for, 4 363, 362 fig. 364
Telstar satellite-communication ground transverse-electric (TE) lines, 75
terminal, at Andover, Maine, transverse-electric-magnetic (TEM) lines,
322–323, 327 75
TEM. see transverse-electric-magnetic transverse-magnetic (TM) lines, 75
thinning, 210. see also tapering trapezoidal tooth antenna, 297, 298 fig.,
tin-hat lens, 273 305
TM. see transverse-magnetic “trapping,” 35
top loading traveling wave antennas, 136
procedure, 131 traveling waves, 136
and radiation resistance, 315 triangular dipole, 294
short-monopole antenna, 131 fig. triply folded horn reflector antenna, 324
towers, 88–91 Tropospheric scatter propagation, 34
T/R. see transmit/receive tuning inductors, and radiation resistance,
transionospheric propagation, 41 315
transmission equation, 42 tuning screws, reactance introduced into
transmission line equations waveguides by, 77
attenuation constant, 64 tuning stub, 66
for determination of load impedance, turnstile antenna, 308
63
special line conditions, 55 UHF. see ultra high frequency
steady-state lossless line equations, 54 ultra high frequency (UHF)
transmission lines absorption in, 13
basic concepts, 49–50 characteristics of, 2
in broad-band array design, 287 frequency allocations for, 4
coupling devices, 77, 80 ultra low frequency (ULF), characteristics
and feed methods, 158–164 of, 3 table
forms of, 69, 71, 70 fig., 75 fig. ultra side lobe array, 367 fig.
impedance matching and power uniform linear arrays, 182–183
division, 65–71 endfire arrays, 186–188
input impedance of, 59 radiation pattern for, 214
Index 501
vacuum waveguide horn radiator, 151

electromagnetic propagation in, 2 waveguides
electromagnetic waves in, 1 circular guides, 78
V antenna, 218, 219 fig. coaxial transmission lines as, 80
variable-index-of-refraction lenses, cutoff frequency for, 74–76
261 forms of, 78, 87
velocity impedance in, 76
in transmission lines, 72, 76 impedance matching in, 76
versus refractive index, 16 junctions, 78, 80 fig.
vertical-tower radiators, 138 modes of propagation for, 75–76
very high frequency (VHF) phase and group velocities for, 72
absorption in, 13 rectangular, 76, 75 fig.
characteristics of, 2 transmission-line-to-waveguide
very low frequency (VLF) coupling, 77
characteristics of, 2 wavelength in, 73
wave-guide-mode propagation, 33 wavelength
VHF. see very high frequency antenna size and, 88
VLF. see very low frequency calculation of, 5
voltage defined, 5
and electromagnetic fields, 1 free-space, 3
in long antennas, 139, 140 fig. wave nomenclature, 40
voltage equation, 54 wave polarization, elliptical, 433. see also
voltage maximum, 61–62 polarization
voltage standing-wave ratio (VSWR), 67 waves. see electromagnetic waves; radio
on impedance charts, 404 waves
and input impedance, 161 wave velocity, in transmission lines, 72
measurement of, 413 weather protection, 90
volume arrays wind forces, and antenna construction,
formation of, 208 255
two-element-deep, 193 wind resistance, of superturnstiles, 292
VSWR. see voltage standing-wave ratio wire, antenna, 90, 330
wireless communications, original, 96
wall cross sections, radome, 463, 464 fig. “wireless,” first use of, 33
water sources, electrical properties of, wire-trapezoidal-tooth antenna, 305
447 table wire triangular tooth antenna, 299,
“wave” antenna, 143 298 fig.
wavefronts
characteristics of, 2 Yagi-Uda antenna, 159
defined, 5 Yagi-Uda array, 147–148
in geometric optics, 233 cross-polarized, 305
in isotropic propagation medium, 10 folded dipoles for, 291
point-source, 9 fig.
in propagation medium, 7 “Zepp” feeder, 161
spherical, 8 zigzag antenna, 159
About the Author
Maurice Long’s interest in antennas and propagation began as a teenager with his amateur
radio station W4GPR. Shortly after World War II, he began research work at Georgia
Institute of Technology on microwave propagation, and that was followed by antenna
developments for microwave and millimeter radar. Activities in multipolarization and
rapid scan antennas lead to his being General Chairman of the Department of Defense
sponsored 1956 Georgia Tech/SCEL Symposium on Scanning Antennas. Subsequently,
much of his work at Georgia Tech focused on antennas, electromagnetic scattering, and
microwave and millimeter radar. Later, as a consultant to Lockheed Martin and other
companies, he worked in radar design and analysis, including the use of large airborne
mechanical and electronic scanning antennas.
At Georgia Tech he held research and academic positions, including Principal Research
Engineer, Professor of Electrical Engineering, Associate Graduate Dean for Research,
and Director of the Engineering Experiment Station (now Georgia Tech Research Insti-
tute). Presently he teaches graduate courses part time at Southern Polytechnic State
University in antenna design and radar systems, works as a radar consultant, and main-
tains an affiliation as a retiree with the Georgia Tech Research Institute.
Dr. Long’s previous books include Radar Reflectivity of Land and Sea, 3rd ed., Artech
House and Airborne Early Warning System Concepts, SciTech Publishing. He is a Life
Fellow of the Institute of Electrical and Electronics Engineers; and member of Academy
of Electromagnetics and Commission F of International Union of Radio Science.
Additional personal information is available in Who’s Who in America, Who’s Who in
Engineering, American Men and Women in Science, and McGraw-Hill Leaders in
Electronics.
503

Coll. - Antennas Fundamentals, Design, Measurement, 3rd Edn (Standard) PDF

Uploaded by

Copyright:

Available Formats

Coll. - Antennas Fundamentals, Design, Measurement, 3rd Edn (Standard) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Coll. - Antennas Fundamentals, Design, Measurement, 3rd Edn (Standard) PDF

Uploaded by

Copyright:

Available Formats

ANTENNAS: FUNDAMENTALS,

SciTech Publishing, Inc

Copyright © 2009 by SciTech Publishing

Editor: Dudley R. Kay

Standard ISBN: 9781891121784

LIBRARY OF CONGRESS Cataloging-in-Publication Data

Printed in the United States of America.

Chapter 2 Transmission Lines 47

Chapter 3 Antenna Parameters 87

3.2 Radiation Pattern 91

Chapter 4 Basic Radiators and Feed Methods 119

4.4 Long-Wire Antennas 138

Chapter 5 Arrays 169

5.5.3 Broadside Linear Array Patterns 197

Chapter 6 Reflectors and Lenses 227

6.3.2 Taper and Side Lobes for Elliptical Apertures 237

Chapter 7 Antennas with Special Properties 285

Chapter 8 Electronically Steered Arrays 337

Chapter 9 Antenna Measurements 371

9.4.2 Cylindrical Near-Field Measurements 387

Answers to Problems 423

B.3 Relationships between Linearly and Circularly

SM 2.0 Spherical Earth Geometry

SM 3.0 Earth Effects on Patterns and Multipath Propagation

SM 4.0 Antenna Radiation Analyses

SM 5.0 Three-Dimensional Pattern Construction, by Aaron Loggins

This is a senior undergraduate or first-year graduate level textbook on antenna funda-

Evolution of the Third Edition

Chapter 1. Electromagnetic Waves

Appendix A. Maxwell’s Equations

It is to be noted that Appendix C discusses complex-variable algebra. Although its con-

Data files of computer scripts

– reflection coefficients for surfaces versus dielectric properties, conductivity, polari-

1.1. Characteristics of Electromagnetic Waves

1.1.1. Wave Velocity

1.1.2. Frequency and Wavelength

TABLE 1-1 The Radio Frequency Spectrum

Designation Frequency Free-Space Wavelength

1.1.3. Space–Time Relationships

TABLE 1-2 Standard Radar Band Designations

Designation Frequency Free-Space Wavelength

1.1.5. Rays and Wavefronts

A wavefront is a surface of constant phase of the wave. As mentioned in connection

1.1.6. Spherical Waves and the Inverse-Square Law

Wavefront A Cubic region

1.1.7. Field Intensity and Power Density

p = (0.707 E0 ) × (0.707 H 0 ) = 0.5E0 H 0 (1–7)

1.1.8. Decibel (Logarithmic) Expression of Attenuation

In terms of decibels the attenuation is

radiation in undesired directions, or to prevent “leakage” of waves from one part of an

AdB = 8.686ANp (1–12)

Nowadays attenuation is commonly expressed in terms of decibels, not nepers.

The evaluation of the variation of g in the atmosphere is a complicated problem. Curves

1.2. Radio-Wave Optical Principles

Original ray direction It is assumed that the wave speed v2 in

In an electrically nonconducting medium, the wave velocity is given by

A Original boundary surface, and when curved wave-

one refracted (transmitted through the Incident Reflected

tion q3 is exactly equal to the angle of FIGURE 1–5.