Slotted Waveguide Array

Antennas
Mario Boella Series on Electromagnetism in Information
and Communication

Piergiorgio L.E. Uslenghi, PhD – Series Editor

The Mario Boella series offers textbooks and monographs in all areas of radio science, with a
special emphasis on the applications of electromagnetism to information and communication
technologies. The series is scientifically and financially sponsored by the Istituto Superiore
Mario Boella affiliated with the Politecnico di Torino, Italy, and is scientifically co-sponsored
by the International Union of Radio Science (URSI). It is named to honour the memory of
Professor Mario Boella of the Politecnico di Torino, who was a pioneer in the development
of electronics and telecommunications in Italy for half a century, and a vice president of URSI
from 1966 to 1969.

Published Titles in the Series

Fundamentals of Wave Phenomena, 2nd Edition
by Akira Hirose and Karl Lonngren (2010)
Scattering of Waves by Wedges and Cones with Impedance Boundary Conditions
by Mikhail Lyalinov and Ning Yan Zhu (2012)
Complex Space Source Theory of Spatially Localized Electromagnetic Waves
by S.R. Seshadri (2013)
The Wiener–Hopf Method in Electromagnetics
by Vito Daniele and Rodolfo Zich (2014)
Higher-order Techniques in Computational Electromagnetics
by Roberto Graglia and Andrew Peterson (2015)
Slotted Waveguide Array
Antennas
Theory, analysis and design

Lars Josefsson and Sembiam R. Rengarajan

The Institution of Engineering and Technology

Published by SciTech Publishing, an imprint of The Institution of Engineering and
Technology, London, United Kingdom

The Institution of Engineering and Technology is registered as a Charity in England &

Wales (no. 211014) and Scotland (no. SC038698).
† The Institution of Engineering and Technology 2018

First published 2018

This publication is copyright under the Berne Convention and the Universal Copyright
Convention. All rights reserved. Apart from any fair dealing for the purposes of research
or private study, or criticism or review, as permitted under the Copyright, Designs and
Patents Act 1988, this publication may be reproduced, stored or transmitted, in any
form or by any means, only with the prior permission in writing of the publishers, or in
the case of reprographic reproduction in accordance with the terms of licences issued
by the Copyright Licensing Agency. Enquiries concerning reproduction outside those
terms should be sent to the publisher at the undermentioned address:
The Institution of Engineering and Technology
Michael Faraday House
Six Hills Way, Stevenage
Herts, SG1 2AY, United Kingdom
www.theiet.org
While the authors and publisher believe that the information and guidance given in this
work are correct, all parties must rely upon their own skill and judgement when making
use of them. Neither the authors nor publisher assumes any liability to anyone for any
loss or damage caused by any error or omission in the work, whether such an error or
omission is the result of negligence or any other cause. Any and all such liability is
disclaimed.
The moral rights of the authors to be identified as authors of this work have been
asserted by them in accordance with the Copyright, Designs and Patents Act 1988.

British Library Cataloguing in Publication Data

A catalogue record for this product is available from the British Library

ISBN 978-1-61353-189-1 (hardback)

ISBN 978-1-61353-190-7 (PDF)

Typeset in India by MPS Limited

Printed in the UK by CPI Group (UK) Ltd, Croydon
 Contents

About the authors xi

Abbreviations and acronyms xiii
List of symbols used xv
Foreword xvii
Preface xix
Acknowledgements xxiii

1 Introduction 1
References 3

2 Review of electromagnetic theory 5

2.1 Maxwell’s equations 5
2.2 Boundary conditions 6
2.3 Energy and power 7
2.4 Reciprocity theorem 8
2.4.1 Source-free region 8
2.4.2 Infinite region or a region bounded by perfect conductors 9
2.5 Vector and scalar potentials 9
2.5.1 Electric vector potential 10
2.6 The image principle 11
2.7 The field equivalence principle 12
2.8 Green’s functions 13
2.8.1 Free-space Green’s function 13
2.8.2 Dyadic Green’s function for the magnetic field of
a magnetic current in a rectangular waveguide 14
References 15

3 History 17
3.1 The early years 17
3.2 The golden years 17
3.3 Waveguide slot antennas 18
3.4 The many shapes of slotted waveguide array antennas 18
References 23

4 The slot antenna 25

4.1 An aperture in an infinite ground plane 25
4.1.1 Problem formulation 25
4.1.2 Far field 26
vi Slotted waveguide array antennas: theory, analysis and design

4.1.3 General field expressions 27

4.1.4 Radiated complex power 29
4.1.5 The aperture admittance 29
4.2 The rectangular slot antenna 30
4.3 Waveguide modes 33
4.3.1 Definitions 33
4.3.2 The rectangular waveguide 34
4.3.3 The TE10 mode case 36
4.4 The longitudinal slot in a waveguide wall 36
4.4.1 Preliminaries 36
4.4.2 The longitudinal rectangular slot 38
4.4.3 Rectangular waveguide and TE10 mode scattering 39
4.4.4 The equivalent slot conductance 40
4.5 Mutual coupling 43
4.5.1 Introduction 43
4.5.2 Mutual coupling calculations 44
4.5.3 Finite ground plane effects 55
References 57

5 Slot models 59
5.1 Modelling principles 59
5.1.1 Using experimental data 59
5.1.2 Equivalent circuit approach 60
5.1.3 Electromagnetic models 62
5.1.4 Finite wall thickness 66
5.2 Integral equation solution 68
5.2.1 The internal field 68
5.2.2 The external field 70
5.2.3 Matrix elements 71
5.3 Longitudinal slot characteristics 72
5.3.1 Slot electric field distribution 72
5.3.2 Slot magnetic field distribution 76
5.3.3 Experimental verification 79
5.3.4 Weakly excited slots 82
5.4 Transverse slots 83
5.4.1 Introduction 83
5.4.2 Theoretical analysis 84
5.4.3 Results 86
References 88

6 The linear slotted waveguide array antenna 93

6.1 Introduction 93
6.1.1 Overview of design approaches 95
6.2 The accuracy of the slot array model 97
6.2.1 Slot model 97
 Contents vii

6.2.2 Network model 97

6.2.3 Ground plane 97
6.2.4 Mutual coupling 97
6.3 The Elliott design equations 99
6.3.1 The first design equation 99
6.3.2 The second design equation 101
6.3.3 The iterative design algorithm for standing wave arrays 102
6.4 Mutual coupling 104
6.4.1 Efficient computation of the external mutual
coupling terms 104
6.4.2 The self-admittance 106
6.4.3 Internal TE20 mode coupling 109
6.4.4 Full wave analysis 111
6.4.5 Mutual coupling compensation 111
6.5 Design example of a standing wave array 112
6.5.1 Slot data needed for the design algorithm 112
6.5.2 Design of a
30 dB sidelobe-level Dolph–Chebyshev array 113
6.6 Design of a travelling wave array 116
6.6.1 Design equations for travelling wave array 118
6.6.2 Design algorithm 118
6.6.3 Example of a 21-element travelling wave array 119
6.7 Array design using the concept of incremental conductance 126
6.8 Array design in terms of scattering parameters 128
6.9 Improvement to Elliott’s design procedure 130
References 132

7 Design of planar slotted waveguide array antennas 135

7.1 Introduction 135
7.2 Elliott’s design procedure for planar standing wave arrays 136
7.2.1 Design algorithm 139
7.2.2 Example of a 6  6 planar array 141
7.3 Design of large arrays using the infinite array mutual
coupling model 145
7.3.1 Infinite array mutual coupling expression 146
7.4 Choice of total slot conductance in radiating waveguides
and resistance in feed waveguides 149
7.5 Sub-array architectures 151
7.5.1 A planar array of 160  160 elements consisting of 10  10
element sub-arrays 151
7.5.2 A slot array with 8  8 radiating elements consisting
of two sub-arrays 151
7.5.3 A slot array with four-element sub-arrays 152
7.6 Examples of planar slot array designs 156
7.6.1 Large array design using the infinite array model 156
7.6.2 An 8  8 slot array for radiometer application 160
viii Slotted waveguide array antennas: theory, analysis and design

7.7 Design of a travelling wave feed for a planar array 162

7.7.1 Design procedure for a travelling wave feed in terms of
scattering wave representation of a coupling slot 164
7.7.2 Computed results 166
7.8 A review of other methods for analysis and design
in the literature 168
7.9 The generalised scattering matrix model for analysis
and design 169
References 172

8 Concepts and models for advanced designs 175

8.1 Coupling slot models 175
8.1.1 Centred-inclined coupling slot 176
8.1.2 Shunt-series coupling slot 179
8.1.3 Transverse/transverse coupling slot 181
8.2 Edge wall slot 182
8.3 Radiating compound broad wall slot 184
8.4 Iris-excited longitudinal slot 186
8.4.1 Elliott’s design procedure for iris-excited slotted
waveguide arrays 189
8.5 Slot arrays in ridge waveguides 190
8.6 Slot arrays with a dielectric cover 191
8.7 Higher-order mode coupling between centred-inclined
coupling slots 193
8.7.1 Theory for the higher-order mode coupling
between centred-inclined coupling slots 195
8.7.2 Procedure to incorporate the higher-order mode
coupling in the design of centred-inclined
coupling slots 196
8.8 Higher-order mode coupling between a centred-inclined
coupling slot and longitudinal radiating slots 197
8.9 Finite ground plane effects 199
8.10 MoM solution to the integral equations of a planar
array and applications 201
8.10.1 Formulation of the coupled integral equations 201
8.10.2 MoM solution 203
8.10.3 Improved design procedure using the MoM solution 204
8.11 Global optimisations for improved designs 209
8.11.1 GA optimisation of a travelling wave linear array 209
8.11.2 GA optimisation of a planar standing wave array
using the MoM analysis 211
8.11.3 Equivalence between a slot with one basis function
and a slot with many basis functions 213
8.11.4 GA optimised results 215
References 217
 Contents ix

9 Antenna systems and special requirements 221

9.1 Phase scanned slotted waveguide arrays 221
9.2 Frequency scanned slotted waveguide arrays 223
9.2.1 Travelling wave linear array 225
9.2.2 Phase reversal travelling array 226
9.2.3 Resonant series feed array 230
9.2.4 Centre-fed travelling array 231
9.3 Dual polarised slotted waveguide arrays 232
9.4 Multiple beam slotted waveguide arrays 237
9.5 Conformal slotted waveguide arrays 239
9.6 Monopulse slotted waveguide arrays 242
9.7 Stripline slotted arrays 245
9.7.1 Introduction 245
9.7.2 Independent excitations 247
9.8 Scattering from slotted waveguide arrays 249
9.8.1 Introduction 249
9.8.2 Scattering by a single slot in a planar ground plane 250
9.8.3 Scattering from an array of slots 252
9.8.4 The feeding network 255
9.8.5 Numerical examples 256
9.8.6 Spatial filters 256
9.9 Second-order lobes 258
9.9.1 Introduction 258
9.9.2 Analysis 258
9.9.3 Reducing the secondary lobes 259
9.10 Omnidirectional slotted waveguide array antennas 266
9.10.1 Slotted rectangular waveguide array 266
9.10.2 Slotted circular waveguide array 266
9.10.3 The pattern ripple problem 269
9.11 On the bandwidth of slotted waveguide array antennas 272
9.11.1 Introduction 272
9.11.2 The slot itself 273
9.11.3 The waveguide-fed longitudinal slot 273
9.11.4 The linear resonant array antenna 275
9.11.5 Overloading 277
9.11.6 Mutual coupling 278
9.11.7 The planar array and feeding arrangements 279
9.11.8 Travelling wave arrays 281
References 282

10 Slot arrays in special waveguide technologies 291

10.1 Slot arrays in circular parallel plate waveguides 291
10.1.1 Circularly polarised radial line slot antenna 291
10.1.2 Slot coupling and the attenuation coefficient
of the propagating wave 296
x Slotted waveguide array antennas: theory, analysis and design

10.1.3 Linearly polarised radial line slot antenna 297

10.1.4 Design of slot coupling coefficients 299
10.1.5 Single-layer radial line slot arrays 301
10.1.6 Radial line antennas for monopulse applications 303
10.1.7 Accurate analysis and design techniques 304
10.2 Slot arrays in rectangular parallel plate waveguides 304
10.3 Slot arrays in substrate integrated waveguides 307
10.4 Slot arrays in gap waveguides 310
References 313

11 Manufacturing aspects 317

11.1 Tolerance theory 317
11.2 Manufacturing methods 322
11.2.1 Metallic materials 322
11.2.2 Carbon fibre-reinforced plastic materials 324
11.2.3 Microfabrication 326
References 328

12 Outlook for the future 331

12.1 Defence systems 331
12.2 Millimetre waves 332
12.3 Spaceborne systems 332
12.4 Communication 333
12.5 Manufacturing 333
12.6 Design tools and software 333
12.6.1 HFSS 334
12.6.2 XFdtd 334
12.6.3 CST Microwave Studio 334
12.6.4 WASP-NET 334
12.6.5 SWANTM 335
References 335

Appendix Standard waveguide frequency bands 339

References 340

Index 341
 About the authors

Lars Josefsson and Sembiam Rengarajan met at the University of California, Los
Angeles (UCLA), USA, where both worked on slotted waveguide array antennas
together with Professor R.S. Elliott, a well-known scientific leader and teacher in
this field. Since then they have kept contact and collaborated on research projects
and courses several times.
Lars Josefsson was born in Norrköping, Sweden. He graduated from The Royal
Institute of Technology in Stockholm (KTH) and got his PhD from Chalmers
University of Technology in Göteborg. A grant from the Marcus Wallenberg
Foundation and a Fulbright Scholarship enabled his stay in 1982–83 as a Visiting
Scientist at UCLA, working on waveguide slotted array antennas. He became
Adjunct Professor in Antenna Technology (1983–86) at Chalmers and obtained the
equivalent position at KTH (1996–2003). He returned to Chalmers again as
Adjunct Professor (2004–07).
For more than 40 years Lars was with Ericsson Microwave Systems AB in
Mölndal, Sweden, where he became responsible for new antenna technology and
systems, internal R&D projects and internal education relating to antennas. In 2004
he received the Thulin Silver Medal from the Swedish Aeronautical Society for his
contributions to several generations of airborne radar antenna developments. He is
fellow of the IEEE since 1999.
In 2003 he formed his own consulting company: Lars Microwave (technical
consulting, research and education). Lars Josefsson is the author and co-author
of more than 100 scientific publications. He is the author (together with Patrik
Persson) of the book Conformal Array Antenna Theory and Design, published by
IEEE/Wiley in 2006.
Sembiam R. Rengarajan received the PhD degree in Electrical Engineering
from the University of New Brunswick, Canada, in 1980. Since then he has been a
faculty member at California State University, Northridge, USA. He has held vis-
iting appointments at Chalmers University of Technology, Sweden; University of
Santiago de Compostela, Spain; University of Pretoria, South Africa; and Technical
University of Denmark. In addition to teaching electromagnetics and antenna
courses, he has performed significant research in topics on applications of elec-
tromagnetics to antennas, scattering, and microwave components. He has also
consulted for many companies and government agencies. He has published more
than 240 journal articles and conference presentations.
Much of Sembiam’s research has dealt with slotted waveguide array antennas.
He collaborated with Prof. Elliott in the 1980s and 1990s for more than a decade.
xii Slotted waveguide array antennas: theory, analysis and design

His research in this area was supported by the University of California, Hughes
Aircraft Company, Lockheed Martin, Saab Ericsson Space, European Space
Agency, Rantec Corporation, DSO Singapore, NASA/Jet Propulsion Laboratory
(JPL), among others. He served as a Distinguished Lecturer for IEEE Antennas and
Propagation Society, lecturing on slotted waveguide array antennas all over the
world. Presently he serves as the Chair of the United States National Committee for
the International Union of Radio Science (USNC-URSI). He has received more
than 20 awards from NASA for his research contributions to JPL. He is a life fellow
of IEEE.
 Abbreviations and acronyms

AF array factor
CEM computational electromagnetics code
CFRP carbon fibre-reinforced plastic
dB decibel
DBF digital beam forming
DOA direction of arrival
EM electromagnetic
FDTD finite difference time domain
FEM finite element method
FSS frequency selective structure
GA genetic algorithm
GHz giga hertz
GO geometrical optics
GSM generalised scattering matrix
GTD geometric theory of diffraction
HFSS high-frequency structural simulator
HTCC high temperature co-fired ceramics
IE integral equation
LEO low earth orbit
LMS least mean square
LTCC low temperature co-fired ceramics
MC mutual coupling
MIMO multiple input multiple output
MIT Massachusetts Institute of Technology
MMIC monolithic microwave integrated circuit
MM mode matching
MoM method of moments
MTI moving target indication
NDT non-destructive testing
xiv Slotted waveguide array antennas: theory, analysis and design

PEC perfect electric conductor

PMC perfect magnetic conductor
PO physical optics
RCS radar cross section
RF radio frequency
SAR synthetic aperture radar
SIW substrate integrated waveguide
SLL sidelobe level
SMI sample matrix inversion
STAP space time adaptive processing
TACAN tactical air navigation
TE transverse electric
TEM transverse electromagnetic
THz terahertz
TM transverse magnetic
TRM transmit/receive module
UCA uniform circular array
ULA uniform linear array
UTD uniform theory of diffraction
Wi-Fi ~WLAN
WLAN wireless local area network
 List of symbols used

Symbol Unit

a width of rectangular waveguide m

A magnetic vector potential Wb/m
b height of rectangular waveguide m
B susceptance Siemens
B magnetic flux density T
d slot spacing in the H-plane m
D electric displacement density Coul/m2
also antenna directivity –
E electric field intensity V/m
f frequency Hz
F electric vector potential Coul/m
gmn mutual coupling integral –
G conductance Siemens
also Green’s function –
also antenna gain –
H magnetic field intensity A/m
J electric current density A/m2
k wavenumber rad/m
k0 wavenumber in free space rad/m
L half length of rectangular slot m
M magnetic current density V/m2
Ms magnetic surface current density V/m
P complex power W
R resistance W
S Poynting vector W/m2
also scattering matrix –
t waveguide wall thickness m
V voltage V
w width of rectangular slot m
X reactance W
Y admittance Siemens
xvi Slotted waveguide array antennas: theory, analysis and design

Symbol Unit

a attenuation constant Neper/m

b phase constant rad/m
g propagation constant m–1
G reflection coefficient
e0 permittivity of free space Farad/m
em Neumann factor, ¼1 for m ¼ 0, –
¼2 for m >0
er relative permittivity –
e permittivity, e ¼ er e0
f azimuth angle in spherical coordinates radians
h0 impedance of free space W
l wavelength m
l0 free space wavelength m
lg guide wavelength m
m0 permeability of free space Henry/m
q tilt angle of a coupling slot radians
also the elevation angle in spherical coordinates radians
s conductivity Siemens/m
ssc scattering cross section m2
w angular frequency rad/s
c2 the coupling coefficient of a coupling slot –
d Dirac delta function –
also slot offset from the centreline –
dij Kronecker delta function –
 Foreword

The Mario Boella series is devoted to textbooks and monographs in all areas of
Radio Science, with a particular emphasis on the applications of electromagnetism
to information and communication technologies. The series is financially spon-
sored by the Istituto Superiore Mario Boella (ISMB), a research entity affiliated
with the Politecnico di Torino, Italy, and is scientifically co-sponsored by the
International Union of Radio Science (URSI). It is named to honour the memory of
Professor Mario Boella of the Politecnico di Torino, who was a pioneer in intro-
ducing electronics and telecommunications into higher education in Italy for half a
century and was a vice-president of URSI from 1966 to 1969.
This research monograph on slotted waveguide antennas is the eighth volume
in the ISMB series. It contains a very comprehensive coverage on a subject that is
of practical importance in antenna and radar systems for a variety of civilian and
military applications. Both theory and design, as well as manufacturing aspects and
future applications, are discussed in depth. The authors are well-known authorities
in the field. Dr Josefsson has decades of experience as a design and project engi-
neer, and has been honoured for his technical achievements. Professor Rengarajan
has taught and done research on antennas and radars for many years; he is highly
regarded in the electromagnetics community for his original contributions. The two
authors have combined their complementary knowledge to produce a monograph
that will be very useful to students, instructors and practitioners.

Piergiorgio L. E. Uslenghi
ISMB Series Editor
Chicago, April 2018
 Preface

The purpose of this book is to present in one volume a comprehensive treatment of

slotted waveguide array antennas from an engineering perspective. This includes
history, concepts, design principles, current state of the art and a discussion about
future applications. The book is addressed to engineers, physicists and students
interested in slotted waveguide array technology. It is written on an intermediate
level. It may serve as a supplementary text for graduate courses in antennas and
applied electromagnetic theory.
Published papers on slotted waveguide array antennas are abundant, starting in
the 1940s. Writing this book has meant a thorough study of the available literature
and it has required a careful analysis of the major contributions to the field. A simple
search on IEEE Xplore gave 2059 publication hits matching the key word ‘slotted
waveguide array antenna’ for the period 1946 to February 2016. Most striking is a
continued increase in the number of publications per year; see the figure below. This
number has an exponential growth. While a detailed analysis and interpretation of
this statistic might not be useful, it is clear that the subject remains important and is
of continued interest for modern applications.

No of hits in IEEE Xplore

2500

‘slotted waveguide array antenna’

2000
No of this: 1946 to year

1500

1000

500

0
1980 1990 2000 2010 2020
Year

Number of hits in IEEE Xplore from 1946 to ‘year’ for the key word ‘slotted
waveguide array antenna’
xx Slotted waveguide array antennas: theory, analysis and design

How do other antenna types such as microstrip patch array antennas and
reflector antennas compare with slotted waveguide array antennas? Without going
into details we try a simple comparison:
Antenna type Characteristics
Microstrip patch arrays
Transmission losses, surface waves
Reflector antennas
Fixed beam, require depth
Slotted waveguide arrays þ Low losses, high precision
There are of course both pros and cons for all types, depending on the appli-
cation. Several books discuss microstrip antennas and/or reflector antennas. The
present book is the first totally devoted to slotted waveguide array antennas.
With specific examples of waveguide array designs, accompanied by detailed
illustrations and antenna characteristics, the book is a must-have reference for
engineers involved in the design and development of slotted waveguide array
antennas. An additional value is provided by a thorough overview of the related
literature in the field and the history of these antennas given in several chapters.
With this book system engineers and design and development engineers in industry
and government will have a single source with a coherent treatment instead of
turning to numerous papers in the literature.
The book goes well beyond some of the commonly discussed topics on slotted
waveguide array design, into areas that include
● higher-order mode coupling and edge effects
● performance optimisation in terms of bandwidth and pattern performance
● special slot excitation methods
● applications such as monopulse and phase steering
● manufacturing techniques and tolerances, etc.
The book is organised as follows:
The first few pages contain a list of abbreviations and acronyms that are
common when dealing with slotted waveguide array antennas and related appli-
cations. There is also a list of symbols with definitions, representing mathematical
and physical quantities that appear in the text. Finally, a table providing definitions
of waveguide bands and designations, with corresponding rectangular waveguide
inner dimensions is included. The following chapters all start with some intro-
ductory material such as the basic concepts that are essential to get an under-
standing of the more advanced concepts to follow.
Chapter 1, ‘Introduction’, is a short overview of the radiating mechanism of
slots in waveguides carrying the fundamental TE10 mode. There is also a short
discussion of equivalent circuits for various types of resonant slots: longitudinal,
transverse and inclined slots.
Chapter 2, ‘Review of electromagnetic theory, starts from Maxwell’s equa-
tions. Important concepts such as boundary conditions, reciprocity theorems, field
equivalence and Green’s functions and other definitions are included. The purpose
is to serve as a reference to be consulted if the need arises when reading the
theoretical chapters that follow.
 Preface xxi

Chapter 3, ‘History’, is an account of the history of early slotted waveguide

array antenna developments from the Second World War and onwards. A few
descriptions of more recent designs are also presented, demonstrating the wide
range of applications of the technology (radar, remote sensing, communication,
etc.).
Chapter 4, ‘The slot antenna’, starts with an analysis of a general aperture in an
infinite ground plane, the near and far fields, and the aperture admittance. It con-
tinues with resonant slots and their radiation conductance. General expressions
for waveguide modes are derived, to be followed by modes in rectangular
waveguides. The case with longitudinal slots is analysed in more detail including
an analysis of mutual coupling among pairs of rectangular slots in the E-plane and
in the H-plane. The array coupling is also calculated, that is, the coupling between
two slots in the presence of other slots in the array. The impact on coupling of edge
diffraction effects in finite ground planes is also analysed.
Chapter 5, ‘Slot Models’, theoretical models for characterising slots in wave-
guides are introduced. An integral equation formulation is presented and expres-
sions for a solution by the Method of Moments are presented. Computed results for
longitudinal and transverse slots are discussed, including the effect of waveguide
wall thickness, frequency dependence, waveguide dimensions, etc.
Chapter 6, ‘The linear slotted waveguide array antenna’. In this chapter algo-
rithms for designing a linear array of slots in a single waveguide are derived. The
influence of mutual coupling is included and typical results are presented. Some
examples of the design of standing wave and travelling wave arrays are discussed.
Chapter 7, ‘Design of planar slotted waveguide array antennas’. Here several
slotted waveguides form a planar array. Mutual coupling plays a large role and a
coupling network is added to feed the waveguides. Examples of planar array design
are presented. Other design methods in the literature are discussed.
Chapter 8, ‘Concepts and models for advanced designs’. The influence of
higher-order modes is treated in more detail, including coupling slots. Designs with
edge wall slots, compound slots, iris-excited longitudinal slots, slots in ridge
waveguides and slots with dielectric cover are treated as well as optimisation
methods.
Chapter 9, ‘Antenna systems and special requirements’. This chapter treats
system aspects such as slotted array antennas with phase and/or frequency beam
steering, monopulse for angular tracking, conformal arrays and polarisation con-
trol. Bandwidth limitations are also addressed, as well as omnidirectional antennas,
spurious lobe generation and elimination, and scattering cross-sectional analysis. A
short section on slotted arrays using stripline technology is also included. The
chapter can be read separately as an introduction into the many system applications
of slotted waveguide array antennas. Several references to the literature on system
applications are included.
Chapter 10, ‘Slot arrays in special waveguide technologies’. In this chapter we
discuss slot arrays in different waveguide technologies such as two types of parallel
plate waveguides propagating the TEM mode, the substrate integrated waveguides
(SIW) and the gap waveguide propagating quasi TEM mode. In the circular parallel
xxii Slotted waveguide array antennas: theory, analysis and design

plate waveguide an electromagnetic wave propagates in the radial direction inward

or outward, while slots cut in one of the parallel plates radiate. Such an array is
convenient for large apertures requiring high gain and mass production. The second
type of parallel plate waveguide has a rectangular shape propagating TEM mode
waveguide with a planar wave front. Slot arrays in SIW have great potential for
easy integration with monolithic microwave integrated circuit devices. Finally slot
arrays in gap waveguides, exhibiting wider bandwidth at millimetre-wave
frequencies and beyond, are presented.
Chapter 11, ‘Manufacturing aspects’. This is a wide subject so we concentrate
on critical parameters such as electrical and mechanical tolerances that relate to
electrical performance. We also discuss joining methods for metal materials, for
example, dip brazing of aluminium. Applications at high frequencies (millimetre
waves) in particular call for very high precision and special methods. An important
area is the use of slotted carbon fibre-reinforced plastic waveguides that offer light
weight and thermal stability, typically required for large antenna systems for space
applications. The technology is also used in some ground and airborne radar sys-
tems. The exciting developments in low-cost fabrication using plastic materials and
metallisation techniques are discussed. This area is related to the microelectronic
area, particularly for high frequencies and highly integrated antenna/microwave
assemblies.
Chapter 12, ‘Outlook for the future’. In this last chapter we try to summarise
the state of the art and discuss expected new applications of slotted waveguide
array antennas. Important areas include integration of antenna and circuit tech-
nology, the trend towards higher frequencies, spaceborne systems, design tools, the
impact of advanced software, new materials, cost reduction and many more.
However, in this final overview it is not possible to mention all the details of the
evolving field; the reader is referred to the respective chapters and the references
for more information.
Extensive literature references are included in each chapter. A Subject Index is
included at the back.
 Acknowledgements

Much pioneering work in the field of slotted waveguide array antennas was done
at the Radiation Laboratory at the Massachusetts Institute of Technology (MIT)
during the years 1940–45. The Radiation Laboratory Series of publications
(28 volumes) remain a valuable source of reference even today.
The book draws on material from several short courses on ‘Slotted Waveguide
Array Antenna Technology’ by the authors. One such course was presented toge-
ther with Dr M Ando and Dr J Hirokawa from the Tokyo Institute of Technology.
Numerous colleagues and friends have supported our work with their advice,
encouragement and contributions. In particular we would like to mention Dr Hans
Steyskal and Dr Bob Mailloux for their valuable comments.
We also remember with gratitude Professor Bob Elliott of UCLA, a mentor
and friend, who introduced us to many exciting aspects of slotted waveguide arrays.
We are grateful for the permission to use results from several studies on slotted
waveguide arrays. This includes Ericsson AB, Saab AB, RUAG Space AB, in
Sweden and many more.
One of us (SR) would like to acknowledge his research sponsors including
Hughes Aircraft, University of California, Saab Ericsson Space, JPL, Rantec,
Lockheed Martin, DSO Singapore and CSUN Foundation, and his many students
and colleagues who have contributed to his work.
Last but not least, we express our appreciation and gratitude to our families for
their encouragement, understanding and patience during the writing of this book.
 Chapter 1
Introduction

The first successful slotted waveguide array antennas were developed in Canada
during the Second World War. The immediate applications were in military ground
and airborne radar systems for target detection and tracking. Later applications
include remote sensing from aircraft and space vehicles and microwave commu-
nication links. Spaceborne synthetic aperture radar (SAR) with slotted waveguide
arrays are used for weather forecasting, environmental monitoring, climate change
studies, etc. The slotted array antenna is also considered in automobile collision
avoidance systems.
For the interested reader not too familiar with slotted waveguide array anten-
nas a few simple concepts are introduced in the following text.
Slots in rectangular waveguides are the most commonly used. The wave-
guide can be standard dimension, or half height or even quarter height, in order to
save space and weight. Many other realisations are possible, however, as will
be discussed in later chapters. The fields inside the waveguide are related to the
currents in the waveguide walls. For the fundamental TE10 mode we have the
current distribution in the waveguide wall as shown in Figure 1.1. Typical slot
positions are illustrated in Figure 1.2. By comparing the two figures we realise
that slot positions c and e are not very useful since these slots do not interrupt any
wall current and hence do not couple to the waveguide mode. Slots a (transverse

Figure 1.1 Surface currents on the waveguide wall for the TE10 mode in a
rectangular waveguide
2 Slotted waveguide array antennas: theory, analysis and design

d
b

q
c a

f
g e

q
Figure 1.2 Possible slot positions in a rectangular waveguide.
Adapted from [1]

broad wall slot) and g, on the other hand, couple strongly, and slot f (inclined
sidewall slot) couples to a degree depending on the inclination angle q. Slot b
(longitudinal broad wall slot) also couples strongly when displaced far from the
waveguide centreline.
The rectangular slot in a large conducting ground plane radiates approximately
as a dipole of same length and width. Thus, the slot in a waveguide has a broad
antenna pattern approximately equal to the corresponding dipole, only that E- and
H-planes are interchanged. The corresponding antenna impedances differ con-
siderably, however (Babinet’s principle) [2, p. 336].
A. F. Stevenson at the National Research Council in Canada [3] found
suitable equivalent networks for the most common slots (at resonance) as shown in
Figure 1.3.
A more detailed analysis of waveguide modes and slot characteristics will be
presented in subsequent chapters. Also array design principles and applications will
be discussed.
New fabrication techniques for slotted arrays such as electroplating on
dielectric materials have demonstrated good performance at frequencies as high
as 100 GHz. Thus, highly directive antennas with small dimensions can be
made, and a very large signal bandwidth becomes possible. Applications at such
high frequencies are mainly for short ranges: covert communication, imaging
radar, collision avoidance, planetary landing, etc. An important step forward is
the possibility to integrate active microelectronics and circuits with the antenna
structure.
Important progress over the recent years has also been made in the design and
optimisation of slotted waveguide array antennas. It is now possible to fabricate
planar arrays in a cost-effective and simplified way in several layers using new
techniques; there are new ways to match and optimise the electrical design,
increase bandwidth, etc.
 Introduction 3

a x1 Longitudinal slot in the broad wall

2 g + jb Equivalent shunt element
W

a W r + jx Transverse slot in the broad wall

2 Equivalent series element

x1

a q Inclined slot in the broad wall

r + jx
2 Equivalent series element

q
Inclined slot in the narrow wall
b g + jb
Equivalent shunt element

Figure 1.3 Approximate equivalent networks for resonant slots.

Adapted from [1]

References
[1] Silver S. ed. Microwave Antenna Theory and Design, MIT Rad. Lab. Series,
Vol. 12, McGraw-Hill, New York, 1949.
[2] Elliott R. S. Antenna Theory and Design. Prentice-Hall, Englewood Cliffs,
NJ, 1981.
[3] Stevenson A. F. Series of Slots in Rectangular Waveguides, National
Research Council of Canada, Radio Reports 12 and 13, 1944.
 Chapter 2
Review of electromagnetic theory

In this chapter, we review the electromagnetic theory and concepts used in later
chapters. We will start with Maxwell’s equations in time harmonic form, followed
by boundary conditions. Expressions for energy and power are derived. We then
discuss the reciprocity theorem. Vector and scalar potentials are derived. Image
principle and the field equivalence principle are presented. Green’s functions are
discussed with a presentation of dyadic Green’s functions for the magnetic current
in a waveguide.

2.1 Maxwell’s equations

Steady-state electromagnetic fields under sinusoidal excitations are expressed in phasor

form. ffiffiffiffiffiffiffi dependence in the form of exp(j wt), where w is the angular frequency and
pTime
j ¼ 1; is assumed and suppressed in Maxwell’s equations (2.1)–(2.4).

r~ ~
E ¼ jwmH (2.1)
~ ¼ jwe~
rH E þ~
J (2.2)
r~
D ¼ rv (2.3)
r~
B¼0 (2.4)
In these equations ~ ~, D
E, H ~ and ~B are the electric field, magnetic field, electric
displacement density and magnetic flux density, respectively. e and m are the per-
mittivity and permeability, respectively. In lossy media e ¼ e0  je00 and m ¼ m0  jm00 .
The sources ~ J and rv are the electric current density and volume charge density,
respectively. The current density term ~ J may consist of both induced currents and
impressed currents.
The constitutive relations are given next, where s is the conductivity of the
medium.
~ ¼ e~
D E (2.5)
~ ~
B ¼ mH (2.6)
~
J ¼ s~
E (2.7)
6 Slotted waveguide array antennas: theory, analysis and design

In lossy dielectric media; s ¼ we00 (2.8)

where e00 is the negative imaginary part of the permittivity.
We will be working mostly with linear, homogeneous and isotropic media in
this book. So the constitutive parameters are scalar quantities and may depend on
frequency. In some equivalent problems there may be an impressed magnetic
~ i , on the right side of (2.1). Such a magnetic current density
current density term, M
results from an impressed electric field, ~ E i , given by (2.9).
~ i ¼ r  ~
M Ei (2.9)

2.2 Boundary conditions

Let us consider two regions labelled 1 and 2 with n^ , a unit vector at the boundary
surface directed towards region 1 as shown in Figure 2.1. At the boundary surface
the tangential components should satisfy the conditions given by (2.10) and (2.11),
where the subscripts 1 and 2 refer to media 1 and 2, respectively.
 
^
n H ~2 ¼ ~
~1  H Js (2.10)
 
^n ~ E1  ~
E2 ¼ M~s (2.11)
where ~ J s and M ~ s are the electric and magnetic surface current densities, respec-
tively, at the boundary. The magnetic surface current density term on the right of
(2.11) is non-zero only in aperture problems where equivalent magnetic surface
currents exist. The boundary conditions for the normal components satisfy (2.12) and
(2.13), where rs is the electric surface charge density.
 
^
n D ~ 2 ¼ rs
~1  D (2.12)
 
n ~
^ B1  ~
B2 ¼ 0 (2.13)
Special cases
If medium 2 is a perfect electric conductor (PEC),

^ ~1 ¼ ~
nH Js (2.14)

1 n^

2
Figure 2.1 Two regions with a boundary surface separating them
 Review of electromagnetic theory 7

and

^ ~ 1 ¼ rs
nD (2.15)
If medium 2 is a perfect magnetic conductor (PMC), introduced in some
equivalent field problems,

n ~
^ ~s
E1 ¼ M (2.16)
and

n ~
^ B 1 ¼ rms (2.17)
where rms is the equivalent surface magnetic charge.

2.3 Energy and power

Maxwell’s equations (2.1) and (2.2) are rewritten with the impressed magnetic
current density M ~ i , and the electric conduction and impressed current density terms
~
J c and ~
J i , respectively, in (2.18) and (2.19).

r~ ~ M
E ¼ jwmH ~i (2.18)
~ ¼ jwe~
rH E þ~
Jc þ~
Ji (2.19)
Let us consider the expression
     
r ~ E H~ ¼ H ~  r  ~
E ~ ~
E  rH (2.20)
where the superscript * denotes the complex conjugate.
Both sides of (2.20) are integrated over a volume v, and the divergence theo-
rem is applied to obtain
  h     i
∯ ~ EH ~   d~s¼∭ H ~  r~ E ~ E  rH ~  dv (2.21)
s v

where s is the surface enclosing the volume v.

The use of (2.18), (2.19), (2.5), (2.6) and (2.7) in (2.21) yields
 
∯ ~ E H~   d~
s ¼ jwm∭ jH ~ j2 dv  jwe∭ j~ Ej2 dv
s vh v i
þ∭ M ~i  H~ þ ~E ~

J i dv þ s∭ j~
Ej2 dv (2.22)
v v

Equation (2.22) is a statement of conservation in terms of complex power.

It may be expressed as
S ¼ j 2wðWm  We Þ þ Pd  Ps (2.23)
where S is the complex power leaving the volume, Wm is the magnetic energy
stored in the volume, We is the electric energy stored in the volume, Pd is the
8 Slotted waveguide array antennas: theory, analysis and design

dissipated power and Ps is the complex power delivered by the sources. If m and e
are complex, additional dissipation terms will be present.

2.4 Reciprocity theorem

Consider a region in space containing two sets of sources. Each set of fields is
related to its sources via Maxwell’s equations as given next.

r~ ~a  M
E a ¼ jwmH ~a (2.24)
~ a ¼ jwe~
rH Ea þ ~
Ja (2.25)
r~ ~b  M
E b ¼ jwmH ~b (2.26)
~ b ¼ jwe~
rH Eb þ ~
Jb (2.27)
Using vector identities, (2.28) and (2.29) are obtained.
     
r  ~ ~b ¼ ~
Ea  H ~b  H
Ea  r  H ~b  r  ~
Ea

¼ jwe~
Ea  ~
Eb þ ~
Ea  ~ ~a  H
J b þ jwmH ~b þ H
~b  M
~a (2.28)
     
r  ~ ~a ¼ ~
Eb  H ~a  H
Eb  r  H ~a  r  ~
Eb

¼ jwe~
Ea  ~
Eb þ ~
Eb  ~ ~a  H
J a þ jwmH ~b þ H
~a  M
~b (2.29)

Subtract (2.29) from (2.28) to obtain

 
r  ~ ~b  ~
Ea  H ~a ¼ ~
Eb  H Ea  ~ ~a  M
Jb  H ~b ~
Eb  ~ ~b  M
Ja þ H ~a (2.30)

Integrating (2.30) over a volume containing all sources and by using the divergence
theorem we obtain
   
∯ ~ ~b ~
Ea  H ~ a  d~
Eb  H s¼∭ ~Ea  ~ ~a  M
Jb H ~b ~
Eb ~ ~b  M
Ja þH ~ a dv
s v
(2.31)

Equation (2.31) is the most general form of reciprocity theorem.

2.4.1 Source-free region

If there are no sources in the volume of interest, terms on the right side of (2.31)
are zero.
 
;∯ ~ ~b  ~
Ea  H Eb  H~ a  d~
s¼0 (2.32)
s

Equation (2.32) is known as the Lorentz reciprocity theorem [1].

 Review of electromagnetic theory 9

2.4.2 Infinite region or a region bounded by perfect conductors

If we consider a volume bounded by a surface S at infinity, the integrand on the left
of (2.32) is zero everywhere. This is easy to prove using the Sommerfeld radiation
condition stated in (2.33) for both sets of fields as
 
r ~E  h0 H~  ^r ! 0 as r ! 1 (2.33)
where ^r is a unit vector in the r direction and h0 is the intrinsic impedance of free
space.
Left of (2.31) becomes zero also for the case of S being a perfect conductor.
Then (2.31) becomes
   
∭ ~Ea  ~ ~a  M
Jb  H ~ b dv ¼ ∭ ~Eb  ~ ~b  M
Ja  H ~ a dv (2.34)
v v

Using the reaction concept [2] the above equation may be stated as
ha; bi ¼ hb; ai (2.35)
The term ha,bi is called the reaction of ‘a’ fields with ‘b’ sources. Reciprocity
theorem has been widely applied in microwave circuits, antennas and scattering
problems.

2.5 Vector and scalar potentials

Let us consider Maxwell’s equations (2.1) and (2.2) in a region of space.

r~ ~
E ¼ jwmH (2.36)
~ ¼ jwe~
rH E þ~
J (2.37)
Since r  ~
B ¼ 0 everywhere ~
B can be expressed as a curl of a vector.

Let ~ ~ ¼ r~
B ¼ mH A (2.38)
where ~
A is the magnetic vector potential [3]. From (2.36) and (2.38) we obtain
 
r ~ E þ jw~
A ¼0 (2.39)
Since the curl of a gradient of any scalar function is zero, we express the electric
field in terms of ~
A and an electric scalar potential f, as given next.

E þ jw~
~ A ¼ rf E ¼ jw~
or ~ A  rf (2.40)

By eliminating ~ ~ in (2.37) using (2.38) and (2.40) we obtain

E and H

1 h i
r r~
A ¼ jwe jw~
A  rf þ J (2.41)
m
10 Slotted waveguide array antennas: theory, analysis and design

Assuming that the medium is homogeneous the above equation may be rewritten as

A ¼ k 2~
rr~ A  jwmerf þ mJ (2.42)

The use of the vector identity r  r  ~ A  r2 ~

A ¼ rr  ~ A, and Lorentz gauge
r  A ¼ jwmef, yields the vector Helmholtz equation (2.43) for ~
~ A. The term
2~ ~
r A is the vector Laplacian of A.

r 2~
A þ k 2~
A ¼ m~
J (2.43)

By performing the divergence operation on (2.40) one can derive the scalar wave
equation
r
r2 f þ k 2 f ¼  (2.44)
e
~ from
One can solve (2.43) for the given source distribution and then determine H
(2.38) and E from

r~A
E ¼ jw~
~ Aþr (2.45)
jwme

2.5.1 Electric vector potential

The dual problem of the electric vector potential produced by a magnetic current is
solved in the charge-free region specified by r  D ~ ¼ 0. Maxwell’s equations in
this case are

r~ ~ M
E ¼ jwmH ~ (2.46)

~ ¼ jwe~
rH E (2.47)

The electric vector potential F is given by

~ ¼ e~
D E ¼ r  ~
F
Using duality [1] we obtain

H ¼ jw~
F  rjm (2.48)
where jm is the scalar magnetic potential

r2 ~
F þ k 2~ ~
F ¼ eM (2.49)

and

~ ¼ jw~ r~F
H F þr (2.50)
jwme
 Review of electromagnetic theory 11

2.6 The image principle

The image principle allows one to work with the electric and/or magnetic current
sources above an infinite planar PEC as shown in Figure 2.2(a). In order to create
an equivalent problem for the half space above the PEC, we place image currents
below the PEC and remove the PEC. Each image current and the corresponding
original current have the same strength and shape, and they are equidistant from the
plane of the PEC as shown in Figure 2.2(b). The original currents and the image
currents make the total tangential electric field zero in the plane where the PEC was
located originally, thus satisfying the uniqueness conditions for the electromagnetic
fields for the half space above the PEC plane [1]. Thus the image principle works to
produce the right equivalence for the upper half space.
In certain equivalent problems one uses an infinite planar PMC. If such a PMC
exists in place of PEC in Figure 2.2, all the image currents will be reversed.

J M

M
J

PEC
(a)

J M

J M

J
M
J

M
(b)

Figure 2.2 Application of the image principle: (a) electric and magnetic
current sources above an infinite planar PEC; (b) equivalent
problem for the upper half space
12 Slotted waveguide array antennas: theory, analysis and design

2.7 The field equivalence principle

Let us consider a set of sources inside a volume called region I, bounded by a
surface S shown in Figure 2.3(a). The medium inside the volume has permittivity e
and permeability m. The exterior region called region II is free space with perme-
ability m0 and permittivity e0. The electric and magnetic fields produced by the
sources are ~
E and H~ , respectively. Love’s [4] equivalence principle allows us to
create an equivalent problem that is valid for the region outside the volume, as
shown in Figure 2.3(b). In the equivalent problem, the region I consists of no
sources and null fields. At the boundary surface equivalent electric and magnetic
currents given by (2.51) and (2.52) are set up.
~
Js ¼ ^ ~
nH (2.51)
~ s ¼ ^
M n ~
E (2.52)
The unit vector ^n is normal to the boundary surface S and is directed into region II.
One could fill up region I with any material in Figure 2.3(b) since it consists of
null fields and no sources. Therefore choosing the region I material same as that of
region II makes it easy to compute the fields of the equivalent currents since they
radiate in a homogeneous medium. One could also fill up region I with PEC or
PMC [5]. These choices will short the electric and magnetic surface currents,
respectively. Consequently, the number of unknowns on S will be reduced. How-
ever, the computation of the fields will be substantially complicated in those cases.
A tutorial on the field equivalence principle with examples of planar cases has been
presented by Rengarajan and Rahmat-Samii [6,7].

II e0, m0
E, H M1
J1
I e, m S

(a)

II e0, m0

E, H
E =0=H n^
I e0, m0 S Js1 = n^ × H
Ms1= –n^ × E
(b)

Figure 2.3 (a) Geometry of the original problem. (b) Love’s equivalence for
region II. (a, b) ’ IEEE 2000, reprinted with permission from [6]
 Review of electromagnetic theory 13

2.8 Green’s functions

Green’s functions are solutions to the wave equations subject to boundary condi-
tions, with point source excitations.

2.8.1 Free-space Green’s function

Let us consider the Helmholtz equation given by
 0  0  0
~0 ~
r2 G R; ~ ~0 ~
R þ k2G R; ~
R ¼ ^z d ~R ~
R (2.53)

where r2 is the vector Laplacian operator, G ~ 0 is the free-space vector Green’s

function for the vector potential produced by the z-directed point source located at
0 0
R :~
~ R and ~
R are position vectors of the field point and the source point, respec-
0 0
jkj~
R ~
tively. The solution should vanish at R ¼ ?. Thus G ~ 0 ð~
R; ~
R Þ ¼ ^z e
R j
0 [1].
4pj~
R~
Rj
0
The magnetic vector potential produced by an electric current distribution, ~
J ð~
R Þ, is
given by
   0  0
~
A ~R ¼ m∭ ~
J ~
R G0 ~R; ~
R dv0 (2.54)
V

It is noted that the vector sign in the Green’s function is left out since the direction
of the vector potential is the same as that of the current density vector. The volume
integral is carried out over the entire region of the current distribution. Alter-
natively we can express the Green’s function as a dyad, noting that only the
diagonal terms in the dyad will contribute to the free space dyadic Green’s func-
tion, that is,
0
~ ~
 0 ejkjRR j
G0 ~R; ~
R ¼ 0 I (2.55)
4pj~
R ~ Rj
satisfies
 0  0  0
r2 G 0 ~R; ~
R þ k2G0 ~ R ¼ I d ~
R; ~ R ~
R (2.56)

where the unit dyad is given by

2 3
^x ^x 0 0
4
I ¼ 0 ^y^y 0 5 (2.57)
0 0 ^z^z
Then (2.54) may be written in terms of the dyadic Green’s function as
   0  0 0
~
A ~
R ¼ m∭ G 0 ~ R ~
R; ~ J ~
R dv (2.58)
V
14 Slotted waveguide array antennas: theory, analysis and design

2.8.2 Dyadic Green’s function for the magnetic field of a

magnetic current in a rectangular waveguide
The field scattered by a slot cut in a rectangular waveguide or coupling between
two waveguides by a slot may be determined with the knowledge of the Green’s
functions in the rectangular waveguide. The slot aperture is shorted out and a
magnetic equivalent current equal to M ~ ¼~ E  ^n is placed in the location of the
~
slot. Here E is the electric field in the slot aperture and ^n is a unit normal vector
directed into the waveguide. The magnetic field produced by M ~ may be obtained
from the integral
   0  0 0
~ ~
H R ¼ ∭ G1 ~R; ~
R M ~ ~
R dv (2.59)
V

The dyadic Green’s function satisfies the following differential equation:

 0  0  0
r  r  G1 ~R; ~
R  k2G1 ~ R ¼ jweI d ~
R; ~ R ~
R (2.60)

and the boundary conditions

 0
n  r  G1 ~
^ R; ~
R ¼ 0 on waveguide walls x ¼ 0; x ¼ a; y ¼ 0 and y ¼ b:
(2.61)

The dyadic Green’s function is readily available [8] and is given next for the time
harmonic field with time dependence in the form exp( jwt).
 0 we X X ð2  d0 Þ
G1 ~R; ~
R ¼  exp½jbmn ðz  z0 Þ
ab m n bmn k 2  b2mn
 n o
1 mp np  
 2 jbmn sx cy ^x jbmn cx sy ^y þ k 2  b2mn cx cy ^z
k a b
n mp 0 0 np   o
 jbmn sx cy ^x  jbmn cx0 sy0 ^y þ k 2  b2mn cx0 cy0 ^z
a b
nnp mp onnp mp 0 0 oi
þ sx cy ^x  cx sy ^y sx0 cy0 ^x  cx sy ^y
b a b a
jwe  0
þ 2 ^z^z d ~ R ~ R
k
(2.62)
where m and n vary from 0 to ? in the double summation except for m ¼ 0 and
n ¼ 0, sx ¼ sin(mp x/a), sx0 ¼ sin(mp x0 /a), cx ¼ cos(mp x/a), cx0 ¼ cos(mp x0 /a),
sy ¼ sin(np y/b), sy0 ¼ sin(np y0 /b), cy ¼ cos(np y/b), cy0 ¼ cos(np y0 /b), in and 
the upper sign is used if z > z0 while the lower sign is appropriate for z < z0 . d0 ¼ 1
 Review of electromagnetic theory 15

if m ¼ 0 or n ¼ 0 and it is 0 for m 6¼ 0 and n 6¼ 0. The singular term containing the

0
delta function is zero if ~
R 6¼ ~R.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bmn ¼ k 2  ðmp=aÞ2  ðnp=bÞ2 for propagating modes and
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bmn ¼ j ðmp=aÞ2 þ ðnp=bÞ2  k 2 for evanescent modes:

For the case of a slot in the broad wall (y ¼ 0), cy ¼ 1 and sy ¼ 0, whereas if y ¼ b,
cy ¼ (1)n and sy ¼ 0. If y ¼ y0 ¼ 0 or b, the Green’s function reduces to the fol-
lowing simple form:
 0
we X X
G1 ~ R; ~
R ¼ 2 ð2  d0 Þexp½jbmn ðz  z0 Þ½gxx ^x ^x þ gxz^x^z þ gzx^z^x þ gzz^z^z 
k ab m n

jwe ~ ~0
þ ^z^z d R  R
k2
(2.63)
where
h i
k 2  ðmp=aÞ2
gxx ¼ sinðmpx=aÞsinðmpx0 =aÞ;
bmn
gxz ¼ jðmp=aÞsinðmpx=aÞcosðmpx0 =aÞ;
(2.64)
gzx ¼ jðmp=aÞcosðmpx=aÞsinðmpx0 =aÞ and

2 
k  b2mn
gzz ¼ cosðmpx=aÞcosðmpx0 =aÞ:
bmn
We have listed the tangential components of interest only here. These results have
also been derived using Green’s theorem by Elliott [9]. For sidewall slot problems
the required Green’s function components can be obtained easily from (2.62).

References
[1] Harrington R. F. Time Harmonic Electromagnetic Fields. McGraw-Hill,
New York, 1961.
[2] Rumsey V. H. ‘The reaction concept in electromagnetic theory’. Physical
Review. 1954;94(6):1483–1491.
[3] Stutzman W., Thiele GA. Antenna Theory and Design. John Wiley and Sons,
Hoboken, NJ, 2013.
[4] Love A. E. H. ‘The integration of equations of propagation of electric wave’.
Philosophical Transactions of the Royal Society London, Ser A.1901;197:1–45.
16 Slotted waveguide array antennas: theory, analysis and design

[5] Schelkunoff S. A. ‘Some equivalent theorems of electromagnetics and their

application to radiation problems’. Bell System Technical Journal. 1936;15:
99–112.
[6] Rengarajan S. R., Rahmat-Samii Y. ‘The field equivalence principle: illus-
tration of the establishment of the non-intuitive null fields’. IEEE Antennas
and Propagation Magazine. 2000;42(4):122–128.
[7] Rengarajan S. R. ‘Errata’. IEEE Antennas and Propagation Magazine. 2001;
43(1):134.
[8] Tai C. T. Dyadic Green functions in Electromagnetic Theory. IEEE Press,
New York, 1994.
[9] Elliott R. S. An Introduction to Guided Waves and Microwave Circuits.
Prentice-Hall, Englewood Cliffs, NJ, 1993, p. 712.
 Chapter 3
History

The most important technical advance in the 1930s was, in the opinion of the
writer, the invention of the resonant slot. This was a device which was both an
aperture radiator and a resonant structure. Its novelty was major. Nothing as
important had appeared since Hertz invented the dipole and the loop and Lodge and
Bose experimented with open-ended waveguide radiators.
The words are from Ramsay [1]. He mentions in particular Alan D. Blumlein
at EMI Central Research Laboratories in the UK as the inventor of the resonant
slot antenna (patent no. GB 515684, 1938). Blumlein also proposed linear arrays
of slots.

3.1 The early years

Antennas for transmitting and receiving electromagnetic waves were developed
already in the nineteenth century. In 1885 Edison patented a wireless commu-
nication system using vertical wire structures as antennas. Hertz and Marconi also
used wire antennas. Much of the early work was made at microwave frequencies.
Hertz demonstrated directional beams using a parabolic reflector at 66 cm
wavelength in 1888. A waveguide radiator was used in 1894 by Lodge. Later the
interest shifted to much longer wavelengths for long-distance communication,
and various types of vertical top-loaded dipole antennas and antenna arrays came
into use. In the 1930s and 1940s vacuum tubes for higher frequencies were
developed and the interest in high frequencies arose again. Microwave directional
antennas were developed for applications such as communication, radar and
navigation. The technology advanced in particular during the Second World War.
It was not until this time that the slotted waveguide array antenna emerged on the
scene [2,3].

3.2 The golden years

The period 1935–45 has been named the golden years for advances in microwave
technology. The reason was an urgent need to develop a radar for the detection of
military targets, and both the United States and Great Britain were heavily involved
18 Slotted waveguide array antennas: theory, analysis and design

in this effort. Radar development was independently carried out also in France and
Germany in the 1930s, while Japan and Russia entered the field later.
In 1940 the United States and Great Britain started to exchange information
about radar developments. The British brought with them an extremely important
invention, the resonant-cavity magnetron, which could produce power in kilowatts
at centimetre wavelengths [4]. At this time the Radiation Laboratory at MIT (1940–45)
was established for the development of microwave radar and navigational
equipment for military purposes. A major legacy even today is the Radiation
Laboratory Series of 28 volumes published after the war by McGraw-Hill,
describing the results of the research at Rad Lab in areas related to microwaves and
radar. Worth mentioning in the present context are in particular the Volumes 10 and
12 edited by Nathan Marcuvitz (Waveguide Handbook) and Samuel Silver
(Microwave Antenna Theory and Design), respectively [5–8].

3.3 Waveguide slot antennas

It was in Canada that the first successful waveguide slot antennas were developed.
Watson and his research group at McGill University in Montreal demonstrated the
usefulness of several slot configurations in rectangular waveguides: broad wall
longitudinal and transverse slots and inclined edge wall slots [9]. An example of
experimental data is shown in Figure 3.1, illustrating the equivalent slot resistance
versus slot offset for longitudinal slots [6].
Stevenson [10] studied the electromagnetic boundary problem resulting from
matching the (external) slot field, assuming resonance and the field in the wave-
guide. From power conservation he could predict the longitudinal slot conductance
and show how it varied with the slot offset from the waveguide centreline. See also
Chapter 4 in this book. Stegen [11] provided a set of carefully measured slot
characteristics that could be normalised and used in array designs.
Further progress on slot modelling includes the variational solution by
Oliner [12] and the moment method solution by Vu Khac [13] using pulse
expansion functions, and many more. The goal was to predict the frequency
dependence and the variations due to slot offset and other physical parameters
such as waveguide dimensions and wall thickness. This subject will be discussed
in detail in Chapter 5.
For the efficient design of slotted waveguide array antennas one has to com-
bine slot models, mutual coupling among slots and models for the waveguide
feeding system. Such an array design model was developed by Elliott [14] in 1983.
Design methods are treated in Chapters 6–8. An overview of the progress in slotted
waveguide array antennas up to 1999 has been presented by Rengarajan et al. [15].

3.4 The many shapes of slotted waveguide array antennas

An interesting mechanically scanned array antenna developed at the Radiation Lab
is the long slotted coaxial line array shown in Figure 3.2. The inner conductor has a
 History 19

100
80

60

40

20

10
R 8
Z0 6

2 x1

1
0.8
0.6
0.5
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Slot displacement from centre of guide to centre of slot, in.

Figure 3.1 Measured longitudinal slot resistance versus offset from centreline
[6]. From J. W. Dodds, E. W. Guptill and W. H. Watson, National
Research Council of Canada

Stub support

Radiating slots

Bearing

Figure 3.2 Coaxial line scanner [16]

20 Slotted waveguide array antennas: theory, analysis and design

Figure 3.3 The principal layout of a slotted waveguide array antenna.

’ 1988 IEEE. Reprinted from [18], with permission

series of eccentric disks, displaced from the centre of the outer slotted conductor.
As the inner conductor rotates, the wavelength in the line changes, causing the
beam to scan.
Other scanners were also developed, for example, the Eagle Scanner, of
which more than 2,000 units were produced before the end of the war. In the
Eagle Scanner radiating dipoles were probe-fed from a rectangular waveguide.
By mechanically changing the width of the waveguide the relative phases
of the dipoles were changed, resembling frequency scanning. 30 of scan
could be realised [17]. Slots in circular waveguides are further discussed in
Section 9.10.
A typical layout of a planar slotted waveguide array antenna, also called a flat
plate antenna, is shown in Figure 3.3. The waveguides act both as a feeding
structure and as structural support. The radiating elements are the openings (slots)
in the front face. The antenna has a flat radiating surface and low thickness. It is
suitable for applications in the range of about 1–100 GHz.
In Figure 3.3 we see eight waveguides with radiating slots forming a rectan-
gular antenna aperture. The cut-out reveals the coupling from a feeding waveguide
with inclined slots in the common wall. With slot dimensions and positions care-
fully optimised excellent radiation performance and very low losses are realised;
more about this in the following chapters. Although the slotted waveguide array
antenna was first used in military radar systems, it is also found in many non-
military applications such as remote sensing, weather radar, navigation, radio links
and several other communication systems. Figure 3.4(a) and (b) shows two slotted
waveguide arrays of very different dimensions.
The SAR antenna shown in Figure 3.4(b) was used on the Earth Resources
Satellite ERS-1 that was launched already in 1991, followed by ERS-2 in 1995.
Several more similar systems have been launched since providing high-resolution
information for land, ocean and atmospheric monitoring [20,21].
There are obviously numerous applications of slotted waveguide array anten-
nas. A single slotted waveguide can act as a linear feed for a cylindrical reflector
antenna. With a non-resonant (series) feed the antenna beam can be scanned by
means of frequency variations. An example of an advanced phase/frequency
 History 21

(a)

(b)

Figure 3.4 (a) Slotted array antenna. ’ 2017 Rantec Microwave Systems Inc.
Reproduced from [19]. (b) The 10-m-long Satellite SAR antenna
(C-band) for remote sensing undergoing near-field tests in an
anechoic chamber. Courtesy of RUAG Space AB

scanned planar array is shown in Figure 3.5(a). It is used in a C-band radar system
(Figure 3.5(b)) for pinpointing the location of hostile artillery. As seen in Figure 3.5(a)
numerous vertical slotted waveguides make up the aperture. The antenna beam is
scanned in azimuth by phase steering and in elevation by frequency variation. This is
22 Slotted waveguide array antennas: theory, analysis and design

(a) (b)

Figure 3.5 (a) Phase/frequency scanned slotted waveguide array; (b) radar
unit. Courtesy of Saab AB

a much more cost-effective solution than two-dimensional phased arrays (active

electronically steered arrays (AESA)) with thousands of phase-controlled antenna
transmit/receive modules (TRMs).
The slotted waveguide array antenna has an attractive form factor: it is thin and
flat. In many environments the flat shape is more acceptable than parabolic dish
antennas. Other flat antennas can of course also be considered, such as microstrip
patch arrays which have an equally attractive form factor. However, they have
higher losses and require perhaps twice the area of the waveguide array to reach
comparable performance. As higher frequencies are used the difference between
the two technologies becomes even more apparent. See, for example, [22,23] for a
discussion of communication slotted waveguide arrays at 20 GHz (Figure 3.6(a)
and (b)).
Today slotted waveguide arrays are also proposed for very high frequencies
(>100 GHz), such as in landing radar systems for space probes sent to distant
planets where atmospheric attenuation is absent. This necessitates of course that the
required high precision can be handled. The continued progress is also due to
advances in electromagnetic modelling techniques for the analysis and design,
improved fabrication techniques and new material technologies.
 History 23

(a)

(b)

Figure 3.6 Radio link antennas: (a) antenna with narrow beam for point-to-point
systems; (b) two antennas (E- and H-plane) with sector beam for
point-to-multipoint systems. Courtesy of Ericsson AB

References
[1] Ramsay J. ‘Highlights of antenna history’. IEEE Antennas and Propagation
Newsletter. December 1981. pp. 8–20.
[2] Van Atta L. C. ‘A history of early microwave antenna development’. IEEE
Microwave Theory and Techniques Newsletter. October 1981. pp. 10–14.
[3] Sobol H., Tomiyasu K. ‘Milestones of microwaves’. IEEE Transactions on
Microwave Theory and Techniques. 2002;50(3):594–611.
[4] Skolnik M. I. Introduction to Radar Systems. McGraw-Hill, New York, 1962.
[5] Burns R. W., ed. ‘The background to the development of the cavity
magnetron. In Radar Development to 1945. Peter Peregrinus, London, 1988.
24 Slotted waveguide array antennas: theory, analysis and design

[6] Silver S., ed. Microwave Antenna Theory and Design. MIT Rad. Lab. Series,
Vol. 12, McGraw-Hill, New York, 1949.
[7] Marcuvitz N., ed. Waveguide Handbook. MIT Rad. Lab. Series, Vol. 10,
McGraw-Hill, New York, 1951.
[8] Flock W. L. ‘The radiation laboratory, fifty years later’. IEEE Antennas and
Propagation Magazine. 1991;33(5):43–48.
[9] Watson W. H. ‘Resonant slots’. Journal of the Institution of Electrical
Engineers. 1946;93(4):747–777.
[10] Stevenson A. F. ‘Theory of slots in rectangular waveguides’. Journal of
Applied Physics. 1948;19(1):24–38.
[11] Stegen R. J. ‘Slot radiators and arrays at X-band’. IRE Transactions on
Antennas and Propagation. 1952;1(1):62–84.
[12] Oliner A. A. ‘The impedance properties of narrow radiating slots in the
broad face of rectangular waveguide’. IRE Transactions on Antennas and
Propagation. 1957;5(1):4–20.
[13] Vu Khac T. A Study of Some Slot Discontinuities in Rectangular Wave-
guides. Ph.D. Dissertation, Monash University, Australia. November 1974.
[14] Elliott R. S. ‘An improved design procedure for small arrays of shunt slots’.
IEEE Transactions on Antennas and Propagation. 1983;31(1):48–53.
[15] Rengarajan S. R., Josefsson L. G., Elliott R. S. ‘Waveguide-fed slot antennas
and arrays: a review’. Electromagnetics. 1999;19(1):3–22.
[16] Ridenour L. N. Radar System Engineering. MIT Rad. Lab. Series, Vol. 1,
McGraw-Hill, New York, 1947.
[17] Cady W. M., Karelitz M. B., Turner L. A. Radar Scanners and Radomes.
MIT Rad. Lab. Series, Vol. 26, McGraw-Hill, New York, 1948.
[18] Petersson R., Ingvarson P. ‘The planar array antennas for the European
remote sensing satellite ERS-1’. Proceedings of the European Microwave
Conference. 1988. pp. 289–294.
[19] http://www.rantecantennas.com, retrieved 2017-04-28.
[20] van’t Klooster K. ‘ERS-1, European remote satellite was launched 20 years
ago’. 21st International Crimean Conference on Microwave and Tele-
communication Technology (CriMiCo’2011), September 2011, Sevastopol,
Crimea, Ukraine. pp. 117–118.
[21] Moreira A. ‘A golden age for spaceborne SAR systems’. 20th International
Conference on Microwaves, Radar and Wireless Communication (MIKON).
2014. pp. 1–4.
[22] Svensson B. ‘Dual use of slotted waveguide array antennas’. IEEE Conference
on Antennas and Propagation for Wireless Communication, November 2000.
pp. 149–152.
[23] Svensson B., Manholm L., Wikgren E. ‘A waveguide sector antenna for
point-to-multipoint systems’. IEEE Antennas and Propagation Society
Symposium. 2003. pp. 1185–1188.
 Chapter 4
The slot antenna

In this chapter, we will consider the single-slot antenna. We will first discuss an
arbitrarily shaped aperture in an infinite conducting ground plane. Then we spe-
cialise to a rectangular resonant slot antenna and derive its radiation conductance.
We continue with the special but important case of a longitudinal slot in the broad
wall of a rectangular waveguide. The normalised modal functions for describing
fields in waveguides are introduced. The equivalent slot conductance as seen from
the feeding waveguide is then calculated. Mutual coupling to other slots and ground
plane edge effects is also discussed.

4.1 An aperture in an infinite ground plane

4.1.1 Problem formulation
An aperture is located in the plane z ¼ 0, as illustrated by the shaded area in
Figure 4.1. The plane z ¼ 0 is assumed to be perfectly conducting, except for the
aperture, and of infinite (or very large) extent. The figure to the right presents a side
view, indicating the electric field in the aperture E a ðx; yÞ. The source of this field is
not specified, but one could think of a waveguide feed from below, ending in the
aperture plane as suggested by the figure.
The field in the upper half space (i.e. for z > 0) can be uniquely determined from
equivalent sources on the plane S (i.e. for z ¼ 0). This follows from Schelkunoff’s
field equivalence principles for a plane surface [1, p. 37]. The equivalent sources are
electric and magnetic surface currents which replace the original sources (at z < 0).
Since we are only interested in the external fields in the upper region we can choose
to impose a null field for z < 0. Let us then place a thin electric conductor to cover all
of the plane z ¼ 0. The electric surface current will be shorted and we are left with
the magnetic current on the (shorted) aperture as the only remaining source. Over the
rest of the ground plane the magnetic current is zero since the tangential electric field
is zero. See also Section 2.7 on the field equivalence principle.
The last step is to combine the magnetic surface current and its image and
remove the infinite ground plane; cf. [2, p. 581]. What remains is the equivalent
source, a magnetic current M s radiating in free space:
M s ¼ 2^ n  Ea (4.1)
^ ¼ ^z is the direction normal to the x/y plane.
Here n
26 Slotted waveguide array antennas: theory, analysis and design

z, z'

Θ r
Ea(x, y)

y, y'

r'
x, x' Φ
(a) (b)

Figure 4.1 (a) A general aperture in a ground plane; (b) cross section

4.1.2 Far field

The radiated far field from the magnetic current distribution is easy to find via the
electric vector potential F [2, p. 120]:
ðð
e ejkR
F¼ Ms dS (4.2)
4p S R
In this general expression the integration is carried out over the magnetic surface
current distribution. R is the distance from any point in the source to the observa-
tion point and e is the permittivity of the medium.1
In the far-field region we can write R ¼ r  ^r  r 0 (see Figure 4.1) which gives
ðð
eejkr 0
M s ejðkx x þky y Þ dx0 dy0
0
F¼ (4.3)
4pr S
where kx and ky are the directional cosines of the wave vector, that is, the far-field
direction. From F we get the electric far field
EQ ¼ jwhFF
(4.4)
EF ¼ jwhFQ
where Er ¼ 0 in the far field, h is the free space impedance.
By inserting (4.1) in (4.3) we obtain the vector potential in the following form:
  ðð  
Fx e ejkr
ejðkx x þky y Þ dx0 dy0
Eay 0 0
¼ (4.5)
Fy 2pr S Eax

Equation (4.5) resembles a Fourier transform, except for the integration limits.
However, let us extend the integration limits to infinity (makes no difference, since

1
We have chosen the most common definition of the potential F from the relation eE ¼ r  F .
However, in [3,4] the relation E ¼ r  F is used.
 The slot antenna 27

the aperture field is zero outside the aperture). Hence, we introduce the following
Fourier transform:
ð1 ð1
 
f t k x ; ky ¼ E a ðx; yÞejðkx xþky yÞ dxdy (4.6)
1 1

where f t is a vector with the components ( fx, fy). Thus, we can write
   
Fx e ejkr fy
¼  (4.7)
Fy 2p r fx
From the x- and y-components of F we get the q- and f-components:
 
FQ ¼ Fx cos f þ Fy sin f cos q
(4.8)
FF ¼ Fx sin f þ Fy cos f

Combining (4.4), (4.7) and (4.8) yields

8
> jwhe ejkr  
>
< EQ ¼ fy sin f þ fx cos f
2pr
(4.9)
>
> jwhe e jkr  
:E ¼ cos q fy cos f  fx sin f
F
2pr
This result shows the electric far field as a function of the Fourier transform of the
aperture field. The inverse transform, cf. (4.6), is
ðð
1  
E a ðx; yÞ ¼ 2 f t kx ; ky ejðkx xþky yÞ dkx dky (4.10)
4p

4.1.3 General field expressions

So far we have only discussed the far field. We will now derive a general solution
for the electric field in all of the source-free half-space above the aperture. The
solution will have to satisfy Helmholtz wave equation in this entire region, both
near and far fields:
 
r2 þ k 2 E ¼ 0 (4.11)

We introduce the following Fourier transform pair

ðð
 
E kx ; ky ; z ¼ E ðx; y; zÞejðkx xþky yÞ dxdy
~ (4.12)

ðð
1  
E ðx; y; zÞ ¼ 2 ~ kx ; ky ; z ejðkx xþky yÞ dkx dky
E (4.13)
4p
28 Slotted waveguide array antennas: theory, analysis and design

and put (4.13) into the wave equation, yielding

 2 
@ 2 ~
þ k z E ¼0 (4.14)
@z2
The solution to this is
   
~ kx ; ky ; z ¼ f kx ; ky  ejkz z
E (4.15)
and thus
ðð
1  
E ðx; y; zÞ ¼ 2 f kx ; ky ejk r dkx dky (4.16)
4p
This solution describes the electric field as a superposition of plane waves [5].
It is valid in the half plane above the aperture, both in the near field and in the
far field.
The integrals in (4.13) and (4.16) extend over the full kx ky plane. kz is impli-
citly included since

k 2 ¼ kx 2 þ ky 2 þ kz 2 (4.17)
That is,
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
< kz ¼ þ k 2  kx2  ky2 when kx2 þ ky2 < k 2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (4.18)
: k ¼ j k 2 þ k 2  k 2 when k 2 þ k 2 > k 2
z x y x y

kz is negative imaginary in the so-called invisible space, characterised by attenuated

waves (see Figure 4.2).
While fðkx ; ky Þ in (4.16) describes the electric field as a spectrum of plane
waves, the corresponding plane wave spectrum for the magnetic field is analo-
gously given by
  1  
h kx ; ky ¼ k  f kx ; ky (4.19)
kh

ky

Invisible

Visible
kx
k

Figure 4.2 Visible and invisible space

 The slot antenna 29

4.1.4 Radiated complex power

For simplicity, let us assume that the aperture electric field is x-directed, that is,
E a ðx; yÞ ¼ Ex ðx; yÞ^x . We have therefore
ðð
1  
Ex ðx; yÞ ¼ 2 fx kx ; ky ejðkx xþky yÞ dkx dky (4.20)
4p

From (4.19) we get the magnetic field in the aperture as

ðð
1 k 2  ky 2  
Hy ðx; yÞ ¼ fx kx ; ky ejðkx xþky yÞ dkx dky (4.21)
4p2 kh kz

since f  k ¼ 0, that is, the plane wave fields are perpendicular to the direction of
propagation, see also [2, p. 624].
Knowing the aperture fields we can now formulate an expression for the
complex power flow out of the aperture:
ðð ð þ1 ð þ1 2
1  1 k  ky2  
fx kx ; ky j2 dkx dky
P ¼ Pr þ jPi ¼ Ex Hy dS ¼ 2 
S2 8p kh 1 1 kz
(4.22)

Inserting Ex and Hy from (4.20) and (4.21) into the expression for the complex
power integral given earlier leads to a sextuple integral. However, it can be reduced
to this form (4.22, right member) using the relation
ð þ1
ejðk1 k2 Þx dx ¼ 2pdðk1  k2 Þ (4.23)
1

An excellent illustration of the complex power from a narrow slot aperture has
been given by Rhodes [6] (see Figure 4.3). The volume within the visible region
kx2 þ ky2 < k 2 corresponds to the radiated power. The reactive power is outside this
region. Note that the reactive, non-radiating (stored) power goes both negative and
positive (electric and magnetic energy, respectively). When they balance each other
the aperture is resonant.

4.1.5 The aperture admittance

Assume that a suitable port with excitation voltage V0 and current I0 is defined.
Then the aperture admittance Y can be defined from the complex power P
according to

V0 I0 jV 2 j
P ¼ Pr þ jPi ¼ ¼ 0 Y (4.24)
2 2
We have so far only discussed the characteristics of a general aperture. In
practice an aperture is backed by some feeding arrangement, cavity or waveguide,
30 Slotted waveguide array antennas: theory, analysis and design

ky
(k2 – ky2)|Fx(kx,ky,k)|2 k 2k 3k 4k
4k
2kZ
8 0{(k
2 – kx2 – ky2)½}*

3k

2k
V02k
22Z0
k

0 k kx

Figure 4.3 One quadrant of the complex power density from a thin half wave
slot antenna with applied voltage V0 at the centre. Reproduced from
[6, Fig. 3.1, p. 50]. By permission of Oxford University Press

etc., and the total admittance is modified by these structures. Before going into
more detail on this we will in the next section analyse the conductance of a narrow
rectangular slot in an infinite ground plane.

4.2 The rectangular slot antenna

Consider a rectangular slot in a large ground plane coincident with the x/z-plane as
shown in Figure 4.4. The feeding voltage is V0 in the slot centre. The slot is narrow,
z-directed, with width w and length 2L.
As in (4.24) we can relate the slot conductance Gslot to the radiated real power Pr:
1
ReðPÞ ¼ Pr ¼ V02 Gslot (4.25)
2
We will find the radiated power by integrating the power pattern from the slot.
Since a narrow slot and a thin dipole have identical radiation patterns (if E- and
H-fields are interchanged) we can use known results for the electric dipole, the dual
problem [3].
In the dipole case with a sinusoidal current Iz fed in the dipole centre
Iz ðzÞ ¼ I0 sin½k ðL  jzjÞ
the pattern is [4]

jhI0 ejkr cosðkL cos qÞ  cosðkLÞ

Eq ¼  (4.26)
2p r sin q
 The slot antenna 31

θ
r

y
2L

Feed line
w
x Φ

Ground plane

Figure 4.4 A rectangular slot antenna in a large ground plane

With a resonant length 2L ¼ l/2 this becomes

p 
jhI0 e cos
jkr cos q
Eq ¼  2 (4.27)
2p r sin q
The corresponding resonant slot is assumed to be excited according to

V0
Esx ðzÞ ¼ cos kz (4.28)
w
The equivalent magnetic current source is

M ¼ 2E  ^y ; (4.29)

which is z-directed, hence

V0
Mz ðzÞ ¼ 2 cos kz (4.30)
w
The resulting far field is
p 
jV0 e cos
jkr cos q
Hq ¼ 2 (4.31)
hp r sin q

The last expression is obtained directly from the dipole case (4.27), by repla-
cing the electric current I0 by the magnetic current 2V0/w and have the impedance
of free space h replaced by 1/h.
32 Slotted waveguide array antennas: theory, analysis and design

The radiated power is obtained by integrating the radiated average power

density in the far field (here we assume that the slot radiates only over one-half
space). The power density is

1  1
jSj ¼ jE  H j ¼ hjHq j2 (4.32)
2 2
thus
ðp ðp
1
Pr ¼ h dF r2 sin qjHq j2 dq (4.33)
2 0 0

We get
 
ð p=2 cos2 p cos q
V02 2
Pr ¼ I; where I ¼ dq (4.34)
hp 0 sin q

This last integral is a bit complicated to evaluate. The solution can be for-
mulated in terms of the cosine integral function Ci(x), see, for example, [7, p. 224],
or be solved numerically.
Changing variables in (4.34) we obtain for the integral I
ð 2p
1 1  cos v
I¼ dv (4.35)
4 0 v

The cosine integral function Ci(x) can be written

ðx
1  cos v
CiðxÞ ¼ g þ ln x  dv (4.36)
0 v
where g ¼ 0.5772 is Euler’s constant. Thus

1
I ¼ ½0:5772 þ lnð2pÞ  Cið2pÞ (4.37)
4
Ci(x) is tabulated in Standard Mathematical Tables [8, p. 532] and we obtain
finally

V02
Pr ¼ 0:609 (4.38)
hp
We have already in (4.25)
1
Pr ¼ V02 Gslot (4.39) = (4.25)
2
0:609 1
From the last two expressions we obtain Gslot ¼ 2
 siemens, thus
60p 1;000
about 1,000 W (or 500 W if radiating on both sides of the ground plane).
 The slot antenna 33

The impedance Zd of a dipole and that of a slot Zs (complementary antennas)

are related as
Zd Zs ¼ h2 =4 (4.40)
(Booker’s relation) [4,9] where h is the intrinsic impedance of free space 120p W.
The typical half-wave dipole has about 73 W impedance. From (4.40) we can
expect the complementary slot to have approximately 35,000/73 ¼ 479 W. The
conclusion is that a slot antenna has a rather high impedance compared to common
transmission line impedances with typically 50 or 70 W characteristic impedance.
This might indicate a potential matching problem. This can be solved in actual slot
antennas and slot arrays as we will see in subsequent chapters.

4.3 Waveguide modes

For the analysis of waveguide-fed slot antennas we need adequate solutions for the
fields in the waveguide. A set of normalised expressions for the waveguide modes
is presented here, starting with a general case and later specialising to TE10 modes.

4.3.1 Definitions
A general expression for the waveguide fields can be written as [1]
X  
E ðx; y; zÞ ¼ Ai e ti ezi^z e
gi z
X  t 
i
(4.41)
H ðx; y; zÞ ¼ Ai h i þ hzi^z e
gi z
i

where Ai is the amplitude of mode i. The superscript t indicates transverse fields,

that is, x- and y-components and z the direction of propagation (Figure 4.5).
4.3.1.1 Sign convention
The time dependence ejwt is assumed. The upper (lower) sign in (4.41) refers to
t
propagation in the positive (negative) z direction. The functions e ti ; h i ; ezi ; hzi are all
independent of the direction of propagation. These conventions are identical to
those in [1] and other places.

y
z

Figure 4.5 A waveguide with arbitrary cross section (S)

34 Slotted waveguide array antennas: theory, analysis and design

4.3.1.2 Modal indices

The mode number i can be understood as a reordering of the triple index pmn,
where p ¼ 1 indicates a TE wave and p ¼ 2 indicates a TM wave. mn is the usual
mode index for variations in the x and y directions, respectively.2

4.3.1.3 Modal admittance

For TE waves: Yi ¼ Y1mn ¼ gi =ðjwm0 Þ

(4.42)
For TM waves: Yi ¼ Y2mn ¼ jwe0 =gi

4.3.1.4 Normalisation
We choose
ðð

0; i 6¼ j
e i  e i dS ¼ dij ¼
t t
(4.43)
S 1; i¼j
where the integration is over S ¼ the waveguide cross section.
Other criteria are sometimes used, for example, in [1,10]:
ðð

t 0; i 6¼ j
e i  h j  ^z dS ¼
t
S 1; i ¼ j
which differs from our choice by a factor ¼ the modal admittance. See also [11].
Our choice according to (4.43) results in
ðð
t
e ti  h j  ^z dS ¼ Yi  dij (4.44)
S

4.3.2 The rectangular waveguide

The waveguide has a cross section a  b with z the direction of propagation (see
Figure 4.6). The materials involved are assumed to be lossless.
Propagation constant gi
gi ¼ ai þ jbi (4.45)
g2i ¼ kc;mn 2  k 2 (4.46)
where
mp2 np2
kc;mn 2 ¼ þ (4.47)
a b
and
2p
k¼
l

2
Note, however, that Collin [1] uses nm.
 The slot antenna 35

b
x
a

Figure 4.6 Rectangular waveguide

gi is positive real for non-propagating waves. It is positive imaginary for propa-

gating waves.

4.3.2.1 Normalised mode expressions

TE case:
rffiffiffiffiffiffiffiffiffih
1 em en np mpx npy mp mpx npy i
e tmn¼ cos sin ^x  sin cos ^y (4.48)
kc;mn ab b a b a a b
rffiffiffiffiffiffiffiffiffih
t gmn em en mp mpx npy np mpx npy i
h mn ¼ sin cos ^x þ cos sin ^y (4.49)
jwm0 kc;mn ab a a b b a b
rffiffiffiffiffiffiffiffiffi
kc;mn em en mpx npy
hzmn ¼ cos cos (4.50)
jwm0 ab a b

TM case:
2 hmp mpx npy np mpx npy i
e tmn ¼ pffiffiffiffiffi cos sin ^x þ sin cos ^y (4.51)
kc;mn ab a a b b a b
2kc;mn mpx npy
ezmn ¼ pffiffiffiffiffi sin sin (4.52)
gmn ab a b
t 2jwe0 hnp mpx npy mp mpx npy i
h mn ¼ pffiffiffiffiffi sin cos ^x  cos sin ^y (4.53)
gmn kc;mn ab b a b a a b

In these expressions em and en are the Neumann factors defined as follows:

(
1 for m ¼ 0
em ¼ (4.54)
2 for m > 0

The special case with m ¼ n ¼ 0 is often neglected since it does not contribute
anything to the power flow in the waveguide [12]. It contributes, however, to the
36 Slotted waveguide array antennas: theory, analysis and design

stored (reactive) power and is sometimes essential in the analysis of slots in the
wall of a waveguide [13].

4.3.3 The TE10 mode case

In this special case the following expressions are valid.
g10 b b l
Wave admittance: Y10 ¼ ¼ 10 ¼ 10 ¼ (4.55)
jwm0 wm0 kh lg h
Propagation constant: g210 ¼ kc;10
2
 k2 (4.56)
p
kc;10 ¼ (4.57)
a
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p2 2p
b10 ¼ k2  ¼ (4.58)
a lg
where lg ¼ l10 is the guide wavelength. The cut-off wavelength is lc ¼ 2a.
rffiffiffiffiffi
2 px
Mode expressions: e t10 ¼  sin ^y (4.59)
ab a
rffiffiffiffiffi
p=a 2 px
h10 ¼
z
cos (4.60)
jwm0 ab a

where x is measured from the waveguide side wall (Figure 4.6).

t
For the transverse magnetic field we have in general h i ¼ Yi^z  e ti and obtain
for TE10
t
h 10 ¼ Y10^z  e t10 ¼ Y10 et10^x (4.61)
As an example we have the transverse magnetic field of a mode with amplitude A10
t
H 10 ¼ A10 Y10 et10^x (4.62)
where again the upper/lower sign indicates propagation in the þ/ z direction,
respectively.

4.4 The longitudinal slot in a waveguide wall

4.4.1 Preliminaries
Figure 4.7 shows a general aperture in a general waveguide.
Suppose the aperture field E s (see Figure 4.7) is known. This field will excite
waves in the waveguide propagating backward (amplitudes A i ) and forward
(amplitudes Aþ i ) passing through the waveguide cross sections S1 and S2. The
amplitudes A i can be found with the aid of the Reciprocity Theorem (see e.g. [14,
p. 24]), applied to the volume enclosed by S ¼ S1 þ S2 þ S3. The method is outlined
next; cf. also Section 2.4.
 The slot antenna 37

n̂
Waveguide Es
S2
S1

n̂ n̂
z

S3

Figure 4.7 An aperture with aperture field E s in a waveguide wall

For two field solutions E 1 =H 1 and E 2 =H 2 in a source-free region the following

relation is true:
ð
 
E 1  H 2  E 2  H 1  ^n dS ¼ 0 (4.63)
S

We assume
E 1 =H 1 ¼ the field produced by the excited aperture
E 2 =H 2 ¼ a testing field ða normalised waveguide modeÞ
and rewrite (4.63) in the following form:
ð
    
^
n  E 1  H 2  ^n  E 2  H 1 dS ¼ 0 (4.64)
S

Now
^
n  E 1 ¼ 0 on S3 except in the aperture
^
n  E 2 ¼ 0 on all of S3 :
Thus
ð ð
   
^
n  E s  H 2 dS þ E 1  H 2  E 2  H 1  ð^z ÞdS
aperture S1
ð
 
þ E 1  H 2  E 2  H 1  ^z dS ¼ 0 (4.65)
S2

Take two modes, i and j with amplitudes Ai and Aj, respectively. Then
ð
E i  H j  ^z dS ¼ Ai Aj Yi dij (4.66)
cross
section

(plus sign if Aj propagates in the positive z direction).

Let E 1 =H 1 be expressed as a summation of modes with amplitudes Aj.
Let E 2 =H 2 be a normalised mode number i (amplitude ¼ 1) propagating in the
negative z direction.
38 Slotted waveguide array antennas: theory, analysis and design

This gives from (4.63) þ (4.64)

ð
 
n  E s  h i egi z dS þ Yi Ai  Yi Ai  Yi Ai  Yi Ai ¼ 0
^ (4.67)
slot

The result is
ð
1
A
i ¼ ð^n  E s Þ  h i
e gi z dS (4.68)
2Yi aperture

where A i is the ith mode amplitude (voltage) propagating in the positive/negative

z-direction; Yi is the modal admittance; ^n is the outward normal; h i ¼

h t þ hzi^z is
the normalised modal function for the magnetic field, propagating in the negative/
positive z-direction.
Thus, if we know the aperture field ðE s Þ we can find the amplitudes of all
modes scattered forward and backward in the waveguide. The expression (4.68) is
quite general; it is valid for many types of waveguides with circular, rectangular
and even ridge-shaped cross sections.

4.4.2 The longitudinal rectangular slot

The aperture will now be specialised to a longitudinal slot with length 2L (see
Figure 4.8). We assume the slot to be narrow enough so that only the transverse
E-field (x-directed) is significant.3 We will furthermore assume the slot to
be resonant, that is, with a cosinusoidal field distribution along the z-direction.
We apply (4.68), noting that the outward normal in the slot region is y-directed, i.e.
 
^
n  E s  h i ¼ Esx  hzi (4.69)
We obtain
ð
1
A
i ¼ Esx ðx; zÞ  hzi ðxÞe gi z dS (4.70)
2Yi slot

L L
w
x0
z

Figure 4.8 A longitudinal slot with coordinates

3
How narrow is ‘narrow enough’? From experience a possible criterion is w/2L < 0.1 to 0.2. For wider
slots the longitudinal E-field may become significant.
 The slot antenna 39

The slot electric field distribution is assumed to be cosinusoidal:

V0 pz
Esx ðzÞ ¼ cos (4.71)
w 2L
where V0 is the slot voltage. Inserting (4.71) in (4.70) and performing the z inte-
gration gives
p
cos bi L ð x0 þw=2
þ  V 0 2k0 L
A i ¼ Ai ¼    2 : hzi ðxÞdx (4.72)
k0 wYi p 2 bi x0 w=2

2k0 L k0

where bi is the propagation constant for mode i in the waveguide and x0 is the slot
position from the side wall. Note that Aþ 
i ¼ Ai since the slot field was assumed to
be symmetric.
This result is valid for longitudinal, narrow slots in arbitrary waveguides. As
the next step we will study the special case with rectangular waveguides and cal-
culate the TE10 scattering, see Figure 4.9.

4.4.3 Rectangular waveguide and TE10 mode scattering

For the fundamental TE10 mode we have the normalised mode function for the
magnetic field (see Section 4.3.3):
rffiffiffiffiffi
p=a 2 px
hzi ¼ hz10 ¼ cos (4.73)
jwm0 ab a

Carrying out the integration in (4.72), neglecting variations across the narrow
slot width, we obtain
pffiffiffi p
V0 jp 2 cos b1;0 L px0
Aþ 
1;0 ¼ A1;0 ¼ pffiffiffiffiffi  2L2  cos (4.74)
b1;0 a ab p a
 b1;0 2
2L

X0

b
a
x

Figure 4.9 Rectangular waveguide with a longitudinal rectangular slot

40 Slotted waveguide array antennas: theory, analysis and design

Inserting the resonant length 2L ¼ l/2 and b1,0 ¼ 2p/lg, where lg is the guide
wavelength, we get finally:
pffiffiffi rffiffiffi    
j 2lg a pl px0
A1;0 ¼ V0 cos cos (4.75)
pl b 2lg a

To summarise we have now a relation between the slot voltage V0, the scat-
tered wave amplitude A 1;0 and the geometrical parameters. The simplifying
assumptions leading to (4.75) were
● narrow rectangular longitudinal slot
● slot voltage V0 given, cosinusoidal field distribution
● slot length 2L ¼ l/ 2
● fundamental TE10 mode
● zero wall thickness
For non-zero wall thickness, V0 applies to the inner slot field. The inner and
outer fields are about the same for a practical wall thickness. Nevertheless, the
thickness has an impact on the slot admittance and the slot resonant length as we
will see later.
We still need to find the relation between the incident waveguide mode
amplitude and the slot excitation.

4.4.4 The equivalent slot conductance

We take a circuit approach and consider the slot to be represented by a shunt
admittance element (shunt conductance Gs if resonant) (see Figure 4.10). The shunt
model is motivated by the symmetrical scattering from a slot with symmetrical
field distribution as already discussed. The slot is excited from an incident wave-
guide mode with amplitude Ainc i .
Only the fundamental mode is assumed to be propagating but higher-order
modes will be excited near the slot. Although they do not propagate in the wave-
guide, they will contribute to the shunt admittance.

Slot

Ainc
i Ainc
i
Gs Yi
– +
Ai Ai

z=0

Figure 4.10 Circuit model with a shunt conductance Gs representing the slot
loading of the waveguide. Ai represent the scattered wave
amplitudes (voltages) referenced to the slot position z ¼ 0. Yi is the
characteristic admittance of the (waveguide) transmission line
 The slot antenna 41

The power radiated by the slot is represented by the power dissipated in the
shunt conductance in our circuit model:
1
Pdis ¼ jVn j2 Gs (4.76)
2

where the mode voltage Vn ¼ Ainc
i þ Ai :
This power must be equal to the radiated power already derived:

V02
Prad ¼ 0:609 (4.77) = (4.38)
hp
The normalised shunt conductance Gs/Yi is related to the reflection coefficient
G ¼ A
i =Ai and we get (assuming resonance)
inc

Gs =Yi ¼ 2G=ð1 þ GÞ ¼ 2A

i =Vn (4.78)
Furthermore, with the slot length 2L ¼ l/2 we have the scattered TE10 mode
amplitude
pffiffiffi rffiffiffi    
j 2lg a pl px0
A ¼ A ¼ V0 cos cos (4.79) = (4.75)
i 1;0
pl b 2lg a

Combining (4.78) and (4.79) and writing Yi ¼ Y10 ¼ G0 we get

pffiffiffi rffiffiffi    
V0 j 2lg a pl px0
Gs =G0 ¼ 2 cos cos (4.80)
Vn pl b 2lg a
The two power expressions (Prad ¼ Pdis ) give us
2
V 0
¼ hp Gs =2 (4.81)
V 0:609
n

Squaring (4.80) leads to

2 2    
Gs2 V0 8lg a 2 pl 2 px0
¼ cos cos (4.82)
G02
V n p2 l2 b 2lg a
b l
Inserting (4.81) in (4.82) and noting G0 ¼ Y10 ¼ 10 ¼ gives the
kh lg 120p
classical result
  px 
Gs a lg 2 pl 0
¼ 2:09 cos cos2 (4.83)
G0 b l 2lg a

This important result, originally derived by Stevenson [15], shows how the
conductance (i.e. the excitation) of a slot depends on the slot position. We repro-
duce the Stevenson result in Figure 4.11.
So far we have discussed the slot conductance for resonant slots. The
assumption has been that the slot length l/2 automatically results in slot resonance
42 Slotted waveguide array antennas: theory, analysis and design

0.70
0.60
0.50
0.40

0.30
Normalised resonant conductance, Gr /G0

Slot width = 0.0625 inch

0.20 a = 0.900 inch
b = 0.400 inch
t = 0.050 inch
v = 9.375 GHz
0.10
0.08

0.06

0.04

0.03

0.02

0.01
0 0.050 0.100 0.150 0.200 0.250
Slot offset in inches

Figure 4.11 Resonant conductance versus slot offset from the waveguide
centreline for a longitudinal slot in standard X-band waveguide,
calculated by Stevenson [15]. Points are experimental results
according to Stegen [16].  1981 J. H. Elliott. Reprinted from [4],
with permission

and that the slot admittance is real. However, resonance also depends on the slot
shape, the waveguide dimensions and the effects of mutual coupling among the
slots in an array. We are also interested in the complete slot admittance and how it
varies over a band of frequencies. A more detailed analysis is therefore necessary in
order to control these parameters and optimise the slot (array) antenna for best
performance including bandwidth.
Consequently it is not necessarily true that a slot length of l/2 results in
resonance. From the more detailed slot modelling examples discussed in Chapter 5,
we will find that both shorter and longer slots (than l/2) can be resonant, depending
on several parameters such as slot offset, waveguide height and dielectric loading.
In order to accurately calculate the slot admittance we need to include the reactive
fields in the slot vicinity. For the external region we already have found one field
representation that includes this: the plane wave spectrum [(4.20) and (4.21)]. For
the internal waveguide region we have similarly the ‘spectrum’ of waveguide
 The slot antenna 43

modes (4.68). Before going into detailed discussions on how these and other
representations can be used in order to find the true slot field distribution and the
slot admittance we will take a look at the mutual coupling effects.

4.5 Mutual coupling

4.5.1 Introduction
The mutual coupling between aperture-type antennas has been discussed by several
authors [3,17]. For the analysis of waveguide slot arrays it has been assumed that an
equivalent array of dipoles could be used as a model for the mutual coupling [18].
An improved model, especially for dielectric filled waveguides, was later devel-
oped by Elliott [19] in which the slot field is assumed to be half cosinusoidal. This
agrees with assuming the slot to have the fundamental TE10 mode excitation. This
is also our approach in our mutual coupling analysis in this chapter.
Consider the small array antenna with 52 slot elements in Figure 4.12. In this
array each slot has its individual stripline feed with a connector and a matched load.
The slot spacing is d/l ¼ 0.625 in a square lattice. In the diagram to the right are
shown two measured E-plane patterns for one of the slots near the array centre,
the isolated pattern and the embedded pattern. In the former case all other slots
were covered with conducting tape, in the latter case all other slots were present
and terminated in matched loads.
We see two important effects. The isolated E-plane element pattern would
ideally be flat over 90 , but exhibits a significant ripple. This is due to the
diffraction from the ground plane edges. The second observation is the drastically
modified embedded pattern due to mutual coupling among the slots. A significant
narrowing and a dip occur at about 35 , as expected from phased array theory

dB
0
Isolated

–5

Embedded

–10
–100° –50° 0° 50° 100°
θc
(a) (b)

Figure 4.12 (a) A slotted stripline array; (b) measured isolated and embedded
E-plane patterns for an element in the centre of the array.
 IEEE 1974. Adapted from [20], with permission
44 Slotted waveguide array antennas: theory, analysis and design

[21]. See also Section 9.1. From the slot spacing the critical ‘blind angle’ qc is
predicted to be at 37 according to the grating lobe angle equation.
sin qc ¼ l=d  1 (4.84)
The H-plane coupling for slots is much weaker than the E-plane coupling, and
can sometimes be disregarded in non-scanned waveguide slotted array designs.
However, our example demonstrates that the array excitation can be dramatically
changed by mutual coupling and that we need to accommodate this effect in a
general design procedure.
A final word on coupling definitions used here: isolated coupling stands for the
case with only two elements present. Array coupling is again coupling between two
elements, but now all other elements are also present and terminated in matched
loads. Active impedance/admittance stands for the situation where all elements are
fed. Isolated (or self) impedance/admittance is for one single element only. Typical
isolated and embedded patterns are compared in Figure 4.12.

4.5.2 Mutual coupling calculations

4.5.2.1 Introduction
In the slotted waveguide array antenna the effect of the waveguide wall thickness
can be modelled as short waveguides connecting the two apertures: the inner slot
and the outer slot. We have discussed the equivalent slot conductance as seen from
the feeding waveguide in Section 4.4.4 based on the inner slot field. We will now
focus on the outer slot field for the mutual coupling calculation. The two regions
will later be connected (Chapter 5).
Figure 4.13 shows a cross section of a number of waveguide openings in a
large ground plane, suggesting a coupled signal from element m to element n. For
the present discussion we assume that each aperture is characterised by one mode
only. This mode will be the TE10 mode with an x-directed electric field (see
Figure 4.13).

Ex
#m #n x
z .

Vm+ Vm– Vn+ Vn–

Figure 4.13 Cross section of a slotted array antenna illustrating the E-plane
coupling between apertures m and n
 The slot antenna 45

4.5.2.2 Equivalent circuit approach

Since we assume a single-mode situation and also have reciprocity and linearity we
can represent the slot array as seen from the feeding side by a simple circuit model
(Figure 4.14).
For the circuit model the currents and voltages are related as follows:
9
J1 ¼ Y11 V1 þ Y12 V2 þ Y13 V3 þ    >
>
>
>
J2 ¼ Y12 V1 þ Y22 V2 þ Y23 V3 þ    =
(4.85)
>
J3 ¼ Y31 V1 þ Y32 V2 þ Y33 V3 þ    >
>
>
;
J4 ¼   
or in matrix form
½J  ¼ ½Y ½V  (4.86)
As an example we can find the matrix element Y13 ¼ J1/V3 provided all the
other voltages V1,V2, V4, . . . ¼ 0, for example, by short circuiting the corresponding
apertures. Repeating this procedure we can find the complete admittance matrix
from the mutual admittances between all pairs of elements! In fact, for a regularly
spaced array with identical elements, the mutual admittance values are only
dependent on the relative positions between the elements (the matrix Y is a Toeplitz
matrix).
We can think of the slot array as one where each slot has a feed port across the
slot centre (cf. Figure 4.4). Shorting the port with a conducting element leaves two
smaller apertures on each side of the short. The small apertures will be very far
from resonance and do not radiate. In practice it would make no difference if the
ground plane covered the slot completely. (A similar type of reasoning applies to
arrays of resonant dipoles when calculating the impedance matrix. In that case the
ports are left open circuited instead.)

J1

V1 +
– Slot
array

N ports

JN

VN +
–

Figure 4.14 Circuit model of a slot array assuming single-mode apertures or

minimum scattering antennas (MSA) [22]
46 Slotted waveguide array antennas: theory, analysis and design

4.5.2.3 Single-slot admittance

We will now analyse the self-admittance of a single slot (which we assign the
number n) in a large ground plane. No other slots are present.
Let the upward and downward propagating modes in the feeding waveguide
just below the slot aperture be Vnþ en and Vn en , respectively (cf. Figure 4.13), where
en is the normalised (electric) modal function. Thus, the total (electric, x-directed)
field at aperture n is Eint ¼ ðVnþ þ Vn Þen . The total (z-directed) magnetic field in
the waveguide just below the aperture is then H int ¼ ðVnþ  Vn ÞY0 en , where Y0 is
the modal admittance of the waveguide. The external magnetic field at the aperture
can be written as H ext ¼ ðVnþ þ Vn ÞHnext ðen Þ, where Hnext ðen Þ is a function
describing the external magnetic field for a given normalised electric aperture field.
We want to determine Vn while Vnþ is assumed known.
We now enforce continuity of the magnetic field at the aperture, by putting
Hint ¼ Hext, leading to ðVnþ  Vn ÞY0 en ¼ ðVnþ þ Vn ÞHnext ðen Þ. An approximate
solution can be found according to the Method of Moments [23] by taking inner
products with the normalised modal function en as testing function. We define this
inner product (or projection) as follows:
ðð
hep ; eq i ¼ ep ðx; zÞ  eq ðx; zÞdxdz ¼ dp;q ; (4.87)
slot

and obtain
 þ   
Vn  Vn Y0 ¼ Vnþ þ Vn hHnext ðen Þ; en i (4.88)
The part inside the brackets h i is the aperture self-admittance: Ynn ¼
hHnext ðen Þ; en i. The reflection coefficient G becomes
Y0  Ynn
G ¼ Vn =Vnþ ¼ (4.89)
Y0 þ Ynn
An accurate evaluation of the self-admittance is rather complicated
(cf. Section 6.4), partly because we need to find and integrate the field inside the
source region. Furthermore, in order to improve the accuracy of the solution; it is
often necessary to include a number of higher-order modes besides the fundamental
mode in the aperture; more about this in Chapter 5.

4.5.2.4 Mutual admittance between two slot apertures

Let us now take a look at the mutual admittance between two slot apertures
(isolated coupling).
The external magnetic field at aperture n due to coupling from the electric field
at aperture m (the source) can be written as
 þ 
Vm þ Vm Hnext ðem Þ þ Vn Hnext ðen Þ (4.90)
The waveguide at the receiving aperture n will be excited with a wave
with internal magnetic field Vn Y0 en . A wave scattered into space will also
 The slot antenna 47

appear. Note that Hnext ðem Þ is the external field at aperture n due to the source at
aperture m.
Matching internal and external magnetic fields at aperture n gives
 
Vn Hnext ðen Þ þ Vmþ þ Vm Hnext ðem Þ ¼ Y0 Vn en (4.91)
Let us also match the magnetic fields at the source aperture m, giving
 þ   
Vm þ Vm Hmext ðem Þ þ Vn Hmext ðen Þ ¼ Y0 Vmþ  Vm em (4.92)
Taking inner products with the modal functions em and en as before yields the
two equations:
  )
Vn Ynn þ Vmþ þ Vm Ynm ¼ Y0 Vn
 þ    (4.93)
Vm þ Vm Ymm þ Vn Ymn ¼ Y0 Vmþ  Vm
Note that Ynm ¼ Ymn ¼ hHnext ðem Þ; en i is the mutual admittance between the
two apertures. Eliminating Vm gives the mutual coupling in terms of the scattering
parameter Smn

Vn 2Y0 Ymn

Smn ¼ þ
¼ (4.94)
Vm Ymn  ðY0 þ Ymm Þ2
2

The coupling between just two elements (no other element present) is called
the isolated coupling. For slot apertures it decays with distance approximately as
1/R in the E-plane. In the H-plane the decay is roughly 1/R2.

4.5.2.5 Array coupling

The scattering from a large array is most easily written using matrices, cf. (4.89):

½S  ¼ ð½Y0  þ ½Y e Þ1 ð½Y0   ½Y e Þ (4.95)

where [S] is the scattering matrix. [Y e] is the mutual admittance matrix (with the
exponent e indicating the external aperture interface) ([Y i] is reserved for the
internal aperture interface). [Y0] is the diagonal matrix with the (identical) modal
admittances of the stub waveguides connected to the apertures:
0 1
Y0    0
B . . C
½Y0  ¼ @ .. Y0 .. A ¼ Y0 ½I ; (4.96)
0    Y0

and [I] is the identity matrix.

The excitation of the aperture array would be represented by the vector [Vþ]
if there were no coupling. With coupling we get

½V þ  þ ½V   ¼ ð½I  þ ½S Þ½V þ  ¼ 2ð½Y0  þ ½Y e Þ1 ½Y0 ½V þ  (4.97)

48 Slotted waveguide array antennas: theory, analysis and design

4.5.2.6 Calculating the mutual coupling

We have defined the mutual admittance between two slots m and n as the inner
product Ynm ¼ Ymn ¼ hHnext ðem Þ; en i. In general terms it can be expressed as the
electromagnetic reaction of the two sources through the scalar product (suitably
defined) of the field of one source on the current of the other [3, p. 340]. The
mathematical formulation can be written as [17]
ðð
1
Ymn ¼ E n  ^y  H m dS (4.98)
Vm Vn slot n

where the normal direction is assumed to be y-directed. E n is the field of aperture n

and H m is the magnetic field at aperture n due to the excitation of aperture m. The
integration is carried out over aperture n, requiring in total the evaluation of a
fourfold integral. The integral can be approximated by neglecting the field varia-
tions across the rectangular apertures in the E-plane dimension, which should work
well for narrow rectangular slots.
Further simplifications aiming at reducing the cost and time of numerical
evaluations have been developed; see Chapter 6 and [24].
We have already discussed the far-field radiation from aperture antennas
(Section 4.1.2). We can take the same approach starting from the vector potential
F , although we now need to include the near-field region. We will first derive the
magnetic field from a single slot. The electric vector potential is
ðð
e ejkR
F¼ Ms dS (4.99) = (4.2)
4p S R

where M s is the magnetic surface current M s ¼ 2^n  E a . The normal is y-directed

(see Figure 4.15). The slot electric field is x-directed and hence the magnetic current
is z-directed:

M s ¼ 2Ea^z (4.100)

The H-field is obtained from the electric vector potential F [1,3]:

1 1
H ¼ rE ¼ rrF (4.101)
jwm jwme

That is,
ðð
1 ejkR
H ¼ rr Ms dS (4.102)
4pjwm S R

We can expand
 
r  r  F ¼ r rF  r2 F (4.103)
 The slot antenna 49

a
x

z Ground plane

Figure 4.15 Slot apertures and major coordinate system

and, since F fulfils the wave equation,

r2 F þ k 2 F ¼ 0 (4.104)
we obtain,
 
r  r  F ¼ r rF þ k 2 F (4.105)
@ 2
F has only a z-component, so rðrF Þ ¼ 2 Fz . Therefore,
@z
 2 
1 @
Hz ¼ þ k 2
Fz (4.106)
jwme @z2
For this case it can be shown that derivation with respect to field coordinates is
the same as derivation with respect to source coordinates. Thus,
ðð  
1 @ 2 ejkR 0 0
Hz ðx; zÞ ¼ Ea ðx0 ; z0 Þ k 2 þ 0 2 dx dz (4.107)
2pjwm0 slot @z R
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where R ¼ ðx  x0 Þ2 þ ðz  z0 Þ2 .
Neglecting variations in the aperture electric field across the slot (x-direction)
we have
ð a=2  
b 0 @ 2 ejkR 0
Hz ðx; zÞ ¼ Ea ðz Þ k þ 0 2
2
dz (4.108)
2pjwm0 a=2 @z R

The electric field variation along the slot is Ea(z0 ) ¼ Ve10(z0 ), where e10(z0 ) is
the normalised modal function:
rffiffiffiffiffi
0 2 pz0
e10 ðz Þ ¼  cos (4.109)
ab a
and V is the slot voltage.
50 Slotted waveguide array antennas: theory, analysis and design

Hence, the magnetic field in an arbitrary point (x, 0, z) becomes

rffiffiffiffiffi ð a=2  0  
b 1 pz @ 2 ejkR 0
Hz ðx; zÞ ¼ V cos k þ 02
2
dz (4.110)
2a pjwm0 a=2 a @z R

A form suited for numerical evaluation is obtained after partial integration,

yielding
rffiffiffiffiffi
jkR1 
b 1 p e ejkR2
Hz ðx; zÞ ¼ V þ
2a p jwm0 a R1 R2
p2  ð a=2  0  jkR )
pz e
þ k  2
 cos dz0 (4.111)
a a=2 a R

The distances R1 and R2 are shown in Figure 4.16.

Weighting the field in slot #n with the normalised modal function and
assuming unit slot voltages (i.e. V ¼ 1) now yields the mutual admittance:
ðð rffiffiffiffiffi  
2 pz
Ynm ¼ hHnext ðem Þ; en i ¼ Hz ðx; zÞ cos dxdz (4.112)
slot#n ab a

Again, neglecting field variations in the x-direction as before, we get

rffiffiffiffiffið pz
2
Ynm ¼ hHnext ðem Þ; en i ¼b Hz ðzÞcos dz (4.113)
ab slot#n a

z'

R1 z
x'

R
(x', z')

Field point
#m R2
Source (x, z)

#n

Figure 4.16 The two-slot geometry for calculating the magnetic field in slot #n
due to coupling from the source slot #m. See also Section 6.4
 The slot antenna 51

and finally

jb
ð pz
p ejkR1 ejkR2

Ynm ¼ cos þ
pawm0 #n a a R1 R2
(4.114)
p2  ð  0  jkR 
pz e
þ k2   cos dz0 dz
a #m a R

In the literature about slotted waveguide array design, the dimension-less

mutual coupling parameter gmn is often introduced [19]. The mutual admittance Ymn
is related to gmn as follows:
jb
Ymn ¼ gmn (4.115)
pah
We need to evaluate the admittance for all unique mn combinations of an array
in order to fill the complete admittance matrix. However, due to the Toeplitz
characteristics, many relative positions repeat. For instance, in a linear array of
N elements we only need to calculate the mutual admittances between element 1
and elements 2 to N. From this row we can fill the complete N  N admittance
matrix. The scattering matrix is then given by

½S  ¼ ð½Y0  þ ½Y e Þ1 ð½Y0   ½Y e Þ (4.116) = (4.95)

The Toeplitz simplification is not valid if the array lattice is not exactly peri-
odic and the slot dimensions are not identical. The consequence is that many more
couplings will have to be calculated. However, the end result is still a scattering
matrix [S] that characterises the array.
It is usually necessary to include the mutual coupling in the design of slotted
waveguide array antennas, but sometimes the coupling can be seen as a minor
correction or be disregarded completely. More about this in Chapters 6 and 7. The
following examples for slot aperture coupling in the E-plane and H-plane illustrate
the magnitude of coupling for several cases. For these calculations periodic arrays
of equal size apertures were assumed.

4.5.2.7 Examples
The mutual coupling between two elements, that is, with no other elements present,
is defined as isolated coupling, just two slots and a common ground plane
(Figure 4.17). In the full array, coupling between two slots will be influenced by the
presence of the other slots, array coupling. One effect is that the other slots will
absorb some power, thus reducing the coupling values. See the following examples
showing isolated and array coupling for a linear array. The calculated results have
been obtained with one aperture mode only, which appears to be sufficient in many
cases. The variations across the slot width have not been included. However, a
correction factor  sinc2(x) has been employed for the E-plane results [25]. This is
not needed for more narrow slots (Figure 4.18).
52 Slotted waveguide array antennas: theory, analysis and design

–15
Computed
–20 Measured
Coupling dB

–25

–30

–35
1 2 3 4 5 6 7 8 9 10
Element number
–120
–125
Coupling degrees

–130
–135
–140
Computed
–145 Measured
–150
1 2 3 4 5 6 7 8 9 10
Element number

Figure 4.17 The magnitude and phase of the isolated coupling (two elements) for
a 10 element E-plane array. The free space delay (phase) has been
subtracted. Element no. 1 is fed. Aperture size: 0.75l  0.20l

–15
Computed
–20
Measured
Coupling dB

–25
–30

–35

–40
1 2 3 4 5 6 7 8 9 10
Element number

–140
–145 Computed
Coupling degrees

Measured
–150
–155
–160
–165
–170
1 2 3 4 5 6 7 8 9 10
Element number

Figure 4.18 The magnitude and phase of the array coupling (all elements) for the
same array as in Figure 4.17. The free space phase delay has been
subtracted. Element no. 1 is fed
 The slot antenna 53

–7.8

–7.85

–7.9

–7.95

–8
S11/dB/

–8.05

–8.1

–8.15

–8.2

–8.25

–8.3
1 2 3 4 5 6 7 8 9 10
Second slot position no.

Figure 4.19 Reflection |S11| in dB of slot no. 1 with a single parasitic slot
at increasing positions 2–10. Dashed line shows also
intermediate positions. Dash-dotted (straight) line: the isolated
reflection. E-plane

We find here that the scattering parameters, that is, elements of the scattering
matrix, do not only depend on the apertures m and n, but also on the surrounding
apertures. Even the self-reflection, that is, the S11 element in the scattering matrix,
is shown in Figure 4.19 to depend on the distance to a second, parasitic slot (only
two apertures present in this case). See also [26]. We have in this situation

Y02  Y11
2
þ Y12
2
S11 ¼ (4.117)
ðY0 þ Y11 Þ2  Y12
2

which becomes the conventional isolated (one element only) scattering if Y12 ? 0.
In terms of scattering parameters we can also write
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
0
S11 1 1 þ S11 0
S11 ¼ þ þ S12 2 (4.118)
2 2
0
where S11 stands for the scattering of the single (isolated) element.
Coupling in an H-plane linear slot array is shown in Figures 4.20 and 4.21.
In the H-plane example the phase agreement (measured vs. calculated) for
array coupling is not as good as in the previous examples (E-plane). It has been
found that higher-order modes are excited in the apertures and should be included if
better agreement is sought. However, the coupled levels are very low so this effort
is probably not needed in practice. As we can see from these results, the H-plane
54 Slotted waveguide array antennas: theory, analysis and design

–20
–30 Computed
Measured
Coupling dB

–40
–50
–60
–70
–80
0 2 4 6 8 10 12 14 16
Element number
50
Computed
Coupling degrees

Measured

–50
0 2 4 6 8 10 12 14 16
Element number

Figure 4.20 The magnitude and phase of the isolated coupling (two elements)
for a 16-element H-plane array. Element no. 1 is fed. Aperture
size: 0.73l  0.30l

–20
–30 Computed
Measured
Coupling dB

–40
–50
–60
–70
–80
0 2 4 6 8 10 12 14 16
Element number

50
Computed
Coupling degrees

Measured

–50
0 2 4 6 8 10 12 14 16
Element number

Figure 4.21 The H-plane array coupling for the same array as in Figure 4.20
 The slot antenna 55

Figure 4.22 In a 2-D array mutual coupling is a multiple coupling phenomenon

Image

Edge

Ground plane

Figure 4.23 Edge reflection effect seen as a contribution from an image slot

coupling is very low for rectangular slots. This also leads to the interesting fact that
isolated and array H-plane coupling (for linear arrays) are at about the same level.
However, in two-dimensional arrays, the ‘H-plane coupling’ can seek other ways
for the interaction. Figure 4.22 shows how elements m and n can interact via several
of the elements in their surroundings.

4.5.3 Finite ground plane effects

The edges of a finite ground plane will cause diffraction effects as we have already
seen in the radiation pattern (cf. Figure 4.12). The effect is of course strongest when
the edge illumination is strong as is typically the case in the E-plane. There is also a
scattering from the edges back to the slots in the ground plane, thus modifying the
array excitation and mutual admittance to some extent. This has been studied by
several authors [27–30].
In an experiment the coupling between two slots in a finite ground plane with
edges was measured [29]. A comparison was made between the cases with the
edges covered by absorbing material and with the edges exposed (not covered). The
geometry and examples of measured and calculated isolated coupling are shown in
Figures 4.23–4.25.
56 Slotted waveguide array antennas: theory, analysis and design

Edge

Ground plane
m n

Edge (covered with absorbers)

Figure 4.24 Set-up for measuring the edge effect on mutual coupling between two
aperture elements m and n. The other apertures were covered with
conducting tape. From [29]. Reproduced by permission of the
Institution of Engineering & Technology

–20
–30
–40
Coupling dB

–50
–60
–70
–80
–90
–100
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
(a) Element number

360
320
280
Coupling degrees

240
200
160
120
80
40
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
(b) Element number

Figure 4.25 Measured (dashed line) and computed (solid line) mutual coupling
including the effect of edge diffraction. The no edge case (ground plane
edges covered with absorbing material) is shown as a dash-dotted line.
(a) is magnitude and (b) is phase. From [29]. Reproduced by
permission of the Institution of Engineering & Technology
 The slot antenna 57

The theoretical calculation is based on geometrical diffraction. Rays from the

source slot are reflected by the exposed straight edge. In principle we introduce a
mirror image of one of the slots. We get coupling contributions from the direct path
and from the reflected path that add together. The reflected contribution is weighted
by an edge diffraction coefficient [31]. In this case the direct path is in the H-plane
and thus the edge influence is clearly visible in the results. More details are found
in the references.

References

[1] Collin R. E. Field Theory of Guided Waves. IEEE Press, New York, 2nd ed.,
1991.
[2] Balanis C. A. Antenna Theory, Analysis and Design. Wiley, New York,
1997.
[3] Harrington R. F. Time-Harmonic Electromagnetic Fields. IEEE Press, New
York, 2001.
[4] Elliott R. S. Antenna Theory and Design. Prentice-Hall, Englewood Cliffs,
NJ, 1981.
[5] Clemmow P. C. The Plane Wave Spectrum Representation of Electro-
magnetic Fields. Pergamon Press, Oxford, 1966.
[6] Rhodes D. R. Synthesis of Planar Antenna Sources. Clarendon Press,
Oxford, 1974.
[7] Kraus J. D. Antennas. McGraw-Hill, New York, 2nd ed., 1988.
[8] Selby S. M. (ed.). Standard Mathematical Tables. The Chemical Rubber
Co., Cleveland, OH, 1973.
[9] Booker H. G. ‘Slot aerials and their relation to complementary wire aerials
(Babinet’s principle)’. Journal of the Institution of Electrical Engineers.
Part. IIIA. 1946;93(4):620–626.
[10] Collin R. E. Foundations for Microwave Engineering. McGraw-Hill, New
York, 1966.
[11] Pathak P. H. ‘On the Eigenfunction expansion of electromagnetic dyadic
Green’s functions’. IEEE Transactions on Antennas and Propagation.
1983;31(6):837–846.
[12] Van Bladel J. ‘Contribution of the Y¼constant mode to the modal expansion in
a waveguide’. IEE Proceedings H – Microwaves, Antennas and Propagation.
1981;128(5):247–251.
[13] Khac T. V., Carson C. T. ‘m¼0 n¼0 mode and rectangular-waveguide slot
discontinuity’. Electronics Letters. 1973;9(18):431–432.
[14] Collin R. E., Zucker F. J. Antenna Theory, Part I. McGraw-Hill, New York,
1969.
[15] Stevenson A. F. ‘Theory of slots in rectangular waveguides’. Journal of
Applied Physics. 1948;19(1):24–38.
[16] Stegen R. J. ‘Slot radiators and arrays at X-band’. IRE Transactions on
Antennas and Propagation. 1952;1(1):62–84.
58 Slotted waveguide array antennas: theory, analysis and design

[17] Borgiotti G. V. ‘A novel expression for the mutual admittance of planar

radiating elements’. IEEE Transactions on Antennas and Propagation.
1968;16(3):329–333.
[18] Elliott R. S. ‘The design of small slot arrays’. IEEE Transactions on
Antennas and Propagation. 1978;26(2):214–219.
[19] Elliott R. S. ‘An improved design procedure for small arrays of shunt slots’.
IEEE Transactions on Antennas and Propagation. 1983;31(1):48–53.
[20] Josefsson L., Moeschlin L., Svensson T. ‘A stripline flat plate antenna
with low sidelobes’, Proceedings of the IEEE Antennas and Propagation
Symposium, Atlanta, GA, USA, 1974, pp. 282–285.
[21] Amitay N., Galindo V., Vu C. P. Theory and Analysis of Phased Array
Antennas. Wiley Interscience, New York, 1972.
[22] Wasylkiwskyj W., Kahn W. K. ‘Theory of mutual coupling among
minimum-scattering antennas’. IEEE Transactions on Antennas and
Propagation. 1970;18(2):204–216.
[23] Harrington R. F. Field Computation by Moment Methods. The MacMillan
Co., New York, 1968.
[24] Rengarajan S. R., Gabrelian E. ‘Efficient and accurate evaluation of external
mutual coupling between compound broad wall slots’. IEEE Transactions on
Antennas and Propagation. 1992;40(6):733–737.
[25] Josefsson L. ‘Mutual coupling effects on the performance of finite radar
antenna arrays’. Proceedings of the International Conference on Radar.
Paris, May 3–6, 1994, pp. 713–716.
[26] Mailloux R. J. ‘Radiation and near-field coupling between two collinear
open-ended waveguides’. IEEE Transactions on Antennas and Propagation.
1969;17(1):49–55.
[27] Bailey M. C. ‘Mutual coupling between circular waveguide-fed apertures in
a rectangular ground plane’. IEEE Transactions on Antennas and Propagation.
974;22(4):597–599.
[28] Commatas C. G., Sahalos J. N. ‘Mutual coupling and radiation patterns of
two slots asymmetrically located on a square plate’. IEEE Transactions on
Antennas and Propagation. 1983;31(3):531–535.
[29] Josefsson L. ‘Mutual coupling calculations including edge effects’. Elec-
tronics Letters. 1994;30(25):2087–2088.
[30] Mazzarella G., Panariello G. ‘Evaluation of edge effects in slot arrays using
the geometrical theory of diffraction’. IEEE Transactions on Antennas and
Propagation. 1989;37(3):392–395.
[31] Kouyoumjian R. G. ‘The geometrical theory of diffraction and its application’,
in R. Mittra (ed.). Numerical and Asymptotic Techniques in Electromagnetics.
Springer-Verlag, Berlin Heidelberg, Chapter 6, 1975.
 Chapter 5
Slot models

In this chapter we will discuss theoretical models for characterising slots in

waveguides. In Chapter 4 we looked at basic theories for slot apertures and found
expressions for the conductance of longitudinal slots in rectangular waveguides.
We will now look at ways of finding the slot susceptance, the effect of slot offset
on resonance conditions, influence of wall thickness, waveguide dimensions, etc.
We will demonstrate how electromagnetic models can tell us about slot behaviour
in greater detail. The longitudinal slot and the transverse slot in the broad wall of a
rectangular waveguide will be treated in particular. We will thus establish a basis
for an efficient design process of high-performance slotted waveguide arrays.

5.1 Modelling principles

5.1.1 Using experimental data
In Chapter 4 we derived the slot conductance for resonant slots. The assumption was
that the slot length l/2 automatically results in slot resonance and that the slot
admittance is real. However, it has been found that resonance also depends on the
slot shape, the waveguide dimensions and the mutual coupling among the slots in an
array. We are also interested in the complete slot admittance and how it varies over a
band of frequencies. A more detailed analysis is therefore necessary in order to
control these parameters and optimise the slot (array) antenna for best performance
including bandwidth. Although much theoretical work was performed on slot
antennas in the years following the Second World War by several authors [1–5], the
design of slotted waveguide array antennas relied in those days to a large degree on
experimental data.
At Hughes Aircraft Co. in California, USA, R. J. Stegen made careful mea-
surements on longitudinal (shunt) slots in rectangular waveguides [6]. He measured
the slot admittance for various slot lengths and slot offsets and assembled the data
in a set of normalised curves that can be used for array design. He introduced the
normalised parameters:
Slot length normalised to the resonant length y ¼ L/Lr
Slot resonant length normalised to the free space wavelength Lr/l0
Slot conductance normalised to the resonant conductance h1 ¼ G/G0
Slot susceptance normalised to the resonant conductance h2 ¼ B/G0
60 Slotted waveguide array antennas: theory, analysis and design

Stegen used the following waveguide and slot dimensions (standard X-band
waveguide):
Slot width 0.0625 inch (1.59 mm)
Waveguide width 0.9 inch (22.86 mm)
Waveguide height 0.4 inch (10.16 mm)
Waveguide wall thickness 0.05 inch (1.27 mm)
Frequency 9.375 GHz
The normalised resonant conductance G/G0 ¼ g(x) as a function of offset (x)
was already introduced in Chapter 4 (Figure 4.11). Additionally, the resonant
length as a function of offset is also needed. The data was presented in curves, and
examples are shown in Figure 5.1(a) and (b).
Figure 5.1(a) h1(L/Lr) and h2(L/Lr) conductance and susceptance vs. slot length
Figure 5.1(b) Lr/l0 resonant length versus offset
Figure 4.11 g(x) normalised conductance versus offset
The curves appear to be quite universal for the given waveguide dimensions.
The data can easily be fitted to polynomials to be used in the array design iterations.
However, new dimensions (half height waveguide, etc.) would require new
measurements.

5.1.2 Equivalent circuit approach

In Chapter 4 the longitudinal slot, as seen from the feeding waveguide, was represented
by an equivalent shunt admittance on a transmission line. This model was plausible
under the assumption that the slot electric field was symmetrical (cosinusoidal
distribution) with equal scattering forward and backward from the slot (Figure 5.2).
The reflection coefficient G is easily measured, and from G the shunt admit-
tance is obtained:
Y =G0 ¼ 2G=ð1 þ GÞ (5.1)
In 1957 Oliner [4] presented a variational solution with equivalent circuits for
shunt and series slots in rectangular waveguides. A correction for wall thickness
was proposed by Yee [5]. The variational approach did not include the effect of slot
offset from the centreline on the resonant length. However, the model was used
successfully in combination with measured data. Additional parameters were
introduced by Lagerlöf and Josefsson [8] (see Figure 5.3).
The parameters in Figure 5.3 represent

B0 Residual susceptance for centred slot

nd Slot displacement coupling factor
Ysext Slot external admittance
Bsint Slot internal susceptance
t Waveguide wall thickness
Z ⇄Y Impedance/admittance converter
 Slot models 61

Ratio of conductance to resonant conductance, G/Gr

Ratio of susceptance to resonant conductance, B/Gr
0.6
Offsets (inches)
h2 x = 0.029
0.4 x = 0.054 1.0
x = 0.074
x = 0.104
0.2 0.8
x = 0.127
x = 0.153
0 0.6

–0.2 0.4

–0.4 a = 0.900 in. b = 0.400 in. 0.2

t = 0.050 in. w = 0.0625 in.
h1
f = 9.375 GHz
–0.6 0
0.90 0.95 1.00 1.05 1.10
(a) Ratio of slot length to resonant length

Slot width = 0.0625 inch

a = 0.900 inch
b = 0.400 inch
0.500 t = 0.050 inch
2lr /λ0

0.490

Experimental points determined by

admittance measurements.
Experimental points from radiation
pattern measurements.

0.480
0.050 0.100 0.150 0.200 0.250
(b) Slot displacement off waveguide centreline, x inches

Figure 5.1 (a) Normalised admittance versus normalised slot length. (b) Resonant
slot length versus slot offset from centreline. (a, b)  1981
J. H. Elliott. Reprinted from [7] with permission

The outer slot admittance Ysext could be determined theoretically [9]. The other
parameters (Bsint, nd and B0) were determined from a set of experimental data on
slot length and offset for several frequencies. Polynomials could then be derived
from the experimental values and subsequently be used in array designs (within the
range of the experimental data).
62 Slotted waveguide array antennas: theory, analysis and design

Y G0
Γ

Figure 5.2 The equivalent shunt admittance

Z Y
jB0 0 j jBsint Ysext
Z=
j 0
1 : nd t

Figure 5.3 Equivalent circuit for a shunt slot in a rectangular waveguide.  1990
Microwave Exhibitions and Publishers Ltd. Reproduced from [8] by
permission of Microwave Exhibitions and Publishers Ltd

Figure 5.4 Test equipment for slot measurements. A wedge absorber and a
waveguide short in the foreground

A test fixture for measuring slot characteristics is shown in Figure 5.4. The
fixture is calibrated using a matched load (wedge absorber) and a moveable short.
Various slot plates with different slot dimensions could be inserted and the offset
carefully monitored with the micrometre shown. The example is from an evaluation
of data for slots in a ridge loaded waveguide where theoretical models are some-
what complex [10].

5.1.3 Electromagnetic models

Theoretical calculations can in principle replace the tedious and costly collection of
experimental data. From the theoretical results polynomials can be defined and
used in the array design just as described above when equivalent circuits were
 Slot models 63

postulated. However, the EM modelling can provide more information such as

fields and currents on the physical structure. When this has been done we can find,
for example, the electric field distribution in the slot aperture (e.g. by solving a
boundary value problem, more about this later) instead of just assuming a particular
distribution. For slot antennas, several authors have contributed in this area.
Khac [11] solved the integral equation for the electric field by the Method of
Moments using pulse expansion and point matching. Lyon and Sangster [12] used
an entire basis with sinusoidal functions and concluded that as few as two basis
functions were sufficient for ‘power calculations’.
For the design of slotted waveguide arrays the resonant length must be known
quite accurately. The slot is by itself narrow band with the bandwidth further
reduced in the array by the mutual coupling among the slots [13,14]. An accuracy
of about half a per cent or better is desired. This is achievable using moment
method techniques as was demonstrated by Stern and Elliott [15]. They compared
theoretical results to carefully obtained experimental data for slots in thin-walled
waveguide with standard inner dimensions. Josefsson [16] presented Moment
Method results and measurements for standard and reduced height waveguides
including wall thickness.

5.1.3.1 The boundary problem

In Figure 5.5 four related boundary problems are illustrated. They can be solved
using similar moment method techniques. No. 1 shows a radiation slot fed from a
waveguide (bold arrow). The two thin arrows in the waveguide represent forward
and backward scattering. In example no. 2 we have coupling between two wave-
guides. No. 3 is the slot antenna in a receive situation, and no. 4 shows a slot
aperture in a metallic screen.

1 2

3 4

Figure 5.5 Four closely related boundary value problems

64 Slotted waveguide array antennas: theory, analysis and design

H ext
z Region e

t Ex Hz

H inc
z Region i
H zint

(a) (b)

Figure 5.6 (a) The radiating slot problem.  IEEE 1987. Adapted from [16], with
permission. (b) Major slot field components

w
t
X" Z

X'
2L

b
a

Figure 5.7 The slot geometry. 2L is the slot length and w the slot width.  URSI
1986. Reprinted from [33], with permission

The four problems in Figure 5.5 are all about a slot interface between two
regions which we call the external (superscript e or ext) and the internal (super-
script i or int); cf. Figure 5.6(a).

5.1.3.2 Zero wall thickness

Let us assume that the wall thickness t is very small and can be disregarded and that
we only need to consider the major Hz and Ex slot field components, Figure 5.6(b).
We will introduce finite wall thickness later. Problem no. 1 is first analysed with
coordinates as shown in Figure 5.7.
With an incident field Hzinc we have the boundary condition for the magnetic
field:

Hze ðEx Þ  Hzi ðEx Þ ¼ Hzinc (5.2)

 Slot models 65

where Ex is the unknown transverse electric field in the slot, expanded in basis
functions:

X
N
Ex ðx0 ; z0 Þ ¼ Ep ep ðz0 Þ (5.3)
p¼1

The solution to (5.2) is found in a projection sense according to the Moment

Method [17], that is, we form inner products with testing functions es(z) resulting in
a system of equations:

hHze ðEx Þ; es i  hHzi ðEx Þ; es i ¼ hHzinc ; es i (5.4)

or, in matrix form:

 e  
½Y   Y i ½E ¼ ½h (5.5)

The matrices [Y e], [Y i] have elements

  )
Ypse ¼ hHze ep ; es i
  (5.6)
Ypsi ¼ hHzi ep ; es i

The excitation vector is

½h ¼ hHzinc ; es i (5.7)

ep(z) and es(z) are the basis and testing functions, respectively. The solution to (5.2)
is in the form of (5.3) with coefficients found in the vector
½ E  ¼ ðE1 ; E 2 ; . . . ; E N Þ (5.8)
obtained as
  1
½E  ¼ ½Y e   Y i  ½h (5.9)

5.1.3.3 Choice of basis and testing functions

Simple functions such as pulse basis (expansion) and point testing have been used.
Very complex functions can be modelled with these. However, at the slot ends the
electric field must go to zero and this is not well represented by pulses. Khac
[11] somewhat arbitrarily added one pulse of length 0.02 l0 to allow for the
expected electric field variation down to zero at the slot ends. A sinusoidal entire
basis may appear more attractive in this respect. In fact, just one term may
sometimes be sufficient. No more than 10 terms have been found necessary even
for accurate resonant length predictions (<0.5% error) [16]. More about the
behaviour of fields near conducting edges (edge conditions) will be discussed in
Section 5.3.
66 Slotted waveguide array antennas: theory, analysis and design

The results presented in this chapter have been obtained with entire basis and
testing functions:
pp 9
basis function ep ðzÞ ¼ sin ðL þ zÞ >=
2L
(5.10)
sp >
testing function es ðzÞ ¼ sin ðL þ zÞ ;
2L
That is, the Galerkin case [17].

5.1.3.4 The other slot problems in Figure 5.5

No. 2: coupling between two waveguides.
Provided the waveguides are identical we have Hzi1 ¼ Hzi2 and thus
1 1
½E  ¼  Y i  ½h (5.11)
2
No. 3: waveguide slot with plane wave incident.
Only a sign change when the excitation has moved:
  1
½E  ¼ Y i  ½Y e   ½h (5.12)
No. 4: slot in screen – no waveguide.
Symmetrical case like no. 2. Hze1 ¼ Hze2 , thus
1
½E ¼  ½Y e 1  ½h (5.13)
2
It is true that the excitation vector is not the same for waveguide mode excitation
and plane wave excitation. However, the difference is not very large; more about
this in Section 9.8.

5.1.4 Finite wall thickness

A rectangular slot in a waveguide wall of thickness t can be seen as a short rectangular
waveguide by itself (Figure 5.8). The field inside the stub waveguide is described as a
superposition of TEi0 modes. We neglect variations across the slot since the wave-
guide height (¼ slot width w) is assumed to be narrow. Our problem can be broken
down into three partial boundary problems as illustrated in Figure 5.8(a)–(c).
We are concerned with matching the tangential fields at the upper (I/IIa) and lower
(IIb/III) interfaces simultaneously. As in the case with zero wall thickness we only
consider the Ex and the Hz field components at the two interfaces; cf. Figure 5.6(b). Our
approach is based on solving the three partial problems formally and then couple the
solutions via the mode expansion inside the stub waveguide.
The upper (external) interface is characterised by a scattering matrix [Re]
relating the downward [E] and upward [Eþ] propagating modes inside the
waveguide:

½ E   ¼ ½ Re  ½ E þ  (5.14)
 Slot models 67

I Halfspace

Y IIa E+
i E–i

Stub
waveguide

(a)

Y
Stub
waveguide

IIb E+
i E–i

III
Main waveguide

(b)

Y
Stub
waveguide

IIb E+
i

III
Hzinc Main waveguide

(c)

Figure 5.8 Slot in thick waveguide wall. The three partial problems (a)–(c).
 URSI 1986. Reprinted from [33], with permission

Likewise, for the lower (internal) interface, we can write

 
½ E þ  ¼ Ri ½ E   (5.15)
These scattering matrices can be written out using appropriate quantities
already defined for the zero thickness case. To summarise we obtain, for example,
for the ith column of the matrix [Re]
 
½Re i:th column ¼ ½Y ew 1  hew
i (5.16)
where the element number (p, s) of the matrix [Y ew] is
ew
Yps ¼ Ypse  Y0 wLdps (5.17)

[Y e] was defined in (5.6), Y0 is the waveguide modal admittance and dps is the
Kronecker delta function. The vector ½hewi  has elements

is ¼  < Hz ðei Þ; ws > Yi wLdis

hew ext
(5.18)
68 Slotted waveguide array antennas: theory, analysis and design

The excitation by a unit incident mode in the main waveguide results in the
formal solution
 1  inc 
½Eþ  ¼  Y iw  h (5.19)
where the matrix [Y iw] has elements

Ypsiw ¼ Ypsi þ Y0 wLdps (5.20)

[Y i] was defined in (5.6) and [hinc] in (5.7). To this solution, (5.19), we will of course
have to add those waves resulting from the E waves incident on the lower interface.
It remains to relate the waveguide modes at the upper interface to those at the
lower interface:

½Eþ upper interface ¼ ½B ½Eþ lower interface (5.21)

and similarly for the downward propagating modes. [B] is a diagonal matrix with
the elements
Bii ¼ expðjbi tÞ (5.22)
where bi is the propagation constant of the TEi0 mode and t is the wall thickness.
Combining everything together we obtain the total tangential electric field at the
lower interface
  1  iw 1  inc 
½E ¼ f1 þ ½B½Re ½Bg Ri ½B½Re ½B  1 : Y h (5.23)
This is the final solution for the field in the slot aperture (lower interface).

5.2 Integral equation solution

In the preceding section, we solved the slot admittance problem in a formal sense.
We have now to derive the appropriate field expressions and compute the matrices
required for the Moment Method solution. The external field is due to an equivalent
magnetic current source in the upper slot aperture. Assuming the ground plane to be
large we can describe this field by a free space Green’s function; cf. Section 4.5.2.
The internal field is a bit more complicated so we start with this.

5.2.1 The internal field

Field solutions for rectangular waveguides are traditionally expressed in terms
of waveguide modes (Section 4.3.2). A complication occurs of course when there
is an aperture (slot) in the waveguide wall. Stevenson [2] described the internal
field including the slot region with the aid of a Green’s function G2(P, P0 ) [18].1

1
Green’s function ¼ ‘field of a point source’ [32]. Georg Green was a British mathematician who in
1828 published ‘An Essay on the Application of Mathematical Analysis to the Theories of Electricity
and Magnetism’ [18].
 Slot models 69

In our case we need the Hz field (TE mode) when there is an Ex field exciting the
slot. Stevenson’s expression for this case is
ð  2
1 0 @
Hz ðPÞ ¼
int
E x ðP Þ þ k G2 ðP; P 0 ÞdS 0
2
(5.24)
jwm0 S @z0 2
Complete expressions including other field components are found in, for
example, [19]. See also Section 2.8.
In (5.24) S 0 denotes the slot region, P and P 0 refer to points in the slot, prime
indicates source points, unprimed field points. The chosen Green’s function is
1 X
X 1 0
egmn jzz j
G2 ðP; P0 Þ ¼ ymn ðx0 ; y0 Þymn ðx; yÞ (5.25)
m¼0 n¼0
2gmn

where ymn(x, y) is the TE modal function

pffiffiffiffiffiffiffiffiffi
em en mpx npy
ymn ð x; y Þ ¼ pffiffiffiffiffi cos cos (5.26)
ab a b
Here em ¼ 1 if m ¼ 0, em ¼ 2 if m > 0. Similarly for en (the Neumann factor). gmn is
the propagation constant:
mp2 np2
gmn 2 ¼ þ  k2 (5.27)
a b
Putting together (5.24) and (5.25) we have
ð  2
1 X 1 X 1
ymn ðx; yÞ @ 0
Hzint ðPÞ ¼ Ex ðP 0 Þymn ðx0 ; y0 Þ 02
þ k 2
egmn jzz j dS 0
jwm0 m¼0 n¼0 2gmn S @z
(5.28)
Note that the summation includes the TE00 mode. Normally, this mode is dis-
regarded in waveguide problems since it does not propagate. However, in the source
region it is essential. This has caught the attention of several authors [11,20–23].
We assume that the slot is in the upper broad wall, consequently y ¼ y0 ¼ b.
Furthermore, we can separate Ex(P 0 ) ¼ Ex(x0 , z0 ) ¼ El(x0 )  E2(z0 ) ¼ E2(z0 ); we assume
that the electric field has no variation across the slot. We will discuss edge conditions
in Section 5.3.
The derivative in 5.28 has to be handled with care when z0 ¼ z, that is, for points
in the source region. Using the Dirac delta function concept we can write [21]
@2 0   0

02
egjz zj ¼ g2  2gdðz0  zÞ egjz zj (5.29)
@z
With this we get
2 ð 
1 X 1 X 1
k þ gmn2 0 gmn jzz0 j 0
Hzint ðPÞ ¼ em en Fm ðxÞ E2 ðz Þe dz  2E2 ðzÞ
4jwm0 m¼0 n¼0 gmn S

(5.30)
70 Slotted waveguide array antennas: theory, analysis and design

where
ð w=2
mpx mpðx0 þ x0 Þ 0
Fm ðxÞ ¼ cos E1 ðx0 Þcos dx (5.31)
a w=2 a

Here x0 is the slot offset from the sidewall. With no variations across the slot
we have
mpx mpx0 mpw
Fm ðxÞ ¼ w cos cos sinc (5.32)
a a 2a
Finally, with the expansion of the electric field variation along the slot; cf. (5.10):

X
N
pp
E2 ðzÞ ¼ Ep sin ðL þ zÞ (5.33)
p¼1
2L

We arrive at the following expression for the internal Hz field:

1 X 1 X 1
Hzint ðx; zÞ ¼ em en Fm ðxÞ ...
jwm0 ab m¼0 n¼0
8
X >
< pp
N
gmn 2 þ k 2  
 Ep  pp2  egmn ðLþzÞ  egmn ðLzÞ
p¼1
>
:4Lgmn g 2 þ
mn
2L
pp2 9
k2  pp >
=
þ 2L ðL þ zÞ (5.34)
pp2 sin
2L >
;
gmn 2 þ
2L
The upper sign in (5.34) is valid for p odd, the lower for p even.

5.2.2 The external field

The magnetic field Hz(x, z) due to the electric aperture field Ea(x0 , z0 ) for an aper-
ture in an infinite ground plane was already derived in Chapter 4. We can start from
(4.107):
ðð 
1 0 0 @2 ejkR 0 0
Hzext ðx; zÞ ¼ E x ðx ; z Þ k þ 2
dx dz (5.35) = (4.107)
2pjwm0 slot @z0 2 R
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2ffi
The distance R ¼ x  x0 þ z  z0 . The Green’s function in (5.35) is a
spherical wave, which can be transformed into a spectrum of plane waves [24,25]:
ð1 ð1
ej½kx ðxx Þþkz ðzz Þþky y
0 0
ejkR 1
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dkx dkz (5.36)
R 2pj 1 1 k 2  kx2  kz2
 Slot models 71

We insert this in (5.35), interchange the order of integration and evaluate the
derivative with the following result:
ð ð
1 1 1 k 2  kz2
Hzext ðx; zÞ ¼  gx ðkx ; kz Þejðkx xþkz zÞ dkx dkz (5.37)
2pwm0 1 1 ky
where gx (representing the radiation pattern):
ðð
1 0 0
gx ðkx ; kz Þ ¼ Ex ðx0 ; z0 Þejðkx x þkz z Þ dx0 dz0 (5.38)
2p slot
With the expansion of the electric slot field as before (5.33) with constant field
across the slot we get finally
X ð ð   p odd
NP
Ep wp 1 1 kz2  k 2 sincðkx w=2Þ cos kz L
Hz ðx; zÞ ¼
ext  pp2   j sin kz L
wm0 4pL 1 1
p¼1 ky kz2  p even
2L
 ejðkx xþkz zÞ dkx dkz (5.39)

5.2.3 Matrix elements

The matrix elements are normalised as shown next
wm0
Yps ¼ hHz ðEp Þ; es i (5.40)
kEp

5.2.3.1 Internal field contribution

Forming the inner product we obtain the elements of the matrix [Y i]

lw2 X 1 X 1
em en
Ypsi ¼  Wm  . . .
pab m¼0 n¼0 4
8
> jpp ps
< gmn2 þ k 2 2
2L 2L
   2   ps2  ðe2gmn L  1Þ
>
:gmn g 2 þ pp
mn gmn þ
2
2L 2L
pp2 9
k2  >
=
2L
þ 2j  pp2  L  dps (5.41)
>
;
gmn2 þ
2L
The function Wm is related to the weighting across the slot. With constant field
we have
mpx  mpw
0
Wm ¼ cos2 sinc2 (5.42)
a 2a
where x0 is the slot offset from the side wall as before.
72 Slotted waveguide array antennas: theory, analysis and design

In (5.41) the upper sign is valid for both p and s odd, the lower sign for both p and s
even. When p is odd and s even or vice versa Ypsi ¼ 0. dps ¼ 1 for p ¼ s, 0 otherwise.

5.2.3.2 External field contribution

We obtain the matrix element Ypse
 2 
 2  2 kx w cos kz L p; s odd
ð ð k  k 2
 sinc 
lpsw2 1 1 z
2 sin2 kz L p; s even
Ypse ¼      ps 2   dkx dkz
8pL2 1 1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pp 2
k 2  kx2  kz2 kz2  kz2 
2L 2L
(5.43)

5.2.3.3 Excitation vector

The excitation vector becomes

ps px0 sin b10 L s even
cos 
L a j cos b L s odd
hinc
s ¼ hHzinc ; es i ¼  ps2
10
(5.44)
b102 
2L
where
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p
b10 ¼ ð2a=lÞ2  1 (5.45)
a

5.3 Longitudinal slot characteristics

5.3.1 Slot electric field distribution
The integral equation solution has proved to be very useful for studies of the
waveguide slot characteristics. A few results will be discussed here, starting with
the electric field distribution in the slot. The results are naturally to a great degree
dependent on the proper selection of basis and testing functions in the Moment
Method. In Section 5.2 we made the following choice:
pp 9
basis function ep ðzÞ ¼ sin ðL þ zÞ > =
2L
(5.46) = (5.10)
sp >
testing function es ðzÞ ¼ sin ðL þ zÞ ;
2L
Going back to the assumption of a shunt admittance model for the longitudinal
slot, we have the relation between the reflection coefficient, G ¼ S11, that is, back
scattering, and the slot admittance:

Y =G0 ¼ 2S11 =ð1 þ S11 Þ (5.47) = (5.1)

 Slot models 73

0.70
Forward scattering

Back scattering
0.65
a = 22.86 mm
w = 1.59 mm
0.60 t =0
2*L/Lambda

f = 9.375 GHz

Half height
0.55

Full height
0.50

0.45
0 1 2 3 4 5 6 7
Offset from centreline (mm)

Figure 5.9 Computed resonant length based on the backscattered and forward
scattered waves. Two waveguide heights are compared. Zero wall
thickness.  IEEE 1987. Reprinted from [16], with permission

Based on the forward scattering (S12) instead, still assuming a shunt repre-
sentation, we would get
Y =G0 ¼ 2ð1  S12 Þ=S12 (5.48)
The two expressions give the same admittance for a pure shunt element since
S12 ¼ 1 þ S11 in that case. Once we have found the electric field distribution in the
slot we can compute the backward and forward scattered wave amplitudes, for
example, from (4.70), and from them calculate the assumed ‘shunt’ admittance
values. Resonance is then defined from the admittance being pure real. An example
is shown in Figure 5.9.
From this example it is clear that the scattering off the slot is not perfectly
symmetrical. It has been found [11] that the electric field variation along the slot
has an asymmetrical (odd) component as well as a symmetrical (even) component.
The odd component gives rise to asymmetrical scattering. Examples of this have
been given by Stern and Elliott [15] and others.
In Figure 5.9 the resonant length based on backscattering and on forward
scattering is shown versus the slot displacement from the waveguide centreline. We
see that the two results differ significantly when the offset is increased. The dif-
ference is even more pronounced for slots in reduced height waveguide. For slotted
arrays with very stringent requirements it may be necessary to represent the slot by
a more complex network (Tee or Pi network) replacing the simple shunt model. See
also Compound Slots in Chapter 8.
74 Slotted waveguide array antennas: theory, analysis and design

1.0

b = 10.16 mm

Ex (z) b = 2.54 mm

0.5

a = 22.86 mm
w = 1.59 mm
t=0
offset = 5 mm
f = 9.375 GHz
0.0
0.0 0.5 1.0
Normalised position along slot

Figure 5.10 Slot electric field distribution for two waveguide heights.  IEEE
1987. Reprinted from [16], with permission

0.50
a = 22.86 mm offset = 3 mm
b = 10.16 mm f = 9.375 GHz
w = 1.59 mm

t = 1.27 mm
2*L/Lambda

0.48

t=0

0.46
0 2 4 6 8 10
Number of basis functions

Figure 5.11 Convergence of resonant length versus number of basis functions.

 URSI 1986. Reprinted from [33], with permission

The computed electric field distribution along a slot, Ex(z0 ), is shown in Figure 5.10
for two cases. The solid line is for a slot in full-height (standard) X-band guide, the
broken line is for a slot in quarter-height guide. The latter case is characterised by a
large asymmetrical component, indicating that the simple shunt model is a poor
representation of the scattering from the slot.
For most of the results presented here nine or ten expansion functions for the
slot field were used. The same number was also used in the stub waveguide to
represent cases with finite wall thickness. Of course fewer terms can be used when
less accurate results are acceptable to save computer time. Figure 5.11 shows for
two wall thicknesses how the computed resonant length depends on the number of
expansion functions chosen.
 Slot models 75

We can see that the zero wall thickness case (t ¼ 0) converges more slowly
than the case with finite wall thickness. It is also observed that the higher-order
even terms in the expansion contribute very little to the computed result. The even
terms represent the asymmetrical part of the slot field (which, however, becomes
more important in reduced height waveguide; Figure 5.10). In general, the electric
field distribution is slightly bowed out compared to the first sinusoidal term,
pffiffiffiffiffiffiffiffiffiffi
resembling the shape of sin q.
The difference between the cases with t ¼ 0 and t ¼ 1.27 mm with respect to
convergence can be understood when the edge condition at the slot ends is con-
sidered [26]. With zero wall thickness the electric field near the slot ends is
expected to behave as r1/2, r being the distance from the edge. This results in an
expected decay rate for the coefficients Ep / p3/2. With finite thickness the right-
angle edge gives the dependence r2/3 and an expected decay rate Ep / p5/3. Thus,
in the latter case, a slightly faster convergence is expected. See Figure 5.12, where
the expansion coefficients are plotted on a logarithmic scale for two wall thick-
nesses. Slot data are the same as in Figure 5.11. pffiffiffiffiffiffiffiffiffiffi
As discussed earlier one might consider adding a term with  sin q shape in
the expansion of the slot field (5.10). This term would have the proper variation at
the edges r1/2 and might reduce the number of terms needed for a good field
solution. A similar term would serve in the thick walled case [27, p. 99].
The entire basis results agree quite well with calculations using pulse expan-
sion and point matching as demonstrated in Figure 5.13. It should be noted, how-
ever, that in the pulse expansion case one extra pulse of length 0.02 l0 was added
somewhat arbitrarily to allow for the expected electric field variation down to zero
amplitude at the slot ends. Including this, the agreement between the two solutions
is very good, with a maximum difference of about 0.003 l0.

1
t=0
t = 1.27 mm
0
Log (Ep)

–1

Odd

–2

Even
–3
0.0 0.5 1.0
Log ( p)

Figure 5.12 Decay of expansion coefficients Ep with increasing p.  IEEE 1987.

Reprinted from [16], with permission
76 Slotted waveguide array antennas: theory, analysis and design

0.51 [11]
This theory

2*L/Lambda 0.49 t = 1.27 mm

a = 22.86 mm
0.47
b = 10.16 mm
t=0 w = 1.59 mm
f = 9.375 GHz

0.45
0 5 10
Offset from centreline (mm)

Figure 5.13 Computed resonant length versus slot offset. Entire basis (solid line)
versus pulse basis (triangles).  URSI 1986. Modified from [33],
with permission

In our expansion of the slot field we have assumed the electric field to be
constant across the slot width. In fact, the expected physical behaviour [26] is a
singularity such that the field Ex varies as 1/r1/2, where r is the distance to the edge
(zero thickness case). A reasonable function modelling this behaviour is the elec-
trostatic field variation in a narrow gap:
w=p
E1 ðx0 Þ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi; jx0 j < w=2 (5.49)
ðw=2Þ2  x0 2

One could also add an odd component (and possibly even higher terms) with
the same edge behaviour [28]:
2x0 =p
E1 ðx0 Þ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi; jx0 j < w=2 (5.50)
ðw=2Þ2  x0 2

Similar edge mode functions have been proposed for the dual case: scattering
from a strip conductor [29]. For slotted waveguide array antennas with narrow slots
it does not seem to improve the results very much to include these edge functions.
However, for transverse slot problems using the Moment Method with pulse
expansion it has been found that an edge mode may be needed in order to obtain
convergence [30].

5.3.2 Slot magnetic field distribution

The purpose of the electromagnetic modelling was to find a solution to (5.2):

Hze ðEx Þ  Hzi ðEx Þ ¼ Hzinc (5.51) = (5.2)

We have focused on the electric field in the slot, Ex(z), and can claim that we
have found good solutions with excellent agreement between measurements and
 Slot models 77

calculations in terms of slot admittance, resonant length, etc. We expect therefore

that (5.51) is fulfilled quite well.
The boundary conditions for the longitudinal H-field at the end points z0 ¼ L
tell that the field shall behave as 1/r1/2 at the end points, that is, be singular. This is
true both for the internal and the external fields. They are oppositely directed, so
they would not cancel each other, but rather add in the zero wall thickness case;
cf. (5.51). This is matched by the electric field going to zero with our choice of
expansion functions, so the electromagnetic energy remains finite in the region
close to the edge [26]. A test case illustrates the H-field singularities:
Standard X-band guide a  b ¼ 22.86  10.16 mm
Wall thickness t¼0
Slot offset x0 ¼ 15.24 mm
Slot width w ¼ 1.587 mm
Slot length 2L ¼ 15.53 mm
Frequency f ¼ 9.17 GHz
The internal and external magnetic fields in the slot, Hzint ðz0 Þ and Hzext ðz0 Þ, were
computed for a widely spaced infinite array (Dx ¼ Dz ¼ 100 mm), approximating the
single slot case. The array situation was chosen in order to improve convergence for
the external field. In total, 2,500 waveguide modes and 2,500 Floquet modes were
included [14,31]. Nine expansion functions were used; see Figure 5.14.

Re(Hz)
Im(Hz)

1.5
Im(Hzext – Hzint )

Re(Hzext – Hzint )
1.0
Im(Hzinc )
0.5 Re(Hzinc )

z
–L –5 0 5 L (mm)

–1.0

–1.5

Figure 5.14 Magnetic field components in the slot [14]

78 Slotted waveguide array antennas: theory, analysis and design

Figure 5.14 shows a comparison based on the assumed identity:

Hze ðEx Þ  Hzi ðEx Þ ¼ Hzinc

The right-hand side is

Hzinc ¼ j cosðpx0 =aÞ  ejb10 z (5.52)

We get in our case

Hzinc ¼ 0:5 sin b10 z  j0:5 cos b10 z (5.53)

In Figure 5.14 the real part and the imaginary part of Hzinc are separately
compared with the corresponding real and imaginary parts of ðHzext  Hzint Þ. We
find that the imaginary parts agree rather well while the real part of ðHzext  Hzint Þ
demonstrates a strong oscillation centred on the real part of the incident field.
The imaginary part, which has a small oscillatory component, is in phase with
the electric field in the slot. It is responsible for the real power radiated from the
slot, that is, the conductance. The real part, in phase quadrature with the electric
field, produces the reactive power. At resonance this is almost zero – the positive
and negative peaks cancel each other. The more terms we use in the expansion of
the electric field the higher will be the frequency of the ripple in the magnetic
fields.
Figure 5.15 shows the magnitude jHzext  Hzint j in the slot and also outside the
slot. In the latter region the phase has stabilised and there is no ripple in the curve.
The field magnitude here corresponds to a concentration of current density at the
slot ends.

|Hzext – Hzint|

10
Relative field strength

0 5 10 15
L
Longitudinal position along and beyond slot, mm

Figure 5.15 Magnetic field along and beyond the slot [14]
 Slot models 79

Now it might seem that we have not verified the boundary equation, from
which we started:

Hze ðEx Þ  Hzi ðEx Þ ¼ Hzinc

We solved this equation by projecting onto the space of the testing functions es(z)
by forming inner products hHzint ðEx Þ; es i as already defined in (5.4). In this space our
boundary condition is fulfilled. Provided that the errors in the assumed field quantities
(expansion functions) are not too large, the errors in the secondary parameters such as
admittance, resonance, reflection, etc., will be of second order [17,32].

5.3.3 Experimental verification

Figure 5.16 shows a comparison between measured and computed resonant lengths for
rectangular slots with 0.38 mm wall thickness. Nine expansion terms were used in this
computation. The difference between theory and experiment is maximum 0.002l0, or
about 0.4%, except for the largest offset where it is about twice that value.
The influence of different wall thicknesses was also studied experimentally.
The resonant slot length versus slot offset was measured for 0.3 and 1.0 mm wall
thickness. The slots had rounded ends, for which we still lack a theoretical model.
However, the difference in resonant length for the two cases can be expected to be
the same as with rectangular slots. As can be seen in Figure 5.17 the measured
difference in resonant length agrees very well with the computed difference in
resonant length.
In order to compensate for the rounded slot ends the following correction
formula can be used:
2L0  2L& ¼ D  w (5.54)

0.51 Measured
Computed

0.49
2*L/Lambda

0.47 a = 22.86 mm
b = 10.16 mm
w = 1.59 mm
t = 0.38 mm
0.45
0 5 10
Offset from centreline (mm)

Figure 5.16 Computed and measured resonant lengths for X-band rectangular
slots in standard waveguide.  IEEE 1987. Reprinted from [16],
with permission
80 Slotted waveguide array antennas: theory, analysis and design

0.03

2* {L (0.3) – L (1.0)} / Lambda

Measured
0.02 Computed

0.01

0.00

–0.01
0 1 2 3 4 5
Offset from centreline (mm)

Figure 5.17 Difference in resonant length for two wall thicknesses, t ¼ 0.3 mm
and t ¼ 1.0 mm, versus slot offset from waveguide centreline.
Rounded slot ends.  IEEE 1987. Reprinted from [16],
with permission

0.4

0.3
Delta factor

0.2

0.1

0.0
0.0 0.2 0.4 0.6
Normalised slot offset

Figure 5.18 Rounded end correction factor from measured data versus slot offset
normalised to half waveguide width.  URSI 1986. Reprinted from
[33], with permission

Here L0 and L& are half the resonant slot lengths for rounded and rectangular slot
ends, respectively, w is the slot width and D is a correction factor. It has been
suggested that slots with equal area, alternatively slots with equal perimeter, should
have equal resonant frequency. Yee [5] proposed an average between the two,
which leads to D ¼ 0.32. More recently the correction factor was determined
empirically from measured data for several cases with rounded slot ends and the-
oretical results for the corresponding rectangular slots, see Figure 5.18. The mea-
surements were made at C- and X-band, and for wall thicknesses from 0.3 to
1.8 mm both in normal and half-height waveguide. Theoretical calculations using
 Slot models 81

0.53 Measured rounded ends

t = 0, rectangular ends
t = 1.27 mm, rectangular ends
t = 1.27 mm, with end correction

2*L/Lambda

0.49

a = 22.86 mm
b = 10.16 mm
w = 1.59 mm
f = 9.375 GHz
0.45
0 5 10
Offset from centreline (mm)

Figure 5.19 Computed resonant length versus slot offset for standard X-band
guide compared with measured data from [35].  IEEE 1987.
Reprinted from [16], with permission

0.52 Measured
Computed

0.50
2*L/Lambda

0.48

0.46
0 3 6
Offset from centreline (mm)

Figure 5.20 Computed and measured resonant length versus slot offset for
C-band slot with rounded ends in half-height waveguide. a ¼ 39 mm,
b ¼ 10 mm, w ¼ 2.8 mm, t ¼ 0.6 mm, f ¼ 5.3 GHz.  URSI 1986.
Reprinted from [33], with permission

MoM/FEM analysis have also been done [34]. Both studies support the assumption
that slots with equal area have approximately equal resonant frequency, that is,
D ¼ 1  p/4 ¼ 0.215. Assuming equal perimeter would lead to twice this correction.
Using this rounded end correction we obtain the results shown in Figure 5.19,
where a comparison is made with Stegen’s measured data [35]. Another example is
given in Figure 5.20.
82 Slotted waveguide array antennas: theory, analysis and design

1.0

Conductance G/Y0

0.5

Computed
Measured

0.0
0.45 0.50 0.55 0.60
(a) Slot length in wavelengths

0.5
Susceptance B/Y0

0.0

Computed
Measured

–0.5
0.45 0.50 0.55 0.60
(b) Slot length in wavelengths

Figure 5.21 (a) Computed and measured slot conductance versus slot length for
X-band slot in half-height waveguide.  IEEE 1987. Reprinted from
[16], with permission. (b) Computed and measured slot susceptance
versus slot length for X-band slot in half-height waveguide.  URSI
1986. Reprinted from [33], with permission

Figure 5.21 shows the computed slot conductance and susceptance versus slot
length compared with measurements. The waveguide wall thickness was 0.3 mm.
The slot had rounded ends and we applied the correction factor to account for this.

5.3.4 Weakly excited slots

A longitudinal slot with no offset, that is, positioned on the waveguide centreline,
will not radiate. Very small offsets will result in weak radiation. This is a typical
situation in long, travelling wave type slotted arrays. With this configuration a wide
bandwidth is possible in spite of the long length, although the beam will squint with
frequency – sometimes a desirable feature (see Section 9.2).
 Slot models 83

A phase/frequency scanned C-band array with long, non-resonant ridge

waveguides has been described by Derneryd and Lorentzon [36]. High accuracy
requirements combined with the small offsets led to an experimental approach in
the slot modelling. Very low normalised conductance values had to be used in the
design. In particular, the waveguide centreline to which slot offsets are referenced
must be known accurately in order to avoid grating lobes [37].
With zero offset the slot will still cause a disturbance which is manifested as a
small reactive loading of the waveguide. This was illustrated in the equivalent
circuit in Figure 5.3 as a shunt element B0, the residual susceptance for a centred
slot. It can typically amount to B0/G0  0.1 or less and can be compensated for in
the hardware with a small capacitive metallic post under the slot [38, p. 271]. Work
on improved scattering models for weakly excited waveguide slots has been
reported by, for example, Rengarajan and Derneryd [39]; see also Chapter 8.

5.4 Transverse slots

5.4.1 Introduction
Transverse slots in the broad wall of a rectangular waveguide have a potential
application in low-thickness array antennas with the polarisation directed along the
waveguide (Figure 5.22).
Transverse slots in the narrow wall of the guide require probes, irises or
similar devices inside the waveguide in order to excite the slots [40,41]. Alter-
natively, one could use inclined narrow wall slots, which, however, radiate a
cross-polarised field component. Watson [1] discussed the transverse waveguide
slot, which he represented as a series element on an equivalent transmission line.
Approximate expressions for the slot impedance were derived by Stevenson [2]
and Oliner [4]. Das [42] reported on expressions for the slot impedance as
a function of slot length, starting from an assumed sinusoidal electric field
distribution in the slot.

b
x
a

Figure 5.22 Transverse slot in a rectangular waveguide

84 Slotted waveguide array antennas: theory, analysis and design

We will here follow a similar procedure as we did when analysing the

longitudinal slot characteristics in the previous sections. A main difference is
that the slot field is z-directed leading to asymmetrical scattering in the
waveguide. The slot can be represented by a series impedance on an equivalent
transmission line.

5.4.2 Theoretical analysis

In order to analyse the slot impedance we make use of an integral equation for the
slot electric field, which we arrive at by equating the magnetic fields on the two
sides of the slot interface. The integral equation is solved by the Method of
Moments. The wall thickness is easily and accurately included without increasing
the number of unknowns. The procedure is the same as for longitudinal slots, the
details are not repeated here [16,42].
The slot field is assumed to be of the form

X
N
pp
Ez ¼ Ep sin ðx  x0 þ LÞ (5.55)
p¼1
2L

with no variation across the slot width, w (see Figure 5.23). The variable x0 is the
off-centre displacement, and L is the half length of the slot. Nine terms (N ¼ 9)
were used for the numerical results presented here. The Ex field along the slot is
neglected as well as the magnetic field Hz across the slot.
The external magnetic field and the associated Moment Method matrix ele-
ments will be the same as in the longitudinal slot case (Section 5.2). The internal
matrix elements, however, are different for the transverse slot.

L L

x0

Figure 5.23 Definition of offset for a transverse slot

 Slot models 85

The appropriate interior Green function for the waveguide region is well
known [2,19]. Limiting ourselves to the Hx component from the slot electric Ez
field, we obtain both TE and TM contributions. They are summed to yield
 
pp mp2 mpx
XN XXem en L  k sin
2
a a
Hx ðx; zÞ ¼ Ep       ...
pp 2 mp 2
p¼1 m n jwm abg2
0 nm 
2L a
0 1p odd
mpx0 mpL
sin cos
B a a C  
B C  1  e gmn w=2
cosh g z (5.56)
@ mpx0 mpL A mn
cos sin
a a p even

Here
mp2 np2
g2mn ¼ þ  k2 (5.57)
a b
and em ¼ 1 for m ¼ 0, while em ¼ 2 for m > 0.
For the Galerkin case we form inner products with the testing functions, ws
ps
ws ¼ sin ðx  x0 þ LÞ (5.58)
2L
to yield the matrix element contributions Ypsint :
  
pp ps mp2 1  egmn w
XX em en k 2
w
L L a gmn
Ypsint ¼         mp2   . . .
pp 2 mp 2 sp 2
m n jwm abg2
0 nm  
2L a 2L a
0 1p odd 0 1s odd
mpx0 mpL mpx0 mpL
sin cos sin cos
B a a C B a a C
B
@ mpx0 mpL
C
A  B
@ mpx0
C
mpL A (5.59)
cos sin cos sin
a a p even a a s even

The driving vector, assuming a normalised incident TE10 field:

rffiffiffiffiffi
2 px
Hx ¼
inc
sin ejb10 z (5.60)
ab a
has elements hs:
0 1s odd
px0 pL
pffiffiffi sp sin cos
4 2 B a a C
hs ¼ hHxinc ; ws i ¼ pffiffiffiffiffi sinðb10 w=2Þ   22L  2 B
@ px
C
pL A
abb10 sp p 0
 cos sin
2L a a a s even
(5.61)
The solution yields in particular the inner aperture field as expressed by the
expansion (5.55), where the coefficients are now known. This is the key to the
86 Slotted waveguide array antennas: theory, analysis and design

equivalent slot impedance. The reflection coefficient inside the waveguide in terms
of the propagating TE10 mode becomes
0 1p odd
pp px0 pL
sin cos
Hxscat sinðb10 w=2Þ X
N Ep B a a C
pp2 p2 B C
G¼ ¼ pffiffiffiffiffi 2L (5.62)
Hxinc wm0 ab p¼1 @ px 0 pL A
 cos sin
2L a a a p even

from which we obtain the equivalent normalised slot impedance:

2G
Z=Z0 ¼ (5.63)
1G

5.4.3 Results
Figure 5.24 shows the computed normalised resistance and reactance versus slot
length compared with previous results by Khac [11], who used a pulse expansion
with point matching. These results are for zero wall thickness.
Figure 5.25 shows computed results compared with measured results for a
transverse slot in a reduced-height waveguide. The measured slot had rounded
ends. The length was therefore reduced by D ¼ 0.215w to represent an equivalent
rectangular slot; cf. Figure 5.18.
The effect of the waveguide height is demonstrated in Figure 5.26. The resis-
tance is approximately inversely proportional to the waveguide height. Figure 5.27

1.6 0.8
Resistance R/Z0

Reactance X/Z0

0.8 0.0

0.0 –0.8
0.40 0.46 0.52
Slot length in wavelengths

Figure 5.24 Transverse slot impedance in zero wall standard X-band waveguide
versus normalised slot length 2L/l. Rings and crosses denote
computed results (present method). Lines are from [11].  IEEE
1993. Reprinted from [43], with permission
 Slot models 87

1.25

4.8 GHz

Reactance X/Z0

0.00

5.8 GHz

–1.25
0.0 2.5
Resistance R/Z0

Figure 5.25 Transverse slot impedance for a 28 mm long slot with zero offset from
4.8 to 5.8 GHz. a ¼ 39 mm, b ¼ 10 mm, w ¼ 3 mm, t ¼ 1.9 mm. Solid
curve with crosses is computed, rings denote measured values.
 IEEE 1993. Modified from [43], with permission

2.5

10 mm
2.0
Resistance R/Z0

1.5
20 mm

1.0

0.5

0.0
4.8 5.3 5.8
Frequency (GHz)

Figure 5.26 Computed slot resistance for two waveguide heights b ¼ 10 mm and
b ¼ 20 mm, other parameters as in Figure 5.25. Curve with rings is
measured results for b ¼10 mm.  IEEE 1993. Reprinted from [43],
with permission
88 Slotted waveguide array antennas: theory, analysis and design

1.5

0 mm

1.0
Resistance R/Z0

5 mm
0.5

0.0
4.8 5.3 5.8
Frequency (GHz)

Figure 5.27 Computed slot resistance for two offset values, x0 ¼ 0 mm and 5 mm.
b ¼ 20 mm, other data as in Figure 5.26.  IEEE 1993. Reprinted
from [43], with permission

shows the dependence on the offset. Only minor changes can be realised using this
parameter.
There is no simple way of varying the excitation of a transverse slot. There is
also no phase reversal technique available, such as alternating the offset direction
or changing the sign of the inclination angle as in other types of slots. In arrays, this
will lead to the generation of grating lobes if resonant slot spacing shall be used,
that is, one guide wavelength for broadside radiation. Parallel plates in the form of
baffles can be used to suppress the grating lobes in the E-plane of an array of
transverse slots; more about this in Section 9.3.

References

[1] Watson W. H. ‘Resonant slots’. Journal of the Institution of Electrical

Engineers, Part IIA. 1946;93(4):747–777.
[2] Stevenson A. F. ‘Theory of slots in rectangular waveguides’. Journal of
Applied Physics. 1948;19(1):24–38.
[3] Wait J. R. ‘On the conductance of slots’. IRE Transactions on Antennas and
Propagation. 1956;4(2):124–127.
[4] Oliner A. A. ‘The impedance properties of narrow radiating slots in the
broad face of rectangular waveguide, Part I, II’. IRE Transactions on
Antennas and Propagation. 1957;5(1):4–20.
 Slot models 89

[5] Yee H. Y. ‘Impedance of a narrow longitudinal shunt slot in a slotted

waveguide array’. IEEE Transactions on Antennas and Propagation.
1974;22(4):589–592.
[6] Stegen R. J. ‘Slot radiators and arrays at X-band’. IRE Transactions on
Antennas and Propagation. 1952;1(1):62–84.
[7] Elliott R. S. Antenna Theory and Design. Prentice-Hall, Englewood Cliffs,
NJ, 1981.
[8] Lagerlöf R., Josefsson L. ‘A phased array radar antenna for airborne early
warning’. Proceedings Military Microwaves. London, July 11–13, 1990,
pp. 257–261.
[9] Lewin L. Advanced Theory of Waveguides. Iliffe & Sons, Ltd., London,
1951.
[10] Falk K. ‘Admittance of a longitudinal slot in a ridge waveguide’. IEE Pro-
ceedings H – Microwaves, Antennas and Propagation. 1988;135(4):263–268.
[11] Khac T. V. A Study of Some Slot Discontinuities in Rectangular Waveguides.
Ph.D. Dissertation, Monash University, Australia, November 1974.
[12] Lyon R. W., Sangster A. J. ‘Efficient moment method analysis of radiating slot
in a thick-walled rectangular waveguide’. IEE Proceedings H – Microwaves,
Antennas and Propagation. 1981;128(4):197–204.
[13] Yee H. Y., Richardson P. N. ‘Slotted waveguide antenna arrays’. IEEE
Antennas and Propagation Society Newsletter. 1982;24(6):4–8.
[14] Josefsson L. Slots in Waveguide Arrays Analyzed by Expanding the Fields in
Waveguide Modes and Floquet Modes. UCLA Rpt. No. AP-202, EE Dept.,
USA, August 1983.
[15] Stern G. J., Elliott R. S. ‘Resonant length of longitudinal slots and validity
of circuit representation: theory and experiment’. IEEE Transactions on
Antennas and Propagation. 1985;33(11):1264–1271.
[16] Josefsson L. ‘Analysis of longitudinal slots in rectangular waveguides’.
IEEE Transactions on Antennas and Propagation. 1987;35(12):1351–1357.
[17] Harrington R. F. Field Computation by Moment Methods. The MacMillan
Co., 1968.
[18] George Green G. An Essay on the Application of Mathematical Analysis to
the Theories of Electricity and Magnetism. T. Wheelhouse, Nottingham,
1828.
[19] Elliott R. S. An Introduction to Guided Waves and Microwave Circuits.
(Appendix C: Stevenson’s Green’s functions for rectangular waveguide).
Prentice-Hall, Englewood Cliffs, NJ, 1993.
[20] Collin R. E. ‘On the incompleteness of E and H modes in wave guides’.
Canadian Journal of Physics. 1973;51:1135–1140.
[21] Rahmat-Samii Y. ‘On the question of computation of the dyadic Green’s
function at the source region in waveguides and cavities’. IEEE Transactions
on Microwave Theory and Techniques. 1975;23(9):762–765.
[22] Van Bladel J. ‘Contribution of the Y¼constant mode to the modal expansion
in a waveguide’, IEE Proceedings H – Microwaves, Optics and Antennas.
1981;128(5):247–251.
90 Slotted waveguide array antennas: theory, analysis and design

[23] Eshrah I. A., Yakovlev A. B., Kishk A. A., Glisson A. W., Hanson G. W.
‘The TE00 waveguide mode – The ‘‘complete’’ story’. IEEE Antennas and
Propagation Magazine. 2004;46(5):33–41.
[24] Clemmow P. C. The Plane Wave Spectrum Representation of Electro-
magnetic Fields. Pergamon Press, Oxford, 1966.
[25] Stinson D. C. Intermediate Mathematics of Electromagnetics. Prentice-Hall,
Englewood Cliffs, NJ, 1976.
[26] Meixner J. ‘The behavior of electromagnetic fields at edges’. IEEE Trans-
actions on Antennas and Propagation. 1972;20(4):442–446.
[27] Amitay N., Galindo V., Vu C. P. Theory and Analysis of Phased Array
Antennas. Wiley Interscience, New York, 1972.
[28] Butler C. M., Wilton D. R. ‘General analysis of narrow strips and slots’.
IEEE Transactions on Antennas and Propagation. 1980;28(1):42–48.
[29] Richmond, J. H. ‘On the edge mode in the theory of TM scattering by a strip
or strip grating’. IEEE Transactions on Antennas and Propagation. 1980;
28(6):883–887.
[30] Park K. P., Stern G. J., Elliott R. S. ‘An improved technique for the eva-
luation of transverse slot discontinuities in rectangular waveguide’. IEEE
Transactions on Antennas and Propagation. 1983;31(1):148–154.
[31] Rengarajan S. R., Josefsson L., Petersson R. ‘Recent developments in broad
wall slots in rectangular waveguides for array applications’. Seventh Inter-
national Conference on Antennas and Propagation, ICAP 91 (IEE). 1991,
pp. 729–732.
[32] Harrington R. F. Time-Harmonic Elctromagnetic Fields. IEEE Press,
New York, 2001.
[33] Josefsson L. ‘Analysis of longitudinal slots in rectangular waveguides
including the effect of wall thickness’. Proceedings of the URSI Electro-
magnetic Theory Symposium. Budapest, Hungary. August 1986, pp. 367–369.
[34] Zhang M., Hirano T., Hirokawa J., Ando M. ‘Method of Moments analysis
of a waveguide round-ended wide slot by using numerical-eigenmode
basis functions’. IEEE Topical Conference on Wireless Communication
Technology. 2003, pp. 360–361.
[35] Stegen R. J. Longitudinal Shunt Slot Characteristics. Hughes Tech. Memo
No. 261, November 1951.
[36] Derneryd, A. G., Lorentzon T. C. ‘Design of a phase/frequency scanned
array antenna with non-resonant slotted ridge waveguide elements’.
Proceedings of the IEEE Antennas and Propagation Symposium, London,
Ontario, June 1991, pp. 1728–1731.
[37] Karlsson I. ‘Applications of waveguide arrays in commercial and military
radars’. Proceedings of the 1993 Antenna Applications Symposium, Allerton
Park, IL, USA, pp. 1–22, 1993.
[38] Marcuvitz N. (ed.) Waveguide Handbook. MIT Rad.Lab.Series, Vol. 10,
1951.
[39] Rengarajan S. R., Derneryd A. ‘Improved scattering model for weak broad-
wall slots’. Microwave and Optical Technology Letters. 1993;6(8):504–507.
 Slot models 91

[40] Hashemi-Yeganeh S., Elliott R. S. ‘Analysis of untilted edge slots excited by

tilted wires’. IEEE Transactions on Antennas and Propagation. 1990;
38(11):1737–1745.
[41] Dudley D. G. ‘An iris-excited slot radiator in the narrow wall of rectangular
waveguide’. IRE Transactions on Antennas and Propagation. 1961;
9(4):361–364.
[42] Das B. N., Sinha M. ‘Impedance characteristics of series slots’. Proceedings
of the Institution of Electrical Engineers. 1974;121(11):1360–1362.
[43] Josefsson L. ‘A waveguide transverse slot for array applications’. IEEE
Transactions on Antennas and Propagation. 1993;41(7):845–850.
 Chapter 6
The linear slotted waveguide array antenna

Based on the analysis of the single-slot antenna in the previous chapters we can
now approach the problem of designing linear and planar slot arrays. We will first
study longitudinal slot arrays and start with the simple case with one row of slots in
a rectangular waveguide. For this case mutual coupling appears mainly in the
H-plane and can in many cases be neglected. Some examples of computed and
measured performance will be presented. Procedures for designing linear arrays of
slots, including some examples, will be discussed.

6.1 Introduction

The longitudinal slot in a waveguide can be represented by an equivalent shunt

admittance according to [1] as already mentioned in Chapter 1. In Chapters 4 and 5
mathematical expressions were derived for modelling this shunt admittance. It
leads to the network representation of an array of slots according to Figure 6.1.
Figure 6.1 shows a resonant slot array where the waveguide is fed at one end
and shorted at the other end. With a spacing from the last slot to the short of a
quarter of a guide wavelength (corresponding to the phase delay jN) the short is
transformed into an open circuit at the location of the slot. The other spacings shall
be half a guide wavelength so that all slots radiate in phase. Note the alternating
offsets (d in Figure 6.1) that provide the missing 180
of phase shift. This means
that all admittances appear in parallel as seen from the feeding end. A match is
obtained when the sum of all admittances equals the waveguide admittance G0:
X
N
G0 ¼ yi (6.1)
i¼1

At the design frequency the slots are resonant so that their admittances are real.
With a uniform amplitude distribution (i.e. with equal slot conductances ¼ G0/N) a
narrow broadside beam is obtained in the H-plane with first sidelobes about
13 dB below the main beam peak. With a tapered distribution the slot con-
ductances shall be chosen according to the desired slot voltage distribution; the slot
conductance is assumed to be proportional to the square of the slot voltage as an
approximation. The relationship between the slot conductance and slot voltage will
be further discussed in Section 6.3.
94 Slotted waveguide array antennas: theory, analysis and design

φ1 φ2 φN

y1 y2 y3 yN

Figure 6.1 A linear slot array and its network representation with shunt
admittances

A simple model for estimating the return loss bandwidth as a function of N,

the number of radiating elements in an end-fed standing wave array consists of a
transmission line periodically loaded by N shunt conductances as shown in
Figure 6.1. The spacing between adjacent loads is a half wavelength at the centre
frequency and the value of each conductance normalised to the characteristic
admittance of the transmission line is kept as 1/N for simplicity. A short circuit is
placed at the end of the transmission line at a distance of a quarter wavelength from
the last load. Using the transmission line equations we obtain the normalised
admittance across the ith conductance looking towards the short, where i ¼ 1 is the
load near the short circuit. Expressions for the total normalised admittance at
section i given by Elliott [2] and reproduced in (6.2) and (6.3) are obtained from
standard circuit theory and transmission line equations [3,4, p. 151].

Yi ðYi1 =G0 Þ cosðbd Þ þ j sinðbd Þ

¼ þ 1=N (6.2)
G0 cosðbd Þ þ jðYi1 =G0 Þ sinðbd Þ
Y1
¼ j cotðbd Þ (6.3)
G0

In (6.2), G0 is the characteristic admittance of the transmission line and b is its

phase constant. Using (6.2) recursively from i ¼ 1 through N, one can determine the
normalised admittance at the input port, and hence the reflection coefficient. This
model does not include the waveguide dispersion, the frequency response of each
shunt element or the mutual coupling between radiating elements. Therefore the
results are to be used in a qualitative sense. Figure 6.2 shows the approximate value
of the reflection coefficient in dB achievable for a specified percentage bandwidth
for different values of N. If we use 10 dB reflection coefficient or approximately
a VSWR of 2 as a reference, an array with eight elements in the end-fed config-
uration will produce about 8% bandwidth. In other words over 8% bandwidth, the
reflection coefficient would be 10 dB or better. For greater values of N, the
10 dB reflection coefficient bandwidth is found to be lower, for example, a
16-element array exhibits only about 4% bandwidth.
 The linear slotted waveguide array antenna 95

–10

Reflection coefficient in dB
–20

–30
N=2 N=4
N=8 N = 16
–40
–10 dB ref

–50
0 4 8 12 16
Per cent bandwidth

Figure 6.2 Reflection coefficient bandwidth of a standing wave array for different
numbers N of slot elements (10 dB ref is a baseline reference)

A centre-fed array is generally preferred over an end-fed array so as to achieve

a greater bandwidth, since the distance between the feed and the short circuit is
halved in a symmetric centre-fed array. Centre-fed arrays will be discussed in
Chapter 7.
Selecting the sum of slot conductances higher than the waveguide conductance
G0 requires an impedance transformer at the input port. The scheme is called
overloading and can be shown to increase the VSWR bandwidth of the array. This
is discussed in more detail in Chapter 7, Section 7.5, and in Section 9.11.

6.1.1 Overview of design approaches

Linear slotted waveguide array designs have been reported since the 1950s [5–8]
using approximate transmission line models with lumped shunt or series elements
for the slots. These early designs assumed that the conductance of a shunt slot or
the resistance of a series slot is proportional to the square of the slot voltage. Elliott
has shown that the above-mentioned assumption can lead to errors [9]. A design
technique for small arrays was presented by Elliott and Kurtz by incorporating the
external mutual coupling between slots by using a dipole analogy [10]. Subse-
quently Elliott dispensed with the dipole analogy and derived the mutual coupling
expressions using a half cosinusoidal distribution for the slot aperture electric field
[11]. His derivation of the external mutual coupling expressions employed the
reciprocity principle. This design procedure, also applicable to slots in dielectric
filled waveguides, uses accurate models for isolated slots and for various coupling
mechanisms. It has been widely used in the design of slot arrays. The Elliott design
equations are derived in Section 6.3. In the first equation, radiating longitudinal
slots are modelled in the form of equivalent shunt admittances in a transmission
line. In the second equation, the active admittance of a longitudinal radiating
slot, in the presence of all other slots in the array, is derived in terms of its
96 Slotted waveguide array antennas: theory, analysis and design

self-admittance, that is, that of an isolated slot, and geometrical parameters such as
offsets and lengths of all slots in the array. This expression includes external mutual
coupling between all slots. Internal TE20 mode coupling between adjacent slots is
derived later. The TE20 mode coupling between the slot closest to the short circuit
and the short may be included using a similar approach. The design procedure uses
an iterative technique as discussed in Sections 6.3, 6.5 and 6.6. Convergence is
reached generally after a few iterations.
The design approach using the incremental conductance, discussed in Sec-
tion 6.7, is especially useful in relatively large arrays in geometries where it is
difficult to determine the self and mutual admittances. The incremental con-
ductance is defined as the difference between the total input conductance of two
arrays, one having N resonant slots and the other N þ 1. In both cases the slots are
identical (same values of offset, length or tilt) with adjacent offsets or tilts alter-
nating positive and negative values. Incremental conductance data is computed or
measured for a range of values of slot offsets or tilts. It is then possible to design
arrays that have slowly varying aperture distributions. It is possible to use this
approximate technique as a starting design, which may be improved with more
rigorous design techniques, if necessary.
The Elliott design technique models longitudinal radiating slots in the form of
shunt admittances in a transmission line. For arrays having slots with large offsets
in reduced height waveguides and compound slots characterised by offsets and tilts,
a better slot model will be in the form of a Tee or Pi network. For such slots a
design procedure using scattering parameters of the slot is discussed in Section 6.8.
The Elliott design procedure requires scattering characteristics of isolated
longitudinal radiating slots as a function of slot offsets and lengths. Ideally a slot in
a waveguide should be modelled as a four-port network, with two ports on the
waveguide on either side of the slot and two ports for the two apertures of a slot
when wall thickness is included in the model. For practical values of wall thickness,
the voltages of both slot apertures are nearly the same. Therefore, the wall thickness
is ignored in the Elliott design procedure except when determining the scattering
data of isolated slots. It should be mentioned that the resonant length of a slot is
sensitive to the wall thickness. A two-port model of a slot with both ports in the
waveguide, with the knowledge of the slot voltage is adequate. The slot aperture
distribution is assumed to be a half cosine distribution with a complex coefficient,
and it is related to the type of excitation such as the incident TE10 mode, higher-
order modes from adjacent slots or external radiation from all other slots. With the
availability of many commercial computational electromagnetic (CEM) codes it is
possible to compute slot data very accurately. The mutual coupling may be com-
puted accurately as discussed in Section 6.4.
The design of slot arrays involves the determination of waveguide dimensions
and slot parameters to achieve the required radiation characteristics, for example,
gain, sidelobe level and cross-polarisation level, and the return loss as a function of
frequency. On the other hand, in the analysis process one is interested in deter-
mining the radiation characteristics and the return loss, given the waveguide and slot
parameters. Design techniques generally use transmission line models with lumped
 The linear slotted waveguide array antenna 97

circuit loads representing the slots. While some analysis models also use similar
loaded transmission line models, rigorous analyses employing finite element meth-
ods, integral equations/method of moments (MoM) and mode matching techniques
provide very accurate solutions. The analysis procedure employing the MoM solu-
tion of the integral equations for the slot aperture fields is discussed in Chapter 8.

6.2 The accuracy of the slot array model

In early designs a few simplifications were made.

6.2.1 Slot model

Accurately measured slot data represents the real world and is often preferred if
available. Computer simulations usually involve some simplifications, more or less
justified. However, a very accurate slot field distribution is not always required.
Simply stated, conductance or admittance is related to power and is thus stationary
and insensitive to small variations in the assumed field shape [12, p. 428]. This was
also illustrated in relation to Figure 5.14.
In array design and analysis the field distribution of individual slots is often
assumed to be symmetric and sinusoidal. A non-symmetric component is some-
times observed, especially in reduced height waveguide; cf. Figures 5.9 and 5.10.
This means that the simple shunt element model is not exactly valid and may
have to be replaced with a Tee or Pi network to account for the asymmetrical
scattering [13]. Alternatively a scattering model for the slot is useful in the design
as explained in Section 6.8.

6.2.2 Network model

The network model (Figure 6.1) is assumed to carry only the fundamental TE10
mode. Higher-order modes are indeed excited near the slot but often assumed to
decay rapidly. However, in some slot arrays, for example, when slot offsets are
small, TE20 mode generated at a slot can couple a non-negligible amount into a
neighbouring slot. This is discussed in Section 6.4. The problem with higher-order
waveguide modes is also apparent in some planar arrays between adjacent coupling
slots and at coupling junctions between a coupling slot and two radiating slots in the
immediate vicinity. These scenarios are discussed in Chapter 8, Sections 8.4 and 8.5.

6.2.3 Ground plane

The slot models often assume an infinite, perfectly conducting ground plane. Edge
diffraction effects were discussed in relation to Figures 4.12 and 4.24 and also in
Section 8.9.

6.2.4 Mutual coupling

In Section 6.1 mutual coupling was only briefly mentioned. There is always some
coupling in the H-plane but usually quite small. The examples in Figures 4.20 and
98 Slotted waveguide array antennas: theory, analysis and design

0
With MC
Pattern level in dB –10 Without MC
–20

–30

–40

–50

–60
–90 –60 –30 0 30 60 90
Theta in degrees

Figure 6.3 Computed H-plane pattern of two eight-element uniformly excited

end-fed slotted waveguide arrays. Solid line represents the array
designed including external mutual coupling; broken line corresponds
to the array designed without mutual coupling

4.21 showed that the coupling level in the H-plane beyond the nearest slot was less
than 40 dB. However, for stringent pattern shapes, for example, low sidelobes,
the coupling in the H-plane may have to be included, cf. [14]. In the E-plane the
coupling is much stronger and should be accounted for. Therefore the mutual
coupling has typically to be included in the design of planar arrays, where both
E-plane and H-plane coupling is present, see Chapter 7.
An illustration of the coupling effect on an eight-element equal amplitude
H-plane linear slot array is shown in Figure 6.3. The array was designed first
without including the mutual coupling effects, and in a subsequent design mutual
coupling was included. A moment method analysis of both designs yielded the
patterns shown in Figure 6.3. The MoM uses nine entire domain sinusoidal basis
functions, similar to the ones discussed in Chapter 5, for each slot aperture. In
addition, the MoM solution accounts for external mutual coupling and internal
higher-order mode coupling. A further discussion on MoM for slot arrays is found
in Chapter 8. As seen in Figure 6.3, the mutual coupling effect on the radiation
pattern is very small and of no importance in this case. The example shown is for a
half-height X-band (9.3 GHz) waveguide with slot widths 1.6 mm and a wall
thickness of 0.5 mm and sum of conductances G0 ¼ 1.0. The principles for
including the mutual coupling in the theoretical design are elaborated in
Section 6.3.
Let us look at an example with a tapered excitation for lower sidelobes where a
greater impact of the mutual coupling could be expected. A result for a 30 dB
Dolph–Chebyshev pattern is shown in Figure 6.4. Here the mutual coupling effect
is noticeable in the first sidelobe. Although the effect of the error is to reduce the
first sidelobe level in this particular example, it may have an undesirable effect in
other examples by increasing the sidelobe levels. Further details of the design of
 The linear slotted waveguide array antenna 99

–10
Pattern level in dB MC
–20
No MC
–30

–40

–50

–60
–90 –60 –30 0 30 60 90
Theta in degrees

Figure 6.4 Theoretically computed H-plane patterns of two eight-element slotted

waveguide arrays with 30 dB Dolph–Chebyshev distribution. Solid
line represents the array designed including external mutual coupling;
broken line corresponds to the array designed without mutual
coupling

this array are found in Section 6.5. The sidelobe taper is due to the slot element
pattern. Slight asymmetry in the pattern is found in the array designed without
mutual coupling. This asymmetry is caused by the internal TE20 mode coupling
generated at the slot closest to the short, which couples to the same slot after getting
reflected by the short. The internal coupling will be discussed in Section 6.4.
The asymmetry is not found for the array designed using the mutual coupling
effects since the above-mentioned TE20 mode coupling was taken into account in
the design. Therefore the slot voltages and the pattern are close to the ideal for
this design.

6.3 The Elliott design equations

A method to iteratively design slotted waveguide array antennas including the
external mutual coupling among the slots has been presented by [10]; see also [15].
The calculations were based on an equivalent dipole array, knowing that the current
distribution on a resonant dipole is the same as the electric field distribution on a
resonant slot. The method was later modified [11] to be valid directly for slot
arrays. The modified design equations are easily derived from our knowledge about
the slot modelling as a shunt element that we discussed in Chapter 4, Section 4.4.

6.3.1 The first design equation

Let us study the ith slot in the array. As previously stated the slot loading the
waveguide can be represented by a shunt admittance, Figure 6.5(a) represents the
waveguide, and Figure 6.5(b) an equivalent transmission line.
100 Slotted waveguide array antennas: theory, analysis and design

Slot # i

A+i A
Yi Ys Vn G0
Ai– B

(a) (b)

Figure 6.5 (a) Slot in waveguide causes symmetric scattering as if loaded by a

shunt admittance. (b) Equivalent transmission line, with characteristic
admittance G0, loaded by the equivalent slot admittance Ys. Vn is the
modal voltage at Ys

The reflection coefficient must be the same in the two representations, thus

G ¼ A þ
i =Ai ¼ B=A (6.4)

We can write for the second case

G0  ðYs þ G0 Þ Ys
G ¼ B=A ¼ ¼ (6.5)
G0 þ ðYs þ G0 Þ Ys þ 2G0
Thus
Ys 2G 2B=A 2B 2B
¼ ¼ ¼ ¼ (6.6)
G0 1þG 1 þ B=A AþB Vn
where Vn is the modal voltage A þ B.
For equal reflected power in the two models we must have

1 2 1
jA j Yi ¼ jBj2 G0 (6.7)
2 i 2
That is,
pffiffiffiffiffiffiffiffiffiffiffiffi
B ¼ Yi =G0 A i (6.8)

This gives
rffiffiffiffiffiffi
Ys 2 Yi 
¼ A (6.9)
G0 Vn G0 i
We have already an expression for the scattered mode amplitude A
i in (4.74).
Thus we obtain

Ys Vs
¼ K1 f ðx0 ; LÞ (6.10)
G0 Vn
 The linear slotted waveguide array antenna 101

This expression relates the slot admittance Ys, the slot voltage Vs and the
modal voltage Vn in the first design equation of Elliott [11, Eq. (10)]. Equation
(6.10) is valid for the isolated slot admittance as well as the active slot admittance.
In the expression for the isolated slot admittance the slot voltage does not include
mutual coupling whereas for the active admittance, mutual coupling is accounted
for. An explicit expression for the mutual coupling is derived in the second design
equation.
The constant
rffiffiffiffiffiffi pffiffiffi
Yi jp 2
K1 ¼ 2 pffiffiffiffiffi ; (6.11)
G0 b10 k0 a ab
is the same as that in [11, Eq. (11)], while
p
cos b10 L px0
2kL
f ðx0 ; LÞ ¼     2 cos (6.12)
p 2 b10 a

2kL k
Equation (6.10) differs from Elliott’s equation (10) in [11] by a minus sign
because of sign differences in the normalisations for the TE10 mode in the two deri-
vations. The sign difference does not cause any problem in the design of slot arrays
since we always work with the ratio of active admittances to realise the aperture
distribution, for example, (6.16). We also observe that the term (k/k0) in the numerator
inside the square root of Elliott’s equation (11) in [11] for K1 should not be present so
that the corrected expression will be valid for dielectric filled waveguides also.

6.3.2 The second design equation

We will now derive the second design equation which includes the effect of the
external mutual coupling. The situation is depicted in Figure 6.6.
We will use the superscript ‘a’ to indicate the active situation with several slots
present; thus G a ¼ G active ¼ A/Aþ. There are two contributions to the scattered
wave A, namely the contribution from the isolated (single) slot and the contribu-
tion from the mutual coupling only. Thus
 
G a ¼ A 
isol þ Acoupl =A
þ
(6.13)

Coupling

Slot
A+
Waveguide
A–

Figure 6.6 Slot excited internally from the waveguide and externally from the
mutual coupling
102 Slotted waveguide array antennas: theory, analysis and design

The term A is proportional to the slot voltage Vs, which likewise can be
divided into two components: Vs ¼ Vsisol þ Vscoupl.
We can write
     
G a =G ¼ A  
isol þ Acoupl =Aisol ¼ Vs =Vsisol ¼ Vs = Vs  Vscoupl ¼ 1= 1  Vscoupl =Vs

(6.14)

In this expression G without the superscript a is the reflection coefficient for

the isolated slot.
Now

Ya 2Ga 2
¼ ¼
G0 1 þ G a 1 þ 1=G a

Using (6.14) and (6.6) in the above expression we obtain

Ya 2
¼
G0 2 1 Vscoupl (6.15)
þ 
Ys =G0 G Vs

which essentially is the second design equation according to Elliott [11, Eq. (33)].

6.3.3 The iterative design algorithm for standing wave arrays

We will describe the design algorithm for standing wave arrays. In Section 6.6 the
design algorithm for travelling wave arrays will be presented. The designer chooses
the number of radiating slots, N, and excitation coefficients from system con-
siderations such as the gain, and pattern shape. Excitation coefficients are chosen to
be pure real for simplicity of implementation. The required input data for the design
are the values of the waveguide inner dimensions, wall thickness, the slot width and
the centre frequency for the design.
The design equations will be solved iteratively. Initially all the radiating slots
are assumed to be resonant length. The values of slot offsets, alternating positive
and negative, are assigned some initial values that are not critical, since they will be
changed later. In the first iteration all the mutual coupling terms are ignored. In
subsequent iterations the mutual coupling terms are fully accounted for, if mutual
coupling is not negligible.
There are two unknowns for each slot, offset and length. Therefore in the linear
array there are 2N unknowns. For all but the Nth radiating slot two equations are
generated, one for the real part and the other for the imaginary part of (6.10) in
conjunction with (6.15). Equation (6.10) is expressed as a ratio of active admit-
tances of the nth and Nth slots as shown next.

Yna fn Vns VN
¼ (6.16)
YNa fN VNs Vn
 The linear slotted waveguide array antenna 103

The subscript n refers to the slot number n while N refers to the Nth slot, the one
closest to the short used as a reference. Note that fN and fn are defined in (6.12).
Vns and VNs are slot voltages while Vn and VN are mode voltages. For standing wave
arrays with a slot spacing of half guide wavelength, the mode voltage magnitudes
are the same and their signs alternate and so do the signs of fn s. For a travelling
wave array, expressions for the mode voltages and the admittances are described in
Section 6.6. For the Nth slot, that is, n ¼ N, (6.16) is not relevant. Enforcing the
active admittance of the Nth slot to be real yields one equation. One more equation
is obtained by enforcing the total active admittance at the input port to the char-
acteristic admittance of the TE10 mode, G0 so as to achieve a match at the input
port. Thus there are 2N non-linear equations in 2N unknowns for the array design.
The simultaneous non-linear equations may be solved iteratively by a quasi-
Newton technique [16]. Convergence is reached when the changes in the values of
slot offsets and lengths between two successive iterations are less than the desired
manufacturing tolerance, D, or if the desired performance is met. Figure 6.7 shows
the flow chart for the design algorithm.

Start

Choose N,
a, b, f, w, t
Read slot data
Iter = 1

Solve 2N
non-linear
Iter = Iter+1 equations

Iter >1
and
No ∆ < tol ?

Yes

Stop

Figure 6.7 Flow chart of the design algorithm

104 Slotted waveguide array antennas: theory, analysis and design

6.4 Mutual coupling

6.4.1 Efficient computation of the external mutual coupling terms
The external mutual coupling expression has received much attention in the lit-
erature since it takes up most of the computational effort in the design of slot
arrays, especially in planar arrays. Design of planar arrays will be discussed in
Chapter 7. An approximation that works very well for slot lengths close to half
wavelength in free space was presented by Mazzarella and Panariello [17]. They
employ a Taylor series expansion around half wavelength long slots, with one or
two terms for very good accuracy for the slot aperture distributions in the form of a
half cosinusoid. For air-filled waveguides, the aperture distribution is similar to that
of the current in a dipole antenna. For that case Mazzarella and Panariello provide
fast and accurate expressions in terms of familiar sine and modified cosine inte-
grals. The integral in the mutual coupling expression was carried out in the Fourier
transform domain by Nauwelaers and van de Capelle [18]. By expressing the
Fourier transform of the aperture fields in terms of a power series in the spectral
variable, the integrals are carried out analytically in terms of spherical Hankel
functions. Recursive relations are used to evaluate the spherical Hankel function.
The results are interpreted in terms of multipole expansions. This method works
very well when the spacing between slots is not too small.
A coordinate transformation in terms of difference coordinates has been
employed to reduce the fourfold integral in the mutual coupling expression into a
numerical double integral by Rengarajan and Gabrelian [19]. For narrow slots,
a further approximation in terms of numerical single integrals is given later.
Elliott has shown that the voltage induced in slot n due to external mutual
coupling from all other slots is given by the expression

1 Ys =G0 X N
0 s
Vscoupl ¼ Vs ext ¼ jðb10 =k Þðk0 bÞða=lÞ3 V gmn ðLm ; Ln ; Xmn ; Zmn Þ
fn Ys =G0 þ 2 m¼1 m
2

(6.17)

where Lm and Ln are the half lengths of slots m and n, respectively, and Xmn and Zmn
are the spacings between slots m and n along the E-plane and H-plane directions,
respectively [11, Eq. (29)]. The prime in the summation means that the m ¼ n term
is excluded. The integral gmn cast as a dimensionless quantity in terms of normal-
ised lengths by Elliott [11] is related to the mutual admittance between apertures in
a ground plane, Ymn, in (4.115), repeated next.
jb
Ymn ¼ gmn (6.18)
pah
For the convenience of calculation of Ymn it has been assumed that Lm ¼ Ln. The
use of (6.17) in (6.15) will yield the second equation given by Elliott [11, Eq. (33)].
It may be seen that the mutual admittance derived in Chapter 4, (4.85) and (4.86) treat
the case of slots on a ground plane. The presence of the waveguide under the slot n
in (6.17) is responsible for the additional terms in (6.17) [11, Eqs. (29) and (33)].
 The linear slotted waveguide array antenna 105

We consider the mutual coupling integral gmn specified in [11] and defined in
(4.114) and (4.115) in Chapter 4. Equation (4.114) is used here to express gmn in
terms of slot length 2L and width w. Since slot lengths in arrays vary by a small
amount, one may calculate gmn for slots m and n by keeping their lengths equal by
averaging the lengths of slot m and n. There are two terms in gmn shown next, one
containing a single integral and the other a double integral. We will simplify gmn2 to
a single integral so that gmn can be computed rapidly by numerically evaluating the
single integrals.

gmn ¼ gmn1 þ gmn2 (6.19)

p L
ð  pz  expðjkR Þ expðjkR Þ

1 2
gmn1 ¼ cos þ dz (6.20a)
2k0 L L 2L R1 R2

and
  p 2  1 ð L  pz ð L  0
pz expðjkRÞ 0
gmn2 ¼ k02  cos cos dz dz
2L k0 L 2L L 2L R
(6.20b)
where
h i
2 1=2
R ¼ Xc2 þ ðZc þ z0  zÞ ; (6.21a)
h i1=2
R1 ¼ Xc2 þ ðZc þ L  zÞ2 (6.21b)

and
h i1=2
R2 ¼ Xc2 þ ðZc  L  zÞ2 : (6.21c)

Figure 6.8 shows the geometry of the coupling problem. Two sub-domains of
the double integral in gmn2, evaluated in terms of difference coordinate t ¼ z0  z,
are shown in Figure 6.9.

Z'

Z
X'

w Zc
X 2L

Xe

Figure 6.8 Geometry of the coupling problem

106 Slotted waveguide array antennas: theory, analysis and design

z'
τ = 2L
L

τ=0

2
–L z
L
1

–L τ = –2L

Figure 6.9 Two sub-domains of the coupling integral; t < 0, and t > 0

The double integral in gmn2 is denoted I2 such that

ðL  pz ð L  0
pz expðjkRÞ 0
I2 ¼ cos cos dz dz (6.22)
L 2L L 2L R
Using a coordinate transformation in terms of difference coordinates t ¼ z0  z
we express I2 as shown in (6.23). Since R is dependent on t it is independent of z0 for
a fixed value of t. Figure 6.9 shows the two domains of the integral labelled 1 and 2.
I2 is broken into two integrals, one for each domain as follows.
ð0 ð
expðjkRÞ Lþt hp i pz0 
0
I2 ¼ cos ðz  tÞ cos dz0 dt
2L R L 2L 2L
ð 2L ð
expðjkRÞ L hp i pz0 
0
þ cos ðz  tÞ cos dz0 dt (6.23)
0 R Lþt 2L 2L

where R ¼ ½Xc2 þ ðZc þ tÞ2 1=2 .

The z0 integrals in (6.23) are evaluated analytically and the resulting expression
for I2 is shown next.
ð 2L    
expðjkRÞ pt L pt
I2 ¼ cos ðL  t=2Þ  sin dt (6.24)
2L R 2L p 2L
In  and , the upper sign is used when t > 0 while the lower sign holds when t < 0.

6.4.2 The self-admittance

The mutual coupling analysis in Chapter 4 deals with rectangular apertures, each
aperture fed by a waveguide with the same cross section as the aperture dimen-
sions. Each aperture can be seen as a rectangular slot and the feeding waveguide is
naturally present in the finite wall thickness of the waveguide (Figures 4.13 and
6.10). The analysis resulted in the mutual admittance between two apertures
 The linear slotted waveguide array antenna 107

Aperture
number: 1 2 n N

V+ V–

Figure 6.10 Waveguide array cross section with slot apertures and feeding
waveguides

(alone, and in an array) and their dependence on their separation. An expression for
the calculation of the mutual admittance Ymn was given by (4.114).
The scattering matrix can be computed once we know the mutual admittance
matrix [Y e] between the slot apertures. However, to fill [Y e] we also need the self-
e
admittance Ynn of a single slot-aperture before we can make the calculation:

½S  ¼ ð½Y e  þ ½Y0 Þ1 ð½Y0   ½Y e Þ (6.25)

where [Y0] is the diagonal matrix with the feeding waveguide admittance; cf.
Chapter 4.
The self-admittance was in fact studied in the single slot analysis model of
Chapter 5, (5.43). The difference is that here, in the coupling analysis model, we
usually simplify and assume a sinusoidal electric field distribution in the slot
aperture, thus putting the coefficients p ¼ s ¼ 1 in (5.43).
The self-admittance of rectangular apertures has been studied by Lewin [20]
and Compton [21]. The Compton solution (his Equation (29)) is reproduced next:
   
2 kx a ky b
ð1 ð1 2 2 sin cos 2
2b k  k y 2 2
Y self ¼ ð2pÞ2   2 dkx dky (6.26)
awm0 1 1 p kx kz
2 2
p2  ky2 b2

Note that in Compton’s expression b is the total width of the waveguide (¼ slot
length) and a is the total height of the waveguide (¼ slot width), just the opposite of
the definitions used in this book. It is easily shown that this solution is identical to
our (5.43) with p ¼ s ¼ 1 inserted.
Compton transforms his solution into the following form, which can be eval-
uated numerically more easily.
ð ða pffiffiffiffiffiffiffiffiffi
ejk x þy
2 2
16 jb b
Y self
¼ g ðxÞhðyÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi dxdy (6.27)
awm0 p y¼0 x¼0 x2 þ y2
The parameters are
8
> p
< g ðxÞ ¼ ða  xÞ
2 (6.28)
> py py
: hðyÞ ¼ D1 ðb  yÞcos þ D2 sin
b b
108 Slotted waveguide array antennas: theory, analysis and design

where
8  
>
> 1 k2 p
> D ¼ 
< 1 b2 4p 4b2
  (6.29)
>
> 1 k2 p
>
: D2 ¼ þ
pb 4p 4b2
In the graphs by Compton the aperture admittance values are normalised to
the wave admittance of the medium outside the aperture. Lewin’s graphs are
normalised to the mode admittance of the feeding waveguide. In Figure 6.11 the

1.2
b/λ = 0.6

b/λ = 0.4
0.8
Conductance

0.6
b/λ = 0.2
0.4

0.2 b/λ = 0.1

0
0.5 0.7 0.9 1.1 1.3 1.5
(a) a/λ

0.6

0.5
b/λ = 0.2
b/λ = 0.4
0.4
Susceptance

b/λ = 0.1
0.3
b/λ = 0.6

0.2

0.1

0
0.5 0.7 0.9 1.1 1.3 1.5
(b) a/λ

Figure 6.11 Aperture conductance (a) and susceptance (b) versus slot length a/l
and width b/l. The aperture field is assumed to be that of a TE10
mode. Values are normalised to the TE10 modal admittance
 The linear slotted waveguide array antenna 109

× 10–3
2.5

2
Conductance
1.5
Admittance
1

0.5

0
Susceptance
–0.5

–1
0.4 0.5 0.6 0.7 0.8 0.9 1
a/λ

Figure 6.12 Calculated aperture admittance (Siemens) for a square aperture fed
from a same size square waveguide, thus a ¼ b. Dashed lines are
measured. Reprinted from [24]

calculated admittance for some typical apertures sizes is plotted. (The definitions
of aperture dimensions a/l (length) and b/l (width) are the ones chosen in this
book.)
This relatively simple theory assuming just the fundamental TE10 mode gives
quite good agreement with more complex models and measured results [22,23].
One further example is shown in Figure 6.12, where the admittance is given in
Siemens (i.e. not normalised).

6.4.3 Internal TE20 mode coupling

The internal higher-order mode coupling between adjacent radiating slots, pri-
marily from the evanescent TE20 mode, may be non-negligible. The TE20 mode
scattered by a slot close to a short circuit, after reflection, may also couple a non-
negligible voltage in that slot. Ignoring such coupling terms may contribute to
errors in the slot aperture distribution, especially for small offset slots [25]. The
contribution to slot voltage due to internal mutual coupling from the TE20 mode has
been reported by Elliott and O’Loughlin [26]. We first evaluate the TE20 mode
fields produced by the electric field in slot n given by
0
!
V s
p z
~E n ¼ ^x n cos
n n
(6.30)
w 2Ln

using the waveguide Green’s function (2.61). This can also be obtained from (4.72)
for the TE20 mode of the rectangular waveguide.
110 Slotted waveguide array antennas: theory, analysis and design

The TE20 mode fields propagating in the z direction in the waveguide are
given next.
~ wm0
E b ¼ ^y Ba sinð2p x0 =aÞexpðg20 z0 Þ (6.31)
2p=a
 
~ b ¼ Ba ^z j cosð2p x0 =aÞ  ^x jg20 sinð2p x0 =aÞ expðg20 z0 Þ
H (6.32)
2p=a
The TE20 mode back scattering coefficient of slot n is
ð2p=aÞ2
Ba ¼ Vns hn (6.33)
g20 k0 h0 abk
where
p coshðg20 Ln Þ
hn ¼ 2 cosð2pdn =aÞ   (6.34)
2kLn 2 p 2
ðg20 =k Þ þ
2kLn
and k ¼ k0 if the waveguide is air-filled. The TE20 mode forward scattering coef-
ficient is determined as Ca ¼ Ba.
s
The TE20 mode induced voltage, Vn;20 in the nth slot is obtained by comparing
s
the TE10 mode induced voltage Vn10 as follows. Equation (7) in [11], combined
with (6.5), yields (6.35)
Gn n
s
Vn10 ¼ A (6.35)
Kfn 10
where fn is defined as f (x0, L) in (6.12).
s
Similarly Vn20 , the voltage induced by the TE20 mode wave of amplitude An20 is
given by
s
Vn10 I20
s
Vn20 ¼ (6.36)
I10
where
ðð  
pzn
I10 ¼ Hz10 cos ds0 (6.37)
slot n 2Ln
and
ðð  
pzn
I20 ¼ Hz20 cos ds0 (6.38)
slot n 2Ln

Hz10 is the magnetic field of the TE10 mode wave and Hz20 is that of the TE20 mode
wave in the region of slot n. Combining (6.35) through (6.38) we obtain

b10 eg20 d Ys =G0 

Vns int ¼ Vn20
s
¼ j s
hn hn1 Vn1 þ hn hnþ1 Vnþ1
s
(6.39)
g20 fn Ys =G0 þ 2
2
 The linear slotted waveguide array antenna 111

The internal coupled voltage is added to the external coupled voltage, (6.17),
thus
Vscoupl ¼ Vs ext þ Vns int (6.40)

6.4.4 Full wave analysis

The term ‘full wave’ is generally used to denote a rigorous electromagnetic analysis
that solves Maxwell’s equations subject to all boundary conditions such as the
finite element method, the MoM solution to the integral equations and the finite
difference method. However, the so-called full wave codes use some approxima-
tions to make the solution tractable. The procedure described earlier to include the
TE20 mode coupling between adjacent slots can be extended to include as many
evanescent modes as needed to obtain the desired accuracy in the design procedure.
An MoM solution of the coupled integral equations for the electric fields of all slot
apertures would provide a rigorous and accurate solution. Other computational
techniques such as the finite element method would provide an accurate full wave
solution as well. However, the full wave analysis techniques are generally not
convenient for the design of slot arrays. Advanced design optimisation techniques
using full wave moment method analysis are discussed in Chapter 8.

6.4.5 Mutual coupling compensation

When the scattering matrix S has been found from the aperture admittance matrix,
see (6.25), we can study the impact of the coupling on the aperture excitations.
Denote the aperture voltages travelling towards the apertures by the vector Vþ and
those reflected back by the vector V, both referenced to the aperture plane (Fig-
ure 6.10). They are related as

V ¼ S  Vþ (6.41)

and the total aperture excitation is

V ot ¼ ðI þ S Þ  V þ (6.42)

where I is the identity matrix. Now, suppose we want to have a desired excitation
V des. We should then feed with

V þ ¼ ðI þ S Þ1  V des (6.43)

The matrix (I þ S)1 provides the desired correction [27,28]. It could in theory
be realised as a network (Figure 6.13).
This corrective matrix would restore the excitation of the slot radiators to the
desired values in the presence of mutual coupling. It would work even for different
tapers and even phase steering of the array, etc. In practice it can hardly be realised
in analogue form. However, for receive only as in digital beam forming it could be
done [27].
112 Slotted waveguide array antennas: theory, analysis and design

Possible corrective
Feed network Radiators
network

Figure 6.13 The insertion of a mutual coupling correction matrix

For one specific excitation a correction factor can be implemented as follows.

The correction vector is V corr ¼ (I þ S)1V des  V des. In terms of a correction factor
we get

corrfac ¼ ððI þ SÞ1 V des Þ:=V des (6.44)

where the symbol ./ stands for element by element division. The correction is
dependent on the desired excitation.
This procedure could be applied to the design problem in the following way.
Starting values for slot offsets and lengths provide the input to the mutual coupling
calculation, yielding the scattering matrix as described earlier. Equation (6.43) then
gives a new feeding vector which requires new offsets and lengths. Thus, the
geometry is changed slightly, and an update of the mutual coupling is done. The
procedure is iterated a few times.

6.5 Design example of a standing wave array

Standing wave arrays provide high efficiency but suffer from narrow return loss
bandwidth. In standing wave arrays, the return loss bandwidth is in general more
limiting than the bandwidth for pattern performance.

6.5.1 Slot data needed for the design algorithm

The design algorithm requires the scattering characteristics of isolated slots at the
design frequency as the input data for the assumed waveguide internal dimensions
a and b. The required slot data are the self-admittance of radiating slots as a
function of slot offset and length. With the availability of CEM codes such as the
MoM solution to the pertinent integral equations, the finite element method or the
finite difference solution of Maxwell’s equations subject to the boundary condi-
tions, accurate slot data may be computed. It has been shown in Chapter 5 that the
MoM using entire domain basis functions provides accurate results.
 The linear slotted waveguide array antenna 113

The self-admittance is determined from S11 of the two port scattering matrix of
an isolated radiating slot as shown next.
Y 2S11
¼ (6.45)
G0 1 þ S11
Thus the design data needed are (i) resonant lengths, 2Lres, of longitudinal
radiating slots computed at the design frequency for a range of values of slot off-
sets, |d| up to a value typically less than 0.35a. In addition, for each value of |d| we
need computed values of (ii) |S11| as a function of the normalised slot length,
L ¼ L=Lres , in the range of 0.9–1.10 and (iii) the phase of S11 for the same range of
offsets and normalised slot lengths.
In the iterative design, the required S11 for any offset and length is determined
from a two-dimensional interpolation of |S11| and arg Sl1 from the computed (or
measured) data as a function of offset and normalised length. The self-admittance
of the slot may then be determined from (6.45). Alternatively the slot admittance
may be plotted in terms of Stegen normalisation as shown in Figure 5.1(a).
The conductance and susceptance curves may be fitted into polynomials for rapid
calculations for any given length and offset.

6.5.2 Design of a 30 dB sidelobe-level Dolph–Chebyshev array

Table 6.1 shows the waveguide inner dimensions, a and b, slot width and slot
thickness used for this design example. The relative slot voltages V1s through V8s for
30 dB Dolph–Chebyshev aperture distribution are also specified. In this theore-
tical example, the design frequency is 9.3 GHz.
Table 6.2 shows the resonant length as a function of the magnitude of the slot offset
from the centreline, and Tables 6.3 and 6.4 show the magnitude and phase of S11 as a
function of normalised slot length and offset for an isolated slot. The data were com-
puted using the MoM solution to the integral equation of the slot aperture field.
The values of slot offsets and lengths obtained from two array designs, one
without external mutual coupling and the other with mutual coupling are listed in
Table 6.5. The slot offsets and lengths are found to be symmetric with respect to the
centre but for alternating positive and negative offsets. The two sets of data have
small differences near the centre, whereas significant differences are noted for
weak slots at the ends. Amplitude and phase of slot excitations for the two designs
obtained from full wave MoM analyses are also shown in Table 6.5. The design
with mutual coupling produces excitations close to the required excitations for the

Table 6.1 Parameters of the eight-element Dolph–Chebyshev array

a ¼ 22.86 mm b ¼ 5.08 mm Slot width ¼ Wall thickness ¼

1.5875 mm 0.508 mm

V s1 V s2 V s3 V s4 V s5 V s6 V s7 V s8
0.262 0.519 0.812 1.0 1.0 0.812 0.519 0.262
114 Slotted waveguide array antennas: theory, analysis and design

Table 6.2 Resonant length of radiating slots

Magnitude of slot offset, Resonant length,

|d|, in mm 2lres, in mm
0.444 14.99
0.889 15.01
1.333 15.05
1.778 15.09
2.222 15.15
2.667 15.23
3.111 15.31
3.556 15.40
4.000 15.50

Table 6.3 Magnitude of S11

L ¼ L=Lres 0.94 0.955 0.97 0.985 1.0 1.015 1.03 1.045 1.06
jdj
0.444 0.0016 0.0019 0.0022 0.0024 0.0024 0.0023 0.0022 0.0020 0.0018
0.889 0.0065 0.0076 0.0086 0.0093 0.0096 0.0093 0.0086 0.0078 0.0071
1.333 0.0146 0.0169 0.0190 0.0206 0.0211 0.0204 0.0190 0.0173 0.0157
1.778 0.0255 0.0294 0.0331 0.0358 0.0366 0.0355 0.0332 0.0304 0.0276
2.222 0.0392 0.0449 0.0503 0.0542 0.0540 0.0538 0.0505 0.0464 0.0424
2.667 0.0557 0.0634 0.0705 0.0754 0.0768 0.0747 0.0702 0.0648 0.0594
3.111 0.0740 0.0837 0.0924 0.0984 0.1000 0.0975 0.0921 0.0854 0.0787
3.556 0.0941 0.1057 0.1159 0.1227 0.1245 0.1216 0.1153 0.1075 0.0995
4.000 0.1156 0.1288 0.1402 0.1476 0.1495 0.1461 0.1391 0.1304 0.1213

Table 6.4 Phase of S11 in degrees

L ¼ L=Lres 0.94 0.955 0.97 0.985 1.0 1.015 1.03 1.045 1.06
jdj
0.444 228.7 219.5 207.9 194.4 180.0 166.5 155.1 146.0 138.9
0.889 228.4 219.2 207.7 194.3 180.1 166.7 155.4 146.3 139.2
1.333 227.6 218.4 207.0 193.7 179.8 166.7 155.6 146.6 139.6
1.778 227.2 218.0 206.8 193.9 180.3 167.5 156.5 147.6 140.6
2.222 226.2 217.1 206.1 193.5 180.3 167.9 157.2 148.5 141.5
2.667 224.9 215.8 204.8 192.5 179.8 167.9 157.6 149.1 142.3
3.111 223.7 214.7 204.0 192.1 179.9 168.5 158.5 150.3 143.6
3.556 222.4 213.4 203.0 191.6 180.0 169.0 159.4 151.4 144.8
4.000 220.8 212.0 201.9 190.9 179.8 169.4 160.2 152.5 146.0

Dolph–Chebyshev array shown in Table 6.1 whereas the design without mutual
coupling is found to have small amounts of error, especially near the edges. There
is a slight asymmetry in the excitations obtained from the analyses because
the structure is not strictly symmetric with a short at one end of the waveguide.
 The linear slotted waveguide array antenna 115

Table 6.5 Offsets and lengths in mm of slots of the 30 dB sidelobe Dolph–
Chebyshev array, with and without mutual coupling (MC) (slot
excitation phase is in degrees)

Slot Offset, Length, Amplitude and Offset, d Length, Amplitude and

# d 2L phase of slot no MC 2L phase of slot
withMC withMC excitation (MC) no MC excitation
(no MC)
1 0.638 15.45 0.261, 1.1 0.58 15.44 0.246, 0.1
2 1.26 15.59 0.515, 0.9 1.17 15.52 0.493, 2.5
3 1.96 15.80 0.809, 0.5 1.86 15.69 0.796, 1.0
4 2.42 15.98 1.0, 0.3 2.34 15.85 1.0, 0.
5 2.42 15.98 1.0, 0.0 2.34 15.85 1.0, 0.1
6 1.96 15.80 0.811, 0.6 1.86 15.69 0.796, 0.6
7 1.26 15.59 0.517, 0.2 1.17 15.52 0.494, 1.9
8 0.638 15.45 0.262, 0.1 0.58 15.44 0.247, 4.31

Table 6.6 Directivity and beam squint of the end-fed Dolph–

Chebyshev array (design with MC)

Frequency (GHz) Directivity (dB) Beam squint (
)

9.07 12.80 1.3
9.3 13.03 0
9.50 13.025 0.9

The higher-order modes generated by the slot closest to the short get reflected
by the short and couples to the slot. The asymmetry is slightly greater for the design
without mutual coupling.
Figure 6.4 in Section 6.2 shows the computed H-plane pattern produced by the
two designs at 9.3 GHz. The sidelobe levels decrease away from broadside because
of the element pattern of the slot. There is a small difference between the two
designs at some of the sidelobes. Table 6.6 shows the pattern characteristics at the
centre frequency and at the band edges, where the return loss is 10 dB for the end-
fed Dolph–Chebyshev array designed with mutual coupling. At the higher end of
the band, the directivity is almost the same as that at 9.3 GHz. The increase in the
electrical size of the aperture at 9.50 GHz is offset by the phase errors in the
aperture distribution due to the design at a different frequency (9.3 GHz).
The centre-fed standing wave arrays do not exhibit beam squint with frequency.
However, there is a small amount of beam squint with frequency in the end-fed
array, with positive values corresponding to the squint towards the short at higher
frequencies while negative values of squint are found at lower frequencies. At the
design frequency we have a broadside beam. At other frequencies, the aperture
phase distribution deviates from being equal, thereby creating a squint. The travel-
ling wave arrays to be discussed in Section 6.6 exhibit significant beam squint as a
116 Slotted waveguide array antennas: theory, analysis and design

function of frequency. The beam squint effect has been eliminated in a recent work
by placing a meta-material, exhibiting negative values of permittivity and perme-
ability in the frequency range of interest, between slots in the waveguide [29].

6.6 Design of a travelling wave array

The general characteristics of travelling wave arrays are discussed in Chapter 9,
Section 9.2. A travelling wave array is designed for element spacing d 6¼ lg/2 and
for small amounts of coupling between the propagating wave in the waveguide and
each radiating element. The waveguide is terminated with a matched load. The
reflection from each radiating element is small and the sum of all the scattered
waves from the slots nearly cancel out at the input port, thereby providing a good
match. Travelling wave arrays exhibit greater frequency bandwidth for return loss
compared to the standing wave arrays but have less efficiency because of the power
dissipated in the load. As the size of a travelling wave array increases, its return loss
performance becomes better. However, since the main beam scans with frequency,
for larger arrays, the gain bandwidth becomes small. In some applications, the
beam scan with frequency may be used as a beam steering technique.
For large arrays, many elements near the feed have very small amounts of
coupling, and these slots are weakly excited. The coupling characteristics of such
slots are sensitive to their offsets and lengths and they require tight manufacturing
tolerances. Thus the array size is a compromise between return loss bandwidth and
realisable coupling values and gain bandwidth. It is generally recommended that the
travelling wave array be designed to have the beam direction at least two beamwidths
away from broadside.
Figure 6.14 shows the transmission line model of a linear travelling wave array
with shunt admittances representing longitudinal broad wall slots and a match
termination at the right end.
Travelling wave arrays may be designed with all radiating slots on the one side
of the broad wall centreline or with alternating offsets [2,8,30]. In the former array
the aperture phase distribution is approximately linearly progressing with a phase
delay between adjacent elements in the propagation direction equal to b d where
b is the phase constant of the TE10 mode and d is the element spacing. The beam
pointing direction is essentially independent of element spacing d, given by

q0 ¼ cos1 ðb=k0 Þ (6.46)

YNa Y2a Y1a G0

Figure 6.14 Transmission line model for a travelling wave array

 The linear slotted waveguide array antenna 117

where q0 is measured from the z-axis, pointed along the incident TE10 wave propa-
gation direction. As an example, if b/k0 ¼ 0.707, q0 ¼ 45
. The element spacing
should be greater than the typical slot length so that adjacent slots do not overlap in
their z-coorodinates or introduce substantial amount of higher-order mode coupling
but small enough to avoid grating lobes. For a travelling wave array the element
spacing is chosen such that the reflected waves from all slots very nearly cancel which
results in good input match.
For a travelling wave array consisting of slots with alternating offsets, the
beam pointing angle is given by

q0 ¼ cos1 ½b=k0  l0 =ð2d Þ (6.47)

Possible values of q0 are limited to a narrow angular range on either side of
broadside as shown in Figure 6.15, where the required values of d in terms of

0.9

0.85
d in wavelength

0.8

0.75

0.7
83 84 85 86 87 88
(a) Beam pointing angle in degrees

0.7
d in wavelength

0.6

0.5

0.4
93 98 103 108 113 118 123
(b) Beam pointing angle in degrees

Figure 6.15 Element spacing in terms of wavelength for the required beam
pointing: (a) beam towards the load; (b) beam towards the feed port
118 Slotted waveguide array antennas: theory, analysis and design

wavelength are shown for the assumed value of b ¼ 0.707 k0. For q0 < 83
,
d becomes large enough to allow grating lobes whereas for large values of q0 the
smaller spacing might create severe higher-order mode coupling effects.

6.6.1 Design equations for travelling wave array

The design equations (6.10) and (6.15) derived previously for the standing wave
array are valid also for travelling wave arrays. However, the mode voltages are not
equal in magnitude since the spacing is not equal to a half guide wavelength in this
case. The mode voltages related to slot voltages and active admittances of slots as
given by the transmission line equations below follow Elliott’s derivations [2].
Thus, the mode voltages Vn and Vn1 at the location of slot n and slot n  1,
respectively, are related as
 
Yn1
Vn ¼ Vn1 cosðb10 d Þ þ j sinðb10 d Þ (6.48)
G0
where d is the spacing between elements. Yn1 is the total admittance at the location
of slot n  1, that is, the sum of its active admittance and the total admittance at the
location of slot n  2 transformed through waveguide of length d. The total
admittance at the location of slot n, Yn (n  2) is given in (6.49a) while (6.49b) is
valid for n ¼ 1. Yn is the active admittance of slot n, Yna plus the input admittance of
the loaded transmission line between slot n to the matched load.
Yn Yna ðYn1 =G0 Þ cosðb10 d Þ þ j sinðb10 d Þ
¼ þ (6.49a)
G0 G0 cosðb10 d Þ þ jðYn1 =G0 Þ sinðb10 d Þ
Y1 Y1a
¼ þ1 (6.49b)
G0 G0
Equations (6.16), modified to include mode voltages, and (6.48) may be combined to
give the first design equation for a travelling wave array as given in (6.50) where n  2.
fn Vns Yn1
a
=G0
cosðb10 d Þ þ jðYn1 =G0 Þ sinðb10 d Þ ¼ (6.50)
fn1 Vn1 Yn =G0
s a

Equation (6.50) is the first design equation for the travelling wave array, similar to
(6.16) for the standing wave array. The second design equation used for standing
wave arrays, (6.15) with coupled voltages specified by (6.17) and (6.39) is valid for
the travelling wave arrays also.

6.6.2 Design algorithm

The mode voltage at the location of the match termination may be assumed to be
unity, without loss of generality. The power dissipated in the load is G0 since the
voltage is unity, and the mode voltage at the location of slot 1 is V1 ¼ exp(jb10d).
The offset of slot 1, d1 is chosen arbitrarily and its length is determined so as to
make its active admittance determined by using (6.15), (6.17) and (6.39) pure
real. The choice of d1 is not critical since it can be changed subsequently. We then
 The linear slotted waveguide array antenna 119

determine the mode voltage of slot 2 using (6.48) and the offset and length of slot 2
are determined using (6.15), (6.17) and (6.39) so as to satisfy (6.50). The mutual
coupling term is computed assuming that slots 3 to N have their desired excitations
and are centred in the waveguide and that their lengths are equal to half wavelength
in free space. These computations will be more accurate in subsequent iterations
when the parameters get updated. This process continues until we reach slot N
when we can determine VN. The total normalised input admittance, YNa =G0 , is
determined using (6.49) recursively. The input reflection coefficient and incident
wave power are now calculated. This process is iterated a few times starting from
slot 1 again, until convergence is reached. Between iterations, d1 is adjusted to have
a good compromise between the load power and the input match condition.
A greater value of d1 may reduce the load power but increase the input reflection
coefficient. As a rule of thumb, the number of slots and the spacing may be chosen
such that the scattered wave voltage phasors from all the slots span the complex
plane an integer number of times as suggested by Elliott [2]. However, since the
scattering from different slots have different amounts; this is not a strict rule.

6.6.3 Example of a 21-element travelling wave array

A 21-element travelling wave array with a 30 dB Dolph–Chebyshev pattern was
designed, built and tested by Elliott [2]. The longitudinal radiating slots of this
array were offset on the same side of the broad wall centreline of a standard X-band
waveguide. The amplitude distribution for the Chebyshev pattern is given in
Table 6.7. The aperture phase distribution is assumed to be a linearly progressive
phase delay corresponding to that of the TE10 waveguide mode. The array
was designed using experimentally measured slot data and included the external
mutual coupling. Elliott demonstrated the significance of the contribution of
external mutual coupling to the slot admittance [2]. From (6.15) it is noted that the
mutual coupling effect is significant for a weakly excited slot that has a small value
of the reflection coefficient in the isolated case. In a travelling wave array with
many weakly excited slots, the mutual coupling effects cannot be neglected.
Internal higher-order mode coupling effects were not accounted for in the design.

Table 6.7 The amplitude distribution of a 21-element 30 dB Dolph–Chebyshev

array

Element # 1 2 3 4 5 6 7
Amplitude 0.334 0.279 0.378 0.485 0.595 0.701 0.800

Element # 8 9 10 11 12 13 14
Amplitude 0.883 0.947 0.986 1.000 0.986 0.947 0.883

Element # 15 16 17 18 19 20 21
Amplitude 0.800 0.701 0.595 0.485 0.378 0.279 0.334
120 Slotted waveguide array antennas: theory, analysis and design

Table 6.8 Offsets and lengths of the radiating slots, designed by including TE20
mode coupling

Element # 1 2 3 4 5 6 7
Offset (mm) 3.000 1.134 1.714 2.225 2.736 2.690 2.413
Length (mm) 15.425 15.368 15.347 15.416 15.442 15.391 15.378

Element # 8 9 10 11 12 13 14
Offset (mm) 2.400 2.521 2.380 2.022 1.812 1.765 1.657
Length (mm) 15.420 15.434 15.383 15.349 15.367 15.378 15.337
Element # 15 16 17 18 19 20 21
Offset (mm) 1.381 1.128 0.989 0.873 0.682 0.366 0.829
Length (mm) 15.296 15.306 15.320 15.288 15.240 15.275 15.093

The measured reflection coefficient was low over 5% frequency band. However,
measured patterns exhibited sidelobe levels in the range of 22 to 25 dB. Elliott
attributed part of the discrepancy between theory and experiment to the internal
coupling between slots via the TE20 mode that was not accounted for in the design
[2]. In addition, manufacturing errors and inaccuracy in the slot data used in the
design procedure and errors in measurement could also account for the discrepancy
between theory and experiment.
In order to test the accuracy of the design procedure for the travelling wave array
and to understand the higher-order mode coupling effects, we investigated the same
21-element array using accurate slot data generated by the MoM solution to the integral
equation of the slot aperture electric field, similar to those shown in Tables 6.2–6.4 in
Section 6.4 but calculated at 9.375 GHz. In addition, an accurate MoM analysis pro-
gram was used to assess the accuracy of the design procedure and various phenomena
that cause discrepancy between the design and analysis. This MoM analysis program
had been verified against very carefully measured experimental data and also with the
commercial code HFSS (www.ansys.com) which is used as an industry standard [31].
We designed two slot arrays. In the first design the internal higher-order mode
coupling via the TE20 mode was included. The second one did not include the
internal TE20 mode coupling to mimic the process that was used in [2]. The values
of the slot spacing (0.389 lg or 17.4 mm) and the first slot offset were chosen to be
the same as in [2].
Using (6.46) we obtain the beam direction measured from the z-axis to be 44.4
.
Table 6.7 shows the relative values of slot voltages, that is, the aperture distribution,
of the array to produce the 30 dB Dolph–Chebyshev pattern. The offsets and
lengths of slots obtained by the accurate design that included the TE20 mode cou-
pling are given in Table 6.8. The slot lengths in Table 6.8 assume square ended slots
used in the MoM computations while those in [2] refer to round ended slots used in
the measured slot data. In addition, our design included internal TE20 mode cou-
pling while Elliott did not account for internal higher-order mode coupling.
 The linear slotted waveguide array antenna 121

0
Dsn incl 20
–10 Dsn excl 20
Pattern level in dB

Ideal
–20

–30

–40

–50
0 30 60 90 120 150 180
Angle from the waveguide axis

Figure 6.16 Computed far field pattern of the 21-element travelling wave array.
Dotted line – design with TE20 mode; solid line – design without TE20
mode, both analysed using full wave MoM including all higher-order
mode coupling; ideal – pattern of a 21 element collinear slot array
with ideal aperture distribution

Figure 6.16 shows two radiation patterns produced by the full wave MoM
analysis for the two array designs mentioned above. The MoM solution employs
nine entire domain sinusoidal basis functions for each slot aperture (cf. Section 5.2)
and is discussed in detail in Chapter 8 for the case of planar arrays. The pattern with
the legend ‘Dsn incl TE20’ refers to the array designed with the TE20 mode cou-
pling while ‘Dsn excl TE20’ refers to the array design that did not account for the
internal TE20 mode coupling. Both MoM solutions account for external coupling
between slots and accurately model the self-term of each slot as well as its exci-
tation by the TE10 mode in the waveguide. Clearly the design that includes the TE20
mode coupling is closer to the ideal patterns and demonstrates the need for
including the TE20 mode coupling between adjacent slots. The directivities of the
three patterns had small differences and therefore in order to compare the patterns,
all pattern peaks were normalised to 0 dB.
Figure 6.17 shows two radiation patterns produced by the MoM for the array
that was designed including the TE20 mode coupling. The pattern with the legend
‘TE20 MoM’ was analysed with TE20 mode only for the internal coupling. The
other, with the legend ‘full wave MoM’, was analysed with all higher-order modes
for internal coupling. The two patterns had slightly different directivities. In order
to compare the patterns, the peaks were normalised to 0 dB in both cases. Small
deviation between the two patterns shows the contribution of higher-order modes
other than TE20 that have not been accounted for in the design procedure. Among
the contribution from all higher-order modes for this array, TE20 accounts for about
a third while TE01, TE11 and TE21 modes account for about half. Modes up to m ¼ 4
and n ¼ 2 captured most of all higher-order mode coupling effects. Most standing
wave array designs work well with TE20 mode alone accounted for and occasion-
ally one needs a few other modes such as TE01, TE11, and TE30. Sidelobes in the
angular region q < 45
are substantially lower since the element pattern approaches
122 Slotted waveguide array antennas: theory, analysis and design

–10 Full wave MoM

Pattern level in dB

–20 TE20 MoM

–30

–40

–50
0 30 60 90 120 150 180
Angle from the waveguide axis in degrees

Figure 6.17 Computed far field patterns of the 21-element travelling wave array
designed with TE20 mode coupling. Dotted line – full wave MoM
analysis including coupling via all modes; solid line – TE20 mode
MoM analysis including internal coupling between adjacent slots via
TE20 mode only

a null as q ? 0, whereas sidelobes in the angular range q > 45
keep increasing in
levels up to 90
and then keep going down because of the element pattern.
Figure 6.18(a) shows the analysed aperture amplitude distributions using the
full wave MoM, and TE20 MoM of the array designed by including the TE20 mode
internal coupling. Both distributions are compared to the ideal 30 dB Dolph–
Chebyshev distribution. The ideal aperture distribution is symmetric, whereas
the other two exhibit slight asymmetry because of imperfections in the design. The
maximum amplitude is found at different elements for the three cases. All aperture
distributions have been normalised to a peak value of unity. We notice that all
amplitude distributions are nearly the same with some deviations between the ideal
and the other two distributions. The effect of higher-order modes other than the
TE20 mode is found to be very small.
Figure 6.18(b) compares the computed values of the aperture phase distribu-
tion using MoM with the ideal linearly progressive phase of the TE10 mode wave
subtracted out. The difference between the phase data shown in Figure 6.18(b) of
the full wave MoM and TE20 MoM is found to be as high as 17
for some slots. The
variation from the ideal linear phase is as high as 66
over a phase range of about
3,000
. It appears difficult to realise the ideal linear phase distribution because of
the presence of a reflected TE10 mode wave. Figure 6.19 exhibits the input
reflection coefficient, and the load power relative to the incident wave power, both
in dB as a function of frequency. These values were also computed by the full wave
moment method.
Approximate values of the beam pointing angle determined by (6.46) have
been used to find the beam squint as a function of frequency plotted in Figure 6.20.
The beamwidth is about 7.8
so this array can be used over 5% bandwidth without
significant loss of gain. While an acceptable design of a travelling wave array can
be accomplished using the procedure described here, further improvements can be
 The linear slotted waveguide array antenna 123

1.00

Relative amplitude 0.80

0.60

Full wave
0.40 TE20 MoM
Ideal

0.20
1 6 11 16 21
(a) Element number

40

30
Phase error in degrees

20

10

0 Full wave

–10 TE20 MoM

–20
1 6 11 16 21
(b) Element number

Figure 6.18 Computed values of the aperture amplitude distribution and phase
error relative to the ideal aperture phase: (a) amplitude distribution
computed by two analysis methods; (b) phase error relative to the
ideal aperture phase computed by two analysis methods. Squares
correspond to the full wave MoM analysis, and diamonds correspond
to the TE20 MoM analysis

made using the global optimisation technique with a full wave MoM code, dis-
cussed in Chapter 8. Such an optimisation technique has been used to reduce
the power dissipated in the load substantially in the traveling wave array [32].
Mode matching technique and generalised scattering matrix method discussed in
Chapter 7 can also be used for accurate design and optimisations.
Elliott’s equations were cast in a form for the analysis of a slot array by
Hamadallah [33]. The external mutual coupling was incorporated in Hamadallah’s
analysis [33] but the internal higher-order mode coupling was not. Reasonably
124 Slotted waveguide array antennas: theory, analysis and design

–5

Load power, reflection

coefficient in dB –15

Load power

–25 Ref. coef.

–35
9.1 9.2 9.3 9.4 9.5 9.6
Frequency in GHz

Figure 6.19 Reflection coefficient and load power relative to incident power in dB

2
Beam squint in degrees

–1

–2

–3
8.9 9.1 9.3 9.5 9.7 9.9
Frequency in GHz

Figure 6.20 Beam squint in degrees as a function of frequency

good agreement was obtained between measured and computed patterns at the
design frequency as shown in Figure 6.21.
Hamadallah’s analysis equations are repeated next by including the TE20 mode
coupling.
The slot voltages of an N-element linear array are calculated from the fol-
lowing matrix equation:
k2
½ G  ½V s  ¼ V1 ½ E  (6.51)
k1
where
k expðg20 d Þ
Gmn ¼ gmn þ hn fdm n1 hn1 þ dm nþ1 hnþ1 g if m n (6.52)
g20 ðk0 bÞða=lÞ3
 The linear slotted waveguide array antenna 125

00
Computed
–10
Measured [2]
–20
dB
–30

–40

–50
00 30 60 90 120 150 180
Angle (deg.)

Figure 6.21 Computed and measured patterns of the 21-element travelling wave
array.  IEEE 1989, reprinted from [33]

and, if m > n,
k expðg20 d Þ
Gmn ¼ gmn þ hn fdm n1 hn1 þ dm nþ1 hnþ1 g
g20 ðk0 bÞða=lÞ3

 jk2 fm fn sinfðm  nÞb10 d g (6.53)

dm n1 and dm nþ1 are Kronecker delta functions (¼1 if the two subscripts are equal,
otherwise 0).
[V s] is the column matrix containing the N slot voltages.
The nth element in the column matrix [E] is given by
sin½b10 fd0 þ ðn  1Þgdg
En ¼ fn (6.54)
sinðb10 d0 Þ
for a standing wave array. The distance from the short to the centre of the closest
slot is d0.
For a travelling wave array with a match termination
En ¼ exp½ jb10 ðn  1Þd  fn (6.55)
2kG0
k2 ¼ (6.55a)
jb10 ðk0 bÞða=lÞ3
and
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 2G0
k1 ¼ (6.55b)
jða=lÞ hðb10 =k ÞðkaÞðkbÞ

All other parameters are defined in Sections 6.3 and 6.4.

126 Slotted waveguide array antennas: theory, analysis and design

The mode voltage V1 may be set to unity as a reference.

A solution of (6.51) yields all slot voltages. In order to determine the load
power of a travelling wave array and the input reflection coefficient, we need to
determine the active admittance of each slot and the total admittance at each slot
section looking towards the load. Equations (6.48)–(6.50) may be used to determine
the slot voltages. For a standing wave array (6.49b) becomes
Y1 Y1a
¼  j cotðb10 d0 Þ (6.56)
G0 G0
While this analysis technique is useful for determining the performance of a
linear standing wave or travelling wave array, it suffers from the assumptions made
in Elliott’s design technique. These assumptions are listed here once again. The slot
aperture distribution is a half cosinusoid equiphase. The slot is modelled as a shunt
admittance in an equivalent transmission line. Wall thickness is ignored except for
the use of scattering data of isolated slots such the resonant length, and self-
admittance. Full wave analysis techniques such as HFSS, MoM, mode matching,
and generalised scattering matrix method can provide more accurate analyses and
optimisations.
Modern antenna pattern measurements, for example, spherical near-field
technique and CEM codes such as MoM and HFSS show extremely close agree-
ment, often in the order of 0.1–0.2 dB even at low sidelobe levels [31]. This is a
testament to the maturity and accuracy of both the analysis and measurement
techniques. The full wave MoM results shown here would correspond to very
accurately measured experimental results such as what one obtains from a well-
maintained and well-calibrated modern spherical wave range. By comparing the
full wave MoM results and reduced MoM model in which TE20 mode coupling
alone is accounted for among the higher-order mode coupling, we have a good
understanding of the reasons for the discrepancy between the design and analysis
(or measurement).
A technique to reduce the movement of beam pointing direction with
frequency of a travelling wave array was patented by [34] and is described in
Section 9.2.4 in Chapter 9.

6.7 Array design using the concept of incremental conductance

This concept, introduced originally by Watson, is described by Silver [35] for the
design of edge slot arrays, shown in Figure 6.22, based on experimental measure-
ments of incremental conductance since it was difficult to determine the slot self-
admittance and mutual coupling effects [36]. This is a simple approximate method
and provides adequate accuracy for linear arrays that are not too short with nearly
uniform aperture distribution or slowly varying distribution. For other arrays, this
method may provide a good starting approximation which can be improved using
better computational models. Although many CEM codes can be used to obtain
self-admittance data, mutual coupling computation is more complicated.
 The linear slotted waveguide array antenna 127

Edge slots

Figure 6.22 A linear array of side wall slots

The term ‘incremental conductance’ originates from the measurement of the

input admittance of a large array of uniform spacing. The difference between the
input admittance of an array of N identical slots and a second array of N þ 1 slots is
the incremental admittance. All slots have equal lengths and the same magnitude of
tilt with respect to the vertical direction but alternate in sign. For a particular value
of slot length, the incremental admittance becomes pure real, yielding incremental
conductance. This is a difficult experimental process but it is easier to compute
with a CEM code, if available.
Let us consider a standing wave array of N edge slots. All slots have the same
tilt angles with respect to the vertical but alternating signs and same lengths. The
measured difference between the input admittance of the array with N and N þ 1
elements gives the incremental admittance if N is reasonably large, about 25 or
greater. Using a good CEM code for an infinite periodic linear array, it is possible
to solve the aperture distribution of two slots in a unit cell which, when repeated in
the exterior region, would produce an infinite array coupling effect [35]. However
inside the waveguide, there is only one unit cell with two elements. A two-element
unit cell is needed when adjacent slots have alternating offsets or alternating tilts as
in the case of edge wall slot arrays. For each tilt angle, one obtains the reflection
coefficient inside the waveguide for a range of values of slot lengths. From the
reflection coefficient, one can determine the active admittance of a single slot. Thus
we generate the active admittance data for a range of values of slot tilts and lengths.
The relationship between the slot voltage, the mode voltage and the active
admittance of a side wall slot, similar to (6.10), may be cast in the following form
similar to Elliott’s design equation (1).
Yna
¼ K fn ðqn ; Ln ÞVns =Vn (6.57)
G0
where qn is the tilt angle of slot n with respect to the vertical and 2Ln is the slot
length. The incremental conductance is approximately the same as the active con-
ductance for slots near the centre of an array. Incremental conductance can be
obtained from a CEM code as a function of slot tilt. The design equation for a
standing wave array is expressed in the form of the ratio (6.58) since the mode
128 Slotted waveguide array antennas: theory, analysis and design

voltage magnitudes are equal with alternating signs for half wavelength spacing;
cf. (6.16).
Yna Y1a j fn j Vns
= ¼ (6.58)
G0 G0 j f1 j V1s
Equation (6.58) serves as the design equation for a linear array. One needs to
obtain the values of active admittances of slots so as to realise the required aperture
distribution and an input match. The aperture distribution may be complex but the
amount of phase variation achievable with rotated slots is limited. This method may
be used for an initial design of an array and subsequently one can improve the
design using a full wave analysis code and an optimisation procedure. It has been
shown that for linear uniformly excited arrays this method provides good accuracy
whereas for nonuniform excitations better methods are needed; see e.g. [37].

6.8 Array design in terms of scattering parameters

The representation of a longitudinal slot in the form of a shunt element becomes poor
for large offsets and for small height waveguides. A Tee or Pi-type circuit would be a
better representation for such slots. Compound slots with an offset from the wave-
guide centreline and tilt with respect to the waveguide axis may also have to be
modelled in the form of Tee or Pi circuits. However, it is easier to work with
scattering wave representations for such slots. In the following we derive a design
procedure for a linear array of slots in the standing wave configuration. A design
procedure employing scattering wave representation of slots was used to find the
coupling levels of a linear array of compound coupling slots in the feed waveguide of
a planar array [38]. Such a design procedure is also applicable to a travelling wave
array. A generalised design procedure using the scattering matrix formulation of a
finite array has been presented in the literature [39]. The following scattering wave
representation considers TE10 mode in the waveguide and a half cosinusoidal aperture
distribution for the slots. It can be generalised to include higher-order modes in
the waveguide and additional modes for the slot aperture distribution in a manner
similar to the generalised scattering matrix method discussed in Section 7.9 in Chapter 7.
Figure 6.23 shows a linear array of longitudinal radiating slots, but they could
also be compound slots. It is assumed that the isolated slot data in terms of the
scattering matrix representation as a function of physical parameters, for example,
offset, length and tilt, have been computed and made available during the design
process. We will assume that there is a short at the end but the design procedure is
valid also for a match termination also. For simplicity we ignore the internal
higher-order mode coupling. It is easy to incorporate it in the mutual coupling term.
We explain the procedure assuming TE10 mode waves incident on each slot from
both directions and scattered TE10 mode waves.
1. We assume the parameters of the Nth slot, length, offset, and tilt in the case of
a compound slot. These estimates are not critical since they will be changed
 The linear slotted waveguide array antenna 129

ds

Slot N–1 Slot N

d

D10,N–1 D10,N
A10,N–1 A10,N
B10,N–1 B10,N
C10,N–1 C10,N

Figure 6.23 Incident, transmitted and scattered TE10 mode waves of slots

later. The scattering parameters of the isolated slot N are determined from
computed slot data for the assumed values of offset and length.
2. We then obtain the contribution Vs N coup due to external mutual coupling from
all other slots assuming that all other slots are centred in the waveguide with
their voltages specified by the aperture distribution and lengths equal to half
wavelength in free space. These values will be updated later in the iterative
process. The isolated and active reflection coefficients of slot N are related as
shown in (6.14) which is repeated below.
GaN =GN ¼ VsN =ðVsN  VsNcoupl Þ (6.59)
The slot voltage due to mutual coupling, VsNcoupl , may be obtained for com-
pound slots in a manner similar to the derivation for longitudinal slots.
3. Steps 1 and 2 are repeated until slot N is resonant in an active sense, i.e. in the
presence of mutual coupling. We then determine the active two-port scattering
matrix ½SNa . We set A10;N þ C10;N to 1 as a reference.
4. If there is a match termination at the end, D10,N ¼ 0 whereas if there is a short,
D10;N ¼ ½A10;N þ C10;N expð2b10 ds Þ (6.60)
5. By converting ½SNa to a transmission matrix, we can determine the TE10 wave
amplitudes and phases at the input side of slot N, A10,N and B10,N þ D10,N. The
slot voltage VsN is also determined.
6. We now determine the parameters of slot N  1 such that its voltage VsN 1
relative to VsN satisfies the required aperture distribution after including the
mutual coupling. Its scattering parameters in the absence of mutual coupling
are obtained from computed slot data.
7. Use the procedure stated in step 2 for slot N  1 to calculate, VsN 1 , and ½SN 1a .
8. The TE10 mode waves incident and leaving the output side of slot N  1 are
obtained from the corresponding wave amplitudes and phases at the input side
130 Slotted waveguide array antennas: theory, analysis and design

of slot N using approriate phase advance or phase delay. Then by converting

½SN 1a to a transmission matrix, we calculate the wave amplitudes and phases
at the input side of slot N  1.
We continue this process until we reach the first slot at the feed end. The input
reflection coefficient is given by Gin ¼ [D10,1 þ B10,1]/A10,1. If Gin is high, in the
next iteration we adjust the parameters of slot N so that it couples a greater amount
of power to external radiation. After a few iterations the process converges and
produces good return loss and correct slot excitations.
The scattering wave model is a convenient design approach for an array
of compound slots, discussed in Chapter 8 instead of a Tee or Pi-type circuit
representation used by [40]. In Section 6.9 a variation of Elliott’s design proce-
dure employing a Tee equivalent circuit for the longitudinal radiating slot is
discussed [41].

6.9 Improvement to Elliott’s design procedure

Elliott’s design procedure uses a shunt admittance model for the longitudinal
radiating slots. For slots in reduced height waveguides and for slots with large
offsets, the shunt model has been found to introduce errors. A Tee network for each
radiating slot may be derived from the computed scattering parameters of an iso-
lated slot. Such a Tee network consists of a dominant term which is the shunt
admittance and two series reactances as shown in Figure 6.24.
The series and shunt elements of the Tee network are readily obtained from the
two-port scattering matrix parameters of the slot. The ABCD matrix of the Tee
network is obtained by multiplying the ABCD matrices of the series and shunt
elements in Figure 6.24 as given in (6.61) [42]. By comparing the elements of the
ABCD matrix of the Tee network to the scattering parameters of the slot, we obtain
the values of the network elements given in (6.62) [42].
       
A B 1 Z 1 0 1 Z 1 þ ZY 2Z þ Z 2 Y
¼ ¼ (6.61)
C D 0 1 Y 1 0 1 Y 1 þ ZY

Z = jX Z = jX

Figure 6.24 Tee network model of a longitudinal radiating slot

 The linear slotted waveguide array antenna 131

1 ð1  S11 Þð1  S22 Þ  S12 S21

Y¼ (6.62)
Z0 2S21
ð1 þ S11 Þð1  S22 Þ þ S12 S21
1 þ ZY ¼ (6.63)
2S21
In (6.62) the characteristic impedance of the transmission line, Z0, is the same
as that of the TE10 mode in the waveguide. If the aperture distribution has a small
amount of odd symmetric variation in the longitudinal direction, the series element
is small and it is reactive as shown by Costanzo et al. [41]. In that case a two-step
design is carried out [41]. Initially the slot array is designed by ignoring the series
elements and by assuming a shunt model for the slots. Subsequently, the slot array
is designed by including different spacings between adjacent slots, instead of the
conventional half guide wavelength spacing. The mutual coupling calculation is
carried out using Elliott’s model. The circuit model used by Constanzo et al. is
shown in Figure 6.25.
The line lengths obtained from the above-mentioned procedure for an eight-
element slot array in a dielectric filled waveguide deviated about 3% from half
guide wavelength. The authors have not compared the performance of their design
with the original Elliott’s design. Because of the presence of the series elements
and the variation of slot lengths relative to half guide wavelength, the new design is
expected to be better. However, if the series impedances are not very small this
procedure may not be accurate. Generalised scattering matrix models for array
analysis and design will be discussed in Chapter 7.
The Elliott’s design procedure or variations discussed in this chapter serve as
a framework for designing any type of array antenna. One needs to include all the
physics such as the characterisation of each radiating element, the mutual cou-
pling effects and excitations. Examples in the literature include slot arrays fed by
boxed striplines and microstrip dipole array fed by microstrip transmission line
[43–45].

jXT2 jXT2 jXT1 jXT1

YS2 YS1

θS2/2 θS2/2 θS1/2 θS1/2

Figure 6.25 Circuit model of a linear array of slots modelled as Tee networks
with different spacings.  IEEE 2010, reprinted from [41]
132 Slotted waveguide array antennas: theory, analysis and design

References
[1] Stevenson A. F. Series of Slots in Rectangular Waveguides. National
Research Council of Canada Radio Reports 12 and 13, 1944.
[2] Elliott R. S. ‘On the design of traveling-wave-fed longitudinal shunt slot
arrays’. IEEE Transactions on Antennas and Propagation. 1979;27(5):
717–720.
[3] Montgomery C. G., Dicke R. H., Purcell E. M. Principles of Microwave
Circuits, MIT Radiation Laboratory Series, vol. 8. McGraw-Hill, New York,
1948.
[4] Collin R. E. Foundations for Microwave Engineering. McGraw-Hill, New
York, 1966.
[5] Stegen R. J. ‘Slot radiators and arrays at X-band’. IRE Transactions on
Antennas and Propagation. 1952;1(1):62–84.
[6] Stegen R. J., Reed R. H. ‘Arrays of closely-spaced nonresonant slots’. IRE
Transactions on Antennas and Propagation. 1954;2(3):109–113.
[7] Kaminow I. P., Stegen R. J. Waveguide Slot Array Design. Hughes Aircraft
Company, Culver City, CA, Tech. Memo No. 348, 1954.
[8] Dion A. ‘Nonresonant slotted array’. IRE Transactions on Antennas and
Propagation. 1958;6(4):360–365.
[9] Elliott R. S. ‘A note on the relation between slot conductance and slot
voltage’. IEEE Transactions on Antennas and Propagation. 1984;32(10):
1133–1134.
[10] Elliott R., Kurtz L. ‘The design of small slot arrays’. IEEE Transactions on
Antennas and Propagation. 1978;26(2):214–229.
[11] Elliott R. S. ‘An improved design procedure for small arrays of shunt slots’.
IEEE Transactions on Antennas and Propagation. 1983;31(1):48–53.
[12] Harrington R. F. Time-Harmonic Electromagnetic Fields. IEEE Press,
New York, 2001.
[13] Gulick J., Stern G., Elliott R. S. ‘The design of linear and planar arrays
of waveguide-fed longitudinal slots’. IEEE International Antennas and
Propagation Symposium Digest. 1986;2:685–688.
[14] Kay A. F., Simmons A. J. ‘Mutual coupling of shunt slots’. IRE Transactions
on Antennas and Propagation. 1960;8(4):389–400.
[15] Elliott R. S. Antenna Theory and Design. IEEE Press, Hoboken, NJ, 2003.
[16] Press W. H., Teukolsky S. A., Vetterling W. A., Flannery B. P., Metcalf M.
Numerical Recipes: The Art of Scientific Computing, 3rd Edition. Cam-
bridge University Press, Cambridge, 2007.
[17] Mazzarella G., Panariello G. ‘On the evaluation of mutual coupling between
slots’. IEEE Transactions on Antennas Propagation. 1987;35(11):2189–
2192.
[18] Nauwelaers B., van de Capelle A. ‘Series expressions for mutual coupling
between magnetic currents’. Proceedings of the International Conference on
Aerospace Application, Torino, Italy, 1989, pp. 199–202.
 The linear slotted waveguide array antenna 133

[19] Rengarajan S. R., Gabrelian E. ‘Efficient and accurate evaluation of external

mutual coupling between compound broad wall slots’. IEEE Transactions on
Antennas and Propagation. 1992;40(6):733–737.
[20] Lewin L. Advanced Theory of Waveguides. Iliffe and Sons, Ltd., London,
1951.
[21] Compton R. T. The Aperture Admittance of a Rectangular Waveguide
Radiating into a Lossy Half-Space. The Ohio State University Antenna
Laboratory Report. No. 1691-1, 1963.
[22] Mautz J. R., Harrington R. F. ‘Transmission from a rectangular waveguide
into half space through a rectangular aperture’. IEEE Transactions on
Microwave Theory and Techniques. 1978;26(1):44–45.
[23] MacPhie R. H., Zaghloul A. I. ‘Radiation from a rectangular waveguide with
infinite flange – exact solution by the correlation method’. IEEE Transac-
tions on Antennas and Propagation. 1980;28(4):497–503.
[24] Cohen M., Crowley T., Levis K. The Aperture Admittance of a Rectangular
Waveguide Radiating into Half Space. The Ohio State University Antenna
Laboratory Report. ac 21114 S. R. No. 22, 1953.
[25] Rengarajan S. R., Nardi D. ‘On internal higher order mode coupling in slot
arrays’. IEEE Transactions on Antennas and Propagation. 1991;39(5):
694–698.
[26] Elliott R. S., O’Loughlin W. R. ‘The design of slot arrays including internal
mutual coupling’. IEEE Transactions on Antennas and Propagation. 1986;
34(9):1149–1154.
[27] Steyskal H., Herd J. ‘Mutual coupling compensation in small array anten-
nas’. IEEE Transactions on Antennas and Propagation. 1990;38(12):1971–
1975.
[28] Derneryd A. G. ‘Compensation of mutual coupling effects in array anten-
nas’. IEEE International Antennas and Propagation Symposium Digest.
vol. 2, 1996, pp. 1122–1125.
[29] Oh, S.-S, Shafai L. ‘Beam squint minimization in series-slotted waveguide-
array antennas using double negative materials’. Proc. International Sym-
posium on Antennas and Propagation Digest. 2005. pp. 507–510.
[30] Yee H. Y. ‘Slot antenna arrays’, in Johnson R. C., Jasik H. (eds.). Antenna
Engineering Handbook. McGraw-Hill, New York, 1984.
[31] Rengarajan S. R., Zawadzki M. S., Hodges R. E. ‘Waveguide-slot array
antenna designs for low-average-sidelobe specifications’. IEEE Antennas
and Propagation Magazine. 2010;52(6):89–98.
[32] Rengarajan S. R., ‘GA-MoM Optimization of Slot Arrays’. Proceedings of
the International Applied Computational Electromagnetics Society Sympo-
sium. 2018 (to be published).
[33] Hamadallah M. ‘Frequency limitations on broad-band performance of shunt
slot arrays’. IEEE Transactions on Antennas and Propagation. 1989;37(7):
817–823.
[34] Karlsson E. R. Waveguide element for an electrically controlled radar
antenna. US patent no. 4788552, 1986.
134 Slotted waveguide array antennas: theory, analysis and design

[35] Silver S. Microwave Antenna Theory and Design. MIT Radiation Laboratory
Series. vol. 12, McGraw-Hill, New York, 1949.
[36] Yee H. Y. ‘The design of a large waveguide array of shunt slots’. IEEE
Transactions on Antennas and Propagation. 1992;40(7):775–781.
[37] Bucci O. M., Mazzarella G., Panariello G. ‘A note on edge-shunt slot array
design’. IEEE International Antennas and Propagation Symposium Digest.
vol. 3, 1997. pp. 1946–1949.
[38] Rengarajan S. R., Derneryd A. G. ‘Application of compound coupling slots
in the design of shaped beam antenna patterns’. IEEE Transactions on
Antennas and Propagation. 1993;41(1):59–65.
[39] Rubio J., Corcoles J., de Aza M. A. G. ‘Inclusion of feeding effects in the
generalized scattering matrix formulation of a finite array’. IEEE Antennas
and Wireless Propagation Letters. 2009;8:819–822.
[40] Montesinos-Ortego I., Zhang M., Sierra-Perez M., Hirokawa J., Ando M.
‘Systematic design methodology for one-dimensional compound slot-arrays
combining Method of Moments, equivalent circuit model and forward
matching procedure’. IEEE Transactions on Antennas Propagation. 2013;
61(1):453–458.
[41] Costanzo S., Casula G. A., Borgia A., et al. ‘Synthesis of slot arrays on
integrated waveguides’. IEEE Antennas and Wireless Propagation Letters.
2010;9:962–965.
[42] Pozar D. M. Microwave Engineering. John Wiley and Sons, Hoboken, NJ,
2012.
[43] Shavit R., Elliott R. ‘Design of transverse slot arrays fed by a boxed
stripline’. IEEE Transactions on Antennas and Propagation. 1983;31(4):
545–552.
[44] Robertson R., Elliott R. ‘The design of transverse slot arrays fed by the
meandering strip of boxed stripline’. IEEE Transactions on Antennas and
Propagation. 1987;35(3):252–257.
[45] Elliott R. S., Stern G. J. ‘The design of microstrip dipole arrays including
mutual coupling’. IEEE Transactions on Antennas and Propagation. 1987;
35(3):252–357.
 Chapter 7
Design of planar slotted waveguide
array antennas

In this chapter, we will present Elliott’s design procedure for planar standing wave
slot arrays including a detailed design example with computed and measured
results. For large planar arrays a modification to Elliott’s design procedure using an
infinite array model is sometimes preferred. Large standing wave arrays can be
broken up into sub-arrays with a parallel feed network to improve bandwidth.
Important parameters are the total normalised slot conductance of radiating wave-
guides and the total normalised resistance of feed waveguides. Additional examples
of slot array designs will be presented, including a procedure for designing a tra-
velling wave feed to excite radiating waveguides with either standing wave or
travelling wave slot arrays. Other design and analysis methods in the literature will
also be reviewed.

7.1 Introduction

Planar slot arrays are employed in applications requiring pencil beams by stacking
several linear arrays or waveguide sticks containing radiating slots next to each
other. They are preferable to conventional parabolic reflector antennas since they
can be designed to provide greater control of the pattern, especially of the sidelobe
levels. They are used in numerous radar and communication systems in ground and
space applications. The use of all metal waveguides in slot arrays is attractive in
deep space applications where severe radiation environment is encountered. Since
slot arrays occupy small volumes, they are easy to deploy.
Figure 7.1 shows a planar array consisting of six radiating waveguides stacked
next to each other with six longitudinal radiating slots cut in the broad wall of each.
Each radiating waveguide is shorted at a distance of a quarter guide wavelength
from the centre of the radiating slot at each end. The radiating slots are cut on a flat
plate which will form the top broad walls of all the radiating waveguides stacked
next to each other. Therefore a planar array like this is also called a flat plate array.
The radiating waveguides are fed by coupling slots cut on top broad wall of a feed
waveguide placed underneath the radiating waveguides and orthogonal to them.
Resonant centred-inclined coupling slots are commonly used in the design. Other
types of coupling slots will be discussed in Chapter 8.
136 Slotted waveguide array antennas: theory, analysis and design

Figure 7.1 A planar array of longitudinal radiating slots cut in rectangular

waveguides

In planar arrays, mutual coupling between radiating slots in the E-plane is sig-
nificant and needs to be included in the design and analysis. Inclusion of mutual
coupling is generally a challenging process. Planar slot arrays have been reported in
the literature since 1958 [1–4]. The design of a planar array was presented by Elliott
[5] that included the model of the coupling slots between the feed waveguide and the
radiating waveguides. The method of accounting for the internal higher-order mode
coupling, discussed in Chapter 6 may be applied to the design of planar arrays as
well. We will discuss Elliott’s procedure for planar slot arrays in detail since it
provides an accurate procedure for the designs of slot arrays. For large arrays
the computational burden of Elliott’s method becomes impractical, because of the
large amount of external mutual coupling computations. The method based on the
infinite array mutual coupling model is a viable alternative. Large arrays are often
broken into sub-arrays. The sub-array architecture is discussed with some examples,
followed by travelling wave feeds used for planar arrays. The chapter concludes with
a discussion on various design and analysis methods found in the current literature.

7.2 Elliott’s design procedure for planar standing wave arrays

The design procedure for planar standing wave arrays described by Elliott in 1988
[5] is presented here with the addition of internal higher-order mode coupling
effects [6]. This is the first rigorous procedure introduced for the design of planar
arrays and it has been used successfully in numerous designs. The design procedure
allows one to determine the offsets and lengths of the longitudinal radiating
slots and lengths and tilt angles of the centred-inclined coupling slots, which are
commonly used to excite the radiating waveguides. It is assumed that the excitation
coefficients of radiating slots are pure real but the design procedure is applicable
 Design of planar slotted waveguide array antennas 137

6
(6,1) (6,3) (6,5)
(6,2) (6,4) (6,6)

(5,1) (5,3) 5 (5,5)

(5,2) (5,4) (5,6)

4
(4,1) (4,3) (4,5)
(4,2) (4,4) (4,6)

(3,1) (3,3) 3
(3,2) (3,4) (3,5)
(3,6)

2
(2,1) (2,3) (2,5) (2,6)
(2,2) (2,4)

(1,1) (1,3) 1 (1,5)

(1,2) (1,4) (1,6)

Figure 7.2 Ordering of radiating and coupling slots in a planar array

for complex excitations as well. However, it should be remembered that the amount
of phase control achievable for resonant elements is limited, generally within
60
from the resonant values.
Let the radiating waveguides be numbered from i ¼ 1 through M with the first
one near the input end of the feed waveguide. There is a short circuit at a distance
equal to a half guide wavelength from the centre of the last coupling slot. The
centred-inclined coupling slot behaves like a series impedance because of its odd-
symmetric scattering, as explained in Chapter 8. Therefore a short is placed a half
guide wavelength away from the last coupling slot in the standing wave array.
Radiating slots in each radiating waveguide are numbered with two indices, the first
one, i being the number of the radiating waveguide while the second, j ¼ 1 through
N, denotes the radiating slot number in waveguide i. The first radiating slot is on the
left of the feed waveguide when the feed waveguide propagation direction is upward
as shown in Figure 7.2. For simplicity the value of N is assumed to be the same for
each radiating waveguide but the design procedure is valid even if it is not the case.
138 Slotted waveguide array antennas: theory, analysis and design

The first design equation for linear arrays given in (6.16) becomes (7.1) for
planar arrays.

Yija =G0 j fij jVijs c1

¼ (7.1)
a
Y11 =G0 j f11 jV11s
ci

The subscripts i and j denote the radiating waveguide number and the slot number in
that waveguide respectively. The reference slot 11 is the first slot in the first radiating
waveguide. The magnitude of the mode voltage of all slots in a radiating waveguide
is the same because of the half guide wavelength spacing between adjacent slots at
the design frequency. Therefore mode voltages do not appear in (7.1). A resonant
coupling slot acts as a transformer with the coupling coefficient ci given by

c2i ¼ S11 =ð1  S11 Þ (7.2)

where S11 is the reflection coefficient of the four-port centred-inclined coupling slot
with all the ports match-terminated. In Chapter 8 coupling slot models are dis-
cussed in detail. Clearly the active admittance of a radiating slot is proportional to
the slot voltage or the excitation coefficient whereas it is inversely proportional to
the coupling coefficient of the coupling slot in that radiating waveguide. Obviously
if the coupling coefficient is greater, a smaller value of active admittance is needed
to realise a given slot excitation.
The terms fij and Yija =G0 in (7.1) have been previously defined in (6.12) and
(6.15), respectively, with a single subscript for each. They are reproduced in (7.3)
and (7.4) in terms of double subscripts corresponding to those of each radiating
slot. Equation (7.4) is a combination of (6.15), (6.17) and (6.39).
p
cosðb10 Lij Þ
2kLij
fij ¼ fij ðx0ij ; Lij Þ ¼ cosðp x0ij =aÞ
2
2 (7.3)
p b10

2kLij k

where x0ij ¼ a=2 þ dij .

Yija 1
¼ (7.4)
G0 1
G0 =Yij þ MCij
2fij2

The mutual coupling term in (7.4) is obtained by combining (6.17) and (6.39),
and it is given by

ða=lÞ3 X M X N
0 s
MCij ¼ jðb10 =kÞðk0 bÞ V gmn;ij ðLmn ; dmn ; Lij ; dij Þ
Vij m¼1 n¼1 mn
s

b h i
þ j 10 eðg20 dÞ hij hi; j1 Vi;s j1 þ hij hi; jþ1 Vi;s jþ1 (7.4a)
g20
 Design of planar slotted waveguide array antennas 139

where

p coshðg20 Lij Þ
hij ¼ 2 cosð2pdij =aÞ
 (7.4b)
2kLij p 2
ðg20 =kÞ2 þ
2kLij

and gmn;ij is the expression for external mutual coupling between radiating slots mn
and ij, defined previously in (6.18) and (4.115). The prime in the summation
indicates that m 6¼ i and n 6¼ j simultaneously. In a planar array, the E-plane
coupling between slots in different waveguides generally becomes substantial.
The normalised self-admittance of the longitudinal radiating slot ij, Yij =G0 , is
obtained from the two-port scattering parameter of the slot with the reference plane
passing through the centre of the slot. S11ij is given by (7.5) which is the same as
(6.45) with double subscripts.

S11ij
Yij =G0 ¼ 2 (7.5)
1 þ S11ij

7.2.1 Design algorithm

Figure 7.3 shows the flow chart for the design algorithm. The procedure starts with
an initial determination of the array geometry, the number of radiating waveguides,
M, the number of radiating slots, N, waveguide interior dimensions, a, and b, slot
spacing, d, slot width, w, wall thickness, t, the aperture distribution and the design
frequency, f. Waveguide dimensions are chosen such that the half guide wavelength
of the feed waveguide at the centre frequency is equal to the ‘a’ dimension of the
radiating waveguide plus its sidewall thickness. This allows the radiating wave-
guides to be stacked next to each other and excited in phase by keeping the cou-
pling slots half guide wavelength apart, with alternating positive and negative tilt
angles providing the required 180 phase reversal. Scattering parameters of isolated
radiating slots are computed for a range of values of slot offsets and lengths nor-
malised to the resonant length for each offset, just as it was done for linear arrays in
Section 6.5. For coupling slots, the resonant length and the value of |S11| at reso-
nance are generated for a range of tilt angles. For simplicity coupling slots are all
assumed to be resonant.
For radiating waveguides 2 through M, there are 2N þ 1 unknowns, the offsets
and lengths of N slots and the coupling coefficient of the coupling slot exciting that
radiating waveguide normalised to the coupling coefficient of the first radiating
waveguide. By separating (7.1) into real and imaginary parts, two equations are
obtained for each slot, thereby producing 2N equations for N slots in each of these
radiating waveguides. An additional equation is obtained when one sets the total
normalised active conductance, G  tot of each radiating waveguide to a specific value
in the design, such as 2, thus producing 2N þ 1 equations in as many unknowns for
each of the radiating waveguides 2 through M. Since the normalised value of the
coupling coefficient is unity for the first waveguide, there is one less unknown or a
140 Slotted waveguide array antennas: theory, analysis and design

Start

Choose M, N,
a, b, f, w, t
Read slot data
lter = 1

Solve 2N or 2N + 1
simultaneous eqns. for rad
lter = lter +1 waveguides 1 through M

lter > 1
and
No Δ < tol?

Yes

Stop

Figure 7.3 Flow chart for the design of a planar array

total of 2N unknowns for this waveguide. For the slot 11 (i ¼ 1 and j ¼ 1), (7.1) is
not relevant. However, we obtain one equation for slot 11 by choosing the ima-
a
ginary part of the active admittance of Y11 to be zero. Thus for the first radiating
waveguide we have 2N equations in 2N unknowns. Thus there are as many equa-
tions as the number of unknowns for each radiating waveguide. However, the
simultaneous equations are nonlinear. The total normalised slot conductance G  tot
mentioned above is discussed in [7] and in Section 7.4.
At the start of the first iteration, the initial values of alternating slot offsets for
all radiating slots are chosen arbitrarily and their lengths have been set to equal
their resonant values. The choice of these values is not critical since they will be
updated to more accurate values in subsequent iterations. Recall that in Section 6.3
there was a discussion on the iterative design for linear standing wave arrays where
it was stated that 2N equations in 2N unknowns were solved using a quasi-Newton
method. The same technique is used in the planar array by solving the non-linear
equations of one radiating waveguide at a time [8]. After solving for the unknowns
in all radiating waveguides, we iterate the process. The iteration is stopped when
 Design of planar slotted waveguide array antennas 141

changes in linear dimensions of slots between successive iterations are less than
achievable values of manufacturing tolerance. In order to determine the manu-
facturing tolerance required to achieve the desired antenna performance, designers
evaluate how changes in slot parameters such as lengths, offsets and tilts on
the antenna characteristics affect the antenna performance, for example, gain,
sidelobe level and cross-polarisation. Based on such studies it may be possible to
set the tolerance needed to achieve the required performance. At low frequencies,
tolerance may not be an issue. The stopping criterion for iterations may be when
change in linear dimensions is below a predetermined level, obtained by a process
similar to that used to determine the tolerance. Experience has shown that good
design solutions are obtained typically after five iterations. The iterative design
process yields the offsets and lengths of all slots and all the coupling coefficients
relative to that of the first coupling slot. By enforcing a match at the input port or by
equating the total normalised resistance of the feed waveguide to unity, we deter-
mine the coupling coefficient c1 from (7.6), and hence all ci as follows.

X
M
 in ¼ c21
R  tot;i c
G  2i ¼ 1 (7.6)
i¼1

where

 i ¼ ci =c1
c (7.7)

is the normalised coupling coefficient of the coupling slot i relative to that of the
coupling slot 1 and

X
N
 tot;i ¼
G Yij =G0 (7.8)
j¼1

Using (7.2), the scattering parameter S11i of the ith coupling slot in a four-port
coupler is determined. Then from the original computed data of coupling slots one
can interpolate the resonant length and the tilt angle corresponding to each S11i .

7.2.2 Example of a 6  6 planar array

We demonstrate the design procedure for the 6  6 planar array shown in Figure 7.1.
A separable aperture distribution with excitation coefficients in each principal
direction is given by 0.513:0.814:1:1:0.814:0.513. The array factor of a linear array
multiplied by the element pattern of a typical slot with this aperture distribution
yields sidelobe levels of 19.5 and 22.5 dB in the E-plane and H-plane, respec-
tively. The design is carried out at 9.3 GHz. The interior dimensions of the feed
waveguide are 22.86 mm  10.16 mm while those of the radiating waveguides are
21.713  5.08 mm. The choice of half-height radiating waveguides helps in mini-
mising the effect of internal higher-order mode coupling problem in the junction
regions involving each coupling slot and a pair of straddling radiating slots in the
immediate vicinity [9]. The guide wavelength of the feed waveguide at 9.3 GHz is
142 Slotted waveguide array antennas: theory, analysis and design

45.46 mm. The wall thickness of each radiating waveguide is 0.508 mm while the
coupling slot thickness is 1.27 mm. When radiating waveguides are stacked next
to each other spacing between centres of adjacent guides is equal to 21.71 þ 2 
0.508 ¼ 22.73 mm. This is exactly equal to half guide wavelength in the feed
waveguide at 9.3 GHz and is also the spacing between adjacent coupling slots. There
is a short at a distance of half guide wavelength from the coupling slot farthest from
the feed end. The radiating slots have alternating offsets with half guide wavelength
spacing, which is 24.055 mm. At each end of every radiating waveguide, there is a
short at a distance of a quarter guide wavelength from the nearest slot centre.
We obtain computed slot data using the moment method solution to the perti-
nent coupled integral equations of isolated radiating or coupling slot. Such data may
be obtained from any computational electromagnetics code or from experimental
measurements. For a range of tilt angles of coupling slots, resonant lengths and |S11|
are determined with all the four ports match-terminated. The reference planes of all
the four ports pass through the centre of the coupling slot. Table 7.1 shows computed
data for coupling slots. For radiating slots, we determine the resonant length for a
range of values of offsets. Then for each offset we determine |S11| and arg S11 for a
range of values of slot lengths normalised to the resonant length. Radiating slot data
in Tables 7.2–7.4 are similar to those in Tables 6.2–6.4 in Chapter 6. In the design,
the total normalised slot conductance in each radiating waveguide was set at 2.8 so
that magnitudes of slot offsets would not be too small. The total normalised

Table 7.1 Resonant length in mm and |S11| of coupling slots

Tilt ( ) |S11| Resonant Tilt ( ) |S11| Resonant

length ¼ 2Lres length ¼ 2Lres
8.0 0.0015 15.82 18.0 0.0071 15.86
10.0 0.0024 15.83 20.0 0.0086 15.87
12.0 0.0033 15.83 22.0 0.102 15.88
14.0 0.0045 15.84 24.0 0.119 15.90
16.0 0.0058 15.85 26.0 0.137 15.91

Table 7.2 Resonant length of radiating slots

Offset (mm) Resonant length ¼

2Lres in mm
0.333 15.57
0.667 15.60
1.0 15.65
1.333 15.73
1.667 15.82
2.0 15.93
2.333 16.07
2.667 16.22
3.0 16.40
 Design of planar slotted waveguide array antennas 143

Table 7.3 Magnitude of S11 as a function of offset (mm) and normalised length
(mm)

 ¼ L=Lres 0.94
L 0.955 0.97 0.985 1.0 1.015 1.03 1.045 1.06
jdj
0.333 0.0027 0.0031 0.0034 0.0037 0.0038 0.0037 0.0034 0.0032 0.0029
0.667 0.0107 0.0122 0.0136 0.0146 0.0148 0.0144 0.0136 0.0126 0.0115
1.000 0.0239 0.0271 0.0300 0.0320 0.0326 0.0318 0.0300 0.0278 0.0256
1.333 0.0421 0.0474 0.0521 0.0553 0.0562 0.0548 0.0519 0.0483 0.0446
1.667 0.0646 0.0722 0.0789 0.0833 0.0845 0.0826 0.0786 0.0735 0.0683
2.000 0.0908 0.101 0.109 0.115 0.116 0.114 0.109 0.102 0.0957
2.333 0.121 0.133 0.142 0.149 0.150 0.147 0.141 0.134 0.126
2.667 0.153 0.166 0.177 0.184 0.185 0.182 0.176 0.168 0.159
3.000 0.187 0.201 0.212 0.219 0.221 0.217 0.211 0.202 0.192

Table 7.4 Phase of S11 in degrees as a function of offset and normalised length

 ¼ L=Lres
L 0.94 0.955 0.97 0.985 1.0 1.015 1.03 1.045 1.06
jdj
0.333 225.5 216.4 205.4 193.0 180.1 167.9 157.5 148.9 142.0
0.667 225.0 216.0 205.1 192.8 180.1 168.2 157.8 149.3 142.5
1.000 224.2 215.2 204.5 192.5 180.2 168.6 158.5 150.1 143.3
1.333 222.8 213.9 203.3 191.7 179.8 168.6 158.9 150.8 144.2
1.667 221.5 212.7 202.4 191.3 179.9 169.3 160.0 152.1 145.6
2.000 220.0 211.3 201.4 190.8 180.1 170.1 161.1 153.6 147.2
2.333 217.9 209.4 199.9 189.9 179.9 170.5 162.1 154.9 148.8
2.667 216.1 207.9 198.9 189.4 180.1 171.4 163.5 156.7 150.8
3.000 213.8 205.9 197.4 188.6 180.0 171.9 164.6 158.2 152.6

Table 7.5 Tilt angles and lengths of coupling slots

obtained from the design

Coupling slot Tilt angle ( ) Slot length (mm)

number, i
1 10.9157 15.8300
2 16.1483 15.8523
3 19.2417 15.8683
4 19.2417 15.8683
5 16.1483 15.8523
6 10.9157 15.8300

resistance in the feed waveguide was set to unity to achieve a match. Table 7.5 shows
the tilt angles and lengths of the six coupling slots. The values of offsets and lengths
of radiating slots are shown in Table 7.6. All data in Tables 7.5 and 7.6 were
obtained using the design procedure mentioned in Section 7.2.1.
144 Slotted waveguide array antennas: theory, analysis and design

Table 7.6(a) Offsets and lengths of radiating slots obtained from the design

Slot Offset Length Slot Offset Length Slot Offset Length

number (mm) (mm) number (mm) (mm) number (mm) (mm)
11 1.271 15.980 21 1.283 15.856 31 1.161 15.877
12 2.326 16.141 22 1.846 15.872 32 1.820 16.143
13 2.240 16.362 23 2.584 16.304 33 2.466 16.367
14 2.988 16.457 24 2.375 16.108 34 2.247 16.343
15 1.860 16.195 25 2.009 16.024 35 1.903 16.084
16 1.428 15.895 26 1.167 15.736 36 1.219 15.981

Table 7.6(b) Offsets and lengths of radiating slots obtained from the design

Slot Offset Length Slot Offset Length Slot Offset Length

number (mm) (mm) number (mm) (mm) number (mm) (mm)
41 1.219 15.980 51 1.167 15.736 61 1.428 15.895
42 1.903 16.084 52 2.009 16.024 62 1.860 16.195
43 2.248 16.343 53 2.375 16.108 63 2.988 16.457
44 2.466 16.367 54 2.584 16.304 64 2.240 16.362
45 1.820 16.143 55 1.846 15.872 65 2.326 16.141
46 1.161 15.877 56 1.283 15.856 66 1.271 15.979

0 H-plane realised
H-plane ideal
–10 E-plane realised
Pattern level in dB

E-plane ideal
–20

–30

–40

–50
–90 –60 –30 0 30 60 90
Theta in degrees

Figure 7.4 E-plane and H-plane patterns, ideal and designed

Slot ends become rounded or semi-circular when fabricated by milling. All

square ended slot lengths are increased by an amount equal to 0.215 slot width, to
account for square end to round end equivalence based on equal area criterion. The
equivalence between round ended and square ended slot has been discussed in
Section 5.3.3. See also (5.54).
Figure 7.4 shows the E-plane and H-plane patterns of the 6  6 array computed
using the method of moments (MoM) analysis used in this work described in
 Design of planar slotted waveguide array antennas 145

E-plane
–10
H-plane
Pattern level in dB
–20

–30

–40

–50
–90 –60 –30 0 30 60 90
Theta in degrees

Figure 7.5 Measured patterns of the 6  6 array

Chapter 8. It is assumed that the slots are embedded in an infinite ground plane and
therefore radiation is confined to the front half space only. The ideal patterns in
Figure 7.4 are obtained with the specified aperture distribution with the actual slot
locations. The electric field of each slot is assumed to have a half-cosine amplitude
and uniform phase distribution in the ideal pattern whereas in the realised pattern
the computed aperture distribution for each slot is used. Small discrepancies
between the ideal and realised patterns are attributed to imperfections in the design
procedure and implementation. Figure 7.5 shows the measured pattern of an array
fabricated with the use of design data shown in Tables 7.5 and 7.6. The discrepancy
between computed and measured patterns, especially in the far out sidelobe region,
is attributed primarily due to diffraction from the edges of the finite ground plane.
Manufacturing tolerances may also account for some discrepancy between theory
and measurement.

7.3 Design of large arrays using the infinite array mutual

coupling model

In large arrays with slowly varying aperture distributions, all slots except the
ones near the edges are surrounded by approximately identical neighbouring
elements but for alternating signs for the offsets. The external mutual coupling
expression for a representative slot n is then obtained using a simplified infinite
array model. Infinite array mutual coupling model has been used in a number of
antenna arrays. In microstrip reflectarrays the infinite array mutual coupling
model has been used successfully for many designs and analyses [10]. It has
been shown that the infinite array model for slot arrays yields errors for only a
couple of rows of slots near the E-plane edges. The accuracy of the model has
been established for moderate to large arrays [11,12]. In this model, slot n is
embedded in two doubly periodic infinite arrays, one with positive offsets and
146 Slotted waveguide array antennas: theory, analysis and design

a/2 + d Slot n dx

dz

Figure 7.6 Infinite array mutual coupling environment for slot n

the other with negative offsets, both in rectangular lattice with period 2dz along
the radiating waveguides and dx across. Figure 7.6 shows a few of the slots in the
infinite array model. The magnitude of the offset of all slots, dn, and their half
lengths, Ln, are those of slot n.

7.3.1 Infinite array mutual coupling expression [11,13]

The term associated with the external mutual coupling experienced by slot ij
expressed in (7.4) is given next where the prime in the summation means that m 6¼ i
and n 6¼ j.

ð1Þ ða=lÞ3 XM X N
0 s
MCij ¼ jðb10 =kÞðk0 bÞ V gmn;ij ðLmn ; dmn ; Lij ; dij Þ (7.9)
Vijs m¼1 n¼1 mn

In the infinite array environment, it is assumed that all slot voltages are equal or
slowly varying. In the latter case all slot voltages are approximated to be equal.
Therefore all slot lengths are the same and the magnitudes of slot offsets are equal.
Then (7.9) becomes

ð1Þ
X
1 X
1
0
MCij ¼ jðb10 =kÞðk0 bÞða=lÞ3 gmn;ij ðLij ;
dij ; Lij ; dij Þ (7.10)
m¼1 n¼1

The sign for the offset of slot mn is chosen based on its location because of alter-
nating slot offsets. We express (7.10) as follows.

ð1Þ 4ph0
MCij ¼  ðb10 =kÞðk0 bÞða=lÞ3 T (7.11)
w2
where
X
1 X
1
0
T¼ Imn;ij ðLij ;
dij ; Lij ; dij Þ (7.12)
m¼1 n¼1
 Design of planar slotted waveguide array antennas 147

and Imn;ij ðLij ;
dij ; Lij ; dij Þ ¼ Ymn Lij w=2. Ymn is specified in (4.113) for a pair of
slots with single subscripts whereas we have double subscripts for the planar
array. It is depicted in Figure 4.15 with slot lengths a and widths b whereas in
(7.12) the half length and width are Lij and w, respectively.
Equation (7.12) may be expressed as
ðð

pzij
T¼ HV cos ds0 (7.13)
2Lij
slot ij

where the integral is carried out over slot ij, zij is its local coordinate along the axis
of the slot, and cosine weighting corresponds to its aperture distribution. HV is the
magnetic field along the location of slot ij, produced by half cosinusoidal magnetic
currents of unit peak strength at the location of all other slots, radiating in free
space in the absence of the ground plane. Clearly each term inside the summation in
(7.12) corresponds to the integral in (7.13) when the contribution to HV due to unit
magnetic current in slot mn only is considered.

2  ðð

1 @ pzmn
HV ¼ þ k 2
S cos ds0 (7.14)
j4pwm0 @V2 0
2Lij
slot mn

where
X X expðjk0 R1 Þ expðjk0 R2 Þ

S¼ þ (7.15)
m n
R1 R2
h i
2 1=2
R1 ¼ fV  ðV0 þ 2mdz Þg þ fx  ðx0 þ ndx Þg
2
(7.16)
h i
2 1=2
R2 ¼ fV  ðV0 þ ð2m þ 1Þdz Þg þ fx  ðx0  2d þ ndx Þg
2
(7.17)

and the integral is carried out over all the slot apertures except that of ij. In the
double summation in (7.15) all positive and negative integer values and zero
are used for m and n except m ¼ 0 and n ¼ 0 in (7.16). Using Poisson’s sum formula
the double summations S may be expressed in terms of double infinite Floquet
series [13] as shown in (7.18), where all positive and negative integer values and
zero are used for p and q.
  
pl0 0 ql0 0
exp jk ðV  V Þ þ ðx  x Þ
j2p X X
0
2dz dx
S¼ "
2
2 #1=2
2dz dx p q
pl0 ql0
k0 1  
2dz dx


4pd
 1 þ ð1Þp exp jq (7.18)
dx
148 Slotted waveguide array antennas: theory, analysis and design

Using (7.18) in (7.14) we obtain

  
pl0 0 ql0 0

2  ðð X X exp jk0 ðV  V Þ þ ðx  x Þ
l0 @ 2dz dx
HV ¼ þ k 2
"

 #1=2
8pwm0 dz dx @V2 0 p q pl0 2 ql0 2
slot mm
1 
2dz dx



4pd pzmn
 1 þ ð1Þp exp jq cos ds0 (7.19)
dx 2Lij

The integral over slot mn yields

p


Lij
ppLij
w
2
2 cos (7.20)
p pp dz

2Lij dz

@2
After using (7.20) in (7.13), the integral over slot ij and the operation @V2 þ k02
yields

" p

2 #

pl0 Lij ppLij
k02 1 w
2
2 cos (7.21)
2dz p pp dz

2Lij dz

From (7.19) through (7.21) and (7.13) we obtain (7.22).



 # 1 þ ð1Þp exp jq 4pd
"

l0 k02 w2 X X pl0 2 dx
T¼ 1 "

 #1=2
8pwm0 dz dx p q 2dz
pl0 2 ql0 2
1 
2dz dx

2
p


Lij 2 ppLij
 "

2 #2 cos (7.22)
p 2 pp dz

2Lij dz

Using (7.22) in (7.11) we obtain the expression for the external mutual coupling for
an infinite array model. It is possible to obtain scattering data for a slot embedded
in an infinite array as a function of offset and length and use such data to design a
slot array. For a large array such a design works very well and for a moderate size
array it may still yield acceptable results.
 Design of planar slotted waveguide array antennas 149

7.4 Choice of total slot conductance in radiating waveguides

and resistance in feed waveguides

In the design of a planar standing wave array, the total slot conductance of each
radiating waveguide, G, is one of the design parameters. A linear array can be
designed with a specified value of G and with an impedance transformer to match
the input port. The frequency behaviour of linear arrays has been analysed by
considering equally spaced constant series or shunt loaded transmission lines [7].
The frequency response of the shunt or series load is ignored since the bandwidth of
the array is usually small compared to that of a load. The analysis employs the
transmission matrix of each unit cell composed of a length of transmission line
equal to the spacing between successive slots and the load impedance or admittance
[14]. A larger value of G yields a greater bandwidth for return loss whereas
the bandwidth for pattern performance such as the sidelobe level is lower.
A smaller value of G will produce the opposite effect. In the case of a planar array,
the total normalised shunt conductance G and the total normalised series resistance
R of the feed waveguide are design parameters. If R is not equal to 1, there is a need
to use a transformer. Derneryd and Petersson [7] have calculated the effects of
overloading on the return loss and sidelobe performance over 5% bandwidth for a
slot array of 10 wavelength diameter circular aperture designed to provide a 30
dB sidelobe level. We reproduce their results in Figures 7.7–7.9.

–20
Sidelobe level (dB)

–25

–30
0 1 2 3 4
Overloading G

Figure 7.7 First sidelobe level in the H-plane at the upper frequency limit as a
function of overloading in radiating waveguides. Reprinted from [7],
with permission
150 Slotted waveguide array antennas: theory, analysis and design

–20

Sidelobe level (dB)

–25

–30
0 4 8 12 16
Overloading R

Figure 7.8 First sidelobe level in the E-plane at the upper frequency limit as a
function of overloading in the feed waveguide. Reprinted from [7],
with permission

4
Voltage standing wave ratio

1
0 10 20 30 40
Overloading G∙R

Figure 7.9 Voltage standing wave ratio at the upper frequency limit as a function
of overloading of the waveguides. Reprinted from [7], with permission

Figures 7.7 and 7.8 show that in order to produce good pattern performance,
especially low sidelobes over the desired bandwidth G and R have to be small.
However, a better VSWR over the desired bandwidth is obtained for a value of the
product GR ¼ 8 as shown in Figure 7.9. A choice of G ¼ 2 yields the best
 Design of planar slotted waveguide array antennas 151

compromise for the performance of VSWR and sidelobe level over frequencies.
Therefore it is recommended that the values of G and R be chosen as 2 and 4,
respectively. These values are guidelines for designs since Derneryd and Peters-
son’s approximate analysis did not consider the frequency response of radiating and
coupling slots and that of the impedance transformer.

7.5 Sub-array architectures

The standing wave arrays are generally designed at the centre of the required fre-
quency band. At frequencies away from the design frequency the slot excitations
deviate from their ideal design values. Therefore the return loss and pattern char-
acteristics such as gain and sidelobe levels degrade. For large arrays the return loss
bandwidth is more limiting than the bandwidth for pattern performance. It was
shown in Figure 6.2 that the return loss bandwidth deteriorates as the number of
slots in a linear standing wave array increases. A similar behaviour is found for
planar arrays also. It is possible to improve the performance of large arrays by
breaking them into several sub-arrays, with each sub-array fed by a corporate
power divider. The sub-array architecture is based on a compromise between the
superior performance achievable by small sub-arrays and the cost and complexity
of the required power divider network when an array is made up of a large number
of sub-arrays. In this section we show some examples of sub-array architectures.
Section 9.11.5 in Chapter 9 also has some discussions on the sub-array architecture.

7.5.1 A planar array of 160  160 elements consisting of 10  10

element sub-arrays
Glacier and land ice surface topography interferometer (GLISTIN), consisting of two
large Ka band 4 m  1 m slotted waveguide array antennas was proposed for a
synthetic aperture radar for mapping ice sheets and glaciers [15]. For demonstrating
the technology a 1 m  1 m array was developed and built. It consisted of 160  160
elements and was made up of 16 so-called sticks each with a separate waveguide
feed port [15]. Received signals from the sixteen sticks was combined to produce
digital beams in the receive mode in the H-plane or elevation. A 10  160 element
array was used as a transmit stick [15]. Figure 7.10 shows the back of the 17 sticks.
Each stick is made up of four 40  10 element modules, one of which is shown in
Figure 7.11. The back of the 40  10 module has four ports for adjusting each
10  10 sub-array for possible phase differences between sub-arrays introduced by
the manufacturing imperfections. The return loss and pattern performance of this
array was found to be satisfactory for the radar application even though the strong
mutual coupling between the radiating slots in the E-plane reduces the bandwidth
substantially. Section 7.6.1 presents the design and measured results of this system.

7.5.2 A slot array with 8  8 radiating elements consisting

of two sub-arrays
Figure 7.12 shows a radiometer array consisting of 8  8 radiating slots divided
into two sub-arrays [16]. Each 8  4 sub-array is centre-fed in each direction and
152 Slotted waveguide array antennas: theory, analysis and design

Figure 7.10 Back of the 1 m  1 m demo array antenna consisting of 16 receive

sticks and one transmit stick (at the bottom).
IEEE 2009, reprinted
from [15], with permission

hence the maximum length of line in two orthogonal directions corresponds to one
and two guide wavelengths respectively. This architecture with small sub-arrays
was able to meet low sidelobe levels and low levels of reflection coefficients over
approximately 5% bandwidth. This array designed with an even symmetry for the
slot offsets in the H-plane helped in reducing the grating lobes called butterfly lobes
produced by the conventional slot arrays with alternating positive and negative
offsets. Butterfly lobes are discussed in Section 9.9. Section 7.6.2 presents results
for sidelobe levels of the slot arrays exhibiting symmetry in the two halves of the
H-plane and for the array with alternating slot offsets along the entire H-plane.

7.5.3 A slot array with four-element sub-arrays

Figure 7.13 shows a two-layer 16  16 slot array consisting of 64 sub-arrays
designed to operate at 60 GHz that was built using the diffusion bonding process
[17]. Each sub-array has four longitudinal offset slots in two radiating waveguides.
Figure 7.14 shows a schematic view of a quadrant of the array.
 Design of planar slotted waveguide array antennas 153

(a)

(b)

Figure 7.11 One of four modules of a stick, each consisting of four 10  10 sub-
arrays: (a) Radiating side of the module. Four modules are combined
in one stick. (b) Rear side of one module showing power dividers.

IEEE 2009, reprinted from [15], with permission

Offset radiating slot

Angled
Radiating slot layer, coupling slot
single-clad braze
sheet, AI Brazing short
Main body,
double-sided AI Offset input slot
hog-out Power divider,
Input sheet, double- single-sided AI
clad braze sheet, AI hog-out

Figure 7.12 An 8  8 array divided into two sub-arrays.
IEEE 2010, reprinted
from [16], with permission

The lower layer contains the corporate power divider made up of H-plane
T junctions. A single H-plane T junction shown in Figure 7.15(a) divides
the incident power into the desired power ratio. In this array all T junctions are
designed to provide equal power division. The centred septum length lf and the iris
width wf are chosen to provide a match at the input port as long as the branch ports
154 Slotted waveguide array antennas: theory, analysis and design

75 mm
Aperture size
16 × 4.2 = 67.2 mm

76 mm
Figure 7.13 A 16  16 array divided into two sub-arrays.
IEEE 2011, reprinted
from [17], with permission

Radiating slot

Cavity

Coupling aperture

Full-corporate-feed
waveguide

Feed aperture
(WR15)

z Radiating slot

Coupling aperture x Cavity

Feed waveguide Feed aperture (WR15)

Figure 7.14 A schematic view of a quadrant of the array with H-plane T junctions
in the lower layer in black and the upper layer consisting of radiating
slots.
IEEE 2011, reprinted from [17], with permission
 Design of planar slotted waveguide array antennas 155

Port 4 Port 2

lf
wf

(a) (b) Port 5 Port 1 Port 3

Figure 7.15 H-plane power dividers: (a) a single H-plane T-junction divider;
(b) three H-plane T’s for four-way power.
IEEE 2011,
reprinted from [17], with permission

Magnetic field

Figure 7.16 Excitation of two radiating waveguides, each containing two

radiating slots.
IEEE 2011, reprinted from [17], with permission

are matched. It is generally possible to design such a T junction with greater than
20 dB return loss over a broad bandwidth. Figure 7.15(b) shows that a combination
of three H-plane T’s can provide one to four way power division. With the use of
additional H-plane T junctions one can achieve the required power division for the
entire array.
Figure 7.16 shows the method of exciting each of the two radiating waveguides
in each sub-array. The two radiating waveguides are excited by a coupling window
which is a longitudinal offset slot in the waveguide in the lower layer. The coupling
window with its magnetic field in the longitudinal direction will excite each of the
156 Slotted waveguide array antennas: theory, analysis and design

two radiating waveguides equally in amplitude but with 180 out of phase.
Therefore, in order to excite the four slots with equal phase, the offsets of slots in
the same waveguide are on the same side of the centreline whereas the offsets of
slots in the two waveguides are on opposite sides. The slot spacing is between
0.812 and 0.925 wavelengths in the frequency range of operation. In the array
design, external mutual coupling was modelled based on an infinite array of four-
element sub-arrays. Uniform aperture distribution was desired but because of lim-
itations in the infinite array mutual coupling model the sub-arrays near the edges
had significant error. This array achieved a return loss better than 14 dB over 8.3%
bandwidth and better than 80% efficiency. Efficiency is the ratio of the antenna
gain to the ideal directivity of the antenna assuming uniform aperture distribution.
An efficiency of 80% corresponds to a value of gain equal to 1 dB below the ideal
directivity. It is possible to achieve such a broadband performance in a slot array if
many small sub-arrays are used. However, for large arrays containing thousands of
elements, the power divider may become very complicated and one may have to
use somewhat larger sub-arrays.

7.6 Examples of planar slot array designs

7.6.1 Large array design using the infinite array model [11,15]
In Section 7.5.1 we discussed a large Ka band slot array consisting of 16 receive
sticks, each with 160 radiating slots in the E-plane and 10 radiating slots in the
H-plane. Each stick has four modules of 40  10 radiating slots each module
consisting of four sub-arrays of 10  10 radiating slots. Figures 7.11 and 7.12 show
the back of the full array and the 40-element module. This array was designed for
uniform aperture distribution. The infinite array mutual coupling model was
employed in the design. Internal higher-order coupling effects were found to be not
significant for this array and therefore all radiating slots were identical but for
alternating signs of offsets. Similarly all coupling slots were also identical but for
alternating tilts. In each sub-array containing 10  10 radiating slots G  tot was set to
2.8 since a lower value of G  tot ¼ 2 would provide rather small offsets. In addition,
the bandwidth for the pattern performance is not critical here since the return loss
bandwidth is more sensitive than the pattern bandwidth (see Section 7.4). The inner
dimensions of the radiating waveguides are 5.707 mm  1.43 mm. Slot width and
wall thickness are both equal to 0.508 mm. At the design frequency of 35.66 GHz a
slot length of 4.163 mm and offset of 0.4 mm yields a normalised active con-
ductance of 0.28 so that G  tot in each radiating waveguide would be 2.8. The total
normalised resistance of each feed waveguide has been chosen to be 4.0 and a shunt
series coupling slot is used to match it to an input waveguide in another layer.
Therefore the coupling coefficient of each coupling slot is given by G  tot c2 ¼ 0:4 or
c ¼ 1=7. The coupling slot models are discussed in Chapter 8.
2

A value of c2 ¼ 1=7 ¼ 1S S11

11
corresponds to the S11 of each resonant coupling
slot ¼ 0.125. The feed and input waveguide inner dimensions are 5.707 mm 
2.85 mm. The coupling slot is found to have a length of 4.140 mm and a tilt of 24
 Design of planar slotted waveguide array antennas 157

Radiating Input
slot slot

Coupling
slot

Feed
guide

Radiating Input
guide guide

Figure 7.17 A sub-array containing 100 radiating slots, 10 coupling slots and an
input shunt-series coupling slot

with respect to the axis of the feed waveguide. At the centre of the feed waveguide
of each 10  10 sub-array, there is a shunt-series coupling slot that matches a
normalised resistance of 4.0 to the input port on one side. The other side of the
input waveguide is shorted at a distance of a quarter guide wavelength from the
centre of the slot. The shunt series coupling slot has an offset of 2.081 mm and
length equal to 4.242 mm. It is centred transverse to the feed waveguide. Fig-
ure 7.17 shows the slots in a sub-array.
Figures 7.18 and 7.19 show the E-plane and H-plane patterns of a 10  40
module, respectively. The computed patterns are obtained from the moment
method analysis of an infinite array of 10  10-element sub-arrays with uniform
excitations of all sub-arrays. From the moment method solution we determine the
pattern of each sub-array and then determine the pattern of four sub-arrays in the
10  40 module. Measured patterns are in very good agreement with computed
plots except in the far-out sidelobe regions. The discrepancies may be attributed to
the fact that the computation assumes an infinite ground plane and therefore it does
not account for the edge diffraction from the finite ground plane. In addition,
imperfections in the fabrication process, especially the 1 mil (25 m) mechanical
tolerance in the manufacturing process may account for some error. Reflection
158 Slotted waveguide array antennas: theory, analysis and design

0
Meas, 35.66 GHz...... calc
–10
Relative magnitude (dB)

–20

–30

–40

–50

–60
–90 –60 –30 0 30 60 90
Theta (deg)

Figure 7.18 Azimuth (E-plane) patterns of a 10  40 module at 35.66 GHz.

IEEE 2009, reprinted from [15], with permission

0
Meas, 35.66 GHz...... calc
–10
Relative magnitude (dB)

–20

–30

–40

–50

–60
–90 –60 –30 0 30 60 90
Theta (deg)

Figure 7.19 Elevation (H-plane) patterns of a 10  40 module at 35.66 GHz.

IEEE 2009, reprinted from [15], with permission

coefficient plots shown in Figure 7.20 indicate some discrepancy, especially at the
frequency where the 10  40 module is tuned, partly because of errors in the fab-
rication process. Figure 7.21 shows the H-plane patterns of 16 sticks and the pattern
computed by the moment method. The light traces corresponding to the patterns of
 Design of planar slotted waveguide array antennas 159

0
Measured
10 × 10s

–10
Refl. coefficient (dB)

Calculated
–20 10 × 10

Measured
–30 power divider

–40
33.66 34.16 34.66 35.16 35.66 36.16 36.66 37.16 37.66
f (GHz)

Figure 7.20 Reflection coefficient of each sub-array and the power divider of a
10  40 module.
IEEE 2009, reprinted from [15], with permission

–10
dB

–20

–30

–40
–9 –6 –3 0 3 6 9
Theta (deg)

Figure 7.21 E-plane patterns of 16 sticks (light) and computed (dark) at

35.66 GHz.
IEEE 2009, reprinted from [15], with permission

different sticks are not identical because of the repeatability problem associated
with the fabrication process. Grating lobes near
8 are due to the wall thickness
between sub-arrays. A key requirement of the array design and fabrication was to
align the 16 beams in the E-plane radiated by each of the 16 sticks, within 1/10th of
the beamwidth or 0.042 . This objective was met.
160 Slotted waveguide array antennas: theory, analysis and design

Table 7.7 Average sidelobe specification

Elevation angles ( ) Average SLL (dB)

20–30 34
30–40 36
40–70 38
70–90 39
90–150 40
150–180 30

7.6.2 An 8  8 slot array for radiometer application

Figure 7.13 shows an 8  8 array employed in a radiometer. Three such antennas,
all 8  8 arrays operating at S, C, and X bands were developed, operating at 2.6, 5.2
and 10 GHz, respectively. The three antennas had the same specifications and
similar designs. Waveguide wall thickness for the three frequencies were 0.762,
1.016 and 1.27 mm and machining tolerances were 0.127, 0.0508 and 0.0254 mm,
respectively. The 3-dB beamwidth of the antenna had to be <13 . The sidelobe
levels averaged over all azimuthal values is specified in different elevation angular
regions in Table 7.7. The pattern performance had to be satisfied over 4% band-
width. The return loss had to be no less than 10 dB with the average value better
than 15 dB in the 4% frequency band.
A separable 35 dB Taylor n ¼ 4 aperture distribution sampled from the con-
tinuous distribution was employed in both principal planes of the 8  8 array. The
amplitude taper in each direction was 13.86 dB with 27.7 dB dynamic range.
The 64-element slot array is divided into two sub-arrays of 8  4 elements with two
feed waveguides. Each feed waveguide consists of eight coupling slots and eight
radiating waveguides with four radiating slots per radiating waveguide. Elliott’s
design procedure described in Section 7.2 was used to design both sub-arrays,
sequentially in the iterative process. The two sub-arrays couple through external
mutual coupling between radiating slots. The total normalised slot conductance in
each radiating waveguide was chosen to be 2 and the total normalised slot resis-
tance in each feed waveguide was set to 1.5. A smaller value of the total normalised
slot resistance has been chosen to achieve a good sidelobe performance over the
bandwidth. A shunt-series coupling slot is used in the input waveguide to excite
each feed waveguide with the specified return loss performance.
The conventional array architecture shown in Figure 7.22 introduced grating
lobes called butterfly lobes at about q ¼ 80 in the H-plane. Such grating lobes
have been treated widely in the literature [18,19] and also discussed in Section 9.9
in this book. The butterfly lobes may be eliminated in slot arrays with centred
longitudinal slots excited by compound irises [20] or by alternating asymmetric
ridges in the waveguide [21]. Iris excited slots will be discussed in Chapter 8. In
the array configuration shown in Figure 7.23 the butterfly lobes are reduced by
designing a symmetric arrangement of slots with respect to a vertical centre line.
 Design of planar slotted waveguide array antennas 161

Figure 7.22 Conventional array architecture with alternating slot offsets.

IEEE 2010, reprinted from [16], with permission

Uniform phase distribution of slot apertures was achieved with 180 phase
difference between the excitations of the two feed waveguides as shown in
Figure 7.23. Tilt angles of coupling slots in the two sub-arrays are in the same
direction in Figure 7.22 whereas they are in opposite directions in Figure 7.23.
Table 7.8 shows the sidelobe levels achieved for the two arrays shown in Fig-
ures 7.22 and 7.23. The symmetric design yields lower sidelobes over the angular
ranges beyond 30 whereas it increases the sidelobe level slightly for angular
region below 30 .
Far-field patterns of the S-band antenna, both computed by the finite element
technique HFSS and measured using the spherical near-field technique are found
to have very good agreement over nearly 60 dB dynamic range [16]. They are
reproduced in Figures 7.24 and 7.25. The spherical far field pattern is mapped onto
a circle, with q varying from 0 to 180 radially and f varying from 0 to 360
azimuthally. q ¼ 180 point in the far field sphere is stretched into the cir-
cumference. Average sidelobe levels in different angular ranges obtained from
computed and measured data are compared in Table 7.9. Excellent agreement
between two sets of data is noted and the design margin is found to be more than
adequate.
162 Slotted waveguide array antennas: theory, analysis and design

Figure 7.23 Symmetric array architecture.
IEEE 2010, reprinted from [16],
with permission

Table 7.8 Average sidelobe for the X-band antennas (
IEEE

2010, reprinted from [16], with permission)

Elevation angles ( ) Average SLL (dB) Average SLL (dB)

(Figure 7.22) (Figure 7.23)
20–30 37.98 36.06
30–40 37.60 39.16
40–70 40.56 44.07
70–90 41.36 45.70

7.7 Design of a travelling wave feed for a planar array

A planar array of slots can be designed to have a travelling wave feed [22] in one
or both principal directions. Figure 7.26 shows a travelling wave feed in which
centred-inclined coupling slots excite 80 radiating waveguides, each containing
8 radiating slots. The feed waveguide is match terminated at the far end. The
radiating waveguides consist of longitudinal radiating slots in the standing wave
 Design of planar slotted waveguide array antennas 163

180° 0
150
–10
100

–20
50
θsinφ

dB
0° –30

–50
–40

–100
–50
–150
–180° –60

180°
–180°

–150 –100 –50 0° 50 100 150

θcosφ

Figure 7.24 Far-field pattern of the S-band antenna computed by HFSS at

2.6 GHz.
IEEE 2010, reprinted from [16], with permission

180° 0
150
–10
100

–20
50
θsinφ

dB

0° –30

–50
–40

–100
–50
–150
–180° –60
–150 –100 –50 0° 50 100 150
–180°

180°

θcosφ

Figure 7.25 Far-field pattern of the S-band antenna measured at 2.6 GHz.

IEEE 2010, reprinted from [16], with permission

configuration. Coupling slots in the feed waveguide may be designed independent

of the radiating slots if we assume a certain amount of total active conductance,
Gtot, in each radiating waveguide at the design frequency. For the specified value of
Gtot, for example, Gtot ¼ 2, each coupling slot may be considered as a four-port
164 Slotted waveguide array antennas: theory, analysis and design

Table 7.9 Average sidelobe for the S-band antennas (
IEEE 2010, reprinted
from [16] with permission)

Elevation Specification (dB) Average SLL in dB Average SLL

angles ( ) computed by HFSS in dB measured
20–30 34 36.7 36.4
30–40 36 38.4 38.3
40–70 38 43.5 43.0
70–90 39 46.0 45.8
90–150 40 50.8 52.1

Figure 7.26 A planar array of longitudinal slots with a travelling wave feed.

IEEE 2011, reprinted from [22], with permission

network with the two ports in the radiating waveguides match-terminated. If we

have a centred feed waveguide with travelling wave type of radiating slots in the
forward-fire backward-fire arrangement (discussed in Chapter 9) and if each
radiating waveguide is properly matched, each four-port coupling slot sees match
terminations in the radiating waveguide. Therefore one needs to design two-port
coupling slots in the feed waveguide for a match at the feed port and to achieve the
desired excitations at each radiating waveguide. It is assumed that the excitation
coefficients are pure real but for the phase added by the propagating travelling
wave and therefore all coupling slots are assumed to be at resonance. The proce-
dure given here for the centred-inclined coupling slots can be modified for other
types of coupling slots such as shunt-series coupling slots that are offset long-
itudinal in the feed waveguide and transverse to the radiating waveguide as well as
compound coupling slots that are offset and inclined in the feed waveguides.

7.7.1 Design procedure for a travelling wave feed in terms

of scattering wave representation of a coupling slot
A centred-inclined slot coupler at resonance can be represented by a four-port
coupler where all ports are planes passing through the centre of the slot, two in the
 Design of planar slotted waveguide array antennas 165

Y
4

f
X
1 2

Figure 7.27 Slot coupler as a four-port device

feed waveguide and two in the radiating waveguide. The scattering matrix of such a
coupler at resonance (see Figure 7.27) is given by [23]
2 3 2 3
S11 S12 S13 S14 r p q q
6 S21 S22 S23 S24 7 6 p
7 6 r q q 7 7
½S  ¼ 6
4 S31 S32 S33 S34 5 ¼ 4 q q p (7.23)
r 5
S41 S42 S43 S44 q q r p
where p ¼ (1  r) and q ¼ [r(1  r)]1/2. If the S-matrix for a positive value of tilt
angle f is S(r,p,q), then for a negative value of f the S-matrix becomes S(r,p,q).
The tilt angle f is shown in Figure 7.27 with ports 1 and 2 in the feed waveguide
and 3 and 4 in the radiating waveguide. If ports 3 and 4 are terminated in reflection
coefficients G3 ¼ G4 ¼ G (¼0 if ports 3 and 4 are matched), the coupler may be
represented as a two-port network.
" 0 #
r 1  r0
½S2  ¼ (7.24)
1  r0 r0
where

r0 ¼ r þ 2q2 ½1=G  1 þ 2r1 (7.25)

The scattering matrix (7.24) may be expressed in terms of a transmission matrix,
relating the incident and reflected waves, a1n, b1n at the input port to the corre-
sponding waves, a2n, b2n at the output port of the nth slot.
   
a1n a2n
¼ ½Tn  (7.26)
b1n b2n
where
 0 
1 r 1
½Tn  ¼ (7.27)
1  r 1  2r0
0
r0
166 Slotted waveguide array antennas: theory, analysis and design

The transmission matrix of a waveguide of length d equal to the spacing

between adjacent coupling slots is given by
 
0 expðjb10 dÞ
½ Tw  ¼ (7.28)
expðjb10 dÞ 0
where b10 is the phase constant of the TE10 mode in the waveguide. The slot spacing
d is chosen to produce the main beam at an angle, typically two to three beamwidths,
away from broadside. The aperture distribution is chosen to produce the desired
pattern shape. The parameter r0 of the Nth coupling slot closest to the matched load is
assigned an arbitrary value. This value is not critical since it will be changed in
subsequent iterations. A larger value of r0 will reduce the power dissipated in the
load while increasing the reflection coefficient at the input port whereas a smaller
value will produce the opposite effect. The complex phasor representing the wave
going towards the load, b2N is initially set at a reference value of unity. The complex
phasor a2N ¼ 0 since there is no reflection from the matched load. The excitation of
the radiating waveguide N is determined readily. From the transmission matrices
½Tw  and ½Tn  of slot N, and a2N and b2N , we can determine the complex phasors a1N
and b1N , and hence a2N 1 and b2N 1 . It is then possible to choose the value r0 of slot
N  1 so that the radiating waveguide N  1 has the desired amplitude of excitation.
We then determine the complex phasors a1N 1 and b1N 1 , and hence a2N 2 and
b2N 2 . One proceeds this way towards the source end, until all the coupling slot
parameters r0 and hence the scattering parameter of each four-port coupler, r are
known. From the values of complex phasors a11 and b11 of slot 1, the input reflection
coefficient is determined. The power dissipated in the load relative to the power
incident in the input port is 1=j a211j. The procedure is somewhat similar to the design
of travelling wave arrays discussed in Chapter 6.
Subsequent iterations with greater or smaller values of r0 of slot N will even-
tually yield a desirable solution with an optimum value of load power and input
reflection coefficient. From an interpolation of the computed moment method data
for the slot parameters, all the coupling slot tilt angles and lengths are obtained.
From the excitations of the radiating waveguides computed in this exercise,
radiating slot parameters may be obtained so as to attain the desired total normal-
ised active slot conductances, using Elliott’s procedure discussed in Section 7.2.
This process will yield the offsets and lengths of radiating slots. In this discussion,
all coupling slots are assumed to be resonant and hence the phases of the radiating
waveguides are not constrained. The phase distribution is nominally that of the
incident TE10 mode waveguide with 180 phase reversals due to the alternating slot
offsets. With non-resonant coupling slots, an additional degree of freedom may be
gained at the cost of increased complexity of the design.

7.7.2 Computed results

Computed results for the 80  8 element array with a travelling wave feed are shown
in Figures 7.28–7.30. The aperture distribution is assumed to be uniform in ampli-
tude and the coupling slot spacing is chosen to be d/lg ¼ 0.464 where lg is the guide
 Design of planar slotted waveguide array antennas 167

0
–10
Pattern level in dB
–20
2.2% PL
–30 3.6% PL
Desired
–40

–50

–60
–90 –60 –30 0 30 60 90
Theta in degrees

Figure 7.28 E-plane pattern of the planar array with an 80-element travelling
wave array feed (PL ¼ load power/incident power in per cent).

IEEE 2011, reprinted from [22], with permission

360
Excitation phase (deg.)

270

2.2% PL
180
3.6% PL

90

0
0 20 40 60 80
Element number

Figure 7.29 Excitation phase of the radiating waveguides (PL ¼ relative power in
the load).
IEEE 2011, reprinted from [22], with permission

35
30
Tilt angle (deg)

25
2.2% PL
20 3.6% PL
15
10
5
0 20 40 60 80
Element number

Figure 7.30 Tilt angles of coupling slots (PL ¼ relative power in the load).

IEEE 2011, reprinted from [22], with permission
168 Slotted waveguide array antennas: theory, analysis and design

wavelength ¼ 1.4741 l0 and l0 is the free space wavelength. In this case the main
beam in the E-plane will be at q ¼ 93 or 3 off normal if the radiating slots in each
waveguide are designed to be of standing-wave type with uniform phase aperture
distribution. Figure 7.28 shows the E-plane pattern of the array shown in Figure 7.26
for two designs. The power dissipated in the load relative to that at the input port for
the two cases are 2.2% and 3.6%, respectively. The desired case shows the pattern
with ideal phase distribution of the incident TE10 mode wave but for the 180 phase
reversal due to alternating slot offsets. The actual phase distribution deviates from
the ideal linear phase due to the presence of reflections at each slot. The phase
distributions for the two designs are illustrated in Figure 7.29. The effect of such
phase deviations on the patterns is not significant. Figure 7.30 shows the tilt angles of
coupling slots. Clearly the elements close to the feed have small amounts of coupling
and hence relatively small values of tilt angles as discussed in Chapter 6. Slots with
small values of tilt are sensitive to manufacturing tolerances. In the design of shaped
patterns asymmetric amplitude distributions with larger amplitudes near the feed end
are preferred to reduce the effects of manufacturing tolerances.

7.8 A review of other methods for analysis and design

in the literature
In Chapters 6 and 7 we have primarily used Elliott’s technique and some variations
for the design of linear and planar arrays. For large arrays, the infinite array mutual
coupling model is an efficient alternative to the original Elliott procedure. In this
section we first present other variations of Elliott’s design procedure found in the
literature. In addition, some of the analysis techniques are listed. Subsequently some
generalised scattering wave models used in the design and analysis will be discussed.
In the design examples we used the moment method to obtain the resonant
length and scattering parameters of radiating and coupling slots. Other computa-
tional electromagnetics codes have been used to obtain slot data. Brown employed
Elliott’s design procedure with the scattering characteristics of isolated slots
obtained from the commercial finite element analysis code HFSS [24,25]. The
longitudinal radiating slot is generally modelled as a shunt element with a con-
ductance in parallel with a susceptance.
Solbach [26] recommended a model for the longitudinal radiating slot in the
form of a shunt element consisting of a resistance R in series with an open circuited
stub of characteristic impedance Z0. At the resonant frequency of the slot, R is the
reciprocal of the resonant slot conductance and the stub length is a quarter wave-
length. At other frequencies the stub length is inversely proportional to the wave-
length. We examined this model for slots in standard height waveguides and half
height waveguides in a range of
6% over the resonant frequency. In that
frequency band R varied about
2.5% while the characteristic impedance of the
stub, Z0 varied about
10%. There was no particular advantage of this model over
Lagrangian interpolation of S parameter data and resonant lengths of slots.
Taddei et al. employ Elliott’s design and moment method analysis with a
graphical user interface [27]. Gatti et al. [28,29] extended Elliott’s design method
 Design of planar slotted waveguide array antennas 169

to complex excitations by making the active admittances complex instead of pure

real values. Since resonant slots have limited phase control, the phase value is
constrained to within
60 from the resonant values. Infinite array mutual coupling
model has been found to be applicable in a number of designs [30–32].
Gulick and Elliott have used the moment method solution of the integral
equations for the aperture electric fields of all slots for the design of a small planar
array [33]. Even for a moderate sized array the computational burden of this
method will make it impractical. One may start with an initial design using Elliott’s
procedure and then improve that by using the moment method analysis of the slot
array [34]. In Chapter 8 we will discuss the use of moment method analysis in the
optimisation of a slot array initially designed by Elliott’s method.
Elliott’s design equations were cast in the form of an analysis for a linear array
by Hamadallah [35]. His analysis employed an iterative procedure as well as a
direct matrix solution. Coetzee et al. extended the iterative analysis to a planar
array [36] whereas Casula and Mazzarella used the direct solution to a planar array
by adding the coupling slot model [37]. Although the above-mentioned analysis
techniques may also be used for design optimisation, we recommend two accurate
analysis methods based on generalised scattering matrices, discussed in Section 7.9,
with a potential for design and optimisation discussed at the end of this section.
Manuilov et al. [38] solve the integral equations for the slot aperture field using
the Galerkin’s method with spectral domain Green’s functions and Chebyshev
polynomials of the first and second kind for basis and test functions. Fast and
accurate full wave analysis of large slot arrays is accomplished by enforcing the edge
conditions. Liu and Long analyse a unit cell of an infinite uniformly slit waveguide,
or periodically cut longitudinal slots or crossed slots [39]. The former supports a
leaky wave mode as well as a slot mode (the latter has a propagation constant close
to free space with a near end-fire beam contaminating the leaky wave radiation)
while the periodic longitudinal slot array supports only the leaky wave mode. The
crossed slot array supports both a leaky wave mode and a surface wave mode. RWG
basis functions are used in the slot apertures [40]. The commercial finite element
code HFSS has been found to provide very accurate results for the radiation pattern
and the input impedance of slot arrays for radiometer applications, requiring low
sidelobes [15]. While such general purpose codes are ideal for the analysis, they are
not so convenient for design and optimisation. A genetic algorithm (GA) optimisa-
tion using the moment method analysis was presented in [41]. For that technique the
initial design was carried out by Elliott’s method and subsequently optimised. This
method will be discussed in detail in Chapter 8.

7.9 The generalised scattering matrix model for analysis

and design
Chiarandini and Morini have presented a generalised scattering matrix model for the
analysis and design of slot arrays [42]. In this model, a planar array is modelled in
the form of three blocks as shown in Figure 7.31. B1 represents the main line with
four ports. It is connected to block B2 representing the branch waveguides through
170 Slotted waveguide array antennas: theory, analysis and design

t1 t2
Feed B1 B2 B3

Main line Radiation

Radiating slot

Inclined slot

Figure 7.31 A planar array of slots modelled by three blocks.
IEEE 2001,
reprinted from [42], with permission

[a1i ] [ao1]
Slot 1
[S1]
[b1i ] [bo1] Radiation
Feed
network [a2i ] [ao2] network
[ai] [SF] [SH]
Slot 2
[S2]
[b2i ] [bo2]
[bi] [aH]

[bH]
[api ] [aop]
Slot p
[Sp]
[bpi ] [bop]

Figure 7.32 A planar array of slots modelled by generalised scattering matrices

(GSM).
IEEE 2000, reprinted from [43], with permission

an inclined coupling slot of a certain thickness t1. The coupling slot is treated as a
waveguide of length t1. Block B3 represents the exterior half space region. It is
connected to each branch line or radiating waveguide through radiating slots of
thickness t2. Each block is characterised by generalised scattering matrices deter-
mined by a full wave method such as the commercial code HFSS. The inter-
connection between blocks by slots or waveguides is modelled in terms of a few
modes. Thus the generalised scattering matrix can be used for analysis as well as
design and optimisation. One of the limitations of this model is that the singularity in
the coupling term, if present, cannot be modelled in terms of the eigen function
expansion or modes used in the generalised scattering matrix. The singularity in the
Green’s functions in the coupling terms is present when a coupling slot is projected
on the wall containing radiating slot overlaps with the radiating slot [9]. This is
discussed in detail in Chapter 8.
Figure 7.32 shows the generalised scattering matrix (GSM) model for a planar
slot array developed by Enneking et al. [43]. The GSM of the feed network is
 Design of planar slotted waveguide array antennas 171

–5

–10

–15

–20

dB
–25

–30

–35

–40
–80 –60 –40 –20 0 20 40 60 80
(a) Theta (deg.)

–5

–10

–15

dB
–20

–25

–30

–35

–40
–80 –60 –40 –20 0 20 40 60 80
(b) Theta (deg.)

Figure 7.33 Radiation pattern of a planar array of 6  8 slots at 9.375 GHz

computed by the GSM technique compared to measured results:
(a) E-plane pattern; (b) H-plane pattern.
IEEE 2000, reprinted
from [43], with permission

represented by ½S F , and the GSM of the P slots are denoted by ½S 1  through ½S P .

The GSM of the external half space is ½S H . The slot thickness is accounted for in
the model by treating each slot as a waveguide of length equal to the slot thickness.
The GSM of the feed network is obtained by the mode matching technique. For
waveguides of arbitrary cross sections the eigenvalue problem is solved by the
finite element method. A scattering matrix representation is used for each slot and a
scattering matrix for the external radiation network. Higher-order modes up to a
certain number are accounted for. An arbitrary number of modes is used to allow
coupling between modes of the same slot and modes between different slots. The
172 Slotted waveguide array antennas: theory, analysis and design

exterior waveguide apertures are assumed to be located on an infinite conducting

ground plane with any number of dielectric layers. The scattering matrix is com-
puted from the admittance matrix of the slot apertures in an infinite ground plane in
a manner similar to the mutual coupling calculation discussed in Section 4.5.2 in
Chapter 4. However, for the GSM, each slot aperture has a number of modes,
instead of the single mode used in Chapter 4. The GSM of the slots are obtained by
the mode matching technique. The single GSM’s are determined separately and
then cascaded as shown in Figure 7.32 to obtain the overall GSM.
This analysis code is implemented in the computer programme Wasp-Net as a
computer aided design and analysis package [44]. Figure 7.33 shows very good
agreement between the radiation patterns computed by the GSM technique for a
6  8 array with the corresponding measured results.

References
[1] Saltzman H., Stavis G. ‘A dual beam planar antenna for Janus type Doppler
navigation systems’. IRE National Convention Record. 1958, pp. 240–247.
[2] Simmons A. J., Giddings O. M. ‘A multiple-beam two-dimensional wave-
guide slot array’. IRE International Convention Record. 1963, pp. 56–69.
[3] Miller J. R., Forman R. J. ‘A planar slot array with four independent beams’.
IEEE Transactions on Antennas and Propagation. 1966;14(5):560–566.
[4] Elliott R., Kurtz L. ‘The design of small slot arrays’. IEEE Transactions on
Antennas and Propagation. 1978;26(2):214–229.
[5] Lo Y. T., Lee S. W. (eds.). Antenna Handbook. Van Nostrand Reinhold,
New York, 1988, p. 12–1.
[6] Elliott R. S., O’Loughlin W. R. ‘The design of slot arrays including internal
mutual coupling’. IEEE Transactions on Antennas and Propagation. 1986;
34(9):1149–1154.
[7] Derneryd A., Petersson R. ‘Bandwidth characteristics of monopulse slotted
waveguide antennas’. Proceedings of the Fourth International Conference
on Antennas and Propagation, 1985, pp. 27–30.
[8] Press W. H., Teukolsky S. A., Vetterling W. T., Flannery B. P. Numerical
Recipes: The Art of Scientific Computing, 3rd edition. Cambridge University
Press, Cambridge, 2007.
[9] Rengarajan S. R., Shaw G. M. ‘Accurate characterization of coupling junc-
tions in waveguide-fed planar slot arrays’. IEEE Transactions on Microwave
Theory and Techniques. 1994;42(12):2239–2248.
[10] Pozar D. M., Targonski S. D., Syrigos H. D. ‘Design of millimeter wave
microstrip reflectarrays’. IEEE Transactions on Antennas and Propagation.
1997;45(2):287–296.
[11] Yee H. Y. ‘The design of a large waveguide arrays of shunt slots’. IEEE
Transactions on Antennas and Propagation. 1992;40(7):775–781.
[12] Scharstein R. W. ‘Mutual coupling in a slotted phased array, infinite E-plane
and finite in H-plane’. IEEE Transactions on Antennas and Propagation.
1990;38(8):1186–1191.
 Design of planar slotted waveguide array antennas 173

[13] Mailloux R. J. Phased Array Antenna Handbook. Artech House, Boston,

1994.
[14] Pease M. C. ‘The iterated network and its applications to differentiators’.
Proceedings of the IRE. 1952;40(6):709–711.
[15] Rengarajan S. R., Zawadzki M. S., Hodges R. E. ‘Design, analysis, and
development of a large Ka-band slot array for digital beam-forming appli-
cation’. IEEE Transactions on Antennas and Propagation. 2009;57(10):
3103–3109.
[16] Rengarajan S. R., Zawadzki M. S., Hodges R. E. ‘Waveguide-slot array
antenna designs for low-average-sidelobe specifications’. IEEE Antennas
and Propagation Magazine. 2010;52(6):89–98.
[17] Miura Y., Hirokawa J., Ando M., Shibuya Y., Yoshida G. ‘Double-layer
full-corporate-feed hollow-waveguide slot array antenna in the 60-GHz
band’. IEEE Transactions on Antennas and Propagation. 2011;59(8):
2844–2851.
[18] Gruenberg H. ‘Second-order beams of slotted waveguide arrays’. Canadian
Journal of Physics. 1953;31(1):55–69.
[19] Kurtz L. A., Yee J. S. ‘Second-order beams of two-dimensional slot arrays’.
IRE Transactions on Antennas and Propagation. 1957;5(4):356–362.
[20] Tang R. ‘A slot with variable coupling and its application to a linear array’.
IRE Transactions on Antennas and Propagation. 1960;8(1):97–101.
[21] Green J., Shnitkin H., Bertalan P. J. ‘Asymmetric ridge waveguide ridge
waveguide radiating element for a scanned planar array’. IEEE Transactions
on Antennas and Propagation. 1990;38(8):1161–1165.
[22] Rengarajan S. R. ‘Theory of a traveling wave feed for a planar slot array
antenna’. Proceedings of the XXXth URSI General Assembly and Scientific
Symposium. 2011, Istanbul, Turkey, pp. 1–4.
[23] Rengarajan S. R. ‘Analysis of a centered-inclined waveguide slot coupler’.
IEEE Transactions on Microwave Theory and Techniques. 1989;37(5):
884–889.
[24] Brown K. W. ‘Design of waveguide slotted arrays using commercially
available finite element analysis software’. Proceedings of the IEEE Inter-
national Antennas and Propagation Symposium, 1996, pp. 1000–1003.
[25] http://www.ansys.com. Retrieved March 2018.
[26] Schrank H., Solbach K. ‘Simulation of slot array antenna using standard
network analysis program’. IEEE Antennas and Propagation Magazine.
1991;33(6):45–47.
[27] Taddei R., Guarnieri G., Mauriello G. ‘Automated waveguide array-
integrated design environment (AWA-IDE): An automated and modular tool
for planar waveguide array synthesis and analysis’. IEEE Antennas and
Propagation Magazine. 2013;55(4):204–216.
[28] Gatti R. V., Sorrentino R. ‘Slotted waveguide antennas with arbitrary
radiation pattern’. Proceedings of the 34th European Microwave
Conference, Amsterdam, The Netherlands, 2004, pp. 821–824.
[29] http://www.swan-soft.com. Retrieved March 2018.
174 Slotted waveguide array antennas: theory, analysis and design

[30] Ando M., Zhang M., Lee J., Hirokawa J. ‘Design and fabrication of milli-
meter wave slotted waveguide arrays’. Proceedings European Conference
on Antennas and Propagation, 2010, pp. 1–6.
[31] Volakis J. (ed.). Antenna Engineering Handbook. McGraw-Hill, New York,
2007. p. 9–1.
[32] Enjiu R. K., Protoni M. B., Andre S. ‘Slotted waveguide antenna design
using 3D EM simulation’. Microwave Journal. 2013;56(7):72–83.
[33] Gulick J. J., Elliott R. S. ‘A general design procedure for small slot arrays’.
Proceedings of the IEEE International Antennas and Propagation Sympo-
sium, 1987, pp. 302–305.
[34] Rengarajan S. R. ‘Improved design procedure for waveguide-fed slot arrays
using the method-of-moments analysis’. Electromagnetics. 2012;32(4):
221–232.
[35] Hamadallah M. ‘Frequency limitations on broad-band performance of shunt
slot arrays’. IEEE Transactions on Antennas and Propagation. 1989;37(7):
817–823.
[36] Coetzee J. C., Joubert J., McNamara D. ‘Off-center-frequency analysis of a
complete planar slotted-waveguide array consisting of sub-arrays’. IEEE
Transactions on Antennas and Propagation. 2000;48(11):1746–1755.
[37] Casula G. A., Mazzarella G. ‘A direct computation of the frequency
response of planar waveguide slot arrays’. IEEE Transactions on Antennas
and Propagation. 2004;52(7):1909–1912.
[38] Manuilov M., Lerer V., Sinyavsky G. ‘Fast and accurate full-wave analysis
of large slotted waveguide array antennas’. Proceedings of the 37th
European Microwave Conference, 2007, pp. 360–363.
[39] Liu J., Long Y. ‘A full-wave numerical approach for analyzing rectangular
waveguides with periodic slots’. IEEE Transactions on Antennas and Pro-
pagation. 2012;60(8):3754–3762.
[40] Rao S., Wilton D., Glisson A., ‘Electromagnetic scattering by surfaces of
arbitrary shape’. IEEE Transactions on Antennas and Propagation. 1982;
30(3):409–418.
[41] Rengarajan S. R. ‘Genetic algorithm optimization of a planar array using full
wave method of moments analysis’. International Journal of Computer
Aided RF and Microwave Engineering. 2013;23(4):430–436.
[42] Chiarandini S., Morini A. ‘Full wave CAD oriented technique for the ana-
lysis of airborne flat plate arrays’. Proceedings of the IEEE International
Antennas and Propagation Symposium, 2001, pp. 210–213.
[43] Enneking A., Beyer R., Arndt F. ‘Rigorous analysis of large finite wave-
guide-fed slot arrays including the mutual internal and external higher order
mode coupling’. Proceedings of the IEEE International Antennas and
Propagation Symposium, 2000, pp. 74–77.
[44] http://www.wasp-net.com. Retrieved March 2018.
 Chapter 8
Concepts and models for advanced designs

In this chapter we will discuss a number of concepts and models used in

advanced design and optimisation of slot arrays. Models for coupling slots will
be presented followed by a discussion of the edge wall slot and the compound
radiating slot. Iris-excited longitudinal slot arrays, slot arrays in ridge wave-
guides and slot arrays covered by a dielectric layer will also be discussed.
Higher-order mode coupling between adjacent coupling slots and that between a
coupling slot and radiating slots in its immediate vicinity will be presented.
A method of incorporating the finite ground plane effects in the design and
analysis will be described. The MoM solution to the coupled integral equations
for the apertures of all slots in a planar array will be discussed. Some examples
employing the moment method solutions of slot arrays for improved designs will
be presented.

8.1 Coupling slot models

Coupling slots are used to distribute power from a main waveguide to branch
waveguides or radiating waveguides of a planar array as shown in Figure 8.1. The
main waveguide or the feed waveguide is usually orthogonal to a number of
stacked radiating waveguides consisting of radiating slots. Some of the commonly
employed coupling slots are the centred-inclined coupling slot and the longitudinal-
transverse coupling slot, discussed in Sections 8.1.1 and 8.1.2, respectively. Each of
the above-mentioned coupling slots is cut in the common broad wall of two
orthogonal waveguides. The centred-inclined coupling slot is centred in the com-
mon broad wall and tilted with respect the broad wall centreline. It is also referred
to as a series–series slot since it may be represented as a series impedance in both
the main waveguide as well as the branch waveguide. The second type of coupling
slot is longitudinal with an offset from the centreline in the main waveguide and
transverse and centred in the branch waveguide’s broad wall. This element is
called a shunt-series slot since its representation in the main and branch wave-
guides are shunt admittance and series impedance, respectively. In multi-layer
waveguide power dividers one may use a coupling slot, discussed in Section 8.1.3,
that is transverse to the common broad wall of two parallel rectangular waveguides
176 Slotted waveguide array antennas: theory, analysis and design

Radiating slots
Coupling
Main slot
waveguide

Branch
waveguides

Figure 8.1 A planar array of radiating slots and coupling slots.
IEEE 1994,
reprinted from [5], with permission

Figure 8.2 A centred-inclined slot coupler in the common broad wall of two
orthogonal waveguides, with ports 1 and 2 in the main waveguide
and ports 3 and 4 in the branch waveguide.
IEEE 1989,
reprinted from [1], with permission

with one on top of the other. Such a slot transmits the entire power from a
waveguide in one layer to the next layer. In that case the coupling slot acts as a
one-to-one transformer.

8.1.1 Centred-inclined coupling slot

Centred-inclined coupling slots are commonly used to couple power from a feed
waveguide into a number of radiating waveguides that are stacked orthogonally to
the feed waveguide. Such a coupler may be characterised as a four-port device as
shown in Figure 8.2. The coordinate system in the upper broad wall of the main
waveguide is shown in Figure 8.3.
 Concepts and models for advanced designs 177

2L
w
z

Figure 8.3 Slot coordinates at the upper broad wall of the main waveguide with
ports 1 and 2.
IEEE 1989, reprinted from [1], with permission

Assuming that the reference planes of all four ports pass through the centre of
the slot, a coupling slot at resonance may be characterised by a scattering matrix
with only one unknown, that is, S11 as given in (8.1) [1].
2 3
r p q q
6 p r q q7
6 7
S ðqÞ ¼ f ðr; p; qÞ ¼ 6 7 (8.1)
4 q q p r5
q q r p
where r ¼ S11(q), p ¼ 1 – r and q ¼ [r(1 – r)]1/2.
From symmetry considerations we can obtain the scattering matrix for com-
plementary tilt angles and negative tilt angles as given by (8.2) and (8.3) [1].
S ðqÞ ¼ f ðr; p; qÞ (8.2)
S ðp=2  qÞ ¼ f ðp; r; qÞ (8.3)
Thus a q range of 0 to p/4 radians is sufficient to characterise the coupling slot.
Equation (4.68) in Chapter 4 may be used to determine the TE10 mode wave
scattered by a centred-inclined slot in the broad wall of a rectangular waveguide.
It may be shown that the forward and backward scattered waves are equal in
magnitude but 180 out of phase. A similar odd-symmetric scattering behaviour is
found when a wave is incident in a transmission line loaded by a series impedance.
Therefore a centred-inclined slot in the broad wall of a rectangular waveguide may
be modelled as a series impedance in an equivalent transmission line. The scattered
TE10 mode waves in the backward and forward direction are in phase and 180 out
of phase, respectively, with respect to the incident TE10 mode wave at resonance.
Analysis of the centred-inclined slot coupler using the MoM shows that the slot
may be modelled as a series impedance in the main line as well as in the branch
line with excellent accuracy. If the branch line ports 3 and 4 are match-terminated,
the series resistance normalised to the characteristic impedance of the TE10 mode
wave in the main line may be expressed in terms of S11 as

¼ 2S11 =ð1  S11 Þ
R (8.4)
178 Slotted waveguide array antennas: theory, analysis and design

The coupling coefficient used in (7.2) in Chapter 7 is one-half of the normal-

ised resistance of a resonant coupling slot given in (8.4) and is repeated next.
c2 ¼ S11 =ð1  S11 Þ (8.5)
This can be easily derived from the scattering matrix (8.1) if the load admit-
tances in ports 3 and 4 are known.
If ports 3 and 4 are not matched, the series impedance seen in the main line at
the location of the coupling slot may be derived easily from (8.6) using the scat-
tering matrix (8.1) with the impedance terminations in ports 3 and 4.
½b ¼ ½S ½a (8.6)
where [b] is a column matrix containing the TE10 mode wave amplitudes leaving
the slot from each of the four ports while [a] contains the corresponding quantities
of the waves incident at the slot. The normalised series impedance Z
in the main
line is given by (8.7).
Z
¼ ðZ
3 þ Z
4 Þc2 (8.7)
where Z
3 and Z
4 are the normalised impedances seen by the slot at ports 3 and 4,
respectively. The bars on top imply normalised impedances whereas the
un-normalised impedances are denoted without bars. Characteristics of this type of
coupler have been analysed by the moment method solutions to the coupled integral
equations for the electric fields of apertures of a thick slot [1].
Resonant coupling slots are employed in planar standing wave arrays to realise
the aperture distribution with uniform phase. Data on the resonant length and |S11|
at resonance for a range of tilt angles, generated by a computational electro-
magnetics (CEM) code are needed in such designs. The variation of the resonant
length with the tilt angle of the slot is not significant for standard height wave-
guides whereas for reduced height waveguides it is not negligible.
Figure 8.4 shows a circuit model for a non-resonant series–series slot [2]. The
transformer ratio n, the inductance L and the capacitance C needed to provide the
susceptance in the main waveguide are functions of frequency. A simplified model
has been proposed by Mazzarella and Montisci [2]. In that model, the transformer

3 4
n:1

1 jB 2

Figure 8.4 Circuit model for a non-resonant series–series coupling slot

 Concepts and models for advanced designs 179

ratio n is determined at the resonant frequency. Over a frequency range of 6.6% below
the resonant frequency L ¼ L1 and C ¼ C1 were found to produce accurate values of
the susceptance. Similarly L ¼ L2 and C ¼ C2 yielded accurate susceptances over a
6.6% frequency band above resonance. By choosing the values of L and C between
the two sets they were able to produce an accurate model over a 10% frequency
band [2]. The value of n at resonance is the coupling coefficient c defined in (8.5).

8.1.2 Shunt-series coupling slot

Figure 8.5 shows the shunt-series coupling slot in the common broad wall of two
orthogonal waveguides. The slot is longitudinal and offset from the broad wall
centreline of the main waveguide with ports 1 and 2. In the branch waveguide with
ports 3 and 4, the slot is centred in the transverse direction. Slot length, 2L, width w
and offset d from the centreline are shown. Representation of the slot as a series
element in the branch waveguide is excellent while the shunt representation in the
main line becomes poor for large offsets and reduced height waveguides [3].
However, we refer to it as a shunt-series coupling slot based on its approximate
representation. When the shunt model fails, one may work with the computed
scattering parameters.

1
(a)

w
δ
2L

z
(b)

Figure 8.5 Shunt-series coupling slot.: (a) cut-away view of the slot;
(b) slot parameters and localised slot coordinates.
IEEE 1989,
reprinted from [3], with permission
180 Slotted waveguide array antennas: theory, analysis and design

The scattering matrix of the four-port resonant coupler with all ports defined to
be planes passing through the centre of the slot is given by
2 3
r p jq jq
6 p r jq jq 7
6 7
S¼6 7 (8.8)
4 jq jq p r5
jq jq r p
where r ¼ S11. If the shunt model is acceptable p ¼ 1 – r, and q ¼ [r(1 – r)]1/2. For a
negative offset, q will be replaced by q in (8.8). A resonant shunt-series coupling
slot was used (see Figure 7.12 in Chapter 7) at the centre of a feed waveguide of a
planar array as a transformer to match the total resistance seen by the feed waveguide
into a match at the input port [4]. In that case the normalised input admittance is given
by (8.9) with the coupling coefficient determined by (8.10) using S11 at resonance.

Y
¼ ðZ
3 þ Z
4 Þc2 (8.9)

where

c2 ¼ S11 =ð1 þ S11 Þ (8.10)

For the case of a non-resonant shunt-series coupling slot, although a circuit
model may be derived, it is suggested that the scattering matrix representation be
used. Because of symmetry, reciprocity and power balance considerations the
scattering matrix gets simplified.
The shunt-series coupling slot employed in a planar array (see Figure 8.6) has
the advantage that the higher-order mode coupling between such a coupling slot
and the radiating slots in its immediate vicinity is negligible whereas such coupling

Figure 8.6 A planar slot array with shunt-series coupling slots

 Concepts and models for advanced designs 181

in arrays employing centred inclined or series–series coupling slots must be

accounted for under certain conditions (see Section 8.7) [5,6].

8.1.3 Transverse/transverse coupling slot

Consider two identical rectangular waveguides sharing a common broad wall as
shown in Figure 8.7(a). A centred transverse resonant slot cut in the common broad
wall behaves as a series element in both waveguides [7]. If the slot is resonant at the
operating frequency and it is centred in both waveguides, the scattering matrix
of this four-port device may be derived from reciprocity, symmetry and power
balance considerations as follows.
2 3
0:5 0:5 0:5 0:5
6 7
6 0:5 0:5 0:5 0:5 7
S¼6 6 7 (8.11)
4 0:5 0:5 0:5 0:5 7
5
0:5 0:5 0:5 0:5
Figure 8.7(b) shows short circuits placed in ports 2 and 3 at a distance s from
the centre of the slot. The reflection coefficients at ports 2 and 3, referenced to the
centre of the slot, are equal to ejy, where y ¼ p – 2b10s. In this case the two-port

3
Slot

2
(a)

Short
(b)

Figure 8.7 Transverse/transverse coupling slot in the common broad wall of

parallel waveguides: (a) four-port coupler; (b) ports 2 and 3
shorted at a distance s from the slot centre
182 Slotted waveguide array antennas: theory, analysis and design

scattering matrix of the junction with ports 1 and 4 referenced to the centre of the
slot is
" #
1 þ ejy 1 þ ejy
S ¼ 0:5 (8.12)
1 þ ejy 1 þ ejy

For the special case of s ¼ 0, the slot with half the original width will be in
front of the short. The previous result may be shown to be valid for this special case
by considering the image of the equivalent magnetic current of the slot. In this
case the two-port device has zero reflection at ports 1 and 4 and the transmission
coefficient between ports 1 and 4 is unit amplitude and 180 phase. Such a device is
conveniently used in multi-layer power dividers for total transmission from one
layer to another and also to introduce the required amounts of phase shifts through
extra lengths of waveguides between ports 1 and 4 and the slot centre [8].

8.2 Edge wall slot

Figure 8.8 shows an edge wall slot cut in a rectangular waveguide. Since wall
currents in the edge wall are vertical (see Figure 1.1), in order to disturb the wall
currents, and thus to radiate from the waveguide, such a slot should have an
inclination with respect to the vertical. Greater values of the tilt angle with respect
to the vertical will provide stronger excitations of the slot. In order to provide an
equi-phase excitation from an array of such slots, a slot spacing of half a guide
wavelength is used with adjacent slots having alternating inclinations of positive
and negative angles with respect to the vertical. An array of edge wall slots
with alternating tilts will radiate in the broadside direction with a principal polar-
isation along the axial direction of the waveguide (see Figure 8.9). Thus the long-
itudinal offset slots and edge wall slots offer two orthogonal polarisations. Another
advantage of using edge wall slots is that adjacent waveguides with edge wall slots
may be placed with a spacing in the order of half wavelength in free space.
Therefore it is possible to scan such arrays electronically over a wide angular range
using phase shifters between adjacent waveguides. In the case of longitudinal offset

Figure 8.8 An inclined slot cut in the edge wall of a rectangular waveguide
 Concepts and models for advanced designs 183

slots in the broad walls, the scanning range is limited because of the broad wall
dimension being greater than half wavelength in free space.
The edge wall slot may be modelled as a shunt element since it scatters sym-
metrically. Since the edge wall dimension is less than a half wavelength in free
space, a slot cut in that wall may have to be wrapped around both broad walls
in order to accommodate the required resonant slot length. This presents a problem
in stacking waveguides one on top of another. The tilted slots introduce a certain
amount of cross-polarisation. Untilted edge wall slots excited by probes, wires or
irises do not radiate cross-polarisation. Figure 8.10 shows a concept for producing
dual polarisation. One set of waveguides with longitudinal offset slots cut in the
broad walls produce linear polarisation perpendicular to the waveguide axes. The
second set of waveguides with edge wall slots will produce the orthogonal polar-
isation along the waveguide axes. See also Chapter 9.
In the design of edge wall slot arrays, mutual coupling can be incorporated by
using an infinite array model [9,10]. It is possible to compute the mutual coupling
by the so-called element by element method by considering a pair of slots at a time
using the spectrum of two-dimensional solutions [11,12].

Figure 8.9 The aperture electric field in two edge wall slots with a half guide
wavelength spacing

Figure 8.10 Planar array of slots for dual polarisation

184 Slotted waveguide array antennas: theory, analysis and design

8.3 Radiating compound broad wall slot

A broad wall slot offset from the centreline and tilted with respect to the waveguide
axis, called the compound slot, can be designed to be resonant and still produce any
excitation amplitude and phase. Figure 8.11 shows different broad wall radiating
slots. The aperture electric field of a resonant longitudinal offset slot is 90 with
respect to the TE10 waveguide mode amplitude at the centre of the slot for positive
offsets whereas the phase of the aperture field becomes 270 if the offset is nega-
tive. Similarly the phase becomes 0 and 180 for a centred-inclined resonant slot
for positive and negative tilts, respectively. For a resonant compound slot, the phase
of the aperture electric field with respect to the incident TE10 mode wave may be
anywhere in the range of 0 –360 depending on the sign of the offset and tilt as
shown in Figure 8.12(a)–(d) [13]. Figure 8.12(e) shows the range of possible values
of offsets and tilts bounded by the four curves in the four quadrants. A standard
X-band waveguide with a slot length equal to one-half free space wavelength at
9.3 GHz and a slot width equal to 1/10 of the length were used to generate this figure.
The horizontal axis is the tilt in degrees and the vertical is the offset normalised to
a/2, where a is the waveguide inner width. For a longitudinal slot (q ¼ 0 ) the
maximum possible offset is (a  w)/2  0.92 a/2, where w is the slot width. For a
transverse slot (q ¼ 90 ) the maximum possible offset is (a/2  L)  0.289 a/2,
where L is one-half of the slot length.
For a compound slot, resonance occurs when the forward scattered TE10 mode
wave is 180 out of phase with the incident TE10 mode wave, both referenced to the
centre of the slot. The back scattered TE10 mode wave phase for a resonant slot is
twice that of the aperture electric field.

3
δ
θ

δ 1 – Longitudinal
2 – Centred inclined
t
3 – Compound

Figure 8.11 Broad wall radiating slots.
IEEE 1989, reprinted from [13],
with permission
 Concepts and models for advanced designs 185

(a) (b)

(c) (d)
1

0.5
Normalised offset

Quadrant 2 Quadrant 1
0
–90 –60 –30 0 30 60 90
Quadrant 3 Quadrant 4
–0.5

–1
(e) Tilt in degrees

Figure 8.12 Four ranges of phase for the aperture electric field and possible range
of offsets and tilts for compound slots. (a) d > 0, 0  q  90 , 0 
arg E  90 ; (b) d > 0, 90  q  0 , 90  arg E  180 ;
(c) d < 0, 90  q  0, 180  arg E  270 ; (d) d < 0, 0  q 
90 , 270  arg E  360 ; (e) possible range of normalised slot
offsets (vertical axis) and tilts (horizontal axis) for a compound slot

Although the compound radiating slot was introduced in the 1940s [7], it was not
used in an antenna application until 1960 [14], when a linear travelling wave array of
compound radiating slots was designed, and built. It is ideally suited for travelling
wave arrays such as the fuze antenna of a missile system in which the beam is at an
angle with respect to the missile axis. Compound slots are also useful as coupling
elements. Although an isolated resonant compound slot can be designed to produce any
value of the aperture amplitude and phase, the amount of realisable phase values will be
limited by the reflection coefficient at the output port, with a greater range of phase
values for smaller reflection coefficients. This result was shown for a compound cou-
pling slot in [15] but it is true for compound radiating slots also. Compound slot arrays
may be designed using the procedure described in Section 6.8 in Chapter 6 in terms of
scattering wave parameters. An array of compound slots all with a tilt of 45 but
different offsets has been used in an automotive collision avoidance radar [16].
186 Slotted waveguide array antennas: theory, analysis and design

8.4 Iris-excited longitudinal slot

Longitudinal slot arrays with alternating slot offsets introduce grating lobes called
butterfly lobes [17]. These lobes may be suppressed by splitting the array in the
H-plane in two halves with even symmetry. Such an arrangement was discussed in
Section 7.6.2 and Figure 7.23 in Chapter 7. It is also possible to eliminate the
butterfly lobes by letting the longitudinal offset slots radiate into an exterior region
consisting of baffles [18,19].
No butterfly lobes appear if the slots are collinear in a waveguide. This
is possible with centred longitudinal slots excited by asymmetric ridges (see
Section 9.9 in Chapter 9), probes, wires or irises. In this section we will discuss iris-
excited slots. The compound iris, shown in Figure 8.13, consisting of a combination
of an inductive iris of width I and a capacitive iris of width C, is particularly
attractive since the resonant characteristics of the iris and that of the slot are
complementary and therefore the combination of iris and slot exhibits greater
bandwidth than the conventional offset slot [20]. MoM analysis and experimental
measurements show that the longitudinal slot generally behaves as a shunt element
in the form of a parallel resonant circuit whereas a compound iris behaves also as a
shunt element but in the form of a series resonant circuit. Tang [20] demonstrated
that it is possible to excite the collinear slots with appropriate magnitude and in
phase by different sized irises.
A systematic experimental investigation of iris-excited centred longitudinal
slots in the broad wall of a rectangular waveguides was undertaken by Nyström
[21]. Pairs of values of I and C were determined such that the compound iris would
be resonant at the design frequency of 5.3 GHz in a WR 159 waveguide of inner
dimensions 40.39 mm  20.19 mm. Marcuvitz’s expressions [22] for the inductive
and capacitive reactances were initially used to determine the approximate values
of I and C needed to have a resonant compound iris at 5.3 GHz. Later the C values
were changed slightly so that the measured resonances were at 5.3 GHz for each
pair of I and C values. Table 8.1 shows the values of C needed for each value of I
in order to produce the desired resonance. Smaller values of I and C will produce
smaller amounts of excitation of a centred longitudinal slot while larger values of
I and C will produce larger excitations.

Figure 8.13 The cross section of a compound iris inside a

rectangular waveguide
 Concepts and models for advanced designs 187

Table 8.1 Values of I and C for the iris to

produce resonance at 5.3 GHz

I (mm) C (mm)
5.0 4.7
6.4 6.1
7.5 7.5
8.8 9.1
10.0 10.8
Source:
1991, I. Landmark/Nyström. Modified from [21],
with permission.

Table 8.2 Values of I, C and the length of a centred slot to produce resonance at
5.3 GHz compared to an equivalent longitudinal offset slot producing
the same resonant conductance

Case Value of I Value of C Centred slot Equivalent Equivalent

(mm) (mm) length (mm) longitudinal slot longitudinal slot
offset (mm) length (mm)
A 5.0 4.7 26.5 1.3 26.6
B 6.4 6.1 27.0 2.7 26.7
C 7.5 7.5 27.3 4.1 26.8
D 8.8 9.1 27.7 6.5 27.1
E 10.0 10.8 28.0 10.4 27.6

Subsequently a centred longitudinal slot of appropriate length was cut in the

waveguide such that the iris and the slot combination would be resonant at
5.3 GHz. In the absence of the iris the incident TE10 mode will not excite a centred
slot whereas the asymmetry produced by the iris will excite it.
Table 8.2 shows the resonant lengths of centred slots for different iris widths.
All slot widths were chosen to be 2.0 mm. Each combination of iris widths and
the length of the centred longitudinal radiating slot produces resonance at
5.3 GHz. For each case the offset and the length of an equivalent longitudinal
offset slot producing the same resonant conductance at 5.3 GHz are also shown in
Table 8.2.
Figure 8.14 shows measured values of the slot conductance and susceptance
normalised to the resonant conductance of the iris-excited slot A in Table 8.2.
Normalised conductance and susceptance of an equivalent offset longitudinal slot
whose length and offset are shown for case A in Table 8.2 are also compared. The
iris-excited slot exhibits greater bandwidth. Similarly Figure 8.15 compares the
iris-excited slot and an equivalent conventional offset longitudinal slot for case C.
Once again the iris-excited slot is found to exhibit greater bandwidth. Adjacent iris-
excited slots should have the inductive section on the opposite sides of the broad
188 Slotted waveguide array antennas: theory, analysis and design

0.6

0.4
Normalised admittance

0.2

0
G1
G2
–0.2
B1
B2
–0.4
4.8 5 5.2 5.4 5.6 5.8
Frequency in GHz

Figure 8.14 Susceptance (B1) and conductance (G1) versus frequency for an
iris-excited centred slot and the susceptance (B2) and conductance
(G2) of an equivalent conventional offset longitudinal slot for
case A in Table 8.2.
1991, I. Landmark/ Nyström. Modified
from [21] with permission

0.2

0.1
Normalised admittance

–0.1
G1
G2
–0.2
B1
B2
–0.3
4.8 5 5.2 5.4 5.6 5.8
Frequency in GHz

Figure 8.15 Susceptance (B1) and conductance (G1) versus frequency for an
iris-excited centred slot and the susceptance (B2) and conductance
(G2) of an equivalent conventional offset longitudinal slot for
case C in Table 8.2.
1991, I. Landmark/ Nyström. Modified
from [21] with permission

wall centreline for a standing wave application so as to provide in phase excitation

of slots. In a travelling wave array the inductive sections may be on the same
side of the centreline or alternate, corresponding to the two cases discussed in
Section 6.6 in Chapter 6.
 Concepts and models for advanced designs 189

8.4.1 Elliott’s design procedure for iris-excited slotted

waveguide arrays
For the conventional longitudinal slot, we obtain resonant lengths for different
offsets and then obtain the admittance data and the two-port scattering parameters
of the slot. The slot voltage is usually determined from the scattering parameters or
the admittance data. In the case of an iris-excited slot, we treat the iris dimensions I
and C as a couplet for a range of values as shown in Table 8.1. For each couplet, the
resonant length and resonant conductance may be obtained from a CEM code. We
then obtain the admittance data as a function of the slot length normalised to its
resonant length for a range of slot lengths. In addition, for each of these cases we
can obtain the slot aperture field from the same CEM code, assuming a half cosi-
nusoidal equiphase distribution for the slot aperture. The relationship between the
backscattered and forward scattered TE10 mode amplitude and the voltage of the
nth slot can be expressed as follows.

B10 ¼ C10 ¼ fin Vns (8.13)

In (8.13) fin is a function of iris dimensions, slot length and width. Similar to the
first design equation (6.10) in Chapter 6 and in [23] we derive

Yna Vns
¼ fin Ki (8.14)
G0 Vn

where
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 2wm0 b10 ab
Ki ¼ (8.15)
p=a G0

The second design equation (6.15) derived in Chapter 6 is valid for the iris-
excited slot also and it is reproduced as (8.16).

Ya 2
¼
G0 2 1 Vscoupl (8.16)
þ
Ys =G0 G Vs

However, the voltage coupled to the slot from the external mutual coupling,
given in (6.17), gets modified for iris-excited slots as

jp 1 Ys =G0 X N
0
Vscoupl ¼ Vs ext ¼ 2 f 2 Y =G þ 2
Vms gmn ðLm ; Ln ; Xmn ; Zmn Þ
2ðk0 bÞb10 a ðwm0 Þ i s 0
3
m¼1
(8.17)
Using the design equations given here, one can design linear and planar arrays
of iris-excited slots using computed values of resonant lengths and admittances as
functions of normalised slot lengths for different couplets I and C.
190 Slotted waveguide array antennas: theory, analysis and design

8.5 Slot arrays in ridge waveguides

Planar arrays of longitudinal radiating slots cut in the broad walls of rectangular
waveguides may be electronically scanned in the E-plane by including phase
shifters between adjacent waveguides. Conventional rectangular waveguides typi-
cally have an ‘a’ dimension in the order of 0.7 l0 at the operating frequency. In this
case, it is possible to scan to a maximum angle of 25 off broadside before grating
lobes start appearing in the visible space. In order to scan to larger angles the
spacing between waveguides should be closer to one-half wavelength. One way to
accomplish this is by reducing the waveguide width which can be realised in ridge
waveguides. Figure 8.16 shows cross sections of ridge waveguides.
Single-ridge waveguides are convenient for slot array applications since
longitudinal radiating slots may be cut in the broad wall opposite to the ridge over a
wide range of offsets (see Figure 8.17). Kim and Elliott presented a design proce-
dure for slot arrays in ridge waveguides using experimental data on isolated slot
admittances [24]. For analysis one can use a model for ridge waveguides presented
by Falk [25,26] or use commercial CEM software. In this section we will briefly
discuss the design procedure developed in [24].

(a) (b)

Figure 8.16 Ridge waveguide cross sections: (a) single-ridge waveguide;

(b) double-ridge waveguide

Figure 8.17 Slot array composed of ridge waveguides

 Concepts and models for advanced designs 191

The longitudinal slot cut in the broad wall of a ridge waveguide scatters
symmetrically, just like the longitudinal slot in a rectangular waveguide and may be
modelled in the form of a shunt admittance on an equivalent transmission line.
Using Montgomery’s expressions [27] for the TE10 mode electric field, the first
design equation similar to (6.10) of the longitudinal slot in the conventional
rectangular waveguide is obtained next.
Yna Vs
¼ K1 fn n (8.18)
G0 Vn
where
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2kt kt2
K1 ¼ (8.19)
k0 wm0 b10 G0

and kt is the transverse wavenumber of the TE10 mode given by [27].

The second design equation (6.15) derived in Chapter 6 is valid for the ridge
waveguide slot as well, and it is reproduced in (8.20).

Ya 2
¼
G0 2 1 Vscoupl (8.20)
þ
Ys =G0 G Vs

However, the voltage coupled to the slot from the external mutual coupling,
given in (6.17) gets modified for the ridge waveguide slot as [24]

jb10 k02 1 Ys =G0 X N

0
Vscoupl ¼ Vs ext ¼ V s gmn ðLm ; Ln ; Xmn ; Zmn Þ (8.21)
4pkt4 fn2 Ys =G0 þ 2 m¼1 m

A planar array consisting of two ridge waveguides with eight radiating slots in
each was designed using (8.18) and (8.20), and it was built and tested [24]. The
measured results for the radiation pattern and the voltage standing wave ratio
(VSWR) compared well with the theory.

8.6 Slot arrays with a dielectric cover

In high-power radar antennas, waveguides are pressurised to prevent electrical
breakdown. Slotted waveguide arrays in such applications have often a dielectric
layer on top of the slot apertures to seal the pressurised waveguides. The presence
of the dielectric affects the resonant length and the resonant conductance of the
slots. The problem of dielectric covered waveguide slots was first studied by Bailey
using the variational approach [28,29]. Rexberg and Katehi studied the properties of
dielectric covered slots using the MoM solution to the integral equations with
the appropriate Green’s functions in the waveguide and the exterior region [30,31].
The dielectric cover generally reduces the resonant length of the slot as shown in
Figure 8.18 since the effective dielectric constant seen by the slot increases when a
192 Slotted waveguide array antennas: theory, analysis and design

0.50 Experiment [8]

This analysis without correction
0.46 This analysis with correction

0.42
2Lres
λ0 0.38

0.34

0.30

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
h/λd

Figure 8.18 The effect of the dielectric cover thickness on the resonant length
of the slot.
IEEE 1990, reprinted from [31], with permission

Experiment [8]
1.00 This theory

0.08
Gr
0.06
G0

0.04

0.02

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
h/λd

Figure 8.19 The effect of the dielectric cover thickness of the resonant
conductance of the slot.
IEEE 1990, reprinted from [31],
with permission

dielectric cover is used. The resonant slot conductance also decreases in presence
of the dielectric cover (see Figure 8.19). For both of these plots, the slot length is
12.065 mm, and the slot offset is 2.36 mm. The dielectric constant is 3.31. The
waveguide inner dimensions are 22.86 mm  10.16 mm and the slot width is 1.59 mm.
Casula and Montisci have modified the Elliott’s design procedure for a slot array
with a dielectric cover [32]. The modification is needed only in the second design
 Concepts and models for advanced designs 193

–5
Far-field pattern [dB]

–10
SLL = –13 dB
SLL = –14.1 dB
–15

–20

–25

–30
–90 –75 –60 –45 –30 –15 0 15 30 45 60 75 90
Angle from broadside [deg.]

Figure 8.20 Simulated radiation pattern in the E-plane. Solid line (with
dielectric); dashed line (without dielectric).
IEEE 2009,
reprinted from [32], with permission

equation involving the external mutual coupling. The external mutual coupling is
evaluated in the spectral domain in the form of numerical double integrals in polar
coordinates. The radial integral exhibits highly oscillating terms and has poles due to
surface waves. For a thin dielectric cover, only the TM0 surface wave pole is present.
The residue contribution of the pole has to be considered in addition to its principal
value [33] or the external mutual coupling integral in spectral domain may be deformed
to avoid the pole singularity [30]. The highly oscillating integral is evaluated by the
method of weighted averaging and by the use of Shank’s transforms [34,35].
Casula and Montisci demonstrate the significance of a dielectric cover of
thickness as small as 0.012 free space wavelength and a dielectric constant equal to
2.2 by comparing the radiation pattern and input reflection coefficient of a 10  5
array (five waveguides with ten slots in each) simulated with and without the
dielectric [32]. Those results are reproduced in Figures 8.20 and 8.21.

8.7 Higher-order mode coupling between centred-inclined

coupling slots
Elliott’s design procedure discussed in Chapters 6 and 7 does not incorporate
higher-order mode coupling between coupling slots of a planar slot array (see
Figure 8.22). Higher-order mode coupling between a coupling slot and a pair of
straddling radiating slots in each coupling junction, also not accounted for in the
design procedure, will be discussed in Section 8.8. It has been shown that the
higher-order mode coupling between coupling slots is significant for small tilt
angles or small slot offsets from the centreline [36]. The TE20 mode coupling
accounts for most of this effect in reduced-height waveguides whereas in standard
194 Slotted waveguide array antennas: theory, analysis and design

–5
–10
–15

–20

–25

–30

–35

–40
–45
8.7 8.75 8.8 8.85 8.9 8.95 9 9.05 9.1 9.15 9.2 9.25 9.3
Frequency [GHz]

Figure 8.21 Simulated reflection coefficient. Solid line (with dielectric);

dashed line (without dielectric).
IEEE 2009, reprinted
from [32], with permission

Slot n + 1 Branch
waveguides

Slot n

Slot n – 1

Feed
waveguide

Figure 8.22 Geometry of the higher-order mode coupling between slot n

and slots n – 1 and n þ 1

height waveguides, in addition to the TE20 mode coupling, there may be some
significant TE01 mode coupling as well. The higher-order mode coupling between
adjacent shunt-series coupling slots may be accounted for by using a procedure
similar to that for such coupling between adjacent radiating slots [37]. In this
section we will discuss a method for incorporating the higher-order mode coupling
between centred-inclined coupling slots [38].
 Concepts and models for advanced designs 195

8.7.1 Theory for the higher-order mode coupling between

centred-inclined coupling slots
Using (4.68) in Chapter 4, we can derive an expression for the backward and for-
ward scattered TE10 mode waves in the main waveguide by a centred-inclined
coupling slot [see (8.22)].
B10 ¼ C10 ¼¼ K2 Cn Vns (8.22)
ðp=aÞ2
Here Vns is the voltage of the nth slot, K2 ¼  , and Cn is the reaction
wm0 wb10 ab
integral (8.23) carried out over the axis of slot n with a tilt angle qn with respect to
the longitudinal centreline of the broad wall.
ð Ln  
b10
Cn ¼ w j sinfp=aðz sin qÞg  cosfp=aðz sin qÞg
Ln p=a
 
p
 expðjb10 z cos qÞ cos z dz (8.23)
2Ln
The centred-inclined coupling slot may be modelled as a series impedance, as
mentioned in Section 8.1.1, in an equivalent transmission line having the phase
constant b10 and the characteristic impedance, Z0 equal to the wave impedance of
the TE10 mode wave. We can define an active impedance, Zna for the coupling slot
similar to the active admittance of a longitudinal radiating slot specified by (6.10)
and (6.16) in Chapter 6 as follows.
Cn ðp=aÞ Vns
Zna ¼  (8.24)
wb10 ½ab=21=2 In
The term ‘active impedance’ includes mutual coupling from higher-order modes
between adjacent coupling slots. In the absence of higher-order mode coupling, the
active impedance becomes self-impedance. The term In in (8.24) is the mode cur-
rent of the equivalent transmission line at the location of the active impedance. The
derivation of mutual coupling expression follows the procedure used in deriving
(6.15) in Chapter 6 for radiating slots.
The normalised active impedance is expressed in terms of the self-impedance
and the isolated slot reflection coefficient as follows.
Zna 2
¼ 0 1 s (8.25)
Z0 2ZZ nG Vnh =Vn
n
s

where Gn is the reflection coefficient of the slot without accounting for higher-order
mode coupling, Zn is the impedance of the coupling slot in the absence of higher-
s
order mode coupling and Vnh is the slot voltage induced by the higher-order mode
coupling from adjacent slots. Using (6.36) in Chapter 6 and (8.22) and (8.23) we
s
obtain an expression for Vnh as
s
Vn1 V s Ih Gn
s
Vnh ¼ Ih ¼ n1 ¼ Ih (8.26)
I10 A10 Cn K2 Cn2
196 Slotted waveguide array antennas: theory, analysis and design

where Ih is given by the reaction integral (8.27) with Hhz the magnetic field along
s
the axis of slot n due to higher-order modes scattered from adjacent slots. Vn1 is the
slot voltage induced by a TE10 mode wave of amplitude A10.
ð Ln  
p
Ih ¼ Hhz cos z dz (8.27)
Ln 2Ln

Combining (8.25) and (8.26) we obtain

Zna 1 1
¼  ¼ h s i (8.28)
Z0 Z0 =Zn  2K12 C 2 Ih =Vns Z0 =Zn  1
2 I 1
Vn1
s þ I 2
s
Vnþ1
s
n 2K2 Cn V
n Vn

The dominant contribution to Ih is expected to be from the TE20 mode but it is easy
to account for a few other modes to get an accurate evaluation. Using the dyadic
Green’s function for the magnetic field of the magnetic current given in (2.62) in
Chapter 2 we can determine I1 for the coupling between slots n  1 and n and I2
between slots n þ 1 and n, shown in Figure 8.22 using (8.29). The term I1 is non-
existent for the first coupling slot and likewise I2 is not relevant for the last one.
ð Ln  ð Ln1  
p p
I1 ¼ cos z cos z0
Ln 2Ln Ln1 2Ln1

 G1xx ðx; x0 ; z; z0 Þ sinðqn Þ sinðqn1 Þ
þ G1xz ðx; x0 ; z; z0 Þ sinðqn Þ cosðqn1 Þ
þ G1zx ðx; x0 ; z; z0 Þ cosðqn Þ sinðqn1 Þ

þ G1zz ðx; x0 ; z; z0 Þ cosðqn Þ cosðqn1 Þ dz0 dz (8.29)

where x ¼ a/2 þ z sin qn, x0 ¼ a/2 þ z0 sin qn1, z ¼ z cos qn, z0 ¼ z0 cos qn–1 – D,
and D is the slot spacing. The tilt angles of slots n and n – 1 are qn, and qn–1,
respectively. The expression for I2 is obtained readily from I1 given in (8.29)
by replacing subscripts n – 1 by n þ 1 and –D by D in z0 in (8.29). The terms
G1xx, G1xz, G1zx and G1zz are the ^x^x ; ^x^z ; ^z^x and ^z^z components respectively of the
dyadic Green function G 1 given in (2.63) and (2.64) in Chapter 2.

8.7.2 Procedure to incorporate the higher-order mode coupling

in the design of centred-inclined coupling slots
Initially coupling slots are designed using Elliott’s procedure described in
Chapter 7. It is assumed that coupling slots are resonant at the design frequency and
that they are spaced a half guide wavelength apart at the design frequency for a
standing wave array. The procedure to be described is applicable even if the slots
are not resonant, and for travelling wave type feeds as well. From the scattering
parameter S11 of each coupling slot, treated as a four-port device, we determine its
 Concepts and models for advanced designs 197

self-impedance. The mode current appearing in Equation (8.24) may be assigned an

arbitrary value of unity for the first coupling slot. For a standing wave array all
mode currents are equal in magnitude with alternating positive and negative signs
between adjacent slots. For coupling slots in a travelling wave feed, mode currents
are obtained using transmission line equations in a manner similar to the mode
voltages of a travelling wave array of longitudinal slots in Section 6.6 in Chapter 6.
In (8.24) we initially set the values of active impedances to self-impedances and
obtain the values of slot voltages. Then the tilt angles and lengths of coupling slots
are perturbed so as to make the active impedance of each coupling slot equal to its
self-impedance in (8.28) and (8.24) while the slot voltages remain the same as the
original values. This process converges rapidly since higher-order mode effects are
usually small. Including coupling from a few higher-order modes such as TE20,
TE30, TE11, TM11 and TE01 will provide very accurate results.
In a study of a planar slot array with spacings between adjacent coupling slots
of 0.6 l0 it was found that there was a significant amount of higher-order mode
coupling effect on the reflection coefficient, whereas the radiations patterns were
not affected much [38]. An optimisation procedure that perturbed the coupling slots
by not more than 3% reduction in slot lengths and less than 0.3 reduction in the tilt
angle produced excellent results with very good match at the centre frequency.

8.8 Higher-order mode coupling between a centred-inclined

coupling slot and longitudinal radiating slots
Figure 8.1 shows the geometry of a planar array consisting of coupling junctions,
each with a centred-inclined coupling slot and a pair of straddling longitudinal
slots. In Figure 8.23, the coupling slot is projected on the broad wall containing the
radiating slots. As shown in Figure 8.23 we define two cases of coupling called
hard coupling and soft coupling. Experimental results show that higher-order mode
coupling is greater in the case of hard coupling than in the soft coupling geometry.
Each radiating slot is at a distance of a quarter guide wavelength from the coupling
slot. Therefore higher-order mode coupling is expected to be significant in this type
of coupling junctions. In some cases there is even an overlap between the coupling
slot and the radiating slot when the former is projected on the wall containing the
radiating slot, thereby requiring the use of the singularity in the Green’s function
(when ~ z ¼~z 0 in Equation (2.60) in Section 2.8 in Chapter 2) to evaluate the
coupling.
An extensive investigation of higher-order mode coupling in coupling
junctions was carried out [5] using the MoM solution to the integral equations of
the aperture fields of the three slots in a coupling junction. That study showed that
the higher-order mode coupling can be neglected for half-height branch wave-
guides (a ~ b/4) whereas it is significant for standard height waveguides (a ~ b/2) as
well as quarter height waveguides (a ~ b/8) and it is greater for the hard coupling
geometry compared to the soft coupling. In a large array or sub-array consisting of
many radiating slots in each radiating waveguide, the errors in the slot excitations
198 Slotted waveguide array antennas: theory, analysis and design

Main
guide

θ
1
a
X01 2Lc W
2
X02

Branch
guide
(a)

Main
guide

2
2Lr2

2Lr1
1

Branch
guide
(b)

Figure 8.23 Geometry of the coupling junction: (a) hard coupling; (b) soft
coupling.
IEEE 1994, reprinted from [5], with permission

are primarily confined to the coupling and radiating slots in the junction regions. If
the vast majority of the radiating slots elsewhere have close to the desired excita-
tions, the radiation pattern is expected to be of acceptable accuracy in the H-plane.
However, the errors in the coupling slots may affect the input match and affect the
pattern in the E-plane. In a small array such errors may become substantial since
the percentage of total number of radiating slots in the coupling junction region is
greater.
There is a simple method to include the higher-order mode coupling discussed
earlier in the design of planar arrays. Let us consider the problem of three slots in
the junction region using an MoM solution to the coupled integral equations of the
aperture fields of slots. The calculation can be made for two sets of excitations,
with and without higher-order mode coupling in the junction region. Then one can
perturb the coupling slot’s tilt and length and the offsets and lengths of the radiating
slots, such that the excitations of the perturbed slots with higher-order mode
 Concepts and models for advanced designs 199

coupling are the same as the excitations of the original slot parameters without
higher-order mode coupling. We then replace the original slots in Elliott’s design
by the perturbed slots. The rationale for this design is that higher-order mode
coupling is a localised phenomenon and therefore we can account for it by
considering only the three slots in the junction region, without accounting for
the presence of other slots. In some cases, however, this method does not produce
satisfactory results and one may need to use a full wave method to analyse the
entire slot array to account for the coupling effects. Such a technique is discussed in
Section 8.10 [39].

8.9 Finite ground plane effects

In the design and analysis of slot arrays, it is usually assumed that the radiating
slots are embedded in an infinite perfectly conducting ground plane. While the
perfectly conducting ground plane assumption generally is very good, the effect of
a finite ground plane on the pattern characteristics can be significant. Edge dif-
fraction effects influence the excitation of slots close to the edge, in addition to
affecting the radiation patterns in angular regions far from broadside. The mutual
coupling expression for the infinite ground plane case can be improved by an
additional term to include the edge diffraction effects [40], especially for slots that
are close to the edges of the ground plane. A simple model using Keller’s geome-
trical theory of diffraction (GTD) was introduced by Josefsson [41] to account for
edge effects in mutual coupling calculations, and its accuracy to predict the mutual
coupling between elements of a linear array of slots aligned in the H-plane has been
validated with experimental results. Josefsson’s model is easy to implement in the
design procedure, not only for the mutual coupling terms but for the self-coupling
also, as explained later.
Figure 8.24 shows two slots m and n and one edge of the ground plane. The ray
incident from slot m gets diffracted at the edge and couples a voltage at slot n. The
expression gmn for the external mutual coupling between slots m and n in (5.66) in
Chapter 5 is proportional to the integral of the magnetic field produced by a unit
magnetic current in slot m in the aperture of slot n weighted by the cosine function
of the aperture electric field of slot n [see (5.40) and (5.41) in Chapter 5]. There is

n' ρ
γ s
n ρ

s1
ρ1

Figure 8.24 The geometry of the edge diffraction effect

200 Slotted waveguide array antennas: theory, analysis and design

an additional term for the magnetic field in the region of slot n due to a unit
magnetic current in slot n, due to edge diffraction, derived later.
The magnetic field incident at the edge is given by

H inc ¼ H0 expðjks1 Þ=s1 (8.30)

where H0 is a constant.
The edge-diffracted field is given by
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s1
H d ¼ H inc Dh expðjksÞ (8.31)
sðs þ s1 Þ

Dh is the hard diffraction coefficient given by [42]

" #
expðjp=4Þ sinðp=nÞ 1 1
Dh ¼ pffiffiffiffiffiffiffiffi ff0
þ fþf0
(8.32)
n 2pk sinðgÞ p p
cos n  cos n cos n  cos n

The parameter n is given by the interior wedge angle (2 – n)p. For a ground
plane of zero thickness, n ¼ 2 whereas for a thick edge, the interior wedge angle is
90 and n ¼ 0.5. j and j0 are the angles that the incident and diffracted rays make
with respect to the plane surface, and they are zero in this case. Because of grazing
incidence, a factor ½ has been included in the diffracted field since the total inci-
dent field at grazing incidence is the sum of an incident and a reflected field [43].
Thus we obtain the diffracted field as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
exp ½ jk ð s þ s1 Þ  ð 1  j Þ ðs þ s1 Þ
H d ¼ H0 pffiffiffiffiffiffiffiffi (8.33)
ðs þ s1 Þ 16p krr1

The diffracted field at slot n can be written as the incident field at the image
slot n0 multiplied by the edge diffraction parameters. Therefore the original mutual
coupling term gmn discussed in Section 6.4 in Chapter 6 is augmented by adding an
extra term containing the integral between slots m and n0 (image of n) as follows.
gmn ! gmn þ H0 gmn0 (8.34)
where
rffiffiffiffiffiffiffiffiffiffiffiffi
1 þ j s þ s1
H0 ¼ pffiffiffiffiffiffiffiffi (8.35)
16p krr1
The field of slot n, incident at the edge, after getting diffracted may couple to
itself. This term can be added to the summation in (6.46b) in Chapter 6 as
Vns gnn0 ðLn ; Ln Þ (8.36)
The gnn0 integral will be carried out using the procedure discussed in Section 6.4
in Chapter 6 for slots n and n0 of half-length Ln, separated by a distance 2r in the
E-plane direction and aligned in the H-plane direction, as shown in Figure 8.24.
 Concepts and models for advanced designs 201

8.10 MoM solution to the integral equations of a planar

array and applications

Section 5.2 in Chapter 5 discusses the MoM solution to the integral equation of the
aperture field of a single slot. This is the most efficient technique to compute the
characteristics of a slot since the unknowns are the equivalent magnetic currents in
the slot apertures, as opposed to techniques such as the finite element method and
finite difference method where the unknowns are in a volume. In addition, when
one uses entire domain sinusoidal basis functions for the longitudinal variations of
the magnetic current, the first term, a half cosinusoidal distribution, is an excellent
approximation to the aperture electric field of a slot near resonance. Therefore a
few unknowns are usually adequate to obtain good accuracy.

8.10.1 Formulation of the coupled integral equations

Several simplifying assumptions are made in deriving the integral equations of a
planar slot array. For narrow slots the longitudinal component of the aperture
electric field is ignored. Consistent with this narrow slot approximation, the
boundary conditions are enforced for the transverse component of the electric field
and the longitudinal component of the magnetic field only. The slots are assumed to
be square-ended and embedded in a perfectly conducting infinite ground plane.
The milling process in manufacturing usually yields round-ended slots. Round-
ended slots are not easy to work with in an analysis model. Therefore one uses an
approximate equivalence between round-ended slots in the experimental work with
the square-ended model used in the analysis based on an equal area criterion. The
equal area criterion yields the following relationship between round-ended and
square-ended slots [see (5.54)].
2Lr ¼ 2Ls þ 0:215w (8.37)
where 2Lr is the length of the round-ended slot, 2Ls is the length of the square-
ended slot and w is the width.
In a study of round-ended slots using the MoM/finite element method by
Zhang et al. [44], the aperture equivalent magnetic current was assumed to be even
symmetric and essentially longitudinal to the slot except near the round ends where
an edge based finite element model was used. The authors show good comparison
between computed and measured S21 values. It is not clear how well such a model
works for S11 especially for slots with small scattering. Based on their work the
authors recommend the following expression for square-ended to round-ended slot
length conversion.
2Lr ¼ 2Ls þ 0:245w (8.38)
For typical narrow slots employed in many practical waveguide array antennas
with a width to length ratio of 1/10, the two formulas are within 0.3% of the slot
lengths. For very wide slots with a width to length ratio of ¼, the deviation between
the lengths in the two formulations is 0.7%.
202 Slotted waveguide array antennas: theory, analysis and design

The slot thickness is accounted for by considering the two (inner and outer)
apertures of the slot. The formulation of the coupled integral equations starts with
shorting out each slot aperture by an infinitesimally thin perfectly conducting sheet.
By invoking Schelkunoff’s equivalence principle (see Section 2.7), we place
equivalent magnetic currents on each such conducting sheet. Enforcement of the
continuity of the tangential electric fields across each aperture of the original pro-
blem requires that the magnetic current on either side of the conducting sheet shall
be equal in magnitude but opposite in direction. Figure 8.25 shows the cross section
of a thick radiating slot which we assume is the mth slot.
By enforcing the continuity of the longitudinal component of the aperture
magnetic field for the mth slot we obtain (8.39) and (8.40).
      X  
extþ ~ ext ext ~ ext ext ~ int extþ ~ int
Hzm M m  Hzm M m  Hzm Mm þ Hzm Mq ¼ 0 (8.39)
q

      X  
intþ ~ int intþ ~ ext int ~ int int ~ int
Hzm M m þ Hzm M m  Hzm Mm  Hzm M n ¼ Hzm
inc
(8.40)
n6¼m

Thus, there are two coupled integral equations for slot m, one for the exterior
aperture facing the exterior half space and the other for the interior aperture facing
the waveguide. There are two unknowns for slot m, M ~ ext and M~ int , the magnetic
m m
currents in the exterior (ext) and interior (int) aperture, respectively. The slot m is
coupled to all the radiating slots in the entire array through external mutual cou-
pling in the summation over q. Similarly the slot m is coupled to all the other
radiating slots and the coupling slot in that radiating waveguide, in the summation
over n, through TE10 mode scattering. In addition, slot m is coupled to adjacent
radiating slots and to the coupling slot, if it is in the immediate vicinity, through
higher-order modes. The magnetic field of the magnetic current in each of the three
regions (waveguide interior, exterior half space and the cavity region to account for
the slot thickness) is expressed in terms of integrals over the magnetic current
with the appropriate Green’s function for the waveguide, half space and cavity,
respectively.
For a linear array of standing wave or travelling wave type directly fed by an
incident TE10 mode wave, an additional contribution on the right side of (8.40)
exists. This source term is absent in planar standing wave arrays where all the

ext
+
–
+
–
Thick wall int

Figure 8.25 The internal and external apertures of the mth radiating slot
 Concepts and models for advanced designs 203

radiating waveguides are supposed to be shorted at both ends and excited by cou-
pling slots. A similar pair of coupled integral equations results when the continuity
of the longitudinal component of the magnetic field is enforced across the apertures
of each coupling slot. The incident field term is absent if the feed waveguide
containing the coupling slots is shorted at both ends and excited by an input
slot coupling to another waveguide layer. An example of such an input shunt-series
coupling slot is found in Figure 7.12. The source term then appears in the integral
equation for this slot.

8.10.2 MoM solution

The coupled integral equations (8.39) and (8.40) are reduced to a system of
simultaneous equations in the MoM where we expand each unknown magnetic
current in terms of a set of expansion functions with unknown coefficients and form
the inner product of each integral equation with a set of weighting functions. The
resulting matrix equation is of order 2N  M where N is the total number of slots in
the array and M is the number of unknown coefficients in each. M need not be the
same for all slots. Entire domain sinusoidal expansion is a good choice for long-
itudinal variation of the aperture magnetic current. The transverse variation is
assumed to be uniform. For narrow slots there is no need to use the edge conditions
discussed in Section 5.3. The weighting functions may be chosen same as the
expansion functions, sinusoidal and pulse or sinusoidal and delta. In the latter case
we use point matching in the transverse direction along the centreline of the slot.
In deriving the integral equations one faces a difficulty in obtaining the mag-
netic field in the waveguide cavity region for a standing wave array. Figure 8.26
shows a waveguide with several slots with shorts at both ends. In the coupling
terms between slots m and n the contribution from the forward scattered TE10
mode will become C10/[1GLGR exp(j2b10D)] in place of just C10. GL and GR are
the reflection coefficients of the left and right shorts, b10 is the phase constant of the
TE10 mode and D is the spacing between the shorts at the two ends. This expression
results from the infinite number of reflected waves from both ends. D is an integer
multiple of a half guide wavelength at the design frequency. For a lossless system,
the denominator becomes zero since GL and GR are both –1 for the short. One could
avoid this problem by assuming a small amount of loss in the metallic waveguide,
so that the resulting round trip attenuation will make the sum of the infinite number
of reflected waves finite.
Once the matrix equations are solved, we determine the expansion coefficients
of the magnetic equivalent currents of all slots. The aperture distribution is then

Short Short

m C10 n

Figure 8.26 Slots cut in a waveguide wall with shorts at both ends
204 Slotted waveguide array antennas: theory, analysis and design

known. The radiation pattern may be computed by summing the radiated far fields
of the magnetic currents of all slots. The reflection coefficient at the input port is
found from the scattering from the coupling slots. Directivity is readily found since
the incident wave power and reflected power are known and the power density in
the main beam peak direction can easily be calculated.

8.10.3 Improved design procedure using the MoM solution

In this section we discuss the use of the MoM solution to improve the design of a
planar slot array. We will start with the radiating slots.
The design of a planar array using Elliott’s technique yields offsets and lengths
of all radiating slots and lengths and tilt angles of coupling slots as well (see
Section 7.2 in Chapter 7). A radiating slot with a strong excitation coefficient in the
middle of the array is chosen as a reference element. Let the offset and the half-
length of the nth slot be x0 and L, respectively. The complex voltage of the nth slot
normalised to the reference slot voltage is Vsn (¼u þ jv) while the desired value of
the normalised voltage of the nth slot is Vin. The difference (Vsn – Vin) ¼ du þ jdv is
the
Xerror. The 2mean square error for all radiating slots in the aperture is
1
N jVsn  Vin j where N is the number of radiating slots in the array.
n
By using the MoM analysis program three times, first for the original values of
each slot parameter and then by incrementing the offset plus half the waveguide
width, x0 alone, and then by incrementing the half-length, L alone, we determine
the numerical values of the partial derivatives for slot n. The perturbations for
the parameters of slot n to eliminate the errors are then given by (8.41).
2 @u @u 31 "
" # #
Dx0 6 @x0 @L 7 du
¼6
4 @v
7 (8.41)
DL @v 5 dv
@x0 @L
The partial derivatives obtained numerically are not robust for very small values of
the derivatives. In that case the slot parameters are left unchanged. Since radiating
slot perturbations affect the return loss, coupling slots have to be perturbed also,
as discussed next.
It is assumed that the feed waveguide is excited at one end. The reflection
coefficient G at this input port, consists of the incident TE10 wave that is reflected at
the end of the feed waveguide short with reflection r, and the TE10 mode scattered
from all coupling slots, s. The initial design based on Elliott’s technique will
produce s  –r.
G¼rþs (8.42)
Let a ¼ argðsÞ þ argðrÞ (8.43)
For a match a is zero, assuming that the magnitudes of r and s are nearly equal.
If a is not equal to zero, we need to delay s by a to compensate for the phase error.
 Concepts and models for advanced designs 205

Therefore all coupling slots need to be detuned so that the input reflection coeffi-
cient will be equal to that given by (8.44).

G0 ¼ r þ s expðjaÞ (8.44)

The normalised input impedance, z, then would become

Z ¼ ð1 þ G0 Þ=ð1  G0 Þ (8.45)

Z can be made approximately equal to 1 by introducing a factor by which

coupling coefficients of all slots are scaled so that the new values of the mth
coupling coefficients is given next.

c2m ¼ jZ 1 jS11 =ð1  S11 Þ (8.46)

S11 in (8.46) is the original scattering parameter of the mth four port coupling slot
with match terminations in all ports with the ports referenced to planes passing
through the centre of the slot. Then the perturbed scattering parameter S11m of the
same will become
 
S11m ¼ c2m expðjaÞ= 1 þ c2m (8.47)
The magnitude is perturbed by the scale factor introduced by the coupling
coefficient in (8.46) and the phase change is due to perturbation for s in (8.44).
Since the coupling coefficients of all coupling slots are scaled by the same factor
and the phase of all S11 are perturbed by the same amount there is very little
additional error introduced in the aperture distribution of the radiating slots.
However, because of the approximation in the assumption that the phase of the
scattering parameters of all coupling slots track over a small frequency range near
resonance, the perturbation produces only an approximate match condition and
further iterations are required. Equations (8.46) and (8.47) can be modified such
that the coupling slot perturbations produce the desired value of the total normal-
ised resistance if there is a matching input slot with centred feed as in some arrays
[4,45]. Such a case is not discussed here.
The radiating slot perturbations are carried out one slot at a time, first for all
radiating slots, followed by the perturbations of all coupling slots. We then iterate
the process 3–4 times until the errors in the aperture distribution are small enough
with an acceptable match at the design frequency. Criteria for stopping the iteration
may use the maximum allowable error or the maximum mean absolute error for the
aperture distribution and the desired value of the reflection coefficient. It is possible
to perturb only slots exhibiting large errors in the aperture distribution thereby
speeding up the process.
The improved design process was demonstrated in a theoretical study of a 5  4
array, shown in Figure 8.27 for a radiometer type application designed at 13.285 GHz
[39]. The design parameters are shown in Table 8.3. This desired distribution pro-
duces E-plane sidelobes in the range of 35.5–32 dB below the peak and H-plane
sidelobes in the range of 21.5–31.7 dB below the peak.
206 Slotted waveguide array antennas: theory, analysis and design

Figure 8.27 A 5  4 slotted waveguide array

Table 8.3 Design parameters of the 5  4 array

Feed Radiating Slot Wall Aperture Aperture

waveguide waveguide width thickness distribution distribution
(mm) (mm) (E-plane) (H-plane)
16.33 mm  15.10 mm  1.27 mm 0.508 mm 0.28:0.75:1:0.75:0.28 0.606:1:1:0.606
8.16 mm 7.55 mm

In the initial design using the Elliott’s procedure, the total normalised slot
conductance of all radiating slots in each radiating waveguide was assumed to be 2
with an input match at the feed port enforced. Figure 8.28 shows the E-plane
pattern of this array with the legend ‘original’. A substantial amount of higher-order
mode coupling is present in the slots in the five coupling junctions, thereby affecting
the return loss and the E-plane pattern. It was concluded that the major source of error
in the original design was the higher-order mode coupling in the five coupling
junctions.
Figure 8.28 shows also the E-plane patterns computed with an approximate
method of accounting for the higher-order mode coupling in the junction slots in
the original design. In this method for which results are shown by the legend
‘approximate’, one perturbs each coupling slot and a pair of straddling radiating
slots by considering one junction at a time. The perturbation compensates for the
errors introduced by the higher-order mode coupling in the original design. This is
a major simplification considering only six coupled integral equations of the three
slots for each junction. However, the method is approximate and, as shown by
Figure 8.28, the improvement produced by this method is marginal. Obviously the
iterative method of perturbing all slots using the moment method analysis of the
entire array antenna, shown by the legend ‘improved’, produces very good results.
 Concepts and models for advanced designs 207

20

10 Improved
Approximate
Pattern level in dB 0 Original

–10

–20

–30

–40
–90 –60 –30 0 30 60 90
Theta in degrees

Figure 8.28 E-plane patterns of the 5  4 array using the original,

improved and approximate methods. Reprinted from [39],
with permission

–5

–10
|S11| in dB

–15

–20

–25

–30 Original
Improved
–35
13 13.1 13.2 13.3 13.4 13.5 13.6
Frequency in GHz

Figure 8.29 Magnitude of the reflection coefficient of the original and the
improved designs. Reprinted from [39], with permission

For arrays made up of sub-arrays, the proposed method can be applied by con-
sidering one sub-array at a time while treating all other sub-arrays as additional
source terms or external excitations.
The improved design procedure using MOM applied to this antenna yielded
radiation patterns close to the ideal patterns. Since the higher-order mode coupling
in the junctions did not have a significant effect in the H-plane patterns, they are not
shown here. Figure 8.29 shows the reflection coefficient in dB over a range of
frequencies. The original design has the best match at 1% below the centre of the
desired frequency band. The improved design places the best performance at the
208 Slotted waveguide array antennas: theory, analysis and design

20.3

20.2

20.1
Directivity in dB

20

19.9

19.8
Original
19.7
Improved
19.6
13 13.1 13.2 13.3 13.4 13.5 13.6
Frequency in GHz

Figure 8.30 Directivity of the original and the improved designs. Reprinted
from [39], with permission

Table 8.4 Coupling slot parameters in the original and improved designs

Original design Improved design

Tilt angle ( ) Length (mm) Tilt angle ( ) Length (mm)
6.40 10.53 6.92 10.49
17.45 10.55 19.25 10.51
22.62 10.56 24.96 10.53
Source: Reprinted from [39], with permission.

centre frequency and achieves the desired 10 dB return loss over 4% frequency
band. Similarly, the improved design achieves between 0.1 and 0.235 dB increase
in the values of directivity within the required 4% frequency band compared to
the original design. Figure 8.30 compares the directivity of the two designs.
E-plane patterns of the improved design at band edges, not shown here, were closer
to the desired patterns.
Table 8.4 shows the lengths and tilt angles with respect to the feed waveguide
axis of the coupling slots for the original and improved designs. Since coupling
slots exhibit symmetry with respect to the centre only the three elements closest to
the feed port are specified. The perturbations are found to be small. Table 8.5
shows similar results for the offsets and lengths of radiating slots in the three
radiating waveguides starting from the one nearest to the feed port. In the original
design radiating slot parameters exhibit symmetry with respect to the diagonal, but
for the sign of the offset. In the improved design such symmetry is present in an
approximate sense. Radiating slot perturbations are significant compared to those
of coupling slots.
 Concepts and models for advanced designs 209

Table 8.5 Radiating slot parameters in the original and improved designs

Original design Improved design

Offset Length Offset Length
(mm) (mm) (mm) (mm)
1.240 10.943 1.669 10.997
2.689 11.092 2.553 10.859
2.130 11.074 2.506 11.143
1.661 10.896 1.297 10.744
1.490 10.849 1.192 10.705
2.422 10.923 2.026 10.665
2.483 11.050 1.948 10.709
1.433 10.755 1.316 10.639
1.453 10.774 1.423 10.744
2.372 10.945 2.372 10.945
2.372 10.945 2.358 10.939
1.453 10.774 1.412 10.746
Reprinted from [39], with permission.

8.11 Global optimisations for improved designs

Biology-inspired global optimisations such as genetic algorithms (GA) [46–48] and

particle swarm optimisations [49] have been applied to many antenna problems.
The design procedure discussed previously for linear and planar slot arrays is
generally carried out at the centre frequency of the desired band. Such a design
usually results in performance degradation at band edges. In a monopulse array, the
slot excitations are different for sum and difference patterns. Therefore, the Elliott
design procedure is not readily applicable if the sum mode and difference mode
patterns are to be realised. In a previous work a design was implemented by
averaging the values of offsets and lengths obtained for each of three patterns, sum,
elevation and azimuth difference patterns [50]. Multi-objective and multi-
frequency designs can be carried out as an optimisation exercise using an analysis
program. In this section we show a couple of examples of GA optimisations of slot
arrays. GA is a powerful technique to find optimum solutions without getting stuck
in local maxima or minima. It is a robust process to search a large solution space
efficiently since it uses probabilistic rules.

8.11.1 GA optimisation of a travelling wave linear array

The GA optimisation was carried out for a travelling wave array [51], based on a
previous design [52], also studied in Section 6.6, using an analysis model in terms
of equivalent circuits on a transmission line. The analysis was facilitated by rear-
ranging Elliott’s equations for an N-element array by Hamadallah [53]. The main
analysis equations shown in (6.51)–(6.53) are reproduced here without the TE20
210 Slotted waveguide array antennas: theory, analysis and design

mode internal higher-order mode coupling terms. The slot voltages in the left
column matrix are given by
k2
½G  ½V s  ¼ V1 ½ E  (8.48)
k1
where

Gmn ¼ gmn if n
m (8.49)
Gmn ¼ gmn  jk2 fm fn sinfðm  nÞb10 d g if n < m (8.50)

In (8.48) [G] is an N  N square matrix whereas [V s] and [E] are N  1 column

matrices where N is the number of slots. Column matrix elements [E] are specified
in (6.55). The constants k1 and k2 are specified in (6.55a) and (6.55b) in Chapter 6.
Using (6.49) and (6.50) we can obtain the input admittance and then the VSWR.
The length and offset of each slot are quantised into 64 values, 6 bits or genes.
For the 21-element slot array a set of values of slot lengths and offsets is repre-
sented by a chromosome of 252 genes. A figure of merit is assigned to each array
design (organism) according to the following performance function.
 
a Pload b 1 1
D 1:0  : : (8.51)
Pin VSWRg SLLt
where D is the directivity, SLL the sidelobe level and Pload/Pin is the ratio of power
dissipated in the load to the input power. The directivity is calculated readily from the
power radiated, Pin – Pref – Pload and the power density in the maximum far-field
direction. The population size was kept at 50. The values of a, b, g and t were varied
depending on the parameters that were optimised. When bandwidth was optimised the
performance function was averaged over a number of frequencies within the band.
The parameters of the studied travelling wave array were as follows:
a ¼ 22.86 mm, b ¼ 10.16 mm, wall thickness ¼ 1.27 mm and slot width 1.5875 mm.
The design frequency is 9.375 GHz, the slot spacing 17.4 mm and the beam peak at
45 with respect to the waveguide axis in the H-plane. The slots are all offset on the
same side of the centreline. The result of this optimisation exercise indicates possible
performance improvements using GA. Hamadallah’s analysis equations [53] with
MoM data for Elliott’s travelling wave array gave the following initial results:
VSWR ¼ 1.05, Pload/Pin ¼ 16.7%, SLL ¼ –20.1 dB. The results compare well with
the experimental results in Elliott’s paper [52]: VSWR ¼ 1.05 and SLL ¼ –22 dB.
The slight discrepancy in the SLL may be attributed to the edge diffraction of the
finite ground plane, especially in the E-plane. The value of Pload/Pin could not be
measured but it was computed by Elliott to be 12.3%. The computed results in [51]
are closer to the experimental results. In subsequent optimisations we varied the
values of slot lengths and offsets to optimise the figure of merit.
When VSWR and Pload/Pin are minimised, values of 1.011% and 2.1%, respec-
tively, were obtained. The SLL improved to 22.1 dB. If the sidelobe level alone was
optimised, it reduced to a low value of 25.2 dB, yielding VSWR of 1.016 and Pload/Pin
 Concepts and models for advanced designs 211

of 4.6%. When the design was optimised for 5% bandwidth, the SLL was better than
21.9 dB, while Pload/Pin was below 6.2% within the band. The VSWR was better than
1.02 in the frequency range of interest. For 10% bandwidth optimisation, SLL was
better than –20.9 dB and Pload/Pin was lower than 7.1%. VSWR values are higher than
those for the 5% bandwidth case but still below 1.04 in the entire band.
The optimisation exercises show the power of the GA when combined with an
antenna analysis technique. However, this theoretical exercise did not include the
internal higher-order mode coupling and has the limitations of the Elliott’s design
procedure discussed in Chapter 6.

8.11.2 GA optimisation of a planar standing wave array

using the MoM analysis
In this section we describe a GA optimisation of a planar array to maximise the
directivity and return loss over 5% bandwidth [54]. With GA the moment method
solution of the array is computed thousands of times making the procedure prohibi-
tively time consuming. The computational effort can be reduced substantially, how-
ever, by using only one expansion and test function for each slot aperture in the MoM
solution. The basis function used is a half sine wave and it is a good approximation to
the aperture distribution of a slot near resonance. However, the accuracy of the solution
with only one basis function is not adequate. We will establish the equivalence between
a slot with one basis function and a slot using many basis functions so as to obtain better
results later.
In order to match slotted waveguide arrays reactive elements can be inserted in
the input waveguide port. Irises are easy to design and manufacture and are very
useful as matching devices in waveguides. The inductive irises are preferred over the
capacitive irises (see Figure 8.31) because of their greater power handling capability.
Figure 8.32 shows a waveguide section consisting of two inductive irises. There are

Width

Width

(a) Inductive iris (b) Capacitive iris

Figure 8.31 Cross sections of inductive and capacitive irises in rectangular

waveguides

d2 d1
Output port Input port

Figure 8.32 A waveguide H-plane section with two thick inductive irises
212 Slotted waveguide array antennas: theory, analysis and design

four parameters, widths of the two irises in the H-plane, the spacing between the
irises, d2, and the distance between the input port and the nearest iris, d1, that may be
optimised to maximise the return loss over a frequency band. Iris thickness is usually
fixed. The input port is connected to the slot array to be matched.
Mode matching (MM) is a fast and accurate technique for solving boundary
value problems for waveguide discontinuities [55]. It is ideally suited to the analysis
of waveguides with thick irises. GA optimisation in conjunction with MM may be
used to improve the impedance match over the required frequency bandwidth.
We start with the design of an 8  10 array consisting of eight radiating
waveguides, each containing ten radiating longitudinal slots, using Elliott’s pro-
cedure. The aperture distribution is assumed to be uniform. The dimensions of the
input and radiating waveguides are 5.70 mm  2.85 mm while those of the feed
waveguide are 5.673 mm  2.836 mm. The slot widths and thicknesses are
0.508 mm and the centre frequency is 35.75 GHz. Diagonal symmetry is assumed
in the array for the radiating slots whereas centred-inclined coupling slots are
assumed to be symmetric with respect to the centre but for alternating signs for tilt
angles. In the GA design there were 85 parameters, offset and length for each of the
40 radiating slots, tilt angle for each of the four centred-inclined coupling slots and
offset from the centreline for the input shunt-series coupling slot. Other radiating
and coupling slot data are obtained from symmetry. Coupling slots are all assumed
to be resonant. Binary GA with 7 bits for each parameter is employed so that the
length of each chromosome is 595. This allows a resolution in the variable size to
be less than the machine tolerance, assumed to be 1 mil (~25 m).
For each element of the population (chromosome) GA performs the moment
method analysis of the slot array at nine equally spaced frequency points within a
4% frequency band centred at 35.75 GHz. The lowest values of the broadside gain
and the return loss are weighted to specify the fitness parameter in GA. Thus we
optimise the worst case values of gain and return loss within the bandwidth for a
nominally uniform aperture distribution. In the GA the fitness parameter is not
allowed to increase if the return loss is greater than 12 dB. A population size of 10,
tournament selection, mutation probability of 0.1, and probability of crossover of 0.5
with uniform crossover are used with the best individual replicated into the next
generation [46–48]. Optimum solutions are achieved typically after 600 generations.
The range of allowed values for the offset from the centreline for radiating slots
and the input shunt-series coupling slot is about 1/10th of the waveguide ‘a’ dimen-
sion. The slot length and the distance between the short and the nearest slot centre are
allowed to have a range of about 5% of their nominal values while the tilt angle of
each centred-inclined coupling slot is allowed to vary 5 with respect to the initial
value. The computer time for each GA execution for about 600 generations is about
one week in a dedicated PC with an Intel Core Duo processor T9600 and a clock
speed of 2.8 GHz with 4GB of RAM. The computer time may be drastically reduced
by using many parallel processors or using resources such as cloud computing.
In the case of matching sections with irises, the range of values for the spacing
between irises and the spacing between the array feed port and the nearest iris is a
maximum of half a guide wavelength at the centre frequency. The irises are
 Concepts and models for advanced designs 213

assumed to be 20 mil (~0.5 mm) thick and their widths are allowed to have a range
of 1/3 of the waveguide ‘a’ dimension. There are four variables and each is 10 bits
long so that the length of the chromosome is 40. Other parameters of GA in
this case are identical to the previous GA MOM optimisation. The fitness
parameter is the lowest value of the return loss computed at nine equally spaced
frequency points within 4% bandwidth centred at 35.75 GHz. GA mode matching
(MM) is an extremely fast computational process.

8.11.3 Equivalence between a slot with one basis function

and a slot with many basis functions
Prior experience shows that for the radiating slots and for the input shunt-series
coupling slot nine basis functions are needed for good accuracy while three basis
functions are sufficient for centred-inclined coupling slots. If the slot parameters
obtained by GA are used in a moment method analysis program with nine (or three)
basis functions, the results for gain and input reflection coefficient would deviate
substantially from the corresponding GA results with one basis function for each
slot. Scattering data were computed for a range of values of slot offsets and lengths
for the longitudinal radiating slots and the shunt-series input slot, and for a range of
values of lengths and tilt angles for the coupling slots using the moment method
solution of an isolated slot for the pertinent integral equations. Notice that for the
nth slot the two slot parameters, p1n and p2n, are the offset and the length, respec-
tively, for a radiating slot and the input shunt-series coupling slot whereas they
are the tilt angle and the length, respectively, for a centred-inclined coupling slot.
The desired slot parameters p1n and p2n (m basis functions) are obtained from the
parameters q1n and q2n (one basis function) of the corresponding slot in the opti-
mum GA design such that (8.52) is satisfied.
S11 ðp1n ; p2n ; mÞ ¼ S11 ðq1n ; q2n ; 1Þ (8.52)
In (8.52), the scattering parameter S11 is the reflection coefficient at the input port
of an isolated slot where all the other ports are matched. The right side of (8.52) is
computed with one basis function with slot parameters q1n, and q2n. In order to
determine slot parameters p1n, and p2n, we perform a two-dimensional interpolation
of the computed data with m (nine or three) basis functions over a range of values
of the magnitude and angle of S11 for radiating slots. For resonant centred-inclined
coupling slots and input shunt-series coupling slot, one-dimensional interpolation is
carried out. For slots near resonance, all scattering parameters can be expressed in
terms of S11. Therefore (8.52) provides equivalence. The rationale for the equiva-
lence stated here is further justified by the fact that the array analysis employing
nine or three basis functions, with parameters p1n and p2n for the nth slot, n ranging
from 1 through 89, yields results close to those produced by the GA design using
one basis function for each slot aperture. Ando et al. employ a similar equivalence
between a slot with one basis function and a corresponding slot with nine basis
functions where the length is the only parameter used as a variable [56]. We found
better results with two variables in (8.52).
214 Slotted waveguide array antennas: theory, analysis and design

Figure 8.33 shows gain values for three cases. The original design using Elliott’s
procedure has a gain plot with a wide variation in the frequency range. A GA solution
obtained with one basis function for each slot aperture has nearly a constant gain over
a 4% bandwidth. For each radiating slot and the input shunt-series coupling slot in
this GA optimisation, we obtain an equivalent slot with nine basis functions while for
each coupling slot the equivalent slot with three basis functions is obtained from
Equation (8.52). The gain values calculated for an array with these equivalent slots
is shown using the legend ‘nine bases GA’. The latter two plots are very similar.
Figure 8.34 shows results for the reflection coefficients for the three cases. These
results further justify the equivalence established by (8.52).

27

26.5
Gain in dB

26 Orig design
1 basis GA
9 bases GA
25.5
34.5 35 35.5 36 36.5 37
Frequency in GHz

Figure 8.33 Gain plots of GA designs with one basis function and an equivalent
array with nine basis functions. The legend ‘orig design’ used
Elliott’s procedure. Reprinted from [54], with permission

–5
–10
Reflection coefficient in dB

–15
–20
–25
–30 Orig design
1 basis GA
–35 9 bases GA
–40 –10 dB ref coef

–45
34.5 35 35.5 36 36.5 37
Frequency in GHz

Figure 8.34 Reflection coefficient plots of GA designs with one basis function
and an equivalent array with nine basis functions. The legend
‘orig design’ used Elliott’s procedure. Reprinted from [54],
with permission
 Concepts and models for advanced designs 215

8.11.4 GA optimised results

Figure 8.35 illustrates the computed values of gain as a function of frequency for
the GA optimisation. The GA yields nearly uniform gain over 5% frequency range.
In the GA optimisation the fitness parameter was first specified as a weighted sum
of the return loss and gain. The reflection coefficient was not allowed to
exceed –12 dB. The GA produces significantly better gain values. Figure 8.36
shows the reflection coefficient as a function of frequency. The reflection coeffi-
cient value of the original design is somewhat higher than the nominal –10 dB

27

26.5
Gain in dB

26
Orig design

25.5 GA

25
34.5 35 35.5 36 36.5 37
Frequency in GHz

Figure 8.35 Gain as a function of frequency of different array designs; ‘orig

design’ used Elliott’s procedure and in the GA design the gain
was optimised over a 4% bandwidth. Reprinted from [54], with
permission

–5
Reflection coefficient in dB

–10

–15

–20
Orig design
–25
GA
–30
–10 dB ref coef
–35
34.5 35 35.5 36 36.5 37
Frequency in GHz

Figure 8.36 Reflection coefficient as a function of frequency for different array

designs; ‘orig design’ used Elliott’s procedure and GA optimised
the gain and the reflection coefficient. Reprinted from [54], with
permission
216 Slotted waveguide array antennas: theory, analysis and design

30
Orig design
20
GA
Pattern level in dB

10

–10

–20

–30
–90 –60 –30 0 30 60 90
(a) Theta in degrees

30
Orig design
20 GA
Pattern level in dB

10

–10

–20

–30
–90 –60 –30 0 30 60 90
(b) Theta in degrees

Figure 8.37 Radiation patterns at 35.75 GHz; ‘orig design’ used Elliott’s
procedure and GA optimised the gain and the reflection
coefficient: (a) H-plane pattern; (b) E-plane pattern.
Reprinted from [54], with permission

within the 5% frequency band. The initial GA optimisation that produced no higher
than –12 dB reflection coefficient is further optimised by GA with a separate design
of two irises. Thus Figure 8.36 shows a nearly flat reflection coefficient of –15 dB
and exhibits greater than 6% bandwidth for input match with better than –10 dB
reflection coefficient.
Typical radiation patterns computed from the GA optimised design are shown in
Figure 8.37 for the centre frequency. The pattern peaks are normalised to 0 dB for easy
comparison of sidelobe levels. There is no cross-polarised radiation in the principal
planes. GA patterns generally exhibit lower levels of sidelobe radiation. The GA
optimised the gain over a 5% bandwidth. The pattern with the legend ‘orig design’ was
obtained from the array designed by the Elliott procedure. The GA design produced
lower sidelobes compared to the original design using the Elliott procedure.
 Concepts and models for advanced designs 217

References
[1] Rengarajan S. R. ‘Analysis of a centered-inclined waveguide slot coupler’.
IEEE Transactions on Microwave Theory and Techniques. 1989;37(5):
884–889.
[2] Mazzarella G., Montisci G. ‘Wideband equivalent circuit of a centered-
inclined waveguide slot coupler’. Journal of Electromagnetics and Appli-
cations. 2000;14:133–151.
[3] Rengarajan S. R. ‘Characteristics of a longitudinal/transverse coupling slot
in crossed rectangular waveguides’. IEEE Transactions on Microwave
Theory and Techniques. 1989;37(8):1171–1177.
[4] Rengarajan S. R., Zawadzki M. S., Hodges R. E. ‘Waveguide-slot array
antenna designs for low-average-sidelobe specifications’. IEEE Antennas
and Propagation Magazine. 2010;52(6):89–98.
[5] Rengarajan S. R., Shaw G. M. ‘Accurate characterization of coupling junc-
tions in waveguide-fed planar slot arrays’. IEEE Transactions on Microwave
Theory and Techniques. 1994;42(12):2239–2248.
[6] Senior D. C. Higher-order mode coupling effects in a shunt-series coupling
junction of a planar slot array. Ph.D. dissertation, University of California,
Los Angeles, CA, 1986.
[7] Watson W. H. ‘Resonant slots’. IEE Journal. 1946;93(Part 3A):747–777.
[8] Zawadzki M. S., Rengarajan S. R., Hodges R. E. ‘The design of H- and
V-pol waveguide slot array feeds for a scanned offset dual-polarized
reflectarray’. IEEE Antennas and Propagation Society International
Symposium. 2005;2B:417–420.
[9] Silver S. Microwave Antenna Theory and Design. MIT Radiation Laboratory
Series, vol. 12, McGraw-Hill, New York, 1949.
[10] Yee H. Y. ‘Slot antenna arrays’ in Johnson R. C., Jasik H. (eds.). Antenna
Engineering Handbook. McGraw-Hill, New York, 1984.
[11] Kildal P.-S., Rengarajan S. R., Moldsvor A. ‘Analysis of nearly cylindrical
antennas and scattering problems using a spectrum of two-dimensional
solutions’. IEEE Transactions on Antennas and Propagation. 1996;44(8):
1183–1192.
[12] Rengarajan S. R. ‘Mutual coupling between slots cut in rectangular
cylindrical structures: spectral domain technique’. Radio Science. 1996;31(6):
1651–1661.
[13] Rengarajan S. R. ‘Compound radiating slots in a broad wall of a rectangular
waveguide’. IEEE Transactions on Antennas and Propagation. 1989;37(9):
1116–1123.
[14] Maxum B. J. ‘Resonant slots with independent control of amplitude and
phase’. IRE Transactions on Antennas and Propagation. 1960;8(4):384–389.
[15] Derneryd A., Petersson R. ‘Bandwidth characteristics of monopulse slotted
waveguide antennas’. Proceedings of the Fourth International Conference
on Antennas and Propagation, 1985, pp. 27–30.
218 Slotted waveguide array antennas: theory, analysis and design

[16] Montesinos-Ortego I., Zhang M., Sierra-Perez M., Hirokawa J., Ando M.
‘Systematic design methodology for one-dimensional compound slot-arrays
combining method of moments, equivalent circuit model and forward
matching procedure’. IEEE Transactions on Antennas Propagation. 2013;
61(1):453–458.
[17] Gruenberg H. ‘Second order beams of slotted waveguide arrays’. Canadian
Journal of Physics. 1953;31(1):55–59.
[18] Gruenberg H. ‘Theory of waveguide-fed slots radiating into parallel plate
regions’. Journal of Applied Physics. 1952;23(7):733–737.
[19] Forooraghi K., Kildal P.-S., Rengarajan S. R. ‘Admittance of an isolated
waveguide-fed slot radiating between baffles using a spectrum of two-
dimensional solutions’. IEEE Transactions on Antennas and Propagation.
1993;41(4):422–428.
[20] Tang R. ‘A slot with variable coupling and its application to a linear array’.
IRE Transactions on Antennas and Propagation. 1960;8(1):97–101.
[21] Nystrom I. Excitering av centrerad slits i vågledare. Report R/AG-91:188,
Ericsson Radar Electronics AB, Sweden, 1991.
[22] Marcuvitz N. Waveguide Handbook, MIT Radiation Laboratory Series,
Vol. 10, Boston Technical Publishers, Inc., 1964.
[23] Elliott R. S. ‘An improved design procedure for small arrays of shunt slots’.
IEEE Transactions on Antennas and Propagation. 1983;31(1):48–53.
[24] Kim D., Elliott R. S. ‘A design procedure for slot arrays fed by a ridge
waveguide’. IEEE Transactions on Antennas and Propagation. 1988;36(11):
1531–1536.
[25] Falk K. ‘Conductance of a longitudinal resonant slot in a ridge waveguide’.
IEE Proceedings, part H. 1987;134(1):98–100.
[26] Falk K. ‘Admittance of a longitudinal slot in a ridge waveguide’. IEE
Proceedings, part H. 1988;135(4):263–268.
[27] Montgomery J. P. ‘On the complete eigenvalue solution of ridge waveguide’.
IEEE Transactions on Microwave Theory and Techniques. 1971;19(6):547–555.
[28] Bailey M. C. ‘Design of dielectric-covered rectangular resonant slots in a
rectangular waveguide’. IEEE Transactions on Antennas and Propagation.
1967;15(5):594–598.
[29] Bailey M. C. ‘The impedance properties of dielectric-covered narrow
radiating slots in the broad face of a rectangular waveguide’. IEEE Trans-
actions on Antennas and Propagation. 1970;18(5):596–603.
[30] Rexberg L. ‘Vector Fourier transform analysis of dielectric-covered slot in
the broad wall of a waveguide’. Microwave and Optical Technology Letters.
1988;1(10):360–363.
[31] Katehi P. B. ‘Dielectric-covered waveguide longitudinal slots with finite
wall thickness’. IEEE Transactions on Antennas and Propagation. 1990;
38(7):1039–1045.
[32] Casula G. A., Montisci G. ‘Design of dielectric-covered planar arrays of
longitudinal slots’. IEEE Antennas and Wireless Propagation Letters.
2009;8:752–755.
 Concepts and models for advanced designs 219

[33] Mosig J. R. ‘Integral equation techniques’ in Numerical Techniques for

Microwave and Millimeter-Wave Passive Structures, Itoh, I. (ed.), John
Wiley & Sons, New York, 1989.
[34] Mosig J. R., Gardiol F. E. ‘A dynamical radiation model for microstrip
structures’. Advances in Electronics and Electron Physics. 1982;59:139–237.
[35] Alaylioglu A., Evans G., Hyslop J. ‘The evaluation of oscillatory integrals
with infinite limits’. Journal of Computational Physics. 1972;13:433–438.
[36] Rengarajan S. R. ‘Higher-order mode coupling between coupling slots in a
planar slot array’. IEEE Transactions on Microwave Theory and Techniques.
1991;39(7):1219–1223.
[37] Elliott R. S., O’Loughlin W. R. ‘The design of slot arrays including internal
mutual coupling’. IEEE Transactions on Antennas and Propagation.
1986;34(9):1149–1154.
[38] Rengarajan S. R. ‘An optimization procedure for including the higher order
mode coupling between coupling slots in the design of a planar slot array’.
IEEE Antennas and Wireless Propagation Letters. 2008;7:785–787.
[39] Rengarajan S. R. ‘Improved design procedure for waveguide-fed slot arrays
using the method-of-moments analysis’. Electromagnetics. 2012;32(4):
221–232.
[40] Mazzarella G., Panariello G. ‘Evaluation of edge effects in slot arrays using
the geometrical theory of diffraction’. IEEE Transactions on Antennas and
Propagation. 1989;37(3):392–395.
[41] Josefsson L. ‘Mutual coupling calculations including edge effects’. Elec-
tronics Letters. 1994;30(25):2087–2088.
[42] Keller J. B. ‘Geometrical theory of diffraction’. Journal of Optical Society of
America. 1962;52(2):116–130.
[43] Kouyoumjian R. G. ‘The geometrical theory of diffraction and its application’.
Numerical and Asymptotic Techniques in Electromagnetics, in Mittra, R. (ed.).
Springer-Verlag, Berlin, 1975.
[44] Zhang M., Hirano T., Hirokawa J., Ando M. ‘Analysis of a waveguide with a
round-ended wide straight slot by the method of moments using numerical-
eigenmode basis functions’. IEICE Transactions on Communications. 2004;
E-87B(8):2319–2326.
[45] Rengarajan S. R., Zawadzki M. S., Hodges R. E. ‘Design, analysis,
and development of a large Ka-band slot array for digital beam-forming
application’. IEEE Transactions on Antennas and Propagation. 2009;57(10):
3103–3109.
[46] Goldberg D. Genetic Algorithms in Search, Optimization and Machine
Learning, Addison-Wesley, Reading, MA, 1989.
[47] Haupt R., Haupt S. E. Practical Genetic Algorithms, 2nd edn., Wiley,
Hoboken, NJ, 2004.
[48] Haupt R., Werner D. H. Genetic Algorithms in Electromagnetics. Wiley-
IEEE Press, Hoboken, NJ, 2007.
[49] Kennedy J., Eberhart R. ‘Particle swarm optimization’. Proceedings of the
IEEE International Conference on Neural Networks IV, 1995, pp. 1942–1948.
220 Slotted waveguide array antennas: theory, analysis and design

[50] Schaffner J., Kim D., Elliott R. S. ‘Compromises among optimum sum and
difference patterns for planar waveguide-fed slot arrays’. Alta Frequenza.
1981;L(6):312–319.
[51] Jensen A., Rengarajan S. R. ‘Genetic algorithm optimization of a traveling
wave array of longitudinal slots in a rectangular waveguide’. Applied
Computational Electromagnetics Society Journal. 2006;21(3):337–341.
[52] Elliott R. S. ‘On the design of traveling-wave-fed longitudinal shunt
slot arrays’. IEEE Transactions on Antennas and Propagation. 1979;27(5):
717–720.
[53] Hamadallah M. ‘Frequency limitations on broad-band performance of shunt
slot arrays’. IEEE Transactions on Antennas and Propagation. 1989;37(7):
817–823.
[54] Rengarajan S. R. ‘Genetic algorithm optimization of a planar array using full
wave method of moments analysis’. International Journal of Computer
Aided RF and Microwave Engineering. 2013;23(4):430–436.
[55] Wexler A. ‘Solution of waveguide discontinuities by modal analysis’. IEEE
Transactions on Microwave Theory and Techniques. 1967;17(9):508–517.
[56] Ando M., Zhang M., Lee J., Hirokawa J. ‘Design and fabrication of millimeter
wave slotted waveguide arrays’. Proceedings of the European Conference on
Antennas and Propagation, 2010, pp. 1–6.
 Chapter 9
Antenna systems and special requirements

So far we have mainly discussed flat slotted waveguide array antennas consisting of
rectangular waveguides, typically with slots in the broad wall, radiating a fixed
narrow beam. Many other configurations are possible, however, and even more
suitable in special cases.
In this chapter we will investigate how slotted waveguide arrays are used in
system applications where more advanced antenna functions or special configura-
tions are required. As this is a very broad subject we will concentrate on cases
where advantages and perhaps difficulties are of particular interest.

9.1 Phase scanned slotted waveguide arrays

Slotted waveguide arrays have been used extensively in phase steered antenna
systems, in particular for beam steering in one dimension. In this case each slotted
waveguide acts as one element in a one-dimensional scanning array. This may
seem as a severe limitation, unable to do full beam steering in all directions.
However, there are many situations where one-dimensional electronic scan is
sufficient. The principal layout of a phase steered slotted waveguide array is shown
in Figure 9.1.
The element size
element spacing (d in Figure 9.1) leads to a theoretical
maximum scan angle qmax according to the formula
sin qmax ¼ l=d  1 (9.1)
in order to prevent grating lobes and array mismatch. A typical slotted waveguide
array with longitudinal slots in the broad wall has a waveguide width of roughly
0.7 l. When used as an element in a phased array this ‘element size’ leads to a
maximum scan angle of about 25 according to (9.1). However, at this angle the
grating lobe already enters visible space and there is a considerable mismatch due
to mutual coupling [1,2]. Fifteen degrees can therefore be a more realistic
maximum scan angle, cf. Figure 9.11 (see also Section 4.5.1). An example is shown
in Figure 9.2 for the centre element of a small array of rectangular aperture elements
(E-plane slots) with different spacings.
For scan angles up to 60 it is necessary to use waveguide structures with a
much reduced width for the highest frequency used. Possible alternatives include
dielectric filled rectangular waveguides [3–5], ridge-loaded waveguides [6–9] and
222 Slotted waveguide array antennas: theory, analysis and design

Waveguide elements

Phase shifters

Feed network

Figure 9.1 Phase steered array with slotted waveguides as radiating elements

–1

–2
d/l = 0.7 0.6 0.5
dB

–3

–4

–5
0 10 20 30 40 50 60 70
Degrees

Figure 9.2 Embedded element patterns for the centre element of a 15-element
array with different element spacings d/l

edge slotted waveguides [10–12]. Figure 9.3 shows a typical size reduction using
ridge-loaded waveguides. The width of the ridge guide is in this example 35% less
than the width of the corresponding rectangular guide for the same guide wavelength.
Shnitkin [9] describes an interesting ridge waveguide configuration (see
Figure 9.4), where the waveguide heights alternate on the two sides of the ridge.
The longitudinal slots remain on the physical centreline, and no ‘butterfly lobes’
(see Section 9.9) are generated.
 Antenna systems and special requirements 223

Figure 9.3 Cross sections of a rectangular waveguide (left) and a ridge-loaded

waveguide (right), with the same cut-off wavelength

A B
SEC AA SEC BB l l
High Low

Low High

A B

Electrical
neutral plane
(zero H-field)

Figure 9.4 Asymmetric ridge waveguide with slots along the centreline.

1990 IEEE, reprinted from [9], with permission

Numerical methods have been developed for the analysis of ridge waveguide
characteristics; see, for example, Conciauro et al. [13]. Also slots in the ridge
waveguide can be analysed [14,15]. Slot data for array design can of course also
be obtained from measurements. Commercial codes using finite elements (e.g. HFSS)
are used as well.
Systems with phase steering in elevation combined with mechanical rotation in
azimuth is a common way of covering all coordinates, for example, in a 3D radar.
The antenna may be composed of several slotted ridge waveguides with individual
phase control; see Figure 11.7 in Chapter 11. Another interesting solution is to
combine phase steering with frequency scanning [16,17].
A large S-band radar with wide angle phase steering in azimuth is used in an
airborne early warning (AEW) system [18,19]. About 200 transmit/receive modules
feed two slot arrays on each side of the unit carried by the aircraft. The vertical
slotted ridge waveguides allow almost 360 of azimuth coverage (Figure 9.5).

9.2 Frequency scanned slotted waveguide arrays

A linear waveguide array with, say, longitudinal slots in the broad wall evenly
spaced along the waveguide is a building block of many array systems. When fed in
one end and with a matched termination in the other it will radiate in a direction
which is determined by the frequency, the waveguide wavelength and the spacing
224 Slotted waveguide array antennas: theory, analysis and design

(a)

Air inlet
RF network

Radome

Slotted waveguides

TR modules

Slotted
(b) waveguides

Figure 9.5 AEW radar with a phase steered slotted ridge waveguide
array antenna. (a) The dorsal unit in a near field test chamber.
(b) AEW demonstrator cross section showing TR modules in centre
with slotted waveguide array panels on each side of the dorsal unit.
Courtesy of Saab AB
 Antenna systems and special requirements 225

between the radiating slots. Thus, the beam direction can be steered by changing
the frequency. This type of electronic beam steering is fairly simple and econom-
ical; it requires no electronic phase shifters, and the antenna can be fabricated with
relative ease. Several linear arrays can of course be combined to form a planar
array. Frequency scanned slotted waveguide arrays have been developed for fre-
quencies as high as at least 180 GHz [20].
The general problem of frequency scanning has been treated by many authors
[21–24]. There are several basic types of frequency scanned linear arrays; the most
important are
1. The travelling wave array
with matched termination and no phase reversal
(no alternating slot offsets or slot inclination angles)
2. The travelling wave array
with matched termination and phase reversal
3. The resonant series-fed array
short instead of a matched termination

9.2.1 Travelling wave linear array

This case is illustrated by the single slotted waveguide array in Figure 9.6. See also
Section 6.6, where a detailed design of a travelling wave array is presented. The
slots can be longitudinal, displaced in the same direction (not alternating offsets) or
inclined (in the same direction). The waveguide is fed in one end with a matched
load in the other end.
Assume the slot spacing along the waveguide to be d. Thus, the phase delay
from one slot to the next one is
2p
Df ¼ d (9.2)
lg
where lg is the guide wavelength.

f
Slots θs

Load
d

Figure 9.6 A travelling wave linear slotted waveguide array

(leaky wave antenna)
226 Slotted waveguide array antennas: theory, analysis and design

The beam direction from the normal qs is related to the phase delay:
2p
Df ¼ d sin qs (9.3)
l
yielding the beam direction
sin qs ¼ l=lg (9.4)
Thus, the beam direction is independent of the spacing d, but it depends on the
frequency. In a typical case with lg ¼ l.5l we obtain qs ¼ 42 . The antenna is called
a leaky wave antenna or fast wave antenna. It is interesting to note that the beam
has the same angle versus the y-direction as the constituent TE10 plane waves
versus the x-direction (the waveguide mode can be seen as plane waves inside the
waveguide bouncing between the waveguide side walls).

9.2.2 Phase reversal travelling array

Here we assume the same general configuration as before, but with the slots having
alternating offsets; the same theory applies to a linear array with inclined slots with
alternating inclination angles. The waveguide is terminated in a matched load as
before. The phase delay from one slot to the next becomes
2p
Df ¼ dp (9.5)
lg
where the minus p comes from the phase reversals. The beam direction from the
normal is


l lg
sin qs ¼ 1 (9.6)
lg 2d

hence depending on both the frequency and the spacing d. Differentiating the last
expression gives the approximate scan versus frequency sensitivity:
 
Df
½Dqdegrees
(9.7)
f0 %

that is, about 1 scan for 1% frequency variation. The closer to waveguide cut-
off the larger is the beam squint. An example is shown in Figure 9.7 for 5%
frequency scan yielding about 5 of beam scanning. It is here assumed that the
beam shall be directed normal to the array at the frequency f ¼ f0.
The useful angular regions for this type of array are indicated in Figure 9.8. It
scans backwards (towards the feed input) when the slot spacing is less than lg/2
and forward when the spacing is larger than lg/2. See also Figure 6.15. Three
‘forbidden’ regions are indicated: (1) slot spacing less than the slot length is
assumed not possible, (2) slot spacing d ¼ lg/2 implies that the reflections from
all slots add in phase with a resulting high VSWR at the input and (3) too large
 Antenna systems and special requirements 227

Scan angle [deg] 2

–2

–4

–6

–8
0.94 0.96 0.98 1 1.02 1.04 1.06
f/f0

Figure 9.7 Frequency scanning of a travelling wave array antenna with phase
reversal. Standard waveguide with a dimension ¼ 22.86 mm. Centre
frequencies are: solid line 7.5 GHz, dash-dotted 8.5 GHz and dashed
line 9.5 GHz

Poor VSWR

d < lg/2 d > lg/2

f
Θs
d<L
Grating lobes
L d
x x0
y
Input Matched load

Figure 9.8 Scan limitations of a frequency scanned slotted waveguide array.

Travelling wave with alternating slot offsets for phase reversal.

1990 Microwave Exhibitions and Publishers Ltd. Reproduced
by permission of Microwave Exhibitions and Publishers [2]

spacing causes a grating lobe to radiate. To prevent grating lobe radiation the
requirement is
1:5
d=l < (9.8)
1 þ l=lg
228 Slotted waveguide array antennas: theory, analysis and design

d'

d'

d
d

Figure 9.9 A serpentine feed. Note the increased path d0 compared to the slot
spacing d.
1981 J. H. Elliott. Adapted from [24], with permission

An example of a phased array antenna system using the phase reversal slotted
travelling wave arrays as the phase steered elements was illustrated in Chapter 3
(Figure 3.5). In this case two-dimensional electronic scan was obtained: frequency
scan in elevation and phase scan in azimuth. If the phase delay between slots is
increased a larger scan range is obtained for the same frequency variation. This can
be achieved with a folded, serpentine feed (Figure 9.9). In the figure the serpentine
waveguide has radiating edge slots. It could also feed a second set of slotted
waveguides to form a planar array; an example is shown in Figure 9.10. More than
90 of scan can be obtained with practical serpentines for 10% frequency variation.
With a serpentine (sinuous) feed of length d0 the beam direction can be written


d0l 1 1
sin qs ¼  (9.9)
d lg lg0
where it is assumed that the beam direction qs ¼ 0 when lg ¼ lg0. From (9.9) it
appears that sin qs might exceed unity for large phase delays. However, phase is
ambiguous and we need to pick the right branch, that is inside [0,2p] or [p, þp].
Compare the real versus imaginary space for array patterns in Figure 9.11. The
necessary modification to (9.9) for any beam direction is to include an integer
parameter m as shown in (9.10):


l d0 1
sin qs ¼ m (9.10)
d lg 2
The scanning range with serpentine lengths of 0.5, 3 and 5 guide wavelengths
is illustrated in Figure 9.12.
 Antenna systems and special requirements 229

Figure 9.10 The ITT Gilfillan AN/SPS-48 3D radar antenna. Note serpentine
on the left. Photo courtesy of Troy Prince [25]

l/d l/d
0
dB

Visible space

–10

–20

–30
–3 –2 –1 0 1 2 3
sin q
sin qs

Figure 9.11 The real beam must be inside ‘visible space’

230 Slotted waveguide array antennas: theory, analysis and design

45

30

15
Scan angle [deg]

–15

–30

–45
0.94 0.96 0.98 1 1.02 1.04 1.06
f/f0

Figure 9.12 Frequency scan with different serpentine lengths d0 . Waveguide a

dimension 22.86 mm, centre frequency f0 ¼ 9.0 GHz. Solid line
d0 ¼ 5lg, dash-dotted line d0 ¼ 3lg and dashed line d0 ¼ 0.5lg

θs

Short
d circuit

∆θ

(a) (b) (c)

Figure 9.13 The resonant series feed array

9.2.3 Resonant series feed array

In this array the matched load is replaced by a short circuit. There will be a forward
travelling wave (Figure 9.13(a)) and a reflected backward travelling wave (b) in the
waveguide. Each of the waves will excite one beam and the two beams (c) will
coincide only if the slot spacing is equal to half the guide wavelength. In the example
shown d > lg/2.
 Antenna systems and special requirements 231

The array will produce one single beam only in a narrow frequency range
where the two beams overlap satisfactorily. Assuming a maximum beam separation
of half the three dB beamwidth we get a useful bandwidth BW of about
50
BW ¼ % (9.11)
N
where N is the number of slots. The number 50 is approximately valid for no taper
and will be slightly larger in an amplitude tapered array. We find that the band-
width in per cent approximately equals the beam width in degrees.
We remarked before in relation to Figure 9.7 that scanning in the broadside
direction would cause prohibitively large reflections, since all slot reflections add
up in this case. Especially in long waveguides this could be problematic. However,
certain slot configurations have been found to reduce the reflections considerably.
In one case an inductive post adjacent to the slot is used to tune out the slot
reflection [26]. Another possibility is to use pairs of transverse slots (closely
spaced) that both radiate but the respective reflections are out of phase [27,28].
A variation still is the offset crossed slot radiating circular polarisation which
scatters only in the forward direction [29,30].

9.2.4 Centre-fed travelling array

Returning to the travelling wave array antenna we observed that it is relatively simple
to build and therefore attractive. The beam squint, however, is sometimes not
acceptable. One such case is in frequency agile radar systems where fast frequency
changes would make the beam jump leaving blank sectors in the search pattern.
A more stable beam can be obtained, however, by feeding in the centre of a travelling
wave array with matched loads at both ends [2,31]. This gives two partial beams, one
from each half of the antenna. The slot separations are different in the two antenna
halves, but such that the partial main beams overlap. This means that one beam is
forward fired, the other backward fired. The scheme is illustrated in Figure 9.14.

f f

Figure 9.14 Principle for the squintless antenna.
DSTO Australia (1990),
reprinted from Radarcon 90 [31], with permission
232 Slotted waveguide array antennas: theory, analysis and design

When the frequency is increased both beams scan towards their respective
waveguide loads with the result that the composite beam remains relatively
stable in the far field. The bandwidth realised before the beam splits depends on the
array length, but can in a practical case be around 3–4%.
This antenna has several interesting features from a theoretical viewpoint.
Since the slot spacing is different in the two halves the usual design methods are not
directly applicable. The problem has been addressed by O’Loughlin et al. [32]. See
also Figure 11.7 (Chapter 11).

9.3 Dual polarised slotted waveguide arrays

Dual polarised antennas can significantly improve system performance by
extracting extra information from the incoming signals. Typical examples include
remote sensing systems such as synthetic aperture radars and Multiple Input
Multiple Output (MIMO) systems. Circular polarisation is effective in rain clutter
rejection for radar. Polarisation diversity can improve the suppression of jamming
and interfering signals, etc.
Many special configurations using waveguide slots combined with a second
element type (dipoles, microstrip patches, stripline-fed slots) have been proposed
for dual polarised arrays. One example is a flat plate waveguide array antenna that
has a microstrip dipole array with feed lines bonded to the waveguide array surface,
thus creating a dual polarised array sharing the same aperture [33]. However, we
will in this section concentrate on solutions that employ only waveguide-fed slots
for achieving the dual polarised function.
The slotted waveguide array antenna is primarily a single polarisation device
as are the narrow slot radiators themselves. However, the slots in a rectangular
waveguide can be oriented in many ways [34]; see Figure 1.2 in Chapter 1.
A combination of two sets of arrays with different slot types and feeding the two
arrays from a switching/phasing circuit would enable a variable polarisation
antenna system. The question is of course how to make this in a practical and
efficient way.
A waveguide with square cross section can support both the TE10 mode and
the TE01 mode independently of each other. Crossed slots centred on one of the
waveguide walls would therefore radiate either longitudinal polarisation or trans-
verse polarisation depending on the waveguide mode that is excited [35]. By
control of the mode amplitudes and phases any polarisation state can be radiated.
A major drawback, however, of this scheme is the size of the square waveguide
which does not allow for the tight spacing often required in a phase steered array.
Also, spacing of slots at one waveguide wavelength intervals for in-phase excita-
tion would cause grating lobes to radiate. Some sort of waveguide loading, for
example, dielectric [5] or corrugated [36] could remedy this.
A solution which enables a relatively tight spacing was proposed by Ajioka
et al. [37,38] using a bifurcated waveguide with crossed slots radiating from the
narrow wall (Figure 9.15).
 Antenna systems and special requirements 233

Septum slot

E
Transverse
E currents

Figure 9.15 The bifurcated waveguide. With even excitation (shown) vertical
polarisation is created. Odd excitation creates horizontal
polarisation.
IEEE, adapted from [38], with permission

a1

y h

h1
z

2L

x
a

(a) (b)

Figure 9.16 (a) Breadboard sub-array for polarimetric SAR. Courtesy of

RUAG Space AB. (b) Building block with main dimensions
indicated.
URSI 1992. Adapted from [40], with permission

Odd and even excitations of the dual waveguide provide the means for con-
trolling the polarisation. The waveguide can be low height so that arraying in a
phase scanned antenna is possible. Related schemes, also based on bifurcated
waveguides, have been proposed by Sangster et al. [39]. It remains, however, to
find a solution to the grating lobe problem caused by the lg spacing of the septum
slots for in-phase excitation.
A combination of two separate waveguides with respectively longitudinal slots
and transverse slots could provide the two orthogonal polarisation components
required for polarisation agility. In order to accommodate the waveguide pair in
an array with sufficiently dense spacing several solutions have been proposed.
Figure 9.16 shows one breadboard C-band array studied for a polarimetric SAR [40].
The building block is shown to the right in Figure 9.16. Both polarisations
excite the region between the vertical baffles that are formed by the waveguides
standing on the narrow side. Inside the baffle region the fundamental propagating
234 Slotted waveguide array antennas: theory, analysis and design

dB
0

–10

–20

θ
–80° –40° 0° 40° 80°

Figure 9.17 Measured patterns for a six-element array with transverse slots
spaced 1.4 l0. Broken curve without baffle, and solid curve with
baffle.
IEEE 1991, reprinted from [41], with permission

modes are a TEM parallel plate mode excited by the longitudinal slots and a TE(1)
parallel plate mode excited by the transverse slots. The latter mode has a wavelength
l1 larger than the wavelength lg of the feeding waveguide (since a1 < a, Figure 9.16).
The slots are resonantly spaced at lg distance. Hence the grating lobes inside the
baffle region are evanescent and will be attenuated provided that a > a1 > l0/2, with
l0 the free space wavelength. In a typical case a baffle height of one-half wavelength
gives about 20 dB grating lobe attenuation; see Figure 9.17 [41,42].
The baffles work as a spatial filter, with a pass band centred at normal inci-
dence. Other spatial filter structures could be selected for the same purpose, based
on, for example, dielectric layers or wire grid layers [43].
We give one more example of a proposed dual polarisation waveguide element
using baffles (Figure 9.18). Here lg spaced transverse slots have the grating lobes
suppressed as before using baffles. Longitudinal slots at lg/2 spacing on the
waveguide centreline would normally not be excited. However, in this case the
lower array, also with centred longitudinal slots with post excitation radiates
through the upper longitudinal slots [44].
Similar to Figure 9.16 two waveguides, one for each polarisation could be used
with the lower waveguide having longitudinal slots and the other waveguide
standing on its side having edge slots. This simple principle is illustrated in
Figure 9.19 [45].
Unfortunately, the inclined edge slots will radiate some cross-polarisation due
to the inclination angle. However, this problem can be circumvented by exciting
 Antenna systems and special requirements 235

E5

E3 C
lg

7
6
5
lg
10 2

lg
2
11

2
4 E4 E1 1
8
3
E2 9
B

(a) (b)

Figure 9.18 Dual polarised slotted waveguide using baffle for grating lobe
suppression. A breadboard sub-array is shown on the right.
Courtesy of RUAG Space AB

Figure 9.19 Longitudinal and inclined edge slot radiators in a dual polarised
array.
IEEE 1996, reprinted from [45], with permission

non-inclined edge slots by a pair of wires inside the waveguide. Furthermore, to

allow a large scan range in the transverse plane (vertical) the lower waveguide can
be ridge loaded. A breadboard panel using this modified concept [46,47] is shown
in Figure 9.20. This design allows about 20 vertical scan range.
236 Slotted waveguide array antennas: theory, analysis and design

Figure 9.20 Breadboard panel of a dual polarised array. Courtesy of

RUAG Space AB

Figure 9.21 The DESA X-band SAR subpanel demonstrator. Courtesy of

RUAG Space AB

The technology has been implemented in an X-band SAR Instrument

Demonstrator (DESA) developed by Dornier/Astrium [48,49]. A sub-array panel of
30 dual polarised waveguide elements is shown in Figure 9.21.
 Antenna systems and special requirements 237

9.4 Multiple beam slotted waveguide arrays

A system using two arrays with edge slots, one behind the other, has been proposed
for a dual beam array [50] (Figure 9.22). The rear array radiates through the spaces
between the waveguides of the front array. Two orthogonal linear polarisations
are available. Each of the arrays can be phase steered in one dimension while
the combined antenna is mechanically steered. This allows two independently
steered beams from the same aperture, however with limited polarisation
flexibility.
The aforementioned example is not strictly a multiple beam solution. With
multiple beam arrays we usually mean systems that generate a number of beams in
space from one common aperture. The feed system can be a Butler matrix [51],
a Blass matrix [52] or equivalent. Typically, each beam is associated with a phy-
sical beam port in the feeding system. For the beam ports to carry independent
signals they must be isolated from each other, in fact the isolation requirement is
equivalent to a requirement on the orthogonality of the corresponding radiation
patterns. For two beams with pattern functions E 1 and E 2 the orthogonality
requirement [53] can be written as
ð 2p ð p

df E 1 ðq; fÞ  E 2 ðq; fÞsin qdq ¼ 0 (9.12)
0 0

A waveguide system for generating five independent beams is shown in

Figure 9.23(a). It is shown here as a receive system but works in principle for both
receive and transmit. A large receive only system was once meant to be used in
a height finding air traffic control radar with more than 100 elevation beams
[54,55]. The concept, usually termed Blass matrix [52], has been studied for future
spaceborne SAR applications with two and three beams. A two-beam array, one

Rear array

Front array

Figure 9.22 Crossed edge slot arrays according to Alexander and Sichelstiel.
Reprinted from [50]
238 Slotted waveguide array antennas: theory, analysis and design

Directional couplers
Waveguides

Radiating
elements

(a) To receivers

–10

dB –20

–30

–40
–45 –30 –15 0 15 30 45
(b) (c) Degrees

Figure 9.23 The Blass concept, example with five beam ports.
McGraw-Hill
Education 1962, reprinted from Skolnik [54] with permission.
(b) Dual beam slotted waveguide array. Courtesy of RUAG Space
AB. (c) Measured dual beam patterns.
IEEE 1960, reprinted
from [56], with permission

narrow beam and one shaped broad beam were successfully tested [56,57]; see
Figure 9.23(b) and (c).
In Section 9.2 on resonant travelling wave-type slotted waveguide arrays it was
mentioned that the reflection from the shorted termination causes a second beam to
radiate. Thus, by feeding the two ends of a non-resonant array we can construct a
dual beam antenna with the two beams symmetrically directed relative to broad-
side. However, a high-efficiency symmetrical taper will be difficult to arrange [23].
Classical navigational systems for aircraft normally include an X-band
Doppler radar with four beams pointing towards ground in the directions forward
right and left, and rear right and left. By measuring the Doppler shift of the four
returns the speed and direction of the aircraft can be determined with high
precision. Multi-beam arrays using non-resonant slotted waveguide arrays are
suitable for this, considering their inherent stability and the high precision
 Antenna systems and special requirements 239

Port 4 Port 1

Feeding array

Port 3 Radiating array Port 2

Figure 9.24 Layout of four-beam Doppler navigation array.
1963 IEEE,

reprinted from [58], with permission

manufacturing possible. By feeding a planar array in the four corners as illustrated

in Figure 9.24 the four beams are realised [58].
Switching between two sets of waveguide arrays was proposed by Saltzman
and Stavis [59]. Similar designs have been presented by Miller and Forman [60],
who used square waveguides with TE10 and TE01 modes for better isolation
between the beam ports.

9.5 Conformal slotted waveguide arrays

A conformal antenna has a shape that is dictated not by electromagnetic

considerations but by, for example, aerodynamic requirements such as in aircraft
structures (the IEEE Standard Definition of Terms for Antennas, IEEE Std 145-
1993). Thus the antenna conforms to a given shape which often is not optimum for
the antenna performance. In many cases, however, a particular curved shape is
chosen in order to realise a specific requirement, for example, a spherical shape
for a very wide angular coverage. Such an array is also said to be conformal.
Straight slotted waveguides are not easily used in doubly curved array antennas.
However, they fit quite well to cylindrical geometries, as shown by the example in
Figure 9.25 [61].
The scattering from a cylindrical slotted waveguide array antenna was studied
by Fan and Jin [62] (see Figure 9.26). See also Section 9.8.
240 Slotted waveguide array antennas: theory, analysis and design

20''
Coaxial loads
Waveguide
structure

23''

Radiating
slots

Dielectric
Primary wedges
feed
Geodesic Parallel plate region
lens

Figure 9.25 A cylindrical slotted waveguide array with a lens type rotating
feed.
IEEE 1980, reprinted from [61], with permission

In linear and planar array antennas all elements point in the same direction and
we usually write the antenna pattern function as a product of an element factor and
an array factor. Thus, for a linear array we have
X
EðfÞ ¼ ELðfÞ Vn ejnkd sin f (9.13)
n

where EL(f) is the element pattern function, Vn is the element excitation amplitude,
d is the element spacing and k ¼ 2p/l.
For a corresponding circular ring array (cf. Figure 9.27(b)), an element factor
can in general not be factored out as was done in (9.13). We have instead
X
EðfÞ ¼ Vn ELðf  nDjÞejkR cosðfnDjÞ (9.14)
n

where Dj is the element angular spacing and R is the circular array radius.
 Antenna systems and special requirements 241

Figure 9.26 Curved slotted waveguide array antenna for radar cross section
studies.
IEEE 1997, reprinted from [62], with permission

n
d R
n n∆j f

(a) (b)

Figure 9.27 (a) Linear array; (b) part of a circular ring array

One consequence is that for conformal arrays in general the polarisation is not
constant as the beam is steered to different angular positions. This may result in a need to
adjust the polarisation individually for each radiating element [63]. Cf. also Section 9.3
on dual polarised waveguide arrays. However, in the special case of a circular or
cylindrical array of vertical electric dipole elements parallel to the cylinder axis there is
a common element factor with constant polarisation. For the array in Figure 9.25 we can
see the slots as magnetic current sources, all pointing vertically. This results in a con-
stant polarisation but there is still no common element factor that can be factored out.
Stripline-fed tapered slots [64] have been used in broadband cylindrical arrays
[65]. For more complex shapes the polarisation problem gets worse. Circularly
polarised elements is one possible solution. Linear polarisation from slot elements
on a cone is a difficult case [66] as illustrated by Figure 9.28.
242 Slotted waveguide array antennas: theory, analysis and design

Figure 9.28 Linear slot elements on a cone, oriented to produce vertical

polarisation in the head-on direction (arrow). Adapted from [66]

9.6 Monopulse slotted waveguide arrays

Monopulse antennas are used in tracking systems such as radar and missile seekers
where the angular coordinates of detected targets shall be measured, in addition to
measuring the distance. Early tracking systems used conical scan, typically a
reflector antenna with a rotating feed causing the received signal to be amplitude
modulated. Zero modulation resulted when the antenna was pointed correctly on
the target. With monopulse (or simultaneous lobing) a single pulse provides in
principle the necessary information by processing three simultaneously received
signals from the antenna. This minimises the effect of amplitude noise on the
measurement accuracy [54, p. 175].
Figure 9.29 illustrates the principle for a phase monopulse array antenna. The
aperture is divided into four quadrants which are combined in a comparator network
to provide the sum, the azimuth difference and the elevation difference signals.
The lower part of Figure 9.29 is the comparator consisting of four hybrids
(magic Tees or equivalent) that form the sum and difference of the quadrant sig-
nals. The output signals are labelled S, Daz, and Del.
The antenna patterns of the sum and difference signals depend on the illumi-
nation function. For the difference signal one aperture half is 180 out of phase with
respect to the other half, that is, there is a strong discontinuity across the centre
of the aperture where the amplitude is maximum. Very high sidelobes result.
One could of course design an illumination function which is optimum for the
difference pattern, but that would penalise the sum pattern with high sidelobes
(see Figure 9.30).
A typical slotted waveguide array with its feed network is shown in Figure 9.31
[67]. The four quadrant outputs shall be connected to a waveguide comparator.
Optimum distributions for the sum function are, for example, the Taylor
(sampled Taylor) and Dolph–Chebyshev distributions and for the difference func-
tion, the Bayliss distribution [68,69]. The best solution would be if the optimum
distributions could be realised simultaneously and independent of each other.
This is in theory possible since the distributions are orthogonal to each other.
However, the resulting waveguide circuitry would be very complex.
Derneryd [70] proposed a separate central part of the array which is not
included in forming the difference illumination functions (Figure 9.32). For the
sum function the central part is included together with the quadrant parts. The result
 Antenna systems and special requirements 243

Array aperture

Comparator

Σ Δ Σ Δ
Σ Σ Δel
Σ

Δaz
Δ Δ

Figure 9.29 A dual plane monopulse antenna system with the aperture
divided into four quadrants

Excitation Pattern

Σ Low sidelobes

High sidelobes
Δ

Δ Low sidelobes

High sidelobes

Figure 9.30 Illumination functions and corresponding patterns: (a) optimised

for the sum function; (b) optimised for the difference function
244 Slotted waveguide array antennas: theory, analysis and design

Inclined series-series
coupling slot

Radiating
slot
Quad. 1 Quad. 2

Radiating
waveguide

Quad. 4 Quad. 3

Figure 9.31 A typical slotted waveguide array antenna with one feeding point in
each quadrant. One quadrant is shown to the left; the complete
antenna (rear view) on the right. Courtesy of Microwave Journal,
reprinted from [67], with permission

Figure 9.32 Partitioning of a monopulse array antenna with a fifth sector added.
Reprinted from [70]

is a smoother (projected) difference illumination in the centre of the array

(cf. Figure 9.30) and a sidelobe improvement of about 5 dB.
Quite complex feed networks are sometimes used for planar slotted waveguide
arrays in order to extend the bandwidth of the antenna [71]. For a relative
 Antenna systems and special requirements 245

Figure 9.33 Complex waveguide feed network on the rear of a large planar array.

Raytheon Company. Reproduced from [72], with permission

bandwidth of 10% the number of slots in series of a resonant slotted waveguide

should not exceed about 50/10 ¼ 5 slots, or 10 slots if centre fed (see Section 9.11).
For a large array the aperture is therefore partitioned into smaller sections that are
fed from a rather complex waveguide parallel feed network, sometimes in several
layers (Figure 9.33).

9.7 Stripline slotted arrays

Stripline technology may seem a bit outside the slotted waveguide array subject
that we have discussed so far. However, there are similarities between stripline-fed
slots and waveguide-fed slots that are worth pointing out. Also, complex multilayer
feed circuits can be realised in stripline in a more compact form compared to a
waveguide solution. As an example a stripline flat plate monopulse antenna with
partly independent sum- and difference feeds will be presented.

9.7.1 Introduction
A stripline slotted array antenna has slots in a metal ground plane fed from stripline
circuitry in a layer below the slots [Figure 9.34(b)]. Just like the linear slotted
waveguide array a linear stripline slot array can be envisioned with either long-
itudinal or transverse slots [73]. The slot positions are not bound to straight feeding
waveguides, but can be chosen more freely, depending on the pattern requirements
and the layout of the feeding circuit [74]. Some of the main characteristics of
waveguide and stripline technologies are compared in Table 9.1.
Stripline circuits can be fabricated using printed circuit technology and are
therefore less expensive than waveguide circuits. On the other hand, metal
246 Slotted waveguide array antennas: theory, analysis and design

x0

x0

(a) (b)

Figure 9.34 (a) Waveguide slot, offset from the waveguide centreline.
(b) Stripline slot with the feed line offset from the slot centre

Table 9.1 Waveguide and stripline technologies compared

Parameter Waveguide Stripline

Material Metal, aluminium Plastic, PTFE
Fabrication Dip brazing Printed circuit technology
Power handling Good More limited
Design Rather thick Thin
Losses Low Moderate
Feed network Complex More flexible

Table 9.2 Selected stripline parameters

Ground plane separation b 3.2 mm

Dielectric filling er 2.33
Strip conductor thickness t 17 mm
Characteristic impedance Z0 50 W
Strip conductor width w to be determined

waveguides can withstand higher temperatures. There are also other technologies
such as waveguides made of metallised CFRP, see Chapter 11, and metal-plated
plastic waveguide antennas, described in Chapter 10.
In a stripline array the TEM mode is the dominant transmission line mode. In
fact, the structure is similar to a coaxial line, albeit in a flat configuration. Side
walls are necessary for preventing parallel plate modes to escape. The characteristic
impedance of a stripline transmission line can be calculated or obtained from
published graphs, see, for example, [75]. As an example we assume the following
parameters (Table 9.2).
A calculation gives the result: w ¼ 2.56 mm for 50 W impedance.
The slot can be symmetrically fed from the stripline in the centre of the slot.
An offset feed position reduces the impedance [76]. The impedance depends also
 Antenna systems and special requirements 247

(a) (b)

Figure 9.35 (a) An X-band flat plate antenna in stripline technology.

(b) Feed network. For clarity, both slots and stripline are here
shown in black

on the dimensions of the cavity formed under the slot, and the dielectric material
[77,78]. At resonance a sinusoidal field distribution results along the slot. As we
know from Chapter 4.2, the slot antenna has by itself rather high radiation impe-
dance, typically around 500 W and some way of transforming the impedance to
match the stripline feed is necessary. An example is shown in Figure 9.35, where
the circuit layout of a stripline flat plate antenna [79] is shown. Each slot is fed
from a narrow strip of length lg/4, which acts as an impedance transformer. The
stripline continues as an open stub about lg/4 beyond the slot. The stub length is
adjusted to tune the reactive part of the slot impedance. Alternatively, an off-centre
feed position can be chosen as in Figure 9.34(b). However, this can cause problems
with the circuit layout since a great deal of symmetry is lost.

9.7.2 Independent excitations

By subdividing each antenna quadrant into sectors and combining corresponding
sectors in several comparator circuits a great degree of independence can be rea-
lised. The principle is illustrated in Figure 9.36. As seen in the figure each group of
four sectors are combined in a comparator circuit. The outputs from the compara-
tors are summed separately for E-plane difference, H-plane difference and Sum
functions. These summation networks are optimised for their respective function.
Within each sector the amplitude distribution remains the same for all functions.
In our design example (see the antenna in Figure 9.38) four sectors per quad-
rant were chosen. The amplitude weightings were decided from a detailed opti-
misation study where gain, sidelobe level, difference slope, number of sectors, etc.,
were included. A typical measured difference pattern is shown in Figure 9.37.
The slotted monopulse stripline array (Figure 9.38) is made up of three sepa-
rate functional layers: the slot layer, the comparator layer and the summation layer.
The total thickness is 7 mm. The slot cavities are formed by plated-through holes.
The same technology is used for interconnections between layers. A corresponding
antenna using ordinary metallic waveguides would be difficult to build.
248 Slotted waveguide array antennas: theory, analysis and design

Δ Δ
Σ

Δ Δ
ΔE Σ ΔH

Δ Δ
Σ

Figure 9.36 Feed principle for independent excitations with three sectors
per quadrant. Reprinted from MEDE ’77 [80]

dB
0

In vacuum bag
–10
After bonding

–20

–30

–108° –72° –36° 0 36° 72° 108°

Figure 9.37 Measured difference pattern (H-plane). The dashed line was
measured before bonding of the antenna layers. Reprinted from
MEDE ’77 [80]
 Antenna systems and special requirements 249

Figure 9.38 Slotted monopulse stripline array antenna with three functional
layers. Reprinted from MEDE ’77 [80]

9.8 Scattering from slotted waveguide arrays

9.8.1 Introduction
Antennas are supposed to transmit and/or receive electromagnetic signals with
adequate efficiency. Incoming signals shall be brought to the receiver(s), and not be
scattered in various directions – in particular for stealth applications. It can be
shown, however, that a good receiving antenna often is a good scatterer, as
the following example shows. A waveguide aperture with length 39 mm and width
8 mm in a large conducting cylinder of diameter 600 mm has been analysed by
matching the fields in the aperture using MoM [81]. Figure 9.39 shows the calcu-
lated scattering cross section in the E plane at a frequency of 6.8 GHz (l ¼ 44 mm),
with and without matching of the aperture. Note that the contribution from the
conducting cylinder itself is not included.
This example shows that matching of the aperture, which improves the receive
function, also increases the scattering by about 4 dB. In the unmatched case
the return loss was about 7 dB (transmission loss about 0.9 dB). For this analysis
the single mode assumption was sufficient [81]. The matching was accomplished
by means of an inductive iris a few mm below the aperture.
The scattering cross section s referred to earlier is defined as
jEscat j2
s ¼ lim 4pr2 (9.15)
r!1 jEinc j2
also called radar cross section. It is defined for backscattering – the monostatic case –
where the illuminator and the receiver are in the same place. It is usually given in
square meters, or dB > 1 m2, dBsm. Einc is the electric field strength at the target and
Escat is the field strength at the receiver.
There are also other cross section definitions:
● bistatic cross section – illuminator and receiver in different positions
● total scattering cross section – the sum of power scattered in all directions
● absorption cross section – total power absorbed
● extinction cross section – sum of scattered and absorbed power
250 Slotted waveguide array antennas: theory, analysis and design

–20

–25

–30

σTETE (dBsm)
–35

–40

–45
φ
–50

–55

–60
–180 –140 –100 –60 –20 20 60 100 140 180
(a) (b) φ (degrees)

Figure 9.39 (a) Waveguide aperture in a cylinder. (b) Calculated scattering

cross sections compared. Single mode (TE10) solution. Solid line:
no matching; dashed line: matched aperture.
Wiley 2006.
Adapted from [63], with permission

3 4

Figure 9.40 Slot boundary problems numbers 3 and 4, waveguide slot and
slot in a ground plane

Numerous references to publications on electromagnetic scattering can be

found in the Proceedings of the IEEE, August 1965 (special issue). See also [82]
for a discussion about transmitting, receiving and scattering properties of antennas.
In this section we will discuss very briefly a few aspects of antenna scattering that
are related to slotted waveguide arrays.

9.8.2 Scattering by a single slot in a planar ground plane

Chapter 5 treated boundary problems for slots in rectangular waveguides. Four
typical situations were outlined in Figure 5.5. We pick out no. 3 and no. 4 as
relevant to scattering situations with plane wave incidence (Figure 9.40).
Let us for the time being assume that the wall thickness is very small and can
be neglected. The extension to finite wall thickness is straight-forward, following
the procedure in Section 5.1.4. The important field components are the longitudinal
 Antenna systems and special requirements 251

y k
q

Figure 9.41 Plane wave incident on a waveguide slot

Hz and transverse Ex as discussed in Chapter 5. We denote the external field with

superscript e and internal/lower field with superscript i, respectively. The boundary
condition in mathematical form for problem 3 is therefore

Hzi ðEx Þ  Hze ðEx Þ ¼ Hzinc (9.16) = (5.2)

A unit amplitude plane wave incident from an angle (q, j) is assumed
(Figure 9.41).
The Ex field is expanded as before (Chapter 5). Testing according to the
moment method yields a solution for the electric field in the vector [E] ¼ (E1,
E2, . . . , EN), cf. (5.8), where
  1
½E ¼ H0 Y i  ½Y e   ½h (9.17)
The excitation vector [h] is here normalised; H0 represents the amplitude of
the incident field including the projection onto the slot and possible polarisation
mismatch, thus

Hzinc ðx; zÞ ¼ H0  ejðkx xþkz zÞ (9.18)

The excitation vector is very similar to the case with excitation from the
waveguide. We obtain the elements hs

 ( )
kx w ps=L j sin kz L s even
hs ¼ w  sinc ps 2 (9.19)
2  kz 2 cos k L s odd
2L z

cf. (5.44).
The problem no. 4 in Figure 9.40, the slot in a ground plane (no waveguide),
is easily solved in the same manner. We have Hzi ¼ Hze and thus

½E ¼ ðH0 =2Þ  ½Y e 1  ½h (9.20)

252 Slotted waveguide array antennas: theory, analysis and design

14 GHz

9.375 GHz

0 –5 –10 –5 0
(dB)

Figure 9.42 Bistatic scattering from a longitudinal slot in a waveguide. The angle
of incidence is q ¼ 60

Now that we have the electric field distribution in the slot aperture we can
derive the electric vector potential in the far field

eejkr
Fz ¼ ½E  ½hðkx ; kz Þ (9.21)
2pr
where [h(kx, kz)] is the same as the excitation vector, cf. (9.19), but with (kx, kz) now
representing the outgoing direction. The far field is easily obtained since
 
Hq ; Hj ¼ jw Fq ; Fj (9.22)
The bistatic scattering cross section is defined as s ¼ limr!1 4pr2 Pscat =Pinc ,
where Pinc and Pscat are the incident and scattered power densities, respectively,
yielding

k 2  kz2

2
s¼ ½E  ½h
(9.23)
pZ0
In this expression kz represents the outgoing direction, which is also contained
in [h]. The incidence direction is implicit in the solution for [E].
Figure 9.42 shows a scattering diagram for a case with a waveguide slot
illuminated from (q, j) ¼ (60 , 0 ). The slot length is 16 mm and thus close to
resonance for the frequency 9.375 GHz. At the higher frequency an asymmetric
slot field results due to the presence of higher slot modes. This is illustrated in
Figure 9.43 showing the electric field distribution in the slot [83].

9.8.3 Scattering from an array of slots

The scattering cross section of a planar slotted waveguide array antenna has several
contributions: the array of slots, the ground plane and the feed network behind the
 Antenna systems and special requirements 253

0.8
Electric field amplitude

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
Position along aperture

Figure 9.43 Slot electric field along the slot for the same case as in
Figure 9.42. Solid line is 14 GHz, and broken line is 9.375 GHz.

IEEE 1990. Adapted from [83], with permission

ky/k

l/dy

Visible region

dy
gl inc n r 1 kx/k

l/dx
sin θi dx

inc n
r
gl

Figure 9.44 The grating lobe lattice. The incidence wave (‘inc’) excites the slots
with a phase slope causing a reflected wave (‘r’). In this example the
grating lobe (‘gl’) is close to radiating back towards the illuminator

slots. At oblique incidence from the illuminator the slots are excited with a phase
slope and radiate like a phase steered array. The grating lobe lattice, the reciprocal
of the slot element lattice [84], is shown in Figure 9.44 for an array with slots in a
rectangular grid.
254 Slotted waveguide array antennas: theory, analysis and design

A common spacing between waveguide slots is lg/2 or typically 0.7 l0

which prevents grating lobes to appear in the array antenna pattern. However,
considering the scattering diagram the spacing must be as small as l0/2 to eliminate
grating lobes for all directions of incidence. The general case is described by the
grating equation

l
sin qinc þ sin qgl ¼ m ; (9.24)
d

where m is the order of diffraction. Compare Figure 9.11.

At the blazing angle the grating lobe radiates back in the direction of inci-
dence, that is,

l
sin qinc ¼ sin qgl ¼ m (9.25)
2d

The array ground plane scatters like a metal plate of the same size. Based on a
physical optics approximation (neglecting edges) the monostatic cross section for a
2
plate with area A is s ¼ 4pA
l2
in the normal direction. The approximate angular
dependence for a square plate with side L is
 
sinðkL sin qÞ 2 2
sðqÞ ¼ sð0Þ cos q (9.26)
kL sin q

and for a circular plate with radius a

 
J1 ð2ka sin qÞ 2 2
sðqÞ ¼ sð0Þ cos q (9.27)
ka sin q

see [85,86].
The simple expressions such as (9.26) for the rectangular plate work well near
the normal direction. An analysis based on equivalent currents along the plate
edges can predict the scattering more accurately outside the principal symmetry
planes of rectangular plates, as demonstrated by Polka and Balanis [87].
A circular plate with 40 cm radius has about 3,000 m2 radar cross section at
X-band. This is quite a large cross section. The contribution from the slot array can
be of the same order of magnitude [88].
It is evident that bending of the array defocuses the returns, that is, conformal
arrays have an advantage here. A curved array of slotted waveguides was shown in
Figure 9.26. We compare in Figure 9.45 the scattering from a planar and a
cylindrical array of waveguide apertures, in both cases 40 32 apertures covering
about 120 of the 600 mm diameter (¼13.6 l0) cylinder.
The example in Figure 9.45 is for the H-plane. In the E-plane the comparison is
not so favourable for the conformal case due to the specular return for all j values.
See also [62].
 Antenna systems and special requirements 255

40 z
30 ∞ ∞

20
<σTETE> (dBsm)

10
H-plane
0
–10
–20 E-plane
–30
R = 0.3 m x
R = 100.0 m
–40
0 20 40 60 80 100 120 140 160 180 ∞ ∞
(a) q (degrees) (b)

Figure 9.45 (a) Monostatic radar cross section for a waveguide array with 40
32 slots on a cylinder (solid line) compared with a corresponding
array on a planar surface (dashed line). H-plane, (b) shows the
geometry.
IEEE 2003, reprinted from [81], with permission

9.8.4 The feeding network

The elements in a slotted waveguide array antenna are fed from a network con-
sisting of coupling slots, waveguide splitters and hybrid couplers, etc. This network
is usually not well matched; in fact most of the power dividers are reactive and
mismatched on the output ports. This will cause reflections towards the aperture
and re-radiation as well as a perturbation of the phase and amplitude distribution.
Consider, for example, a non-isolated power divider as symbolically illustrated in
Figure 9.46(a). It is matched on the input port when the admittances are chosen
Y0 ¼ Y2 þ Y3. (An ideal transformer on the output ports is assumed in order to bring
back all signals to the same impedance level.) A symbolic form of the divider is
shown in Figure 9.46(b).
In terms of the power division ratio q ¼ P2/P3 we can write the 3 3 scattering
matrix:
0 rffiffiffiffiffiffiffiffiffiffiffi
q 1 1
0 pffiffiffiffiffiffiffiffiffiffiffi
B 1þq 1þqC
B C
B rffiffiffiffiffiffiffiffiffiffiffi pffiffiffi C
B q 1 q C
Sq ¼ B
B 1þq 1þ q
C (9.28)
B 1þq C C
B pffiffiffi C
@ 1 q q A
pffiffiffiffiffiffiffiffiffiffiffi 
1þq 1þq 1þq
We find that an equal power splitter (q ¼ 1) with no isolation has 6 dB return
loss on the output ports. With a more uneven power split the mismatch increases on
one of the ports, and decreases on the other. Note that transmission line lengths are
not included in the above expression.
256 Slotted waveguide array antennas: theory, analysis and design

2
Y0
2
Y2 P2
1 Y0 1 q
Y3 3
P3
Y0
3
(a) (b)

Figure 9.46 (a) A non-isolated power divider with transmission lines.

(b) Symbolic version

The impact of the feeding system on the scattering from slotted waveguide
array antennas is not much discussed in the literature with few exceptions; see, for
example, Tittensor and Newton [88]. There are also various ways of modelling feed
networks [89].

9.8.5 Numerical examples

There are not many publications dealing with the RCS performance of slotted
waveguide arrays. Fan and Jin [90] and Zhang et al. [91] have calculated the
scattering from fairly large (up to >1,000 slots) slotted waveguide arrays where the
slotted waveguides were either shorted or matched at the waveguide ends and all
slots had in most cases identical dimensions. The theoretical results have also been
compared with measured results. The results verify the expected general behaviour
including the grating lobe patterns.
In [92] a more realistic antenna, albeit small, was analysed. The rectangular
array had six waveguides, each with 12 longitudinal slots, a feeding waveguide
with coupling slots and a matched termination at the feed input. The analysis
was based on solving the integral equations for the problem using MoM, in prin-
ciple as described in Chapters 6 and 7. A typical computed result is shown in
Figure 9.47.
In Figure 9.47 we see the specular reflection from the slot array at 45
including a grating lobe at 32.5 (solid line). The specular return from the ground
plane at 45 is also shown. It can easily be found with the aid of the grating
equation (9.24) that the slot spacing in the H-plane is 0.8 free space wavelengths.
Very little power goes into the antenna matched load at this angle of incidence.

9.8.6 Spatial filters

Covering a planar antenna with a curved frequency selective radome will reduce
the monostatic scattering peak at normal incidence, at least at out-of-band fre-
quencies. Such frequency selective surfaces (FSS) can be designed using wire
grids, dielectric layers, slot or dipole arrays, etc. Pelton and Munk [93] describe a
conical metallic radome with a slotted surface with good transmission properties
within a narrow frequency band. A spherical FSS surface to be mounted on a planar
 Antenna systems and special requirements 257

–10 Total

Field amplitude in dB
PO
–20

–30

–40

–50
–90 –60 –30 0 30 60 90
Theta in degrees

Figure 9.47 Scattered field in the H-plane for a rectangular slotted waveguide
array antenna. Incidence is from qinc ¼ 45 .
IEEE 2007,
reprinted from [92], with permission

Figure 9.48 Frequency selective slotted radome. Courtesy of Saab AB

slotted waveguide array antenna has been described by Stanek and Johansson [94]
(see Figure 9.48).
The FSS radome in Figure 9.48 is built up of two metal skins and a dielectric
support. The 2,000 þ slots in each skin are laid out in a quasi-periodic pattern due
258 Slotted waveguide array antennas: theory, analysis and design

to the curvature, and designed for equal phase delay within 5 and minimum loss
within the operating frequency band. Out-of-band (and for the orthogonal polar-
isation) the radome scatters as a solid spherical surface. It can be viewed as a large
conformal slotted array (but with no waveguides).

9.9 Second-order lobes

9.9.1 Introduction
The typical linear slotted waveguide array has longitudinal slots separated at about
one-half waveguide wavelength (0.5 lg along the waveguide). The slot excitation is
governed by the amount of slot offset from the waveguide centreline, with equal
phase radiation obtained by alternating the offset directions (Figure 9.49).
Since the radiator spacing of 0.5 lg corresponds to about 0.7 l0 in free space
the first grating lobe appears at an angle corresponding to sin qg ¼ 1/0.7
1.4, that
is, outside visible space (cf. Figure 9.11). However, considering the alternating slot
arrangement, the array period is actually twice that of the single slot, that is, the
array element is strictly speaking the slot pair. Thus, a grating lobe will appear at
sin q ¼ 0.7 or circa 45 from broadside. However, in the vertical plane containing
the array symmetry line such a grating lobe will not appear. But outside this plane
the phase differences due to the slot offsets will result in increased sidelobes,
or ‘butterfly lobes’ [95,96] (see Figure 9.50). This phenomenon, also named
‘Gruenberg lobes’, has been studied by several authors [97–101].

9.9.2 Analysis
For the simple case shown in Figure 9.49 the radiation patterns with and without
slot offsets are easily calculated. Assuming the slots to be radiating isotropically

dy

(a)

(b)

Figure 9.49 (a) A linear slot array, with the slot offset from the centreline
indicated (dy). (b) A corresponding real antenna (with tapered
excitation)
 Antenna systems and special requirements 259

z 21

y
19

18 16

12 11 12 11
x

17
20

Figure 9.50 Butterfly lobes (numbers 18, 19, 20 and 21 in the figure) due to slot
offsets showing up outside the cardinal planes. Reprinted from [96]

with identical polarisation we only need to examine the array factor. Two examples
are shown in Figure 9.51.
As predicted, the secondary lobe appears at the grating lobe location kx =k ¼ 2dx l

where dx is the slot spacing along the waveguide. The magnitude of the disturbance
is proportional to the slot offset from the centreline; it follows the cone kx ¼ constant
and increases linearly with the transverse coordinate ky.
It is possible to reduce the secondary lobe amplitude somewhat by breaking up
the symmetry of the array; see Section 7.6. A possible layout from Ahlbom et al.
[18] is shown in Figure 9.52. This type of layout applied to our eight-slot example
(Figure 9.49) given earlier gives the result shown in Figure 9.53. See also [97,101].
A phase steered array with slotted waveguides as radiating elements (as in
Figure 9.1) will also suffer from secondary lobes due to the slot offsets. As the main
beam is steered the secondary lobe will be steered as well. It may then be moved
into a region where the sidelobes otherwise are quite low (see Figure 9.54).
In Figure 9.55 the radiation over the full kx/ky-plane is shown for the same
array, now with a –26 dB Dolph–Chebyshev taper in the E-plane and a slot offset of
0.1 wavelengths. The waveguide width is 0.77 l and the main beam is steered
11.5 in the E-plane (about the maximum possible for normal waveguides; cf.
Figure 9.2).

9.9.3 Reducing the secondary lobes

The larger the offset the larger is the butterfly lobe. The slot offset can be reduced
by designing each slotted waveguide for a higher impedance (less offset) and use an
impedance transformer in the feed system. Also long waveguides with many slots
have small offsets and thus secondary beams are not so problematic.
With slots on the centreline of the waveguide no secondary lobes are gener-
ated. For exciting the slots an internal structure in the waveguide can then be used.
One possibility is ‘meandering’ the waveguide [96,102] while keeping the slots on
260 Slotted waveguide array antennas: theory, analysis and design

–5

–10

–15
dB

–20

–25
1
–30
0
0.5
0.5 kkyy/k
/k
(a) kx/k
1 0

–5

–10

–15
dB

–20

–25
1
–30
0
0.5
0.5 ky/k
(b) kx/k
1 0

Figure 9.51 Radiation patterns in one quadrant with (a) no slot offset,
(b) slot offset dy ¼ 0.1 wavelengths. Eight slots, spacing
along waveguide dx ¼ 0.7 wavelengths

a straight line (Figure 9.56). Another possibility is to use post or iris excited slots
(Figures 9.57 and 9.58).
The iris can be inductive or inductive-capacitive (compound iris) as in
Figure 9.58 (see Section 8.4). The iris can be tuned together with the slot to improve
the overall bandwidth of the radiator [103,104]. The ridge waveguide with alternating
depths discussed in connection with Figure 9.4 represents still another solution.
 Antenna systems and special requirements 261

Figure 9.52 Slot panel. The upper half is a mirror image of the lower half.
Courtesy of Saab AB

–5

–10
dB

–15

–20

–25
1
–30
0 0.8
0.2 0.6
0.4 0.4
0.6
kx/k 0.8 0.2 ky/k
1 0

Figure 9.53 Computed pattern for a single waveguide with mirrored slot
layout. Cf. Figure 9.51(b)

All these examples refer to shunt slots in the broad wall of the waveguide. The
secondary beam problem occurs also in shunt slots in the narrow wall of
the waveguide. In those, instead of controlling the excitation with slot offsets,
the excitation is controlled by rotating the narrow wall slots. Second-order beams
262 Slotted waveguide array antennas: theory, analysis and design

0.8

0.6
Abs

0.4

0.2

0 1
0
0.2 0.5
0.4
0.6 ky/k
(a) kx/k 0.8
1 0

0.8

0.6
Abs

0.4

0.2
1
0
0
0.2 0.5
0.4
0.6 ky/k
(b) kx/k 0.8
1 0

Figure 9.54 Array factor for a planar array of eight slotted waveguides. The
beam is steered 17.5 in the E-plane. (a) No slot offset. (b) With
a slot offset ¼ 0.3 wavelengths. Note the vertical axis scale is
here in linear voltage (magnitude)

are therefore cross-polarised. Non-rotated slots can be excited by probes, wires,

posts or irises inside the waveguide, similar to what we have described for the
longitudinal slot array [47] (cf. Figures 9.20 and 9.59). More about the many dif-
ferent slot types and their characteristics is discussed in Section 8.1.
In an array using slots that are offset or rotated, the secondary lobes can also be
eliminated by using external chokes or baffles. These will filter out the higher
modes that are the source of the secondary radiation (cf. Section 9.3 on Dual
 Antenna systems and special requirements 263

0 G

–5
–10
dB

–15
–20
–25
–30 1
–1 0.5
–0.5 0
0
(a) kx/k 0.5 –0.5 ky/k
1 –1

M
B

0
B
–5
–10
–15
dB

–20
–25
–30 1
–1 0.5
–0.5 0
0
(b) kx/k 0.5
–0.5 ky/k
1 –1

Figure 9.55 (a) Pattern with no slot offset; (b) pattern with slot offset.
M denotes the main beam, steered 11.5 in the y/z plane. G is
the corresponding (part of the) grating lobe. B indicates the
four butterfly lobes

Figure 9.56 Meandering the waveguide keeps the slots on a straight line.
Reprinted from Kaminow and Stegen [102]
264 Slotted waveguide array antennas: theory, analysis and design
lg
2

l
2

Figure 9.57 Post-excited longitudinal slots. Reprinted from [34, p. 301]

Figure 9.58 Iris-excited slot in a rectangular waveguide

polarised waveguide arrays). Baffles for a longitudinal slot array were described in
a patent by Gruenberg [106,107] (Figure 9.60).
For planar arrays with rotated edge wall slots the external space between
adjacent waveguides can act as a choke which reduces the currents associated with
cross-polarisation radiation. An optimum choke depth is about a quarter of a
wavelength [10].
 Antenna systems and special requirements 265

Figure 9.59 Non-tilted edge slots with wire excitation.
IEEE 1990,
reprinted from [105], with permission

Figure 9.60 External baffles for suppression of secondary lobes. Reprinted

from [102]
266 Slotted waveguide array antennas: theory, analysis and design

9.10 Omnidirectional slotted waveguide array antennas

In previous sections we have considered antenna arrays made up of a number of slots
in a linear, planar or curved configuration with the purpose of forming a narrow
directive beam and sometimes to steer the beam(s) over several angles. There are,
however, also applications that require a fixed omnidirectional beam, for example
radio and television broadcasting systems, local area networks (LAN) and radar
beacon systems [108]. An omnidirectional antenna is sometimes also used together
with a directive antenna for sidelobe cancelling functions [54]. With omnidirectivity
we here mean coverage of 360 in one plane, typically the horizontal or azimuth
plane. True omnidirectivity over 4p sr (strictly isotropic) is not realisable for a single
antenna even though it is used as a reference for 0 dB antenna gain [109–111].
In this section we will deal with the problem of creating omnidirectivity using
slotted waveguide arrays. First we will treat slots in rectangular and circular
waveguides and then discuss the general problem of realising omnidirective
patterns from slots in conducting cylinders. This has bearing upon slotted circular
waveguides as well as cylindrical and circular array antennas in general. The
interested reader is referred to the literature on conformal antennas [63]. The
external conductance for resonant slots in conducting cylinders of varying curva-
ture has been analysed by Wait [112,113]. With increasing curvature the
conductance increases for circumferential slots and decreases for axial slots. The
case with a flat infinite ground plane was discussed in Section 4.2.

9.10.1 Slotted rectangular waveguide array

A single longitudinal slot in a rectangular waveguide has a fairly omnidirectional pat-
tern in the E-plane. Several slots along the length of the waveguide will increase gain
and sharpen the H-plane pattern. The region on the rear side of the waveguide will be
poorly covered, however. Adding slots on the rear side of the waveguide just like the
front face can solve this problem [114]. Further improvements are obtained if a reduced
height waveguide is chosen [115,116] and in particular if all slots are aligned along the
centreline of the waveguide. This last alternative will require probe excitation of
the slots or a similar technology (cf. Figures 9.57 and 9.58 in Section 9.9). A satellite
communication antenna using this approach is shown in Figure 9.61.
Many more solutions using special slotted waveguides have been proposed. One
example is the longitudinally slotted rectangular waveguide bent to a circular form
(‘Ring Waveguide’), thus providing omnidirectional coverage in the H-plane [117].

9.10.2 Slotted circular waveguide array

An omnidirectional antenna based on a slotted circular waveguide should perform
even better than the slotted rectangular waveguide due to the rotational symmetry
of the former. To preserve symmetry the chosen waveguide mode should be rota-
tionally symmetric, for example, the TM01 mode (Figure 9.62(a)). The fundamental
coaxial mode can be used as well [118]. Unfortunately, the (non-symmetric) TE11
mode is also supported by the cylindrical waveguide since it has a lower cut-off
 Antenna systems and special requirements 267

Figure 9.61 The omnidirectional rectangular waveguide antenna (breadboard)

for the ISEE-B satellite. Courtesy of RUAG Space AB

(a) (b)

Figure 9.62 (a) The lowest order rotationally symmetric mode in a circular
waveguide is TM01. (b) The lowest order propagating mode in
a circular waveguide is TE11. Solid lines indicate electric field;
broken lines are magnetic field lines [120]
268 Slotted waveguide array antennas: theory, analysis and design

frequency than the TM01 [119,120]. A proper symmetric feeding system is there-
fore required to prevent excitation of this mode.
With TM01 (or a coaxial waveguide) the currents flow along the length of the
guide. For exciting axial slots probes can be used as shown in the example in
Figure 9.63. Here additional screws/probes are also seen in between each set of
radiating slots, claimed to improve impedance matching. One can also use inclined
or crossed slots as was shown by Grabherr and Huder [121] and others.
A beacon antenna [122] for navigational purposes is shown in Figure 9.64. It
has an upper circular slotted array for horizontal polarisation (X-band). The lower
circular array for S-band has just one row of inclined slots covering both horizontal
and vertical polarisations. The biconical horn arrangement provides beam shar-
pening in elevation; the vertical coverage for both antennas is about 20 .

(a) (b)

Figure 9.63 (a) Slotted cylindrical waveguide array; (b) cross section
showing probes for exciting slots [34]

Figure 9.64 A radar beacon antenna system using circular waveguide arrays and
an electronics package and radome for X- and S-band operation.
Courtesy of Telefonaktiebolaget L M Ericsson, reprinted from [122]
 Antenna systems and special requirements 269

9.10.3 The pattern ripple problem

An interesting question is: how many slots are needed around a given cylinder size,
assuming a specified maximum pattern ripple in azimuth? It turns out that ‘the
more the better’ is not always true as we shall see. The problem has been studied by
many authors, mainly for circular dipole arrays [123]. The use of directive elements
instead of omnidirectional dipole elements complicates the analysis somewhat but
improves the bandwidth and pattern stability [124,125].
Let us assume that we have a slot array in a cylindrical body as shown in
Figure 9.65(a).
There are N slots around the cylinder with radius R. Each slot is horizontally
polarised with a TE10-type electric field. The spacing between consecutive slots is
d ¼ 2pR/N, that is, Nd/l ¼ kR where k is the propagation constant, k ¼ 2p/l.
For the circular/cylindrical geometry the excitation function V(j) is rotation-
ally symmetric, that is, it repeats every 360 in azimuth (j-direction). It can
therefore be represented by a Fourier series:
X
þ1
V ðjÞ ¼ Cm ejmj (9.29)
1

The coefficients Cm are phase mode amplitudes; the mth mode has m times 2p
phase variation along the circumference. The radiated far field can similarly be
written as a Fourier series:
X
1
EðfÞ ¼ Am ejmf (9.30)
1

R y
R

x φ

2L

(a) (b)

Figure 9.65 (a) An array of slots in a conducting cylinder; (b) one slot
with coordinate definitions
270 Slotted waveguide array antennas: theory, analysis and design

The radiated pattern from an N-element circular array can also be written as a
summation of the element contributions:
X
EðfÞ ¼ Vn ELðf  nDjÞejkR cosðfnDjÞ (9.31) = (9.14)
n

where EL(f) is the element pattern and Dj ¼ 2p/N is the element angular spacing
around the circle. The element pattern is the same for all elements although the
elements point in different directions. Since it also is periodic, period 2p, the ele-
ment pattern can be written in Fourier series form as
X
1
ELðaÞ ¼ Dp ejpa (9.32)
p¼1

Inserting this into (9.31) yields

ð X1
1 p
EðfÞ ¼ V ðjÞ Dp ejpðfjÞ ejkR cosðfjÞ dj (9.33)
2p p p¼1

Inserting also the excitation V(j) according to (9.29) and taking out just the
contribution from phase mode number m yields the expression
X1 ð
1 p jðmpÞðjfÞ kR cosðjfÞ
Em ðfÞ ¼ Cm ejmf Dp e e dj (9.34)
p¼1
2p p

or
" #
X
1
Em ðfÞ ¼ Cm Dp j mp
Jmp ðkRÞ ejmf ¼ Am ejmf (9.35)
p¼1

where Jm is the Bessel function of the first kind and order m.

This gives us an important relation between the slot excitation, that is, the
excitation modes (Cm, near field) and the radiation modes (Am, far field). Ideally,
for an omnidirectional pattern, we only want one radiation mode, preferably A0.
With discrete elements, however, we most likely will also have many excitation
modes Cm, and our hope is that most of them, including their harmonics, will be
attenuated by the factor in brackets in (9.35).
The coefficients Dp represent a mode spectrum for the radiating element.
Different elements: isotropic, electric dipole, rectangular slot, etc., have different
spectra. For the slot in Figure 9.62, assuming a type TE10 excitation, the far-field
azimuth pattern function can be written [113,126]:

2LV0 ejkr X
1
jn enf
Ef ¼ 0 (9.36)
p3 Rr 1 Hnð2Þ ðkRÞ

where V0 is the peak of the sinusoidal voltage across the slot with length 2L, and
0
Hnð2Þ is the derivative of the Hankel function of the second kind and order n.
 Antenna systems and special requirements 271

The element azimuth pattern according to (9.36) is shown in Figure 9.66(a)

compared with the approximation (1 þ cos f). The corresponding mode spectra are
shown in Figure 9.66(b) and (c).
As we see, the radiation patterns (and the corresponding mode spectra) are
quite similar, from which we can conclude that the (1 þ cos f) function is a
satisfactory pattern approximation for the axial slot case. Using this approximation
it is a simple matter to calculate the full circular array pattern for different numbers
of slots and cylinder sizes. The resulting ripple amplitude for several cases is shown
in Figure 9.67. A contour plot is presented in Figure 9.68.
The results show that increasing the number of slots for a given cylinder size
in general makes the ripple smaller. There are, however regions where, for
example, six slots is better than seven slots. Similar results have been published
by Chu [127] and Croswell and Knop [128]. With omnidirectional elements (i.e.
vertical dipoles) it has been shown that quite often an odd number of elements is
better than an even number. This is not so often the case with directional
elements.

90 40 dB
120 60
30

150 20 30
10

180 0 deg

210 330

240 300
(a) 270

1 1

0.75 0.75
Magnitude

Magnitude

0.5 0.5

0.25 0.25

0 0
–2 –1 0 1 2 –2 –1 0 1 2
(b) Spectral order m/2 (c) Spectral order m/2

Figure 9.66 (a) Axial slot pattern (solid line) and function (1 þ cos f) (dashed
line). Mode spectra for slot (b) and (1 þ cos f) function (c).

Wiley 2006, reprinted from [63], with permission
272 Slotted waveguide array antennas: theory, analysis and design

10

8
4

5
6
Ripple dB

7
4 6
8
9

0
2 3 4 5 6 7 8 9
kR = N*d/λ

Figure 9.67 Maximum ripple versus cylinder size kR for several numbers N of
slots with assumed (1 þ cos j) element patterns

15

10
No of elements

1 dB

3 dB
10 dB

0
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
d/λ

Figure 9.68 Ripple levels indicated by contour lines: solid curve 1 dB, dashed
3 dB and dotted 10 dB. (1 þ cos j) element patterns assumed

9.11 On the bandwidth of slotted waveguide array antennas

9.11.1 Introduction
Slotted waveguide array antennas are not intrinsically broadband since in the most
common designs the radiating slots are fed by a resonant waveguide structure.
Besides, the slot itself is a resonant l/2 long radiator. These limitations may not
 Antenna systems and special requirements 273

always be a problem for applications where bandwidths of the order of 10% or less
are required. In any case, the bandwidth ‘problem’ deserves our attention and many
of the chapters in this book have already addressed this issue.
In this last section of the chapter we discuss bandwidth limiting factors from an
antenna system point of view, without going into the details of specific designs.
Design rules are addressed in a broad sense, so that the reader can get a feeling for
what is important and what phenomena are involved. Cross-references to relevant
parts of the book and references to the literature are given. The index at the end of
the book also provides aid to find more information.

9.11.2 The slot itself

A slot antenna is the dual of the dipole antenna, and they are both about half
wavelength long when working in their resonant mode; cf. Chapter 4. The band-
width of the slot (dipole) alone is typically about 10%. In special cases up to 20%
or 30% has been reported. Some kind of feeding structure or cavity is usually
required which may limit the bandwidth.
The bandwidth is defined from the maximum and minimum frequencies within
which acceptable performance is realised, often in terms of impedance matching:
for example, VSWR less than 2 or return loss better than 10 dB. Other limitations
can be pattern performance (gain, sidelobe level, beam direction, etc.) or polar-
isation performance (cross-polarisation level), etc.
Slots fed by a rectangular waveguide, respectively by a stripline waveguide
were discussed and compared in Section 9.7 (see Figure 9.35). For both types
the cavity formed by the feeding system has a significant impact on the slot
impedance and bandwidth [77,128]. In the stripline case the feeding strip ends at
about a quarter of a wavelength from the slot, transforming the strip open line to an
effective short at the slot location. Hence, this stub imposes a frequency limitation
just as the slot itself. Some increase in bandwidth is possible by adding one or more
(shorter) slots beyond the first slot [129]. Each of the shorter slots sees a shorter
stub, as they should (see Figure 9.69). Altogether a bandwidth of, for example,
9–12.5 GHz in the X-band, that is, about 30%, has been noted for a VSWR less
than 2 [78]. Multiple slots have also been used in waveguides [130]. Another
broadband shape is offered by the dumbbell slot.
Many different shapes of slots have been investigated. The dumbbell slot is
both short and wide (Figure 9.69(b)). The circular ends make manufacturing easier
if the slot is to be machined. The general rule is that the resonant length of a wide
slot is shorter than that of a narrow slot. A very narrow slot resonates at one-half
wavelength, while a thicker slot is somewhat shorter at resonance. There are,
however, exceptions to this rule when the slot cavity or feeding waveguide is taken
into account.

9.11.3 The waveguide-fed longitudinal slot

We expect a wide slot to have a larger bandwidth than a narrow slot, also when
excited from a waveguide. This is exemplified in Figure 9.70 for longitudinal slots
274 Slotted waveguide array antennas: theory, analysis and design

(a) (b)

Figure 9.69 (a) Stripline feed (hatched) with multiple slots for increased
bandwidth; (b) dumbbell slot

0.9
0.8
0.7
Conductance G/Y0

0.6 Slot widths, mm

0.5 1.6
3.2
0.4
6.4
0.3 12.7
0.2
0.1
0
2.7 2.8 2.9 3.0 3.1 3.2 3.3
Frequency × 109 cps

Figure 9.70 Conductance curves for longitudinal slots in a rectangular

waveguide with different widths. From Dodds and Watson,
National Research Council of Canada. Adapted from [34]

in a rectangular waveguide, showing the conductance of slots with widths ranging

from 1=16 in to ½ in (1.6–12.7 mm). As expected, the conductance curve is flatter the
wider the slot. The slot length was kept constant, about 50 mm, and the offset was
19.8 mm. The susceptance follows a similar behaviour [34].
For longitudinal slots the resonant length dependence on the width also
depends on the slot offset. General rules are hard to establish; an example is shown
in Figure 9.71.
 Antenna systems and special requirements 275

1%
4
Wavelengths*103

–4
–1%

–8
0 2 4 6 8 10
Offset (mm)

Figure 9.71 Calculated change in resonant length when the slot width is changed
from 3 mm to 1.6 mm, for different slot offsets. Standard X-band
waveguide with wall thickness 1.27 mm

φ1 φ2 φN

y1 y2 y3 yN

Figure 9.72 Equivalent network for an end-fed linear shunt slot array antenna

9.11.4 The linear resonant array antenna

We assume that the slotted waveguide is fed from one end. An equivalent network
for this case is shown in Figure 9.72. The nominal slot spacing is one-half guide
wavelength corresponding to the electrical length ji and the last slot is a quarter
of a guide wavelength from the short. At the centre frequency all slot suscep-
tances yi are real and sum up to the waveguide conductance G0 – we have a
matched case.
A similar network applies to a series slot array antenna where the waveguide is
loaded with series impedance elements. In both cases the analysis of the network
can be simplified by a suitable matrix representation [131]; here we have chosen
wave matrices [132,133]. A wave matrix relates the wave amplitudes on the input
and output sides of a two-port network as shown in Figure 9.73.
276 Slotted waveguide array antennas: theory, analysis and design

c1+ c2+
W
c1– c2–

Figure 9.73 The input and output ports of a wave matrix

e jφ 0
φ W=
0 e–jφ

1 2+y y
y W=
2 –y 2 – y

z 1 2 + z –z
W=
2 z 2–z

Figure 9.74 Wave matrices for some simple two ports

We can write
! !
cþ cþ
1
¼W  2
(9.37)
c
1 c
2

or in short form

  
c1 ¼ W c2 (9.38)
The total wave matrix for the complete ladder network such as shown in
Figure 9.72 is found by multiplying the matrices of the individual blocks (9.39).
Y
N
 tot ¼
W i
W (9.39)
i¼1

From the total wave matrix the total transmission and reflection are obtained
according to
(
T ¼ 1=W11
(9.40)
G ¼ W21 =W11
The wave matrices for some elementary networks are shown in Figure 9.74.
 Antenna systems and special requirements 277

A short as in Figure 9.72 can be represented by letting c2 ¼ c2þ; cf.

Figure 9.73.
The matrix multiplications are conveniently executed with modern computers.
It can be noted that in the case when all sections of the ladder network are identical
we have simply the matrix of one section raised to the power N corresponding to
the number of sections. This case can be formulated in a compact form using the
theory of iterated networks [134] involving Dolph–Chebyshev polynomials of the
second kind. However, quite often the sections are not identical, for instance when
the array is tapered, and the matrix multiplications (9.39) will have to be executed.
Using the wave matrix approach we have calculated the input return loss of a
linear array as represented in Figure 9.72 for different numbers of slots and dif-
ferent frequencies off resonance. For a given maximum return loss we can then find
the array bandwidth. Note that the slot conductances were identical and assumed to
be constant, that is, not depending on the frequency. This may be a valid approach
if the variation in frequency is small. The mutual coupling was neglected.
The results were compared with calculations based on an approximate formula
according to Takeshima and Isogai [135]. See also Hamadallah [136]. Define df as
(half) the relative bandwidth in per cent and introduce the variable

2ð3 þ q2 Þ
a¼ where q ¼ pNdf =100 (9.41)
qð2 þ q2 Þ
The return loss (reflection factor) according to [135] is then obtained as
pffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ 1 þ a2
G¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi (9.42)
1 þ a2 þ 1 þ a2
The two approaches agree quite well; Figure 9.75, cf. also Figure 6.2.
It must be emphasised that these calculations are approximate. However, they
indicate the general behaviour in terms of frequency sensitivity and number of slots
fed in series. The following effects are not accounted for
● slot admittance is frequency sensitive
● the dispersion of the waveguide itself
● mutual coupling among slots
● impact on radiation patterns
A commonly used expression for the relative bandwidth in % is 50/N, where N
is the number of slots fed in series. Examination of Figure 9.75 reveals that 60/N is
a good measure for the (total) bandwidth if 10 dB return loss is postulated. If
15 dB is required 30/N gives a good match.

9.11.5 Overloading
The term ‘overloading’ refers to choosing the slot conductance values higher than
G0/N where G0 is the waveguide admittance, for instance R  G0/N, with an over-
load factor of R [34]. The input admittance of the array will then be increased and
278 Slotted waveguide array antennas: theory, analysis and design

–5

–10
N=8
Return loss dB

–15
N=6

–20 N=4

–25 N=2

–30
0 1 2 3 4 5
Frequency df %

Figure 9.75 The nominal (half) bandwidth in per cent is obtained from the return
loss versus frequency for arrays with two, four, six and eight slots.
The solid line is according to the wave matrix analysis; the broken
line according to the formula (9.42). No overloading

an impedance transformer is required on the input in order to match the array. The
transformer pcould
ffiffiffi be a waveguide section with a quarter wave length and admit-
tance Gt ¼ RG0 [131]. A similar overloading scheme can be implemented for
arrays with series slots where the slots are represented by series elements.
Results obtained by overloading by a factor of R ¼ 2 are shown in Figure 9.76
and compared with the previous case with no overloading.
Overloading improves the impedance match also for an amplitude tapered
array. The amplitude tapering is usually introduced in order to improve the pattern
characteristics and reduce the sidelobe level. However, in a study by Derneryd and
Petersson [137] regarding a typical planar array antenna it was concluded that
overloading should be avoided when maximum bandwidth from a sidelobe point of
view is desired. In the practical case a careful trade off will have to be made.

9.11.6 Mutual coupling

As mentioned earlier, mutual coupling had been neglected in the bandwidth
calculations. This is justified for linear arrays with longitudinal slots since the
coupling is in the H-plane where the coupling is low; cf. Section 4.5. For a planar
array there will be strong coupling in the E-plane, which will have to be accounted
for in the design. The coupling reduces the bandwidth of the radiating slot com-
pared to the isolated slot as demonstrated by the example shown in Figure 9.77.
 Antenna systems and special requirements 279

–5

–10 8
6
Return loss dB

–15 4

–20
8
6

–25 4

–30
0 1 2 3 4 5 6 7 8 9 10
Frequency df %

Figure 9.76 The return loss versus frequency for arrays with four, six and
eight longitudinal slots. The wave matrix analysis was used;
the solid line is without overloading, the dashed line with an
overloading factor of 2

It has been found that breaking up the periodicity of a planar slot array can
reduce the bandwidth limitation caused by the mutual coupling. Instead of having
identical radiating waveguides with alternating slot offsets, adjacent waveguides
can be designed to be mirror images of each other. In one example [139,140] the
improvement was about a doubling of the reflection bandwidth, from 1.2% to 2.5%
at 15 dB return loss (at the expense of some grating lobes in the diagonal planes).

9.11.7 The planar array and feeding arrangements

A planar array must have some sort of network for feeding the radiating wave-
guides. This network is usually a series feed with coupling slots as exemplified in
Figure 3.3 and other places. For large arrays the aperture will have to be broken
down into sub-arrays resulting in a relatively complex feed network. As already
indicated, a rough bandwidth estimate can be obtained from the formula
Bð%Þ ¼ C1=N (9.43)
where B is the bandwidth in per cent, N is the average number of slots in series and
C1 is a constant, say C1 ¼ 30. For a planar array N shall include both the coupling
slots in the feeding network and the radiating slots. Since the slot spacing is
one-half guide wavelength or about 0.7 l0 where l0 is the free space wavelength we
can estimate the approximate number of slots in series as N ¼ D/1.4l0, where D is
280 Slotted waveguide array antennas: theory, analysis and design

0.2

0.15 Single slot

Conductance

Slot in array

0.1

0.05

0
8.5 9 9.5 10
(a) Frequency GHz

0.1

0.05 Single slot

Susceptance

Slot in array

–0.05

–0.1
8.5 9 9.5 10
(b) Frequency GHz

Figure 9.77 (a) Normalised conductance and (b) normalised susceptance versus
frequency for an isolated longitudinal slot compared to a slot in an
infinite array. Standard X-band (R 100) waveguide, slot width
1.59 mm, total slot length 14.76 mm, offset 2 mm from centreline. Slot
spacing in the array 0.7 l0 along waveguides, 0.75 l0 across [138]

the array diameter, provided that there is only one feeding point for the complete
circular array. If there are more feeding points we can write the bandwidth in GHz as
pffiffiffiffiffiffi
BðGHzÞ ¼ 250 Np =D (9.44)
where Np is the number of feeding points per quadrant and D is the diameter in
millimetre (see Figure 9.78).
In this analysis we have not included the radiation pattern performance,
tapering effects, overloading or the precise shape of the array. In a detailed
 Antenna systems and special requirements 281

Bandwidth (MHz)
1,400

1,200

1,000

800

600

400

200

0
200 300 400 500 600 700 800
Diameter (mm)

Figure 9.78 The approximate bandwidth in MHz as a function of the array

diameter for different number of feeding points per quadrant:
1/4, 1 and 4. See also Figure 9.31

investigation the influence from coupling slots, mutual coupling, waveguide short
design, impedance levels and transformers, etc., will have to be included [141].
The ideal solution could be to use a corporate feed broadband power divider
to feed all radiating waveguides in parallel, or even to feed all individual slots by
such a power divider. In one solution by Zhou et al. [142] a 1:16 waveguide
power divider feeds 2 2 sub-arrays with radiating wide slot apertures. Better
than 30% bandwidth (return loss < 10 dB) was obtained at 12 GHz. Another
interesting feed network technology is based on barline transmission lines which
essentially is a coaxial line technology. It is more broadband and offers better
design flexibility than the traditional waveguide feed using rotated coupling
slots. Dual barline power dividers 1:8 were used to feed the dual polarised
slotted waveguides that make up the panels of a large C-band scatterometer
instrument [47].

9.11.8 Travelling wave arrays

Travelling waveguide arrays differ very much from resonant waveguide arrays. In
Section 9.2 on frequency scanned arrays the performance of linear slotted arrays
with a matched load instead of a short at the waveguide end was discussed. In this
case there is only one forward travelling wave in the waveguide and a good match
is obtained at the input port, except when the slot spacing approaches one-half
waveguide wavelength. Design of both travelling wave arrays and standing wave
arrays with several examples is treated in Chapter 6.
282 Slotted waveguide array antennas: theory, analysis and design

References
[1] Amitay N., Galindo V., Wu C. P. Theory and Analysis of Phased Array
Antennas. Wiley Interscience, New York, 1972.
[2] Josefsson L., Derneryd A., Karlsson E. R. ‘A phase steered slotted wave-
guide array with non-squinting beam’. Proceedings Military Microwaves.
London, July, 1990, pp. 251–256.
[3] Sen B., Ranga Rao K. S. ‘Impedance properties of narrow radiating slots in
broadface of dielectric loaded rectangular waveguide’. Proceedings of the
IEEE Antennas and Propagation Symposium, Vancouver, Canada, 1985,
pp. 191–194.
[4] Rengarajan S. R., Steinbeck M. ‘Longitudinal slots in dielectric filled rec-
tangular waveguides’. Proceedings of the IEEE Antennas and Propagation
Symposium, London, Ontario, Canada, 1991, pp. 1276–1279.
[5] Joubert J., Coetzee J. C. ‘The use of inhomogeneously loaded rectangular
waveguide to overcome grating lobes and distribution taper limitations in
transverse slot arrays’. Proceedings European Microwave Conference. 1992,
pp. 870–875.
[6] Cohn S. B. ‘Properties of ridge wave guide’. Proceedings of the IRE. 1947;
35(8):783–788.
[7] Hopfer S. ‘The design of ridged waveguides’. IRE Transactions on Micro-
wave Theory and Techniques. 1955;3(5):20–29.
[8] Hu A. Y., Lunden C. D. ‘Rectangular-ridge waveguide slot array’. IRE
Transactions on Antennas and Propagation. 1961;9(1):102–105.
[9] Shnitkin H. ‘Unique joint stars phased-array antenna’. Proceedings of the IEEE
Antennas and Propagation Symposium, Dallas, USA, 1990, pp. 678–681.
[10] Hilburn J. L., Kinney R. A., Emmett R. W., Prestwood F. H. ‘Frequency-
scanned X-band waveguide array’. IEEE Transactions on Antennas and
Propagation. 1972;20(4):506–509.
[11] Bostrom D. E., Castaneda J. A., Hodges R. E., Fujii P. ‘Large X-band
electronically scanned array’. Proceedings of the IEEE Antennas and
Propagation Symposium, Los Angeles, CA, USA, 1981, pp. 213–216.
[12] Elsallal M. W., Herting B. J., West B. J. ‘Planar edge slot waveguide antenna
array design using COTS EM tools’. Proceedings of the 2007 Antenna
Applications Symposium, Allerton Park, IL, USA, 2007.
[13] Conciauro G., Bressan M., Zuffada C. ‘Waveguide modes via an integral
equation leading to a linear matrix eigenvalue problem’. IEEE Transactions
on Microwave Theory and Techniques. 1984;32(11):1495–1504.
[14] Falk K. ‘Admittance of a longitudinal slot in a ridge waveguide’. IEE Pro-
ceedings H – Microwaves, Antennas and Propagation. 1988;135(4):263–268.
[15] Garb K., Kastner R., Meyerova R. ‘Analysis of planar arrays of shunt slots in
ridged waveguides’. Electronics Letters. 1994;30(7):533–534.
[16] Spradley J. L. ‘A volumetric electrically scanned two-dimensional microwave
antenna array’. IRE International Convention Record. 1958;6:204–212.
 Antenna systems and special requirements 283

[17] Derneryd A., Lorentzon T. ‘Design of a phase/frequency scanned array

antenna with non-resonant slotted ridge waveguide elements’. Proceedings
of the IEEE Antennas and Propagation Symposium, London, Ontario,
Canada, 1991, pp. 1728–1731.
[18] Ahlbom S., Andersson P., Lagerlöf R. ‘A Swedish airborne early warning
system based on the Ericsson ERTEYE radar’. Ericsson Review. 1995;72(2):
54–63.
[19] Josefsson L. ‘Radar antenna R&D in Sweden’. 7th European Conference on
Antennas and Propagation (EUCAP 2013), Gothenburg, Sweden, April
2013, pp. 3166–3170.
[20] Cullens E. D., Ranzani L., Vanhille K. J., Grossman E. N., Ehsan N.,
Popović Z. ‘Micro-fabricated 130–180 GHz frequency scanning waveguide
arrays’. IEEE Transactions on Antennas and Propagation. 2012;60(8):
3647–3653.
[21] Dion A. (1958), ‘Nonresonant slotted arrays’. IRE Transactions on Antennas
and Propagation. 1958;6(4):360–365.
[22] Ishimaru A., Tuan H.-S. ‘Theory of frequency scanning of antennas’. IRE
Transactions on Antennas and Propagation. 1962;10(2):144–150.
[23] Ramsay J. F., Popovich B. V. ‘Series-slotted waveguide array antennas’.
IEEE International Convention Record. March 1963, pp. 30–55.
[24] Elliott R. S. Antenna Theory and Design. Prentice-Hall, Englewood Cliffs,
NJ, 1981.
[25] midwaysailor.com/midwaymyphotos4/restormisc.html. Accessed November
2016. Courtesy of Troy Prince.
[26] Hossain M. G. S., Tsunemitsu Y., Hirokawa J., Ando M. ‘Grating lobe
suppression by a dipole layer over a waveguide transverse slot array with
reflection-canceling posts’. Proceedings of the IEEE Antennas and Propa-
gation Symposium, Washington, DC, USA, 2005, pp. 720–723.
[27] Solbach K. ‘Below-resonant-length slot radiators for traveling-wave-array
antennas’. IEEE Antennas and Propagation Magazine. 1996;38(1):7–14.
[28] Ando M., Zhang M., Lee J., Hirokawa J. ‘Design and fabrication of milli-
meter wave slotted waveguide arrays’. Proceedings of the fourth European
Conference on Antennas and Propagation (EuCAP 2010), Barcelona, Spain,
April 2010, pp. 1–6.
[29] Simmons A. J. ‘Circularly polarized slot radiators’. IRE Transactions on
Antennas and Propagation. 1957;5(1):31–36.
[30] Getsinger W. J. ‘Elliptically polarized leaky-wave array’. IRE Transactions
on Antennas and Propagation. 1962;10(2):165–171.
[31] Josefsson L., Derneryd A., Lagerlöf R. ‘Electronic scanning with slotted
waveguide arrays’. Proceedings of the Radarcon 90 Conference, Adelaide,
Australia, April 1990.
[32] O’Loughlin W. R., Kim Y. U., Elliott R. S. ‘Pattern synthesis for a forward-
fire/backward-fire linear array’. IEEE Transactions on Antennas and
Propagation. 1989;37(6):721–727.
284 Slotted waveguide array antennas: theory, analysis and design

[33] Sparks R. A. ‘Recent microwave applications to radar in the U.S.’

Proceedings of the 19th European Microwave Conference (EuMA), 1989,
pp. 96–106.
[34] Silver S. ed. Microwave Antenna Theory and Design. MIT. Rad. Lab Series,
Vol. 12, 1949.
[35] Hougardy H. H., Shanks H. E. ‘Arbitrarily polarized slot array’. IRE Wescon
Convention Record. Los Angeles, USA, 1958, pp. 157–160.
[36] Sangster A. J. ‘Circularly polarized linear waveguide array’. IEEE Trans-
actions on Antennas and Propagation. 1973;21(5):704–705.
[37] Ajioka J. S., Joe D. M., Tang R., Wong N. S. ‘Arbitrarily polarized slot
radiators in bifurcated waveguide arrays’. IEEE Transactions on Antennas
and Propagation. 1974;22(2):196–200.
[38] Ajioka J. S., Joe D. M., McFarland J. L. ‘Slot radiators in septated waveguide’.
IEEE Transactions on Antennas and Propagation. 1984;32(3):247–251.
[39] Sangster A. J., Smith P., Jacobs R., Wang H. ‘Moment method solutions for
polarization agile slots in bifurcated waveguide’. Electromagnetics. 1999;
19(1):91–107.
[40] Josefsson L. ‘A dual polarised slotted waveguide array antenna’. Proceed-
ings URSI Electromagnetic Theory Symposium, Sydney, Australia, August
1992.
[41] Josefsson L. ‘A grating lobe filter for transverse slot arrays’. Proceedings of
the IEEE Antennas and Propagation Symposium, London, Ontario, Canada,
June 1991, pp. 1156–1159.
[42] Josefsson L. ‘A waveguide transverse slot for array applications’. IEEE
Transactions on Antennas and Propagation. 1993;41(7):845–850.
[43] Pedersen J. F., Hannan P. W. ‘A metal-grid 5 5 foot angular filter’. Pro-
ceedings of the IEEE Antennas and Propagation Symposium, Albuquerque,
USA, 1982, pp. 471–474.
[44] Josefsson L. Waveguide Antenna which Includes a Slotted Hollow Wave-
guide. US Pat. 5 541 612, 1996.
[45] Wood P. J., Sultan N., Seguin G. ‘A dual-polarized reconfigurable-beam
antenna for the DSAR synthetic aperture radar’. Proceedings of the IEEE
Antennas and Propagation Symposium, Baltimore, USA, 1996, pp. 1716–1719.
[46] Derneryd A., Lagerstedt A. ‘Novel slotted waveguide antenna with polari-
metric capabilities’. Proceedings of the Geoscience and Remote Sensing
Symposium (IGARSS), Florence, Italy, 1995, pp. 2054–2056.
[47] Magnusson P., Di Salvo M., Scarchilli C. ‘A dual polarised slotted
waveguide antenna for satellite based wind scatterometer instruments’. 7th
European Conference on Antennas and Propagation (EuCAP), Gothenburg,
Sweden, April 2013, pp. 3449–3452.
[48] Zahn R., Stangl M., Rostan F. ‘The X-band SAR instrument demonstrator:
intermediate test results’. Proceedings Millennium Conference on Antennas &
Propagation (AP 2000), Davos, Switzerland, April 2000.
[49] Roederer A., Mosig J., Balling P., Vandenbosch G., Keizer W., Kabacik P.
COST Action 260. Summary and Final Report, May 2001.
 Antenna systems and special requirements 285

[50] Alexander D. K., Sichelstiel B. A. ‘A unique waveguide phased array with

independently steered beams’. 1970 Phased-Array Antenna Symposium
(digest), Polytechnic Institute of Brooklyn, NY, June 1970.
[51] Butler J., Lowe R. ‘Beamforming matrix simplifies design of electronically
scanned antennas’. Electronic Design. 1961;9(7):170–173.
[52] Blass J. ‘Multidirectional Antenna – a new approach to stacked beams’. IRE
International Convention Record. 1960, pp. 48–50.
[53] White W. D. ‘Pattern limitations in multiple-beam antennas’. IRE Transac-
tions on Antennas and Propagation. 1962;10(4):430–436.
[54] Skolnik M. I. Introduction to Radar Systems, McGraw-Hill, New York,
1962.
[55] Torrero E. A. ‘The bypass of 3-D radar’. IEEE Spectrum. 1976;13(10):
49–52.
[56] van’t Klooster C. G. M., Karlsson I., Bonnedal M., Albertsen N. C. ‘Multiple
beam slotted waveguide antenna for spaceborne synthetic aperture radar’.
Proceedings of the IEEE Antennas and Propagation Symposium, London,
Ontario, Canada, 1991, pp. 531–534.
[57] Ingvarson P., Zachrisson J. ‘Swedish space antenna projects’. 7th European
Conference on Antennas and Propagation (EUCAP 2013), Gothenburg,
Sweden, April 2013, pp. 3049–3053.
[58] Simmons A. J., Giddings O. M. ‘A multiple-beam two-dimensional wave-
guide slot array’. IRE International Convention Record. 1963, pp. 56–69.
[59] Saltzman H., Stavis G. ‘A dual beam planar antenna for Janus type Doppler
navigation systems’. IRE National Convention Record. 1958, pp. 240–247.
[60] Miller J. R., Forman R. J. ‘A planar slot array with four independent beams’.
IEEE Transactions on Antennas and Propagation. 1966;14(5):560–566.
[61] Bodnar D. G., Barnes W. J., Corey L. E. ‘A lightweight HWL radar
antenna’. Proceedings IEEE International Radar Conference, Arlington,
VA, USA, 1980, pp. 259–262.
[62] Fan G., Jin J-M. ‘Scattering from a cylindrically conformal slotted wave-
guide array antenna’. IEEE Transactions on Antennas and Propagation.
1997;45(7):1150–1159.
[63] Josefsson L., Persson P. Conformal Array Antenna Theory & Design. IEEE
Press/Wiley, 2006.
[64] Lewis L. R., Fassett M., Hunt J. ‘A broadband stripline array element’.
Proceedings of the IEEE Antennas and Propagation Symposium, Atlanta,
GA, USA, June 1974, pp. 335–337.
[65] Rehnmark S. ‘Instantaneous bearing discriminators with omnidirectional
coverage and high accuracy’. IEEE Microwave Theory and Techniques
International Symposium, Washington, DC, USA, 1980, pp. 120–122.
[66] Kummer W. H. Special issue on conformal arrays (Preface). IEEE Trans-
actions of Antennas and Propagation. 1974;22(2):1–3.
[67] Muhs H. P. ‘mm-wave antenna’. Microwave Journal. 1985;28(7):191–194.
[68] Bayliss E. T. ‘Design of monopulse antenna difference patterns with low
sidelobes’. Bell Systems Technical Journal. 1968;47(5):287–289.
286 Slotted waveguide array antennas: theory, analysis and design

[69] Hansen R. C. Phased Array Antennas. Wiley, New York, 1998.

[70] Derneryd A. ‘A technique for low sidelobes in monopulse planar array
antennas’. Proceedings of the 1986 Antenna Applications Symposium,
Allerton Park, IL, USA, 1986.
[71] Watson C. K., Ringer K. ‘Feed network design for airborne monopulse slot-
array antennas’. Microwave Journal. 1988;31(6):129–145.
[72] https://commons.wikimedia.org/wiki/File:An_apg65_h.jpg. Retrieved 9 June
2017.
[73] Robertson R. S., Elliott R. S. ‘The design of transverse slot arrays fed by the
meandering strip of a boxed stripline’. Proceedings of the IEEE Antennas
and Propagation Symposium, Vancouver, Canada, 1985, pp. 111–114.
[74] Barnett R. I., Elliott R. S. ‘A feasibility study of stripline-fed slots arranged
as a planar array with circular grid and circular boundary’. IEEE Transac-
tions on Antennas and Propagation. 1989;37(12):1510–1515.
[75] Cohn S. B. ‘Characteristic impedance of the shielded-strip transmission
line’. IRE Transactions on Microwave Theory and Techniques. 1954;2(2):
52–57.
[76] Breithaupt R. W. ‘Conductance data for offset series slots in stripline’.
IEEE Transactions on Microwave Theory and Techniques. 1968;16(11):
969–970.
[77] Lagerlöf R. ‘Stripline fed slots’. Proceedings of the second European
Microwave Conference, Stockholm, August, 1971.
[78] Lagerlöf R. ‘Optimization of cavities for slot antennas’. Microwave Journal.
1973;16(10):12C–12F.
[79] Josefsson L., Moeschlin L., Svensson T. ‘A stripline flat plate antenna with
low sidelobes’. Proceedings of the IEEE Antennas and Propagation
Symposium, Atlanta, GA, USA, 1974, pp. 282–285.
[80] Josefsson L., Moeschlin L., Sohtell V. ‘A monopulse flat plate antenna for
missile seeker’. Proceedings of the Military Electronic Defence Expo
(MEDE), Wiesbaden, Germany, September 1977.
[81] Thors B., Josefsson L. ‘Radiation and scattering trade-off design for con-
formal arrays’. IEEE Transactions on Antennas and Propagation. 2003;51(5):
1069–1076.
[82] Bach Andersen J., Vaughan R. G. ‘Transmitting, receiving, and scattering
properties of antennas’. IEEE Antennas and Propagation Magazine. 2003;
25(4):93–98.
[83] Josefsson L. ‘Slot coupling and scattering’. Proceedings of the IEEE
Antennas and Propagation Symposium, Dallas, USA, 1990, pp. 942–945.
[84] Diamond B. L. ‘A generalized approach to the analysis of infinite planar
array antennas’. Proceedings of the IEEE. 1968;56(11):1837–1851.
[85] Kerr D. E. (ed.) Propagation of short radio waves, MIT Rad. Lab. Series,
Vol. 13, McGraw-Hill, New York, 1947.
[86] Ross R. A. ‘Radar cross section of rectangular flat plates as a function of
aspect angle’. IEEE Transactions on Antennas and Propagation. 1966;14(3):
329–335.
 Antenna systems and special requirements 287

[87] Polka L. A., Balanis C. A. ‘Non-principal plane scattering from perfectly

conducting, flat, rectangular plates’. Proceedings of the IEEE Antennas and
Propagation Symposium, San José, CA, USA, 1989, pp. 1278–1281.
[88] Tittensor P. J., Newton M. J. ‘Prediction of the radar cross-section of an
array antenna’. Proceedings of the Sixth International Conference on
Antennas and Propagation (ICAP), 1989, pp. 258–262.
[89] Levy R. ‘Determination of simple equivalent circuits of interacting dis-
continuities in waveguides or transmission lines’. IEEE Transactions on
Microwaves, Theory and Techniques. 2000;48(10):1712–1716.
[90] Fan G., Jin J-M. ‘Scattering from a large planar slotted waveguide array
antenna’. Electromagnetics. 1999;19(1):109–130.
[91] Zhang L., Yuan N., Zhang M., Li L-W., Gan Y-B. ‘RCS computation for a
large array of waveguide slots with finite wall thickness using the MoM
accelerated by P-FFT algorithm’. IEEE Transactions on Antennas and
Propagation. 2005;53(9):3101–3105.
[92] Rengarajan S. R. ‘Scattering and receiving characteristics of waveguide
slot array antennas’. Proceedings of the IEEE Radar Conference, Waltham
MA, USA, 2007, pp. 532–534.
[93] Pelton E. L., Munk B. A. ‘A streamlined metallic radome’. IEEE Trans-
actions on Antennas and Propagation. 1974;22(6):799–803.
[94] Stanek T., Johansson F. S. ‘Analysis and design of a hemispherical metallic
radome’. Proceedings of the Workshop on EM Structures, Nottingham,
UK, 1995.
[95] Gruenberg H. ‘Second-order beams of slotted waveguide arrays’. Canadian
Journal of Physics. 1953;31(1):55–69.
[96] Gruenberg H. ‘Microwave Antenna’. U.S. patent 2 639 383, May 19, 1953.
[97] Kurtz L. A., Yee J. S. ‘Second-order beams of two-dimensional slot arrays’.
IRE Transactions on Antennas and Propagation. 1957;5(4):356–362.
[98] McCormick G. C. ‘A two-dimensional slotted array’. IRE Transactions on
Antennas and Propagation. 1958;6(1):26–35.
[99] McCormick G. C. ‘The effect of the size of a two-dimensional array on
second-order beams’. IRE Transactions on Antennas and Propagation.
1958;6(3):297–298.
[100] Derneryd A. ‘Butterfly lobes in slotted waveguide antennas’. Proceedings
of the IEEE Antennas and Propagation Symposium, Blacksburg, VA, USA.
1987, pp. 360–363.
[101] Herting B. J., Elsallal M. W., West J. B. ‘Sub-array mirroring for the
mediation of second-order beams in planar edge slot waveguide arrays’.
Proceedings of the 2007 Antenna Applications Symposium, Allerton Park,
IL, USA, 2007.
[102] Kaminow I. P., Stegen R. J. Waveguide Slot Array Design. Hughes Aircraft
Company, Techn. Memo. No. 348, July 1954 (available as NTIS Report
AD 63600).
[103] Tang R. ‘A slot with variable coupling and its application to a linear array’.
IRE Transactions on Antennas and Propagation. 1960;8(1):97–101.
288 Slotted waveguide array antennas: theory, analysis and design

[104] Dudley D. G. ‘An iris-excited slot radiator in the narrow wall of rectangular
waveguide’. IRE Transactions on Antennas and Propagation. 1961;9(4):
361–364.
[105] Hashemi-Yeganeh S., Elliott R. S. ‘Analysis of untilted edge slots excited
by tilted wires’. IEEE Transactions on Antennas and Propagation. 1990;
38(11):1737–1745.
[106] Gruenberg H. ‘Theory of waveguide-fed slots radiating into parallel-plate
regions’. IRE Transactions on Antennas and Propagation. 1952;3(1):63–66.
[107] Gruenberg H. Microwave Antenna. U.S. patent 2 659 005, Patented
Nov. 10, 1953.
[108] Roberts A. (ed.) Radar Beacons. MIT. Rad. Lab Series, Vol. 3, 1947.
[109] Galindo V., Green K. ‘A near-isotropic circularly polarized antenna for
space vehicles’. IEEE Transactions on Antennas and Propagation. 1965;
13(6):872–877.
[110] Scott W. G., Soo Hoo K. M. ‘A theorem on the polarization of null-free
antennas’. IEEE Transactions on Antennas and Propagation. 1966;14(3):
587–590.
[111] Kraus J. D. Antennas, 2nd edition, McGraw-Hill, New York, 1988, p. 89.
[112] Wait J. R. ‘On the conductance of slots’. IRE Transactions on Antennas
and Propagation. 1956;4(2):124–127.
[113] Wait J. R. Electromagnetic Radiation from Cylindrical Structures.
Pergamon Press, London, 1959.
[114] Gayen R. K., Das S. D., Das A. C. ‘A high gain-bandwidth, nearly omni-
directional waveguide slot array antenna’. Proceedings of the IEEE Antennas
and Propagation Symposium, Vancouver, BC, Canada. 2015, pp. 222–223.
[115] Li J-Y., Li L-W. ‘Analysis of omnidirectional waveguide slots array
antennas’. Proceedings of the 6th International Symposium on Antennas,
Propagation and EM Theory, ISAPE’03, Beijing, October 28–November 1,
2003, pp. 38–41.
[116] Solbach K. ‘Simulation of slot array antennas using standard network analysis
program’. IEEE Antennas and Propagation Magazine. 1991;33(6):45–47.
[117] Croswell W. F., Knop C. M., Hatcher D. M. ‘A dielectric-coated
circumferential slot array for omnidirectional coverage at microwave
frequencies’. IEEE Transactions on Antennas and Propagation. 1967;15(6):
722–727.
[118] Fiet O. O., Scudder R. M. Ultrahigh-Frequency Broadcast Antenna System.
US patent 2 658 143, November 1953.
[119] Lee C. S., Lee S. W., Chuang S. L. ‘Plot of the modal field distribution in
rectangular and circular waveguides’. IEEE Transactions on Microwave
Theory and Techniques. 1985;33(3):271–274.
[120] Marcuvitz N. Waveguide Handbook. MIT. Rad. Lab Series, Vol. 10, 1951.
[121] Grabherr W., Huder B. ‘Omnidirectional slotted-waveguide array antenna’.
Proceedings of the 29th European Microwave Conference, Munich, Germany.
1999, pp. 134–137.
 Antenna systems and special requirements 289

[122] Morwing B. ‘AGA-ERICON – a marine radar beacon’. Ericsson Review.

1981;58(4):180–187.
[123] Knudsen H. L. ‘Radiation from ring quasi-arrays’. IRE Transactions on
Antennas and Propagation. 1956;4(3):452–472.
[124] Davies D: E. N. ‘Circular arrays: their properties and potential applica-
tions’. IEE Proceedings of the Second International Conference on
Antennas and Propagation, April 1981, pp. 1–10.
[125] Rahim T., Davies D. E. N. ‘Effect of directional elements on the directional
response of circular antenna arrays’. IEE Proceedings H – Microwaves,
Antennas and Propagation. 1982;129(1):18–22.
[126] Harrington R. F. Time-Harmonic Electromagnetic Fields, IEEE Press,
New York, 2001.
[127] Chu T-S. ‘On the use of uniform circular arrays to obtain omnidirectional
patterns’. IRE Transactions on Antennas and Propagation. 1959;7(4):
436–438.
[128] Croswell W. F., Knop C. M. ‘On the use of an array of circumferential slots
on a large cylinder as an omnidirectional antenna’. IEEE Transactions on
Antennas and Propagation. 1966;14(3):394–396.
[129] Lagerlöf R., Sjöholm L., Svensson T. Antenna for Strip Transmission
Lines, US patent 3 713 165, filed Oct. 5, 1970.
[130] Ohira M., Miura A., Ueba M. ‘60-GHz wideband substrate-integrated-
waveguide slot array using closely spaced elements for planar multisector
antenna’. IEEE Transactions on Antennas and Propagation. 2010;58(3):
993–998.
[131] Montgomery C., Dicke R., Purcell E. Principles of Microwave Circuits.
MIT Rad. Lab. Series, Vol. 8, McGraw-Hill, New York, 1948.
[132] Collin R. E. Foundations for Microwave Engineering. McGraw-Hill, Inc.,
New York, 1966.
[133] Collin R. E. Field Theory of Guided Waves. IEEE Press, New York, 1991.
[134] Pease M. ‘The iterated network and its application to differentiators’.
Proceedings of the IRE. 1952;40(6):709–711.
[135] Takeshima T., Isogai Y. ‘Frequency bandwidth of slotted array aerial system’.
Electron. Eng. February 1969, pp. 201–204.
[136] Hamadallah M. ‘Frequency limitations on broad-band performance of
shunt slot arrays’. IEEE Transactions on Antennas and Propagation.
1989;37(7):817–823.
[137] Derneryd A., Petersson R. ‘Bandwidth characteristics of monopulse slotted
waveguide antennas’. Fourth International Conference on Antennas and
Propagation (ICAP 85), 16–19 April, 1985, pp. 27–30.
[138] Josefsson L. Slots in Waveguide Arrays Analyzed by Expanding the Fields
in Waveguide Modes and Floquet Modes. Rpt No. AP-202, 1983, EE Dept,
UCLA, USA.
[139] Svensson B., Snygg G., Holmberg P. ‘A low cost, high performance point-
to-point slotted waveguide array’. Microwave Journal, 1999;42(11):1–5.
290 Slotted waveguide array antennas: theory, analysis and design

[140] Rengarajan S. R., Zawadzki M. S., Hodges R. E. ‘Bandwidth enhancement

of large planar slot arrays’. Proceedings of the IEEE Antennas and
Propagation Symposium, Honolulu, HI, USA, 2007, pp. 4405–4408.
[141] Rengarajan S., Chatterjee S. ‘Bandwidth characteristics of waveguide-fed
planar slot arrays’. Proceedings URSI General Assembly, Chicago, IL,
USA, 2008.
[142] Zhou S-G., Huang G-L., Peng Z-H., Li-Jian Y. ‘A wideband full-corporate-
feed slot planar array’. IEEE Transactions on Antennas and Propagation.
2016;64(5):1974–1978.
 Chapter 10
Slot arrays in special waveguide technologies

In this chapter, we discuss slot arrays in two types of parallel plate waveguides
propagating the TEM mode. In the first an electromagnetic wave propagates in the
radial direction inward or outward in a circular parallel plate waveguide, while slots
cut in one of the parallel plates radiate. Such an array is convenient for large
apertures requiring high gain and mass production. The second type has a rectan-
gular shape and propagates the TEM mode. Slots cut in one of the plates radiate.
We then present slot arrays in substrate integrated waveguides (SIW) which have a
great potential for easy integration with planar devices. Finally slot arrays in gap
waveguides are discussed.

10.1 Slot arrays in circular parallel plate waveguides

The concept of a radial line slot array, first introduced by Kelly in 1957 [1] and
published in 1961 by Goebels and Kelly [2], is shown in Figure 10.1. A standing
wave TM10 mode is excited between a pair of circular flat plates by a circular
waveguide feed at the centre. Annular slots cut at certain radial values produce
broadside radiation with the desired polarisation, including dual polarisation and
circular polarisation by exciting the appropriate waveguide modes. In a subsequent
paper [3] they presented a monopulse antenna divided into four 90
sectoral
waveguides as shown in Figure 10.2. Much of the later developments of the radial
line slot antennas were made at the Tokyo Institute of Technology, by Goto, Ando,
Hirokawa and others. We will discuss designs of radial line slot antennas for cir-
cular and linear polarisations. The design of a radial line slot antenna for mono-
pulse applications will also be presented.

10.1.1 Circularly polarised radial line slot antenna

The radial line slot antennas are easy to manufacture, for applications such as direct
broadcast satellite (DBS) receive antennas where inexpensive mass production is
desired. They have been investigated by numerous researchers and have been
published extensively. They have found applications in radar, satellite commu-
nications, wireless local area networks (WLAN), local multiport distribution
systems (LMDS) and plasma etching, among others, at microwave and millimetre-
wave frequencies. The two-layer radial line antenna is well suited to the design of
292 Slotted waveguide array antennas: theory, analysis and design

Figure 10.1 A typical radial line slot antenna.
IEEE 1961, reprinted from [2],
with permission

Figure 10.2 Monopulse radial line slot array.
IEEE 1964, reprinted from [3],
with permission
 Slot arrays in special waveguide technologies 293

nearly uniform aperture distributions required for high gain applications. In this struc-
ture the increase in power density of the wave as it propagates inward is offset by the
power radiated by the slots. In a single-layer radial waveguide antenna power flows
radially outward. It is better suited to the design of a tapered amplitude distribution.
Since a systematic design of the two-layer radial line slot antenna was presented first,
we start with a discussion of that structure. Figure 10.3 shows a radial line slot antenna
consisting of a two-layer circular parallel plate waveguide [4]. The lower parallel plate
waveguide is excited by an outward travelling radial TEM mode. In the upper parallel
plate waveguide consisting of radiating slots on the upper conductor, the TEM mode
propagates inward. It is then terminated in an absorber at the centre.
Figure 10.3(a) shows the two-layer antenna structure. Figure 10.3(b) illustrates
the power flow direction in both parallel plate waveguides. The antenna has a
coaxial cable to radial waveguide adaptor and a 180
E-bend between lower and
upper radial lines. In the upper waveguide orthogonal slot pairs couple power from
the radial waveguide and radiate a circularly polarised broadside beam. An
absorbing material placed near the centre at the radial value rm of the upper

Y D2

Ψ ρ X
O
Sρ

Sφ

ρM
Z
ρm ρ
d

Slow wave Absorber

structure
Coaxial cable
(a)

(b)

Figure 10.3 A two-layer radial line slot array: (a) complete antenna structure;
(b) direction of power flow in the radial waveguides.
IEEE 1985,
reprinted from [4], with permission
294 Slotted waveguide array antennas: theory, analysis and design

#1
#1
β1
ρ1 Θ1 2L ρ1 Θ P1
O X O
1
X
β1 ρ2
P1 Φ3
β3 ρ3 Θ1 #2
Sφ
ρ4 Θ3 P2
ρ2 δ #3
Θ1 Θ3
P3

P2
#4
#2 P4

–Y
(a) (b)

Figure 10.4 Slot arrangements in the upper wall: (a) first slot pair; (b) first two
slot pairs.
IEEE 1985, reprinted from [4], with permission

waveguide dissipates the remaining power in the travelling wave. In order to pro-
duce a good match at the coaxial cable input, the coaxial cable to radial waveguide
adaptor and the 180
E-bend have to be designed to have very little reflection.
In addition, reflection from the slot pairs shall be minimised as well.
The arrangement of slot pairs on the top plate of the upper parallel plate
waveguide is illustrated in Figure 10.4. The two slots of each slot pair are ortho-
gonal to each other with nearly the same amplitude of excitation and 90
phase
difference so as to produce a circular polarisation. The current distribution in the
plate for an inward travelling wave will be in the form of a Hankel function of the
first kind and order zero, whose phase term may be approximated by a cylindrical
phase function. Therefore the radial coordinates of the centres of the slot pair are
spaced a quarter guide wavelength so that their excitation phases are 90
apart as
shown by (10.1) and (10.2) [4].

kg ðr2  r1 Þ ¼ p=2 (10.1)

kg ¼ 2p=lg (10.2)
where lg is the guide wavelength and kg is the radial wavenumber in the upper
parallel plate waveguide. Excitations of slots 1 and 3 lag 90
with respect to those
of 2 and 4, respectively. Similarly all other slot pairs have a quarter wavelength
spacing between the radial coordinates of the centres of the slots. The two slot pairs
shown in Figure 10.4 as well as all others with similar orientation will radiate right
circular polarisation.
In order to eliminate the grating lobes, one may employ a slow-wave structure
in the upper parallel plate waveguide by loading it with a dielectric or by using a
corrugated structure in the lower plate of the parallel plate waveguide. The azi-
muthal spacing between successive slot pairs, Sf is approximately equal to the
 Slot arrays in special waveguide technologies 295

radial spacing Sr. The first slot of each slot pair makes the same angle with respect
to the radial line from the origin to the slot centre (e.g. Q1 ¼ Q3). In order to
produce a broadside beam, the two slot pairs in Figure 10.4(b) should radiate in
phase in the direction normal to the aperture. This requires that the tilt angle of the
second slot pair with respect to the first compensate the excitation phase difference
as specified by (10.3).

kg ðr3  r1 Þ ¼ F3 þ 2np (10.3)

where n is an integer. Slot pairs are arranged over a spiral with one guide wave-
length radial spacing between successive turns of the spiral. The reflections from
different slot pairs cancel out since they originate from different radial values just
like slots in a linear travelling wave array. The slot pairs are non-resonant so that
they couple a small amount of power from the propagating TEM wave.
We will now describe a specific design of the coaxial cable to the radial wave-
guide adaptor and the E-bend presented by Ando et al. [5]. Figures 10.5 and 10.6
show a section of the adaptor and the E-bend, respectively, which are rotationally

Lower
radial waveguide W1
θ
C
A
B

Coaxial cable
εr

r W0

Figure 10.5 Coaxial cable to radial waveguide adaptor (A ¼ 0, B ¼ C ¼ 1.5 mm,

q ¼ 51.3
, W0 ¼ 7.5 mm, W1 ¼ 12.5 mm, r ¼ 3 mm).
IEEE 1987,
reprinted from [5], with permission

Upper θ2
radial W.G. W2

h1 h2
D
Lower
radial W.G. W1 θ1

Figure 10.6 180
E-bend between lower and upper radial lines (D ¼ 12 mm, q1 ¼
43.8
, q2 ¼ 38.7
, W1 ¼ 12.5 mm, W2 ¼ 15 mm, h1 ¼ h2 ¼ 4 mm).

IEEE 1987, reprinted from [5], with permission
296 Slotted waveguide array antennas: theory, analysis and design

symmetric with respect to a vertical axis. The radii of the inner and outer conductors of
the coaxial cable are r and r þ W0, respectively.
Ando et al. [5] used the commercial finite element analysis code HFSS to
design the parameters A, B, C, q, W0, W1 and r of the adaptor and D, q1, q2, W1, W2,
h1 and h2 of the bend at 6 GHz. The reflection coefficients of these devices were
found to be less than 0.1 in the frequency range of 5.5–6.5 GHz. One could use any
computational electromagnetic code to design these devices with a starting value
obtained by scaling these parameters for the frequency range of interest.

10.1.2 Slot coupling and the attenuation coefficient

of the propagating wave
The power coupled by each slot pair from the inward travelling wave for radiation
is designed to provide the required aperture distribution. In many applications, it is
desired to have a uniform amplitude distribution to maximise the gain of the
antenna. Assuming weak slot coupling and neglecting reflections, the incident
wave at each slot pair may be expressed in the form

 pffiffiffi
f ðrÞ ¼ exp jkg r þ ar = r (10.4)
where a is the attenuation coefficient of the propagating TEM wave due to radia-
tion from slot pairs.
Figure 10.7 shows the amplitude distribution f (r) as a function of r for dif-
ferent values of a from 0.03 to 0.15 for a nine wavelength radius aperture. As seen
a value of a ¼ 0.15 produces an approximately uniform distribution while a value
of a ¼ 0.03 provides an 11 dB taper. The slot lengths are designed to have the
desired coupling by using an approximate infinite array model [6], with a unit cell
slot pair as shown in Figure 10.8. The waveguide model has conducting walls at the
top and bottom and periodic walls on the sides. This is an approximate model for
the slot pairs on a spiral. One can use a computational electromagnetics code such
as HFSS to analyse such a structure and obtain the slot lengths to realise the desired

0.0 α = 0.03 α = 0.06

–2.0 α = 0.09 α = 0.12
Amplitude in dB

α = 0.15
–4.0
–6.0
–8.0
–10.0
–12.0
0 2 4 6 8 10
Radial value in wavelengths

Figure 10.7 Amplitude distribution of a nine wavelength radius antenna for

different values of the attenuation coefficient a
 Slot arrays in special waveguide technologies 297

x
Periodic
b wall
y

Figure 10.8 Waveguide model for designing the coupling from a slot pair.
IEE
1990, reprinted from [6], with permission

pffiffiffi
coupling factor and the aperture taper. The factor 1= r and the attenuation term in
(10.4) will not appear in the analysis model as they are not relevant in the rectan-
gular waveguide. Let the coupling coefficient of the nth slot pair be cn and the field
incident at the nth slot pair be En. The power radiated by the nth slot is (Encn)2. It is
easy to show that the coupling coefficient cn is related to the attenuation coefficient
by ð1  c2n Þ1=2 ¼ eaðrn rn1 Þ . If the radial spacing between adjacent slot pairs is the
same, the coupling coefficients are equal.

10.1.3 Linearly polarised radial line slot antenna

The two-layer structure with a coaxial cable feed in the previous section is also
applicable to produce a linearly polarised antenna. In this case slot pairs are
arranged on concentric circles spaced a half-guide wavelength. Figure 10.9 shows
an x-directed linearly polarised radial line slot antenna in the top conductor of the
upper parallel plate waveguide, and Figure 10.10 shows the first two slot pairs. The
electric current in the slot plate induced by the travelling TEM mode is radially
directed. The slot 4 does not radiate since it is along the current direction of the
TEM mode whereas slot 3 radiates x-polarised field. The relationship between
angles q1 and q2 that slots 1 and 2 make with the radial line is derived below for
x-polarised radiation from the slots [7].
The co-polarised and cross-polarised radiations from the slot pairs 1 and 2 are
given by b1 sin(q1 þ f) þ b2 sin(q2 þ f) and b1 cos(q1 þ f)  b2 cos(q2 þ f),
respectively, where the coupling factors b1 and b2 of slots 1 and 2 are proportional
to sin q1 and sin q2, respectively, because of the half-guide wavelength radial
spacing between them. This does not take into account the radial dependence of the
amplitude of the travelling wave field and the associated current. The radial
dependence of the amplitude in the previous discussion of Figure 10.7 and the
waveguide model used for calculating the coupling are valid for this linearly
298 Slotted waveguide array antennas: theory, analysis and design

f X
O
Sρ

Sf

ρM Z
ρm ρ
du
dl
Slow wave Absorber
structure Coaxial cable

Figure 10.9 The slot pairs cut on concentric circles.
IEEE 1988, reprinted from
[7], with permission

Current-flow line
ρ1 E
θ1
1
λg/2 E

ρ2 Desired
2
θ2 Polarisation

f 3
E
4

Figure 10.10 First two slot pairs.
IEEE 1988, reprinted from [7], with
permission

polarised slot array also. The co-polarised and cross-polarised radiations for the slot
pair are given by (10.5) and (10.6).
sin q1 sinðq1 þ fÞ  sin q2 sinðq2 þ fÞ ¼ 1 (10.5)
sin q1 cosðq1 þ fÞ þ sin q2 cosðq2 þ fÞ ¼ 0 (10.6)
 Slot arrays in special waveguide technologies 299

These two equations are satisfied when

q1 ¼ p=2  f=2 (10.7)

and

q2 ¼ p  f=2 (10.8)

The radial distance between adjacent slot pairs is kept one guide wavelength. If
the guide wavelength is kept less than the free space wavelength by dielectric
loading, grating lobes are suppressed. The azimuthal distance Sf between adjacent
slots is kept one guide wavelength, just like that of the radial line slot antenna for
circular polarisation (see Figure 10.3).

10.1.4 Design of slot coupling coefficients

Let us assume that there are N rings of slot pairs with a guide wavelength spacing
between successive rings. Since the parallel plate waveguide is loaded with a
dielectric to make the guide wavelength smaller than the free space wavelength,
grating lobes are avoided. The spacing between adjacent slot pairs in the azimuthal
direction is also less than the free space wavelength for broadside beams, and the
number of slots in each ring is proportional to the value of the radius. Let the field
of the incident wave arriving at the outermost (Nth) ring of radius rN be

 pffiffiffiffiffiffi
AN ¼ exp jkr rN = rN (10.9)

If there are MN slot pairs in this ring and if the coupling coefficient, defined in
Section 10.1.2, of each slot is bN, the incident field in the (N  1)st ring is

 1=2 pffiffiffiffiffiffiffiffiffiffi
AN 1 ¼ exp jkr rN 1 1  MN b2N = rN 1 (10.10)

The number of slots in a ring is proportional to the radius of the ring. The
reflections from the slots are ignored in this analysis. The coupling coefficient bN1
of each slot in this ring is related to CN and CN1, the aperture excitation coeffi-
cients of Nth and (N  1)st rings, respectively, as given in (10.11).
pffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffi 1=2
bN = rN ¼ bN 1 = rN 1 1  MN b2N ðCN =CN 1 Þ (10.11)

Recurrence relations like this are written for all rings up to the first or the
innermost ring. The wave going past the innermost ring of slots will be dissipated
in the absorber.
The value of bN is initially estimated and subsequently updated in an iterative
fashion until the power dissipated in the absorber becomes a small value such as
5% of the power incident in the Nth ring of slots.
Figure 10.11 shows the values of coupling coefficients, bi, where i ¼ 120
denote the ring number. The phase of the aperture distribution is assumed to be
300 Slotted waveguide array antennas: theory, analysis and design

0.10
0.09
0.08
Uniform
0.07 10 dB edge taper
0.06
Coupling

0.05
0.04
0.03
0.02
0.01
0.00
0 2 4 6 8 10 12 14 16 18 20
Radial value in wavelengths

Figure 10.11 Slot coupling coefficient as a function of the radial value for an
inward travelling wave in a 20 wavelength radius double-layer
radial waveguide

uniform whereas two amplitude distributions, uniform and 10 dB taper, have been
considered for the circular aperture [8]. For the outermost slots, coupling increases
as the radius decreases for both uniform and tapered apertures due to decreasing
power in the travelling wave as it propagates inward. For tapered apertures, the
coupling coefficient starts with a smaller value for the outermost ring but the rate of
increase is greater than that of the uniform distribution since the aperture excitation
increases as the radius decreases. The coupling coefficient reaches a peak and starts
decreasing for innermost rings since the ratio rn1/rn decreases rapidly for small
values of n.
The incident travelling wave amplitude in dB in each ring is shown in
Figure 10.12. For both cases the incident power in the outermost ring is the same. For
uniform distribution, the incident wave power decreases more rapidly than for the
tapered distribution as greater power is radiated by the outermost rings in the former.
Both cases were designed to dissipate 14 dB power at the centre relative to the
incident wave power at the 20th ring. Therefore both plots coalesce at the centre.
Since the slot pairs are spaced one wavelength radially, the reflections from
them add up. Therefore the reflection coefficient of this antenna type is high. It is
possible to reduce the reflection coefficient at the input port by introducing addi-
tional pairs of slots called reflection cancelling slots on the radiating wall or on the
opposite wall in the case of a single-layer parallel plate waveguide or by changing
the slot spacing in the radial direction as a function of the azimuthal angle, thereby
producing a squinted beam off broadside. In one case the squinted beam design
produced 22 dB return loss compared to 4 dB of the corresponding broadside beam
design [9]. The improved return loss caused 1 dB increase in gain for the squinted
beam design. It is convenient to produce the squinted beam in the polarisation
 Slot arrays in special waveguide technologies 301

0.00

–2.00
Incident wave power in dB

–4.00

–6.00

–8.00

–10.00 Uniform

–12.00 –10 dB edge taper

–14.00
0 2 4 6 8 10 12 14 16 18 20
Radial value in wavelengths

Figure 10.12 Incident wave power in dB as a function of the radial value for an
inward travelling wave in a 20 wavelength radius double-layer
radial waveguide

plane, for example, for an x-polarised beam pointing at an angle (qt, ft ¼ 0). The
spacing, Sr, between two adjacent slot pairs in the radial direction is given next.

kr Sr þ k0 Sr sin qt cos f ¼ 2p (10.12)

where k0 is the free space wavenumber.

lg
Sr ¼ (10.13)
1 p1ffiffiffi sin qt cos f
er

Sr reduces to the previous value of lg for a broadside beam or qt ¼ 0. It is recommended

that the beam tilt qt be about a couple of beamwidths of the antenna, if possible.

10.1.5 Single-layer radial line slot arrays

Single-layer radial line slot antennas have been designed to produce circular
polarisation or linear polarisation using the same principles as those for the double-
layer radial line slot antennas discussed previously. In the single-layer array the
travelling wave propagates outward while slots couple radiated power. The slots
near the centre have very small amounts of coupling whereas the outer slots have
large amounts of coupling. It is not possible to design this array with an attenuation
coefficient such as that shown in Figure 10.7. Takahashi et al. showed that the
coupling increases very rapidly as the radial value increases [10]. Equations similar
to (10.9)(10.11) are used for outward radially propagating TEM mode. We then
obtain results for coupling coefficients and incident wave power for linearly
polarised slot pairs in concentric rings over a 20 wavelength radius aperture.
The coupling coefficient shown in Figure 10.13 increases rapidly as the wave
propagates outward. Since the incident wave amplitude decreases as the radius
302 Slotted waveguide array antennas: theory, analysis and design

0.09

0.08

0.07
Uniform
0.06 10 dB edge taper
Coupling

0.05

0.04

0.03

0.02

0.01

0.00
0 2 4 6 8 10 12 14 16 18 20
Radial value in wavelengths

Figure 10.13 Coupling coefficient as a function of the radial value for an outward
travelling wave in a 20 wavelength radius single-layer radial
waveguide

0.00

–2.00
Incident wave power in dB

–4.00

–6.00 Uniform

–8.00 10 dB edge taper

–10.00

–12.00

–14.00
0 2 4 6 8 10 12 14 16 18 20
Radial value in wavelengths

Figure 10.14 Incident wave power in dB as a function of the radial value for an
outward travelling wave in a 20 wavelength radius single-layer
radial waveguide

increases, larger amounts of coupling coefficient are needed to produce the desired
aperture distribution even though there are greater number of slots per ring as the
radius increases. For the tapered distributions, the coupling coefficient is larger
than that of the uniform distribution for small radial values while it is less for large
values of the radius.
Figure 10.14 illustrates the incident wave power as a function of the radial
value. For small radial values the slot coupling coefficient and the radiated power
 Slot arrays in special waveguide technologies 303

are small and hence the incident wave power is nearly constant whereas for large
radial values the incident wave power drops rapidly. For tapered distributions, the
power level is lower since greater power is radiated from the centre of the aperture.
Both plots coalesce at N ¼ 20 since they are designed to have the same amount of
power absorbed at the outer end of the parallel plate waveguide.

10.1.6 Radial line antennas for monopulse applications

A single-layer radial waveguide with outward travelling wave is considered here. A
similar analysis can be made for a two-layer radial waveguide slot array. For
monopulse applications we need four feeds, A1, A2, A3, and A4 all having the
same radial value and azimuthal locations of 0
, 90
, 180
and 270
, respectively,
as shown in Figure 10.15 [11].
The self and mutual impedances of the feed probes have been derived in [11]
using the induced emf method. The field distribution of the outward travelling
wave derived by Miyashita and Katagi [11] is used here. The coupling between
feeds i and j is given by

Sij ¼ C0 Zij (10.14)

where

2=Z0
C0 ¼
 (10.15)
ð1 þ Zii =Z0 Þ 1 þ Zjj =Z0
Z0 is the characteristic impedance of each feed line, and Zii and Zjj are the self-
impedances of the input ports i and j, respectively.
For the sum pattern with all four feeds excited equally (A1 ¼ A2 ¼ A3 ¼ A4 ¼ A),
the incident travelling wave is equivalent to that produced by a single centred probe

4 feeding probes at
radius = b

Outer boundary of the

radiating aperture

Figure 10.15 Feed probes for monopulse application.
IEEE 1996, reprinted
from [11], with permission
304 Slotted waveguide array antennas: theory, analysis and design

feed. For the azimuth difference pattern in the xz plane (A1 ¼ A2 ¼ A3 ¼ A4 ¼ A)
the incident travelling wave field is equivalent to that produced by a centred probe
feed weighted by a cos f term. The pattern in the xz plane produces a null in the
broadside direction with a major lobe on each side. Similarly for the elevation dif-
ference pattern in the yz plane (A1 ¼ A2 ¼ A3 ¼ A4 ¼ A) the incident travelling
wave field is equivalent to that produced by a centred probe feed weighted by a sin f
term. The pattern in the yz plane produces a null in the broadside direction with a
major lobe on each side. The design of the radiating slots may use any of the methods
discussed in the previous sections.

10.1.7 Accurate analysis and design techniques

An efficient full-wave method of moments analysis of radial line slot antennas was
presented by Albani et al. [12]. They split the Green’s function of a large circular
cavity into a sum of parallel plate waveguide contributions and a rim effect con-
tribution expanded in terms of radial modes. In a subsequent paper [13] an
approximate asymptotic formula was presented for the mutual admittance of slots
so that the numerical integration is reduced by more than 90% in computing the
mutual coupling between slot pairs. An automatic design of circularly polarised
radial line slot antennas for a specific aperture distribution was demonstrated using
a full wave analysis and optimisation [14].

10.2 Slot arrays in rectangular parallel plate waveguides

Slots cut in a parallel plate waveguide propagating a planar travelling wavefront is
less expensive to manufacture than a conventional planar slot array made up of
several rectangular waveguides, although the latter is preferred for high-performance
designs. Figure 10.16 shows a planar array of slots cut in a parallel plate waveguide
[15]. The TEM mode in the parallel plate waveguide is excited by an array of cou-
pling slot pairs cut in the broad wall of a rectangular waveguide. The rectangular
waveguide is placed underneath the parallel plate waveguide. The spacing between
slots in a pair is kept a quarter guide wavelength so that the reflection from each slot
pair in the feed waveguide at the design frequency is reduced substantially. The
adjacent slots pairs are spaced one guide wavelength so that excitation phases are
equal. The feed waveguide and the parallel plate waveguide are both loaded by
dielectrics such that the guide wavelength of the feed waveguide is lower than that of
the parallel plate waveguide which is in turn less than that of free space wavelength.
The feed waveguide may be excited by a coaxial cable at its centre or it may be
excited by a shunt series type coupling slot from another waveguide.
Figure 10.16 shows the feed waveguide excited by a coaxial cable. The TE10
mode wave propagating from the centre of the feed waveguide towards both ends is
equal in amplitude and phase but the propagation directions are opposite. There-
fore, the coupling slot pairs are tilted in opposite directions on the two sides, while
all the slot pairs on one side are tilted in the same direction. Both ends of the feed
waveguide have matched terminations and so have both ends of the parallel plate
 Slot arrays in special waveguide technologies 305

y y

z ar x
Square plate

λr Radiating slot pair

(linear pol.)

Feed rectangular
waveguide

Feed slot pair

Z Radiating slot
X
Parallel plate waveguide
Absorber
Feed waveguide
Dielectric Feed slot
Coaxial line

Figure 10.16 Slot array fed by a parallel plate waveguide.
IEEE 1992,
reprinted from [15], with permission

waveguide. Although it is possible to have reflection cancelling slots, absorbers at

the ends of the travelling wave arrays provide a better gain bandwidth and pattern
bandwidth than those employing reflection cancelling slots or shorts at the end. The
a dimension of the feed waveguide, af, and the dielectric constant ef are chosen to
satisfy the inequality (10.16) so that only a single mode propagates in the feed
waveguide.

l0 l0
pffiffiffiffi < af < pffiffiffiffi (10.16)
2 ef ef
In order to have a nearly uniform TEM mode to propagate in the parallel plate
waveguide, the spacing between the coupling slot pairs should be less than the
TEM mode wavelength in the parallel plate waveguide, that is, lgf < pl0ffiffiffi
er . Typically

l0 pffiffiffi
lgf ¼ pffiffiffiffi 2 (10.17)
ef
Therefore,
ef
er < (10.18)
2
The design of coupling slot pairs in the feed waveguide is carried out using the
procedure described in Chapter 6. Generally, a uniform aperture distribution is
306 Slotted waveguide array antennas: theory, analysis and design

desired in many applications. The scattering parameters of a given slot pair may be
obtained from a computational electromagnetic code such as HFSS as a function of
slot lengths and spacings. The tilt angle is kept at 45
. The spacing between slots in
a pair is nominally a quarter guide wavelength while the slots are parallel. The
lengths (non-resonant) and spacings are adjusted so that the reflection from the slot
pair is zero in the feed waveguide. For uniform aperture distribution, the coupling
into the parallel plate waveguide is the same from every slot pair and the power
going into the absorber in the feed waveguide should be small.
Figures 10.17 and 10.18 display a four-port coupler and the analysis model of a
feed slot pair, in which the parallel plate waveguide is replaced by a rectangular
waveguide with two periodic side walls. Such a model can also be used for the case

z
y Port 3

x Periodic wall

Port 1 Port 2
af

Feed rectangular waveguide

λf
Port 4

Figure 10.17 Analysis model for the feed slot pair.
IEEE 1992, reprinted from
[15], with permission

Port 3

Periodic wall y

Electric wall x

#2

lf2 df
Port 1 Port 2
lf1

#1
Magnetic field
Magnetic field (TEM mode)
(TE10 mode)

Port 4

Figure 10.18 Feed slot pair in the common wall of a four-port coupler.
IEEE
1992, reprinted from [15], with permission
 Slot arrays in special waveguide technologies 307

y
ar

x
Port 2
Periodic
Adjacent #1
boundary
lr1

#1
#2 dr

lr2
Adjacent #2

Port 1

Figure 10.19 Analysis model for a radiating slot pair.
IEEE 1992, reprinted
from [15], with permission

of a non-uniform aperture distribution as long as the aperture distribution is slowly

varying in both directions. Scattering parameters of the four-port coupler are
computed for a range of values of slot lengths and spacing for use in the design.
Radiating slot pairs in the parallel plate waveguide also produce no significant
reflection at the design frequency when the spacing between the slots in a pair is a
quarter guide wavelength just like the coupling slot pairs. The distance between the
adjacent slot pairs is kept at one wavelength. For a relatively large aperture, an
infinite array mutual coupling model may be employed to design the radiating slot
pairs, as shown in Figure 10.19. The amount of coupling from each radiating slot
pair is designed to produce the desired aperture distribution and to have a small
amount of power going into the absorber at the ends. This design procedure is
similar to that of travelling wave linear arrays discussed in Chapter 6.
The bandwidth of the parallel plate slot array shown in Figure 10.16 for 1 dB
gain drop is about 3%, and it is possible to achieve a return loss greater than 10 dB
in that frequency range. In the centre-fed parallel plate slot arrays such as the ones
discussed here, there are no radiating slots in the centre and hence the sidelobe level
in the E-plane is rather high, about 1011 dB. The end-fed parallel plate wave-
guide array has a major drawback since beam squint with frequency will limit the
gain bandwidth substantially. The sidelobe level may be improved by choosing a
tapered aperture distribution in the centre-fed configuration [16].

10.3 Slot arrays in substrate integrated waveguides

Dielectric-filled rectangular waveguides with sidewalls replaced by an array of

metallised posts are called SIW [17]. They are convenient for easy integration
of planar devices and rectangular waveguide components. The dielectric-filled
308 Slotted waveguide array antennas: theory, analysis and design

rectangular waveguides can be designed with a very small b dimension and are
easily transitioned to microstrip transmission lines. If the number of metallised vias
is greater than 10 per wavelength, such waveguides essentially behave like
dielectric-filled rectangular waveguides with metallic side walls. Based on a study
of dispersion characteristics of the substrate integrated rectangular waveguide, the
cut-off frequencies of the first two modes were obtained as a function of the
structural parameters, as shown in (10.19) and (10.20) [18].
 1
c D2
fc10 ¼ pffiffiffiffi W  (10.19)
2 er 0:95b
 1
c D2 D3
fc20 ¼ pffiffiffiffi W   (10.20)
er 1:1b 6:6b2

where c is the velocity of light in free space, D is the diameter and b is the spacing
of adjacent vias. W is the distance between the two arrays of vias, that is, nominally
the a dimension of the corresponding dielectric-filled waveguide. If the spacing
between vias is sufficiently small, that is, greater than 10 vias per wavelength, the
effective value of the width of the waveguide is given by

D2
Wef f ¼ W  (10.21)
0:95b
Thus an SIW is equivalent to a dielectric-filled metallic rectangular waveguide
with width Weff. Then the values of offsets and lengths of the radiating slots in a
dielectric waveguide can be determined using the design procedures discussed in
Chapters 6 and 7.
An example of a planar array of slots in a two-layer SIW configuration is
shown in Figure 10.20 [19]. They used the conventional design of a slot array in
equivalent dielectric-filled rectangular waveguides. Transitions from microstrip
line to SIW using tapered lines were found to work well. In the lower layer a feed
waveguide is formed using the SIW technology with centred-inclined coupling
slots. Coupling coefficients for such slots are found using (7.2) in Chapter 7.
A single-layer SIW slot array for 60 GHz is shown in Figure 10.21 [20]. The
feed waveguide couples into radiating waveguides through coupling windows at
one end. Even though such an end-fed arrangement has bandwidth limitations, it
has the advantage of being a low-cost device because of the ease of fabrication of
the single-layer device. A conductor backed coplanar waveguide (CBCPW) is used
to feed the antenna, transitioning into a feed SIW.
SIW technology, that is, arrays of metallised vias in place of conducting walls,
has also been employed in parallel plate waveguide slot arrays [21]. The low-
temperature co-fired ceramics (LTCC) manufacturing technique is suited to make
SIW slot arrays, especially at millimetre-wave frequencies and even higher fre-
quencies in the range of hundreds of GHz, with the availability of low-loss
dielectric materials [22,23].
 Slot arrays in special waveguide technologies 309

Top layer
COL6
(slot array)

Coupling
slots

Input
port

Bottom layer
(feed waveguide)

COL1 z^

x^

ROW1 ROW6

Figure 10.20 A two-layer SIW slot array at 60 GHz. Reproduced from [19],
courtesy of The Electromagnetics Academy

Power
divider
50 Ω CBCPW Radiating
slots

Substrate

CBCPW to SIW
transition Metalised
via holes

Figure 10.21 A single-layer SIW slot array at 60 GHz.
IEEE 2010, reprinted
from [20], with permission
310 Slotted waveguide array antennas: theory, analysis and design

10.4 Slot arrays in gap waveguides

Slotted rectangular waveguide array antennas are difficult to fabricate at millimetre-
wave frequencies and above because of manufacturing tolerances. In addition, it is
difficult to ensure good electrical contact between the waveguide wall containing the
slots and the side walls when the slot plate is brazed into the waveguide channel
(see Section 11.2). Although the SIW technology is widely used at high frequencies,
such waveguides exhibit significant losses. Gap waveguides propagate quasi-TEM
mode over a wide bandwidth [24,25]. Power loss in a gap waveguide is comparable
to that in a rectangular waveguide [26], and it is much less than that for an SIW.
Therefore, the gap waveguide technology has a lot of promise at millimetre-wave
frequencies and beyond.
Figure 10.22 shows the geometry of the gap waveguide consisting of two
parallel plates and a ridge in the middle. On each side of the ridge there is a bed of
nails. If the spacing ‘a’ between adjacent nails is very small compared to their
height ‘d’, that is, a  d, and if there are several nails per wavelength, the bed of
nails may be considered as a high impedance surface [27]. In that case the region
above the bed of nails behaves as a stop band for electromagnetic wave propagation
for a broad range of frequencies and a quasi-TEM mode propagates between the
ridge and the conductor above. Therefore, it is possible to have an airgap between
the top plate and the bed of nails, thereby eliminating the need for any electrical
contact in the structure. The cut-off frequency of the first mode is given by d ¼ l/4
while that of the next mode is given by d þ h ¼ l/2. By choosing a small value of h,
one can realise nearly an octave of bandwidth for single-mode propagation.
Zaman and Kildal [28] describe a single-layer low-loss corporate feed network
for a slot array. Figure 10.23 shows the corporate power divider and a four-element
linear array of slots in the top plate shown on the right. The gap waveguide without
the ridge is excited by a coaxial cable probe and the waveguide is transitioned to
ridge gap waveguide by stepped ridges. The power divider is similar to microstrip
power dividers with quarter wave impedance matching sections and mitred corners.
The ends of the ridge waveguides are terminated in Ts before they excite the slots.
Corrugated sections at the end of the E-planes reduce the edge diffracted fields in
the back region.

h x a
d z

w a

Figure 10.22 The geometry of the gap waveguide.
IEEE 2011, reprinted from
[25], with permission
 Slot arrays in special waveguide technologies 311

Figure 10.23 A linear array of four slots excited by a ridge gap waveguide with a
corporate feed.
IEEE 2014, reprinted from [28], with permission

Rectangular
waveguide opening Bottom plate
with radiating slots

Corporate feed
network

Figure 10.24 A planar array of 2  2 slots excited by a ridge gap waveguide with
a corporate feed.
IEEE 2014, reprinted from [28], with
permission

In the design of the bends of the T-junctions, a few pins were removed or
relocated locally. This does not affect the performance of the gap waveguide
structures. For designing the antennas without grating lobes, one row of pins has
been used at some locations. Even with one row of pins the leakage of energy to
neighbouring elements is 20 dB down.
Figure 10.24 shows a 2  2 planar array in the ridge gap waveguides. It has a
90
transition from ridge gap waveguide to rectangular waveguide. Since the ridge
312 Slotted waveguide array antennas: theory, analysis and design

z
y

x
ls
Radiating slot
ws

Top
Coupling slot layer

lc
Cavity ws

Bottom
lT layer
Ridge feed line lm
wm wT

Figure 10.25 The 2  2 array unit cell excited by a cavity.
IEEE 2014,
reprinted from [29], with permission

gap waveguide propagates a quasi-TEM mode just like a microstrip line, the ridge
gap waveguide to rectangular waveguide transition is designed similar to a
microstrip line to waveguide transition.
The four-way power divider is designed by tapering the ridge sections gradually
for the impedance match. This results in 20 dB return loss over 1215 GHz band.
The 2  2 element array is excited in phase and with equal amplitude by a corporate
feed network. The element spacing is about 0.875l at 15 GHz. Both the 1  4 and
2  2 arrays produced about 20% bandwidth for better than 10 dB return loss.
Figure 10.25 shows a unit cell comprising 2  2 slots excited by a cavity [29].
The bottom layer of Figure 10.25 has a ridge waveguide terminated in a T section.
A coupling slot in the middle excites the cavity which in turn excites the four
radiating slots in the top plate. Sixteen such unit cells make up an 8  8 array of
slots in ridge gap waveguide technology with a corporate power divider as illu-
strated in Figure 10.26 [29].
The 8  8 array discussed earlier has uniform aperture distribution and pro-
duces 13 dB sidelobe levels in both the E- and H-planes. It is possible to design
tapered distributions using non-uniform power dividers and non-resonant slots of
different lengths for low sidelobe arrays. The slot arrays in ridge waveguide tech-
nology are expected to find many applications in the future, such as integration of
planar antennas, filters, active devices, etc., up to terahertz frequencies.
 Slot arrays in special waveguide technologies 313

lt wq
Radiating slots
wt lq

Feeding
network
Transition to
WR-15

Figure 10.26 An 8  8 slot array in ridge gap waveguides with a corporate power
divider.
IEEE 2014, reprinted from [29], with permission

References
[1] Kelly K. C. ‘Recent annular slot array experiments’. IRE National Conven-
tion Record, pt. 1, 1957, pp. 144–151.
[2] Goebels F. J., Kelly K. C. ‘Arbitrary polarisation from annular slot
planar antennas’. IRE Transactions on Antennas and Propagation. 1961;9(4):
342–349.
[3] Kelly K. C., Goebels F. J. ‘Annular slot monopulse antenna arrays’. IEEE
Transactions on Antennas and Propagation. 1964;12(4):391–403.
[4] Ando M., Sakura K., Goto N., Arimura K., Ito Y. ‘A radial line slot antenna
for 12 GHz satellite TV reception’. IEEE Transactions on Antennas and
Propagation. 1985;33(12):1347–1353.
[5] Ando M., Ito S., Sakurai K., Goto N. ‘Suppression of reflection in a radial
line slot antenna for 12 GHZ band satellite reception’. IEEE Antennas and
Propagation Symposium Digest, 1987, pp. 898–901.
[6] Hirokawa J., Ando M., Goto N. ‘Analysis of slot coupling in radial
line slot antenna for DBS reception’. IEE Proceeding H. 1990;137(5):
249–254.
[7] Ando M., Numata T., Takada J.-I., Goto N. ‘A linearly polarized radial line
slot antenna’. IEEE Transactions on Antennas and Propagation. 1988;36(12):
1675–1680.
[8] Stutzman W. A., Thiele G. A. Antenna Theory and Design, 3rd edn. John
Wiley and Sons, Hoboken, NJ, 2013.
314 Slotted waveguide array antennas: theory, analysis and design

[9] Davis P. W., Bialkowski M. E. ‘Linearly polarized radial-line slot-antennas

with improved return-loss performance’. IEEE Antennas and Propagation
Magazine. 1999;41(1):52–61.
[10] Takahashi M., Takada J., Ando M., Goto N. ‘Characteristics of small-aperture,
single-layered, radial line slot antennas’. IEE Proceedings-H. 1992;139(1):
79–83.
[11] Miyashita H., Katagi T. ‘Radial line planar monopulse antenna’. IEEE
Transactions on Antennas and Propagation. 1996;44(8):1158–1165.
[12] Albani M., Cono G. L., Gardelli R. ‘An efficient full-wave method of
moments analysis for RLSA antennas’. IEEE Transactions on Antennas and
Propagation. 2006;54(8):2326–2336.
[13] Albani M., Mazzinghi A., Freni A. ‘Asymptotic approximation of mutual
admittance involved in MoM analysis of RLSA antennas’. IEEE Transac-
tions on Antennas and Propagation. 2009;57(4):1057–1063.
[14] Albani M., Mazzinghi A., Freni A. ‘Automatic design of CP-RLSA anten-
nas’. IEEE Transactions on Antennas and Propagation. 2012;60(12):
5538–5547.
[15] Hirokawa J., Ando M., Goto N. ‘Waveguide-fed parallel plate slot array
antenna’. IEEE Transactions on Antennas and Propagation. 1992;40(2):
218–223.
[16] Hirokawa J., Ando M. ‘Sidelobe suppression in 76 GHz post-wall wave-
guide-fed parallel plate slot arrays’. IEEE Transactions on Antennas and
Propagation. 2000;48(11):1727–1732.
[17] Deslandes D., Wu K. ‘Integrated microstrip and rectangular waveguide in
planar form’. IEEE Microwave and Wireless Components Letters. 2001;
11(2):68–70.
[18] Cassivi Y., Perregrini L., Arcioni P., Bressan M., Wu K., Conciauro G.
‘Dispersion characteristics of substrate integrated rectangular waveguide’.
IEEE Microwave and Wireless Components Letters. 2002;12(9):333–335 .
[19] Bakhtafrooz A., Borji A., Busuioc D., Safavi-Naeini S. ‘Novel two-layer
millimeter-wave slot array antennas based on substrate integrated wave-
guides’. Progress in Electromagnetics Research. 2010;109:475–491.
[20] Chen X.-P., Wu K., Han L., He F. ‘Low-cost high gain planar antenna array
for 60-GHz band applications’. IEEE Transactions on Antennas and pro-
pagation. 2010;58(6):2126–2129.
[21] Hirokawa J., Ando M. ‘Efficiency of 76 GHz postwall waveguide fed par-
allel plate slot arrays’. IEEE Transactions on Antennas and Propagation.
2000;48(11):1742–1745.
[22] She Y., Fujino R., Hirokawa J., Ando M., Hanatani D., Fujimoto M. ‘LTCC
oversized rectangular waveguide slot array antenna with air-layer in the
radiating part in the millimeter-wave band’. IEEE Transactions on Antennas
and Propagation. 2013;61(4):1777–1783.
[23] Xu J., Chen Z. N., Qing X. ‘270-GHz LTCC-integrated strip-loaded linearly
polarized radial line slot array antenna’. IEEE Transactions on Antennas and
Propagation. 2013;61(4):1794–1801.
 Slot arrays in special waveguide technologies 315

[24] Kildal P. S., Alfonso E., Valero-Nogueira A., Rajo-Iglesias E. ‘Local

metamaterial-based waveguides in gaps between parallel metal plates’. IEEE
Antennas and Wireless Propagation Letters. 2009;8:84–87.
[25] Polemi A., Maci S., Kildal P.-S. ‘Dispersion characteristics of metamaterial-
based parallel plate ridge gap waveguide realized by bed-of-nails’. IEEE
Transactions on Antennas Propagation. 2011;59(3):904–913.
[26] Pucci E., Zaman A. U., Rajo-Iglesias E., Kildal P. S., Kishk A. ‘Losses in
ridge gap waveguide compared with rectangular waveguide and microstrip
lines’. Fourth European Conference on Antenna and Propagation, 2010.
[27] Silveirinha M. G., Fernandes C. A., Costa J. R. ‘Electromagnetic char-
acterization of textured surfaces formed by metallic pins’. IEEE Transac-
tions on Antennas and Propagation. 2008;56(2):405–415.
[28] Zaman A. U., Kildal P. S. ‘Wideband slot antenna arrays with single-layer
corporate feed network in ridge gap waveguide technology’. IEEE Trans-
actions on Antennas Propagation. 2014;62(6):2992–3001.
[29] Vosoogh A., Kildal P. S. ‘Corporate-fed planar 60-GHz slot array made of
three unconnected metal layers using AMC pin surface for the gap waveguide’.
IEEE Antennas and Wireless Propagation Letters. 2016;15:1935–1938.
 Chapter 11
Manufacturing aspects

In this chapter, we will take a look at manufacturing aspects that are important for
high-quality production of slotted waveguide array antennas. It is a wide subject so
we will concentrate on critical parameters such as mechanical tolerances that relate
to electrical performance. We will also discuss joining methods for metal materials,
for example, dip brazing of aluminium. Applications at high frequencies (milli-
metre waves) in particular call for very high precision and special methods. An
important area is the use of slotted carbon fibre reinforced plastic (CFRP) wave-
guides that offer light weight and thermal stability, typically required for large
antenna systems for space applications. The technology is also used in some mili-
tary ground and airborne radar systems.
Some of the exciting developments in low-cost fabrication using plastic
materials and metallisation techniques are discussed. This area is related to the
microelectronic area, particularly for high frequencies and highly integrated
antenna/microwave assemblies.

11.1 Tolerance theory

When an array antenna is tested experimentally, the measured electrical perfor-
mance will sometimes depart to some extent from the theoretical design due to
excitation errors. The errors can be random and/or systematic. Systematic errors
can in principle be corrected once the cause of the errors has been found. Random
errors could be due to imperfections in the materials, tools, fixtures or other
unknown source. They may be difficult to reduce at a reasonable cost. Errors due to
inadequate design modelling are also important.
Several authors have treated the tolerance problem for general arrays and also
for slotted waveguide arrays [1–4]. When analysing random excitation errors a
common approach is to assume the errors to be uncorrelated from element to ele-
ment, and with a normal distribution of the real and imaginary parts of the error.
We can write the resulting voltage Vn of element n:


Vn ¼ Vn0 þ sVn0 ¼ Vn0 1 þ rn e jan (11.1)

According to this model the phase error an is uniformly distributed within

(0,2p). It can be shown that the relative amplitude error rn is Rayleigh distributed
318 Slotted waveguide array antennas: theory, analysis and design

V0 sV0

Figure 11.1 The nominal element voltage V0 and the error voltage sV0

while the array pattern function has a modified Rayleigh distribution; see, for
example, Rondinelli [3]. The error model is represented graphically in Figure 11.1.
Assuming small errors the errors expressed in phase and amplitude become

sphase ¼ s radians
(11.2)
sdB ¼ 20logð1 þ sÞ dB
As shown here the amplitude error and the phase error are both given by the
same parameter s. This can of course be questioned since in many cases the two
types of error have different sources. Still, the model illustrates the level of influ-
ence from typical errors. For given or assumed phase and amplitude errors we can
in practical cases write
s2 ¼ s2tot ¼ s2phase þ s2ampl (11.3)
The effect on the array gain can be estimated as


G ¼ G0 = 1 þ s2 (11.4)
Normally this effect is quite small. The effect on the sidelobe level is usually
much more serious. The main beam power level can be written as
!2
XN
Gmax / Vn (11.5)
1

while the sidelobe power level is

X
N
GSL / ðsVn Þ2 (11.6)
1

Thus, the sidelobe level relative to the main beam becomes

s2
SL ¼ SL0 þ (11.7)
hN
where h is the aperture efficiency:
X 2
Vn
h¼ X 2 (11.8)
N Vn
 Manufacturing aspects 319

Example:
Number of radiating elements: N ¼ 64
Element error: 0.5 dB and 5 , uncorrelated => stot ¼ 0.11
Pattern function: SL0 ¼ 50 dB sidelobes, h ¼ 0.72 (error-free)
This gives the average sidelobe level due to the errors alone:
s2
SLerr ð hN Þ ¼ 36 dB. Sidelobe peaks will be perhaps 68 dB higher. Clearly,
this is an unbalanced design; the tapering is too strong considering the error level
(too much overdesign).
In this example the 64 elements could be the 64 slots in a single linear wave-
guide array. If we now take 32 identical slotted waveguide arrays of this kind and
combine them into a 32  64 planar array, we would still get an error sidelobe level
of about 36 dB. However, if the 32-element slotted arrays had completely inde-
pendent errors (but at the same average level as before), the sidelobe average level
due to errors becomes 15 dB lower, that is, at about the same level as SL0, a much
more reasonable design.
Figure 11.2 shows the calculated sidelobe level SL versus the error-free side-
lobe level SL0 for different numbers of uncorrelated elements N.
We have here discussed array element errors in dBs and degrees, while the real
error character is slot positions, orientations and lengths, etc., in inches and milli-
metres. Furthermore, the average sidelobe level as discussed earlier is not always
the most suitable parameter. A more relevant measure can be the maximum side-
lobe level that is expected with, say, 95% confidence. This approach was studied by

SL [dB]

–20
s = 2 dB
hN = 37
–30

s = 1 dB
hN = 70
–40

s = 1 dB
hN = 700
–50

SL0 [dB]
–60
–20 –30 –40 –50 –60

Figure 11.2 Average sidelobe level SL (achieved) versus sidelobe level SL0
(ideal) with error s as parameter. N is the number of uncorrelated
radiating elements, and h is the aperture efficiency
320 Slotted waveguide array antennas: theory, analysis and design

1.0

σ = .1 .2 .3 .4
.5 .6
0.8 .7 .8
.9 1.

0.6
Probability

0.4

0.2

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Normalised amplitude

Figure 11.3 Normalised probability function for sidelobe level. Both the
amplitude and the error parameter s are normalised to the design
sidelobe level.
1982 IEEE. Adapted from [4], with permission

Hsiao [4], who derived the probability density function of the sidelobe level (see
Figure 11.3).
The error tolerance problem can also be studied by direct simulation of the
antenna array in question. Lee et al. [5] used the Monte Carlo method for deter-
mining acceptable error levels. The tolerance for each element was set inversely
proportional to the nominal element weight. A Monte Carlo simulation was also
used by Rengarajan et al. [6]. For their simulation random element error levels of
0.5 dB and 5 were assumed. The authors also made a Method of Moments analysis
with assumed machining tolerances of  1 mil for slot positions, lengths and
widths. Finally, a finite element analysis using HFSS code was made before
building and testing an experimental antenna. The results from all these efforts
showed good agreement.
In two early papers [7,8], the authors treated the tolerance problem for linear
waveguide arrays with longitudinal slots. The mechanical errors considered were
slot position errors: along waveguide, across waveguide (i.e. offset errors), and slot
length errors. The errors were assumed to be independent, random and normally
distributed. The error in slot position along the waveguide was found to be of minor
importance compared to the errors in slot offset and slot length. Bailin and Ehrlich
[7] quoted a result by Stegen [9] where the change of phase of the radiated field was
measured as a function of a change in slot length off resonance. Stegen’s curve is
reproduced in Figure 11.4.
We find from Stegen’s graph that 5 degrees of phase error results from a
slot length error of about 0.5%, or less than one-tenth of one mm (X-band).
 Manufacturing aspects 321

80

60

40
Relative phase/degrees

20

–20

–40

0.90 0.95 1.00 1.05 1.10

Ratio of slot length to resonant length

Figure 11.4 Radiated phase as a function of change in slot length off resonance.
The points on the curve indicate slot offsets, from about 0.7 to 3.9
mm. Standard X-band waveguide, slot width 1.6 mm, frequency 9.375
GHz. Adapted from Kaminov and Stegen [10]

(We know from Chapter 5 that an error in slot offset also changes the slot resonant
length as well as the slot admittance.)
The offset of a longitudinal slot is referenced to the waveguide centreline,
that is, the line halfway between the inner side walls of the waveguide. This line is
difficult to establish with high precision from the outside of the waveguide,
especially for long waveguides. If in error the result is a systematic error, that is,
a periodic offset error with a period of about one guide wavelength, giving rise
to grating lobes. The artillery locating radar described in Chapter 3 (Figure 3.5)
has a radiating aperture of frequency scanned non-resonant waveguides, each
with 64 longitudinal slots. It has been estimated that a slight offset error in the
waveguide centreline by as little as 0.04 mm would result in a 20 dB grating lobe
level [11].
Rectangular waveguide sizes are standardised (EIA, IEC)1 including toler-
ances and flange types [12]. Also ridge waveguide dimensions are standardised to
some extent. However, for slotted waveguide arrays using ridge waveguides (e.g. to
reduce the width of the waveguide) non-standard cross sections may be required.
The tolerances for the critical dimensions can be analysed theoretically using HFSS
or special software [13,14]. Figure 11.5 shows a simplified cross section of a ridge
waveguide which is characterised by four main dimensions: width (A), height (B),

1
EIA ¼ Electronic Industries Association, IEC = International Electrotechnical Commission. See also
Appendix.
322 Slotted waveguide array antennas: theory, analysis and design

Figure 11.5 Cross section of a ridge waveguide

ridge width (S) and the separation between the upper wall and the ridge (D). From a
theoretical analysis it was found that the separation (D in the figure) is the most
critical and should be given the tightest tolerance. This is also the region where the
electric field is concentrated.

11.2 Manufacturing methods

We will first discuss manufacturing methods using metallic materials and then
continue with plastic materials. Examples of technical solutions and real arrays as
presented in the open literature are also given.

11.2.1 Metallic materials

Important characteristics of materials for slotted waveguide array antennas are
● Good conductivity
● Low weight
● Corrosion resistance
● Processability
More parameters can be added to the list depending on the application. Low
cost is sometimes very important. For space applications, on the other hand, ther-
mal stability can be the critical parameter. Overviews of manufacturing methods
have been presented [15,16].
Aluminium, copper and silver are typical metals used. Aluminium or alumi-
nium alloys are widely used for frequencies up to about 30 GHz. Copper and silver
are sometimes used for higher frequencies. Standard waveguides are available in
many materials and can be the starting point for fabrication. Typical dimensions for
the X-band WR90/R100 waveguide are inner dimensions 22.86  10.16 mm, wall
thickness 1.27 mm and a tolerance for the inner dimensions of 0.08 mm (0.003
in.). For antennas a reduced height version is often preferred. Instead of starting
from separate waveguides several waveguide channels can be milled out of a block
of the chosen material. Die-casting and electroforming are also used for special
requirements.
 Manufacturing aspects 323

The traditional assembly technique for joining aluminium parts is dip brazing.
The parts are kept together by fixtures, if necessary, and submerged in a bath of
special salt heated to about 600 C. This is close to the melting temperature of the
aluminium alloy. A filler material is added in the joints where it will melt and join
the assembly. After the brazing the unit is hardened and surface treated. Anderson
et al. [17] describe the design and fabrication of X- and Ku-band array antennas
using numerically controlled machining and dip brazing. Joining the various parts
by epoxy bonding has also been reported [18].
A 94 GHz linear array of 21 longitudinal slots was made in copper by photo-
etching the slots on a 0.1 mm copper sheet [19]. The sheet was then brazed onto a
copper block with the waveguide channel. The theoretical analysis indicated a
required positional accuracy for the slot positions of 0.0004 in. (0.01 mm) and even
tighter for the slot length. An array of series inclined slots was also fabricated.
In another study [20] it was shown that a linear slotted waveguide array with
ten slots can be built for 94 GHz from standard WR-10 waveguide. However, the
wall thickness was reduced to facilitate scaling from X-band data. The slots were
milled with a machine tolerance of typically  0.02 mm ( 0.001 in.).
For the machining of rectangular slots the slot ends are often rounded and a
correction for this can be applied as discussed in Chapter 5, Section 5.3.3; cf. also
Section 8.10.1.
Low-cost mass production of 20 and 60 GHz planar arrays was aimed for in an
experimental work by Ando and Hirokawa [21]. Their findings include die-casting
of waveguide channels showed promise. Laser-welding of a slotted plate to the
waveguide channels caused too much distortion while bonding using a conducting
adhesive worked well mechanically, but losses were too high. However, feeding
adjacent waveguides in antiphase (compensated with slot offset directions) was
promising since the currents in the waveguide vertical walls almost vanished.
A crucial step in the fabrication is the joining of a slot plate to the waveguide
channel(s). Screws can only be considered for low frequencies. For high fre-
quencies (millimetre waves) diffusion bonding is an interesting technique where
the parts are pressed together at a high temperature causing the parts to join (no
adhesive). Stainless steel is often used but it has high conduction loss. Copper is
preferred for antennas and was tried by [22,23]. Also, 60% efficiency was obtained
for a 94 GHz 18  18 slotted waveguide array. Alignment of the layers during
bonding is critical. See also reference [24] and Chapter 10.
Solbach [25] reported work on linear and planar arrays at 40 and 60 GHz. The
slots were photoetched on a thin (0.1 mm) copper plate. The slot lengths and widths
showed a spread of about 50100 mm, mainly due to underetching. A thinner plate
or thin metallisation on a dielectric carrier was suggested to improve this. Such a
dielectric carrier was used by Zhao et al. [26] for a ten-element slotted waveguide
array at 93 GHz. They used a teflon substrate (Rogers 5880) with a thickness of
0.254 mm. The slot plate was photoetched and bonded to the milled waveguide
channel using a conductive resin. A related technology is also used in stripline fed
slotted arrays; cf. Section 9.7.
324 Slotted waveguide array antennas: theory, analysis and design

11.2.2 Carbon fibre-reinforced plastic materials

Plastic materials with plated conducting surfaces can be used, similar to printed
circuit techniques; cf. stripline arrays, Section 9.7. The waveguide sidewalls can be
formed by rows of plated-through holes. For applications requiring low weight and
thermal stability CFRP is suitable. CFRP reflects microwaves and is good for
reflector antennas at not too high frequencies (up to about 15 GHz), but the con-
ductivity is not sufficient for use in waveguides. A possible solution for slotted
waveguide designs is to use metallised carbon fibre composites.
Several metallisation techniques are described in the literature [27]. In the gal-
vanic method a thin copper layer is chemically deposited on the composite with the
aid of a catalyst. This thin layer is then reinforced by galvanic metallisation. For
corrosion protection a thin layer of gold can be added. Figure 11.6 shows samples of
CFRP waveguides metallised with this process. Another method is to apply a metal
foil on the composite before curing. Still another method is to metallise a mandrel
and then apply a layer of pre-impregnated carbon fibres (the ‘prepreg’). After curing
the mandrel is removed. In some cases the three methods are combined.
An X-band mobile short range radar with a metallised carbon fibre slotted
waveguide array antenna is shown in Figure 11.7(a) and (b). The antenna is phase

Figure 11.6 Waveguides made of metallised carbon fibre composites (galvanic

method) [28]. Courtesy of Saab AB

(a) (b)

Figure 11.7 Mobile radar system: (a) radar unit; (b) array antenna with
metallised CFRP slotted waveguides [28]. Courtesy of Saab AB
 Manufacturing aspects 325

steered in elevation. The horizontal slotted waveguide runs are fed in the centre
with different slot spacings in the two halves in order to maximise the frequency
bandwidth [29]. The centre feeding technique is discussed in more detail in
Chapters 7 and 9.
With composites it is possible to form several waveguides in one operation, for
example, to produce integrated subpanels; an example is shown in Figure 11.8.
When metallised this design performs electrically as its aluminium counterpart, but
with improved characteristics in terms of weight and thermal stability. The same
technique has also been used for airborne applications [30].
Low weight and thermal stability are critical parameters for space applications.
A metallised CFRP composite material was therefore chosen for the SAR antennas
for the European Earth Resource Satellites ERS-1 and ERS-2. The C-band SAR
antenna is 10 m  1 m when deployed (Figure 3.4(b)). The antennas were manu-
factured by Dornier Satellitensysteme in Germany [32]. Metallised carbon fibre
arrays have also been used for X-band active SAR systems, for example, in the
TerraSAR-X Radar satellite [33]. The active radiating element with the dual CFRP
waveguides (for dual polarisations) is shown in Figure 11.9.
Large parts of modern aircraft structures are made of CFRP panels. It has been
suggested that such panels could be designed to function as, for example, radar
antennas and loadbearing structures at the same time. Several studies along these
lines have been published [34,35], addressing RF performance and structural
performance of test panels: SWASS = Slotted Waveguide Antenna Stiffened

Figure 11.8 An integrated ridge waveguide slotted subpanel made in metallised

CFRP [31]. Courtesy of Saab AB

Figure 11.9 The active radiating element of the TerraSAR array with a
combination of edge slots in the rectangular waveguide and
longitudinal slots in the ridge waveguide, both in CFRP technology.

2003 IEEE. Reprinted from [33], with permission
326 Slotted waveguide array antennas: theory, analysis and design

Structure. The tested panels were not metallised. The reported transmission losses
were somewhat high and probably not acceptable for most applications. However,
losses depend on the orientation of the carbon fibres in the waveguides. If the
technology is successful the result could be large high-performance conformal
slotted waveguide array antennas with minimum impact on drag and weight.

11.2.3 Microfabrication
Microelectronics packaging is a key element in the production of devices and
systems that are becoming indispensable in our modern life. Miniaturisation is
taken for granted in portable electronic devices such as mobile phones, Wi-Fi
networks, automotive radar, etc. The trend towards higher frequencies makes the
design and fabrication a challenging task. Circuits on ceramic substrates have been
used extensively and efforts to combine this technology with antennas have
been reported. The low-temperature co-fired ceramics (LTCC) technology has been
studied for slotted waveguide array antennas at 79 GHz [36] and even at 140 and
270 GHz [37]. One problem is high transmission losses in the substrate material.
More about the LTCC process and the materials used in [38,39].
An eight-element slotted waveguide array antenna was developed for 300 GHz
using four silicon layers and a UV-sensitive epoxy-based photoresist (SU-8). The
contact between layers proved to be a critical factor [40]. The slot length at this
frequency is indeed small, in fact only 0.54 mm.
Photoetching of slots in a thin copper plate and brazing the plate onto a milled
waveguide channel at 94 GHz has been successful [19]. It is a less complicated process,
but lacks the possibility to integrate active and passive circuits in the same structure.
The PolyStrata (literally ‘many layers’) process (http://nuvotronics.com) has
been used for fabrication of micro-coaxial lines and components [41] and recently
also for slotted waveguide arrays at millimetre-wave frequencies, 130180 GHz
[42]. According to this process a photo resist pattern is applied onto a silicon wafer.
Next, a copper layer is electroplated on the wafer. Additional photoresist is applied
(defining the waveguide walls) and a copper layer is again deposited, etc. About 15
layers can be built up this way. The final layer includes the slot elements. At the
last step the photoresist is rinsed away through release holes. For corrosion pro-
tection the copper surface can be gold plated.
The PolyStrata process is illustrated schematically in Figure 11.10 for fabrica-
tion of a micro-coaxial line using five layers [43]. (For a slotted waveguide array no
centre conductor is needed.) Two fabricated slotted waveguide arrays for 130180
GHz with 10 and 20 elements, respectively, are shown in Figure 11.11. The technol-
ogy has been applied to slotted waveguide arrays operating up to 300 GHz [44].
The PolyStrata process is very promising for low-cost fabrication of arrays
and components at very high frequencies. Another possible fabrication method in
the future might be the 3D printing technique, so-called ‘Additive Manufacturing’
[45,46].
 Manufacturing aspects 327

S1 S2 S3 S4

S5 S6 S7 S8

S9 S10 S11 S12

Photoresist
Dielectric
Copper
Silicon

Figure 11.10 The PolyStrata process (five-layer) for fabrication of a micro-

coaxial line. Reprinted from N. Ehsan, PhD thesis, Univ. Colorado
2010 [43]

Figure 11.11 The ends of two PolyStrata waveguide arrays (130180 GHz).
Note the small holes that, together with the slots, are used for
releasing the lossy photo resist.
IEEE 2012. Reprinted from [42],
with permission
328 Slotted waveguide array antennas: theory, analysis and design

References
[1] Ashmead D. ‘Optimum design of linear arrays in the presence of random
errors’. IRE Transactions on Antennas and Propagation. 1952;4(1):81–92.
[2] Elliott R. S. ‘Mechanical and electrical tolerances for two-dimensional
scanning antenna arrays’. IEEE Transactions on Antennas and Propagation.
1958;6(1):114–120.
[3] Rondinelli L. A. ‘Effects of random errors on the performance of antenna
arrays of many elements’. IRE International Convention Record. 1959;
1(Part 1):174–189.
[4] Hsiao J. K. ‘Constraints on low sidelobe phased array’. IEEEAntennas and
Propagation International Symposium, Albuquerque, NM, USA, 1982,
pp. 687–690.
[5] Lee J., Lee Y., Kim H. ‘Decision of error tolerance in array element by the
Monte Carlo method’. IEEE Transactions on Antennas and Propagation.
2005;53(4):1325–1331.
[6] Rengarajan S., Zawadski M. S., Hodges R. E. ‘Waveguide-slot array antenna
designs for low-average-sidelobe specifications’. IEEE Antennas and
Propagation Magazine. 2010;52(6):89–98.
[7] Bailin L. L., Ehrlich M. J. ‘Factors affecting the performance of linear arrays’.
IRE Transactions on Antennas and Propagation. 1952;1(1):85–106.
[8] O’Neill H. F., Bailin L. L. ‘Further effects of manufacturing tolerances on
the performance of linear shunt slot arrays’. IRE Transactions on Antennas
and Propagation. 1952;4(1):93–102.
[9] Stegen R. J. ‘ Slot radiators and arrays at X-band’. IRE Transactions on
Antennas and Propagation. 1952;1(1):62–84.
[10] Kaminov I. P., Stegen R. J. Waveguide slot array design, Hughes Aircraft
Company, Techn. Memo No. 348, July 1954 (available as NTIS Report AD
63600).
[11] Karlsson I. ‘Applications of waveguide arrays in commercial and military
radars’. Antenna Applications Symposium, Allerton Park, IL, USA, 1993,
pp. 1–22.
[12] Brady M. M. ‘A plea for clarity in waveguide designation’. IEEE Transac-
tions on Microwaves, Theory and Techniques. 1965;13(2):247–249.
[13] Conciauro G., Bressan M., Zuffada C. ‘Waveguide modes via an integral
equation leading to a linear matrix eigenvalue problem’. IEEE Transactions
on Microwaves, Theory and Techniques. 1984;32(11):1495–1504.
[14] Falk K. ‘Admittance of a longitudinal slot in a ridge waveguide’. IEE Pro-
ceedings H - Microwaves, Antennas and Propagation. 1988;135(4):263–268.
[15] Sikora L., Womack J. ‘The art and science of manufacturing waveguide
slot-array antennas’. Microwave Journal. 1988;31(6):157–162.
[16] Muhs H. P. ‘Mm-wave antenna’. Microwave Journal. 1985;28(7):91–194.
[17] Anderson T., Yun-Li-Ho, Michalski J. ‘Design and manufacturing tech-
niques for planar slot array antennas for a variety of radar application’.
Proceedings of the IEEE International Radar Conference Radarcon, 1998,
pp. 337–342.
 Manufacturing aspects 329

[18] Singh A. K. ‘A low cost, low side lobe and high efficiency non-orthogonally
coupled slotted waveguide array antenna for monopulse radar tracking’.
IEEE Antennas and Propagation International Symposium, Washington DC,
USA, 2005, pp. 732–735.
[19] Rao B. R. ‘94 Gigahertz slotted waveguide array fabricated by photolitho-
graphic techniques’. IEEE Antennas and Propagation International Sympo-
sium, Houston, TX, USA, 1983, pp. 688–691.
[20] Farrar F. G. ‘Millimeter wave W-band slotted waveguide antennas’. IEEE
Antennas and Propagarion International Symposium, Los Angeles, CA,
USA, 1981, pp. 436–439.
[21] Ando M., Hirokawa J. ‘Single-layer slotted waveguide arrays for DBS
reception and higher frequency applications’. Electromagnetics. 1999;19(1):
23–48.
[22] Hirokawa J., Zhang M., Ando M. ‘Millimeter waveguide fabrication to reduce
transmission loss by diffusion bonding, light-curing resin or dielectric par-
tially-filling’. Asia-Pacific Microwave Conference, APMC. 2008, pp. 1–4.
[23] Zhang M., Hirokawa J., Ando M. ‘Fabrication of a slotted waveguide array
at 94 GHz by diffusion bonding of laminated thin plates’. IEEE Antennas
and Propagation Society International Symposium, Charleston, SC, USA,
June 2009, pp. 1–4.
[24] Ando M., Zhang M., Lee J., Hirokawa J. ‘Design and fabrication of milli-
meter wave slotted waveguide arrays’. 4th European Conference on Anten-
nas and Propagation (EuCAP), Barcelona, Spain, 2010, pp. 1–6.
[25] Solbach K. ‘Some millimeter-wave slotted array antennas’. 14th European
Microwave Conference, Liege, Belgium, 1984, pp. 181–186.
[26] Zhao G., Xu H., Chen Z., Sun H., Lv X. ‘A W-band waveguide slot array
antenna based on print technology’. International Conference on Microwave
and Millimeter Wave Technology (ICMMT), 2012, Vol. 3, pp. 1–4.
[27] Jönsson D. ‘Metallisation methods & applications’. Proceedings of the ESA
Symposium on Space Applications of Advanced Structural Materials,
ESTEC, Noordwijk, March 1990, pp. 35–38.
[28] Dahlsjö O., Ljungström B., Magnusson H. ‘Fibre-reinforced plastic compo-
sites in sophisticated antenna designs’. Ericsson Review. 1987;64(2):50–57.
[29] Karlsson E. R. ‘Waveguide element for an electronically controlled radar
antenna’. US Pat. 4788 552, 1988.
[30] Brunzell S., Magnusson H. ‘Lightweight CFRP waveguide array’. Pro-
ceedings Military Microwaves, London, UK, October 1984, pp. 305–309.
[31] Josefsson L., Derneryd A., Lagerlöf R. ‘Electronic scanning with slotted
waveguide arrays’. Proceedings of the Radarcon 90 Conference, Adelaide,
Australia, April 1990.
[32] Petersson R., Bonnedal M. ‘The slotted waveguide arrays of the European
remote sensing satellites ERS-1 and ERS-2’. Electromagnetics. 1999;19(1):
77–89.
[33] Stangl M., Werninghaus R., Zahn R. ‘The TerraSAR-X active phased array
antenna’, IEEE International Symposium on Phased Array Systems and
Technology, Boston, MA, USA, October 2003, pp. 70–75.
330 Slotted waveguide array antennas: theory, analysis and design

[34] Sabat J. W. Structural Response of the Slotted Waveguide Antenna Stiffened

Structure Components under Compression. M.Sc. Thesis, Air Force Institute
of Technology, Wright-Patterson Air Force Base, OH, USA, March 2010.
[35] Callus P. J., Nicholson K. J., Bojovschi A., Ghorbani K., Baron W., Tuss J.
‘A planar antenna array manufactured from carbon fibre reinforced plastic’.
28th International Congress of the Aeronautical Sciences, ICAS 2012,
Brisbane, Australia, September 2012.
[36] Bauer F., Menzel W. ‘79 GHz LTCC – integrated antennas’, 2013 Interna-
tional Workshop on Antenna Technology (IWAT), 2013, pp. 158–161.
[37] Xu J., Chen Z. N., Qing X. ‘270-GHz LTCC-integrated strip-loaded linearly
polarised radial line slot array antenna’. IEEE Transactions on Antennas and
Propagation. 2013;61(4):1794–1801.
[38] Kulke R., Rittweger M., Uhlig P., Günner C. ‘LTCC – Multilayer ceramic
for wireless and sensor applications’. http://www.ltcc.de/downloads/rd/pub/
10-doc-plus-engl-2001.pdf Retrieved 2017–04–02.
[39] Ullah U., Ain M. F., Mahyuddin N. M., et al. ‘Antenna in LTCC technolo-
gies: a review and the current state of the art’. IEEE Antennas and Propa-
gation Magazine. 2015;57(2);241–260.
[40] Wang Y., Ke M., Lancaster M. J. ‘Micromachined 300-GHz SU-8-based
slotted waveguide antenna’. IEEE Antennas and Wireless Propagation Let-
ters. 2011;10:573–576.
[41] Ehsan N., Cullens E., Vanhille K., et al. ‘Micro-coaxial lines for active
hybrid-monolithic circuits’. IEEE MTT-S International Microwave Sympo-
sium Digest, 2009, pp. 465–468.
[42] Cullens E. D., Ranzani L., Vanhille K. J., Grossman E. N., Ehsan N.,
Popović Z. ‘Micro-fabricated 130–180 GHz frequency scanning waveguide
arrays’. IEEE Transactions on Antennas and Propagation. 2012;60(8):
3647–3653.
[43] Ehsan N. Broadband Microwave Lithographic 3D Components, Ph.D thesis,
University of Colorado, 2010.
[44] Boryssenko A., Vanhille K. ‘300 GHz microfabricated waveguide slotted
arrays’. 39th International Conference on Infrared, Millimeter and Terahertz
Waves (IRMMW-THz). 2014, pp. 1–2.
[45] Nayeri P., Liang M., Sabory-Garcia R. A., et al. ‘3D printed dielectric
reflectarrays: low-cost high-gain antennas at sub-millimeter waves’. IEEE
Transactions on Antennas and Propagation. 2014;62(4):2000–2008.
[46] Nassar I. T. and Weller T. M. ‘An electrically-small, 3d-cube antenna fab-
ricated with additive manufacturing’. IEEE Topical Meeting on Silicon
Monolithic Integrated Circuits in RF Systems (SIRF), 2013, pp. 162–164.
 Chapter 12
Outlook for the future

In this book we have analysed the design of several types of slotted waveguide
array antennas, from theories and optimisation to applications and manufacturing
techniques.
In this last chapter we will discuss the current status in the field and look at
new technologies and new applications recently presented and researched. Still, in
this short overview it is not possible to mention all the details of the evolving
field; the reader is referred to the respective chapters and the references for more
information.

12.1 Defence systems

The development of slotted waveguide array antennas started during the Second
World War driven by the need for radar to detect and track enemy targets. The
development continued during the following years, and the technology was also
applied to non-military systems: communication [1], remote sensing [2,3], navi-
gation [4], weather radar [5], etc.
The flat plate antenna succeeded the reflector antenna and became the standard
antenna type in the nose of military fighter aircraft in combination with a central
transmitter and receiver. The development in the microelectronic area of today has
led to the point where transmit/receive modules can be miniaturised and integrated
with the individual radiating elements of phased array antennas. These AESA
systems (Active Electronically Steered Arrays) enable advanced signal processing
and ‘graceful degradation’. Since the systems are quite complex and costly they are
primarily considered for the most advanced multifunction systems. Still, they are
today being implemented in new fighter versions as well as in some ground and
naval systems [6–8]. We also find examples where (passive) slotted waveguide
subarrays are the elements in systems with the signal processing acting on the
subarray level [9,10]. This passive electronically steered array (PESA) might seem
as a good cost/complexity compromise, especially when the electronic scanning is
mainly required in one dimension only [11].
A sometimes important aspect is the protection of the receiver circuits from
(un)intentional external microwave radiation. The AESA radiating elements have
a broad beam and are not protected by low sidelobes as formed in arrays with a
central receiver. Looking at the other side, creating high-power radiation from
332 Slotted waveguide array antennas: theory, analysis and design

slotted waveguide arrays has been discussed as a possible high-power microwave

weapon [12,13].
An interesting idea is the inclusion of slotted waveguide arrays in the skin of
aircraft or other vehicles: conformal arrays. The waveguides would serve as
structural members and antenna elements at the same time [14]. This is to be
compared with today’s systems like AWACS [15] and ERIEYE [16] where a large
array antenna is carried outside the aircraft body.

12.2 Millimetre waves

In 1984 a literature search revealed only three papers on millimetre-wave slotted

waveguide array antennas (at 60 and 94 GHz) [17]. Today, work on slotted
waveguide arrays at frequencies as high as 300 GHz (0.3 THz) [18] and even higher
has been reported. An LTCC-integrated radial line slot array for 270 GHz was
designed and evaluated by Xu et al. [19].
The benefits of going to millimetre-wave frequencies include smaller
dimensions, narrow beams, large signal bandwidths, high data rates, sometimes
covert communication due to atmospheric losses, etc. However, higher fre-
quencies mean smaller antenna apertures, and hence weaker signals. This leads to
a need for higher antenna gain and arrays both for transmit and receive. Typical
millimetre-wave applications include automotive radar in 76–81 GHz range
[20,21], planetary landing radar in 138–180 GHz [22,23], point-to-point wireless
links [24], etc.
Microwave and millimetre-wave signals penetrate optically opaque materials
such as composites, ceramics, concrete, wood, clothing and interact with their
interior structures. Imaging techniques using various forms of modulation and
signal processing have demonstrated both 2D and 3D images at 24 GHz as shown
by Gahsr et al. [25]. In their ‘microwave camera’ 576 modulated sidewall slot
elements were used. Applications include non-destructive testing and evaluation
(NDT&E) of materials, medical imaging, airport security scanners, concealed
weapon detection, etc.

12.3 Spaceborne systems

SAR stands for synthetic aperture radar, a technology used in airborne and space-
borne systems. By coherent signal processing along the flight path a large
(synthetic) aperture is created in this direction. Together with a large physical
aperture across the flight path very high angular resolution is possible. In
one application [26] scatterometer antennas provide ocean wind data and soil
moisture content. The technology is used in many remote sensing satellites such as
the C-band instruments ERS-1 and ERS-2; cf. Figure 3.4(b). The ERS satellites
were operational almost 20 years [27]. The major antenna technology for these
systems is slotted waveguide arrays. New developments include dual polarisation,
dual and triple beams, beam scanning, X-band and higher frequencies, and also
 Outlook for the future 333

millimetre waves. The new spaceborne SAR systems will provide improved high-
resolution images on a global scale for monitoring the environment, wind speed,
ocean currents, land mapping, climate change, etc. Together with advanced signal
processing techniques, digital beam forming and interferometry, the available
information content will increase at least one order of magnitude [28].

12.4 Communication

Communication based on millimetre-wave radio systems is of increasing impor-

tance. Telecom operators need high flexibility when rerouting of the network
becomes necessary. Narrow beam antennas, point-to-point, are used for connecting
base stations to the main switching office (backhaul). Slotted waveguide array
antennas are the best solution in many cases, enabling bandwidths of about 10%
and an attractive form factor. Short millimetre-wave links are more flexible and
often less costly than conventional cables or fibres.
Steerable beams are desired in future high-capacity wireless local area net-
works. We find many interesting solutions here: adaptive beams, MIMO in WLAN
(WiFi, WiGig) and also in cellular networks such as the coming 5G systems [29].
For many of these applications slotted waveguide arrays have been studied [30,31],
but there are naturally also other technologies that can be considered. The low loss
of waveguide antennas favours this technology.

12.5 Manufacturing
The challenges are to reduce cost and complexity. We have come a long way, for
example, in integrating the feeding waveguides in the same plane as the planar
slotted waveguide array [32]. Traditionally, the antenna used to be a separate unit,
connected to the transmit/receive front end via coaxial cables or waveguides. At
higher frequencies the active front end can be integrated with the slotted waveguide
antenna using the same materials and processes, for example, LTCC, the SU-8
process or the PolyStrata“ process [22,33]. One example is the LTCC substrate-
integrated slotted waveguide array antenna. It is essential to master the tolerance
requirements and possibly modify the design to match the available fabrication
capability, especially at millimetre-wave bands [34].

12.6 Design tools and software

In the early days of slotted waveguide array antennas, say before 1960, the designs
were based primarily on experimental data. In the following years electromagnetic
models and software for design and analysis were used more and more. This
development is reflected by several examples in this book. The first comprehensive
array design method that included external and internal higher-order mode mutual
coupling was due to Professor Elliott in the 1980s; see in particular Chapters 6–8.
334 Slotted waveguide array antennas: theory, analysis and design

The interest in computational electromagnetics increased even more in the

1990s and several commercial software tools suitable for analysis and design of
slotted waveguide arrays became available. The computer codes employ various
numerical and asymptotic techniques such as the Method of Moments (MoM), the
finite element method (FEM), the mode matching technique (MM), the finite dif-
ference time domain technique (FDTD), etc. [35–38]. As computer software
becomes more and more powerful the size and complexity of the structure to be
analysed can be allowed to increase. The continued development in computers and
software leads to better tools for design and analysis, at the same time reducing the
need for expensive testing and measurements.
It shall be emphasised, however, that computers do not design antennas. It still
requires a great deal of engineering efforts, for example,
● analysing system requirements
● optimising antenna configuration
● choice of materials
● manufacturing methods
● testing and verification
There exists today a plethora of computer software for design of array antennas.
The following listing provides a few examples; much more information is available
in the cited papers as well as on the internet.

12.6.1 HFSS
HFSS stands for high-frequency structural simulator, originally developed by
Professor Z. Cendes around 1980 [39]. The code was further developed by Ansoft
Corporation and is today marketed by ANSYS [40]. It is a FEM and includes an
integral equation solver (MoM) which is convenient for slot array problems.

12.6.2 XFdtd
The XFdtd by Remcom [41] uses the FDTD technique, first proposed in 1966 [42].
The problem geometry is simulated by a mesh which can be directly connected to
lumped circuit elements. This has become attractive for simulations of sub-
assemblies that include both MMIC circuits and antenna parts.

12.6.3 CST Microwave Studio

CST Microwave Studio [43] is another software tool, also based on FDTD, for
analysis of high-frequency devices such as antennas, filters, couplers, planar and
multi-layer structures. It has adaptive meshing and includes both a frequency
domain solver and a time-domain solver.

12.6.4 WASP-NET
WASP-NET [44] combines several methods: MM, the FEM and the MoM for
design of slotted waveguide arrays including feed networks.
 Outlook for the future 335

12.6.5 SWANTM
The software SWANTM [45] is a powerful CAD tool for the design and analysis of
very large slotted waveguide arrays. It is based on Elliott’s theory with a number of
significant extensions and improvements. It has a graphical user interface so that
the user can input design and analysis parameters easily.

References
[1] Grabherr W., Huder B. ‘Omni-directional slotted waveguide array antenna’.
29th European Microwave Conference, Munich 1999, pp. 134–137.
[2] Ekengren B. ‘Ericsson SLAR – airborne radar for maritime surveillance’.
Ericsson Review. 1979;56(4):151–157.
[3] Alves A. M. P., Duplat D. N., de Oliveira L. P., Hernández-Figueroa H. E.
‘Aerotransported radar antenna for oil spills monitoring antennas and
propagating’. 4th European Conference on Antennas and Propagation
(EuCAP), Barcelona, April 2010, pp. 1–3.
[4] Morwing B. ‘AGA-ERICON – a marine radar beacon’. Ericsson Review.
1981;58(4):180–187.
[5] Anderson T., Yun-Li-Hou, Michalski J. ‘Design and manufacturing techni-
ques for planar slot array antennas for a variety of radar application’.
Proceedings of the IEEE Radar Conference, Radarcon 98, Dallas, TX, USA,
May 1998, pp. 337–342.
[6] Brookner E. ‘Phased arrays and radars – past, present and future’. Microwave
Journal. 2006;49(1):1–12.
[7] Kopp C. ‘Evolution of AESA radar technology’. Microwave Journal.
2012;55(8):3–12.
[8] Pendergast S. L. ‘Recent advances in radar technology’. Microwave Journal.
2015;58(9):1–14.
[9] Karlsson I. ‘Applications of waveguide arrays in commercial and military
radars’. Proceedings of the 1993 Antenna Applications Symposium, Allerton
House, IL, USA. pp. 1–22, 1993.
[10] EMS White Paper Passive phased arrays for radar antennas. EMS Tech-
nologies Inc., Norcross, GA, USA, December 2005.
[11] Lewis D. J., Lee J. R., McCarty D. K. ‘A single-plane electronically scanned
antenna for airborne radar applications’. Phased-Array Antenna Symposium,
Polytechnic Institute of Brooklyn, Farmingdale NY, 2–5 June 1970.
[12] Coburn W., Litz M., Miletta J., et al. A Slotted-Waveguide Array for
High-Power Microwave Transmission. Army Research Laboratory, Rpt. MD
20783-1197, January 2001.
[13] Baum C. E. Sidewall Waveguide Slot Antenna for High Power. Sensor and
Simulation Note No. 503, Univ. New Mexico. August 2005.
[14] Callus P. J., Nicholson K. J., Bojovschi A., Ghorbani K., Baron W., Tuss J.
‘A planar antenna array manufactured from carbon fibre reinforced plastic’.
336 Slotted waveguide array antennas: theory, analysis and design

28th International Congress of the Aeronautical Sciences, ICAS 2012,

Brisbane, Australia, September 2012.
[15] Skillman W. A. ‘Environmental effects on airborne radar performance’.
IEEE International Microwave Symposium, 5–10 June, 2011, Baltimore
MD, USA, pp. 1–4.
[16] Ahlbom S., Andersson P., Lagerlöf R. ‘A Swedish airborne early warning
system based on the Ericsson ERIEYE radar’. Ericsson Review. 1995;72(2):
54–63.
[17] Solbach K. ‘Some millimeter-wave slotted array antennas’. Proceedings of
the 14th European Microwave Conference, 1984, pp. 181–186.
[18] Wang Y., Lancaster M. J. ‘A micromachined centre-fed slotted waveguide
antenna for mm-wave applications’. IEEE MTT International Microwave
Workshop Series on Millimetre wave Wireless Technology and Applications
(IMWS), 2012, pp. 1–3.
[19] Xu J., Chen Z. N. Qing X. ‘270 GHz LTCC-integrated strip-loaded linearly
polarized radial slot array antenna’. IEEE Transactions on Antennas and
Propagation. 2013;61(4): 1794–180.
[20] Hirokawa J., Ando M. ‘76 GHz post-wall waveguide fed parallel plate slot
arrays for car-radar applications’. Proceedings of the IEEE Antennas and
Propagation Symposium, July 2000, pp. 98–101.
[21] Menzel W. ‘Antennas in automobile radar’. In Handbook of Antenna Tech-
nologies, Springer Verlag, Berlin Heidelberg 2015.
[22] Cullens E.D., Ranzani L., Vanhille K. J., Grossman E. N., Ehsan N., Popovic Z.
‘Micro-fabricated 130–180 GHz frequency scanning waveguide arrays’. IEEE
Transactions on Antennas and Propagation. 2012;60(8):3647–3653.
[23] Samoska L., Kangaslahti P., Pukala D., Sadowy G., Pollard B., Hodges R.
‘A G-band 160 GHz T/R module concept for planetary landing radar’.
Proceedings of the 36th European Microwave Conference, Manchester UK,
September 2006, pp. 757–760.
[24] Frenzel L. ‘Millimeter waves will expand the wireless future’. Electronic
Design, March 6, 2013.
[25] Gahsr M. T., Abou-Khousa M. A., Kharkovsky S., Zoughi R., Pommerenke D.
‘Portable real-time nicrowave camera at 24 GHz’. IEEE Transactions on
Antennas and Propagation. 2012;60(2):1114–1125.
[26] Magnusson P., Di Salvo M., Scarchilli C. ‘A dual polarised slotted waveguide
antenna for satellite based wind scatterometet instruments’. 7th European
Conference on Antennas and Propagation (EuCAP), Gothenburg, Sweden,
April 2013, pp. 3449–3452.
[27] van’t Klooster. ‘ERS-1, European remote-sensing satellite was launched
20 years ago’. 21st International Crimean Conference on Microwave &
Telecommunication Technology, September 2011.
[28] Moreira A. ‘A golden age for spaceborne SAR systems’. 20th International
Conference on Microwaves, Radar and Communication (MIKON), 2014, pp. 1–4.
[29] Dussopt L., Bouayadi O. E., Luna J. A. Z., Dehos C., Lamy Y. ‘Millimeter-
wave antennas for radio access and backhaul in 5G heterogenous mobile
 Outlook for the future 337

networks’. 9th European Conference on Antennas and Propagation

(EuCAP), Lisbon, Portugal, April 2015, pp. 1–4.
[30] Ohira M., Miura A., Ueba M. ‘60-GHz wideband substrate-integrated-
waveguide slot array using closely spaced elements for planar multisector
antenna’. IEEE Transactions on Antennas and Propagation. 2010;58(3):
993–998.
[31] Wen Y., Wang B., Ding X. ‘Wide-beam SIW-slot antenna for wide-angle
scanning phased array’. IEEE Transactions on Wireless Propagation Letters.
2016;15:1638–1641.
[32] Zhang M. Full-Structure Analysis and Design of Single-Layer Slotted
Waveguide Arrays with their Practical Realisation, Doctoral Dissertation,
Tokyo Institute of Technology, Dept. of Electrical and Electronic Engi-
neering, March 2008.
[33] Wang Y., Ke M., Lancaster M. J. ‘Micromachined 300-GHz SU-8-based
slotted waveguide antenna’. IEEE Transactions on Wireless Propagation
Letters. 2011;10:573–576.
[34] Xu J., Chen Z. N., Qing X., Hong W. ‘140-GHz SIW LTCC antenna array
using a large via-fenced and slotted dielectric loading’. 7th European
Conference on Antennas and Propagation (EuCAP), Gothenburg, April
2013, pp. 2861–2864.
[35] Mittra, R. (ed.). Numerical and Asymptotic Techniques in Electromagnetics,
Springer Verlag, Berlin, 1975.
[36] Itoh, T. (ed.). Numerical Techniques for Microwave and Millimeter-Wave
Passive Structures. Wiley Interscience, New York, 1989.
[37] Peterson, A. F., Ray, S. L., Mittra, R. Computational Methods for Electro-
magnetics. Wiley-IEEE, December 1997.
[38] Weiland T., Munteanu I. ‘State of the art and the future in modelling RF
components’. 1st European Conference on Antennas and Propagation
(EuCAP), Nice, France. November 2006.
[39] Cendes Z. ‘The development of HFSS’. USNC-URSI Radio Science Meeting.
Fajardo, Puerto Rico, USA, June 2016. pp. 39–40.
[40] http://www.ansys.com. Retrieved March 2018.
[41] http://www.remcom.com. Retrieved March 2018.
[42] Yee K. S. ‘Numerical solution of initial boundary value problems involving
Maxwell’s equations in isotropic media’. IEEE Transactions on Antennas
and Propagation. 1966;14(3):302–307.
[43] https://www.cst.com. Retrieved March 2018.
[44] http://www.mig-germany.com. Retrieved March 2018.
[45] http://www.swan-soft.com. Retrieved March 2018.
 Appendix
Standard waveguide frequency bands

Standard rectangular waveguides are listed with their common designations and
useful frequency ranges. The waveguide inner dimensions are given in mm.

Band letter* Waveguide Frequency Inner dimensions

designation** range (GHz) (mm)

EIA WR- IEC 153-2

L 650 R 14 1.15–1.72 165.10
82.55
510 R 18 1.45–2.2 129.54
64.77
430 R 22 1.7–2.6 109.22
54.61
340 R 26 2.2–3.3 86.36
43.18
S 284 R 32 2.6–3.95 72.14
34.04
229 R 40 3.3–4.9 58.17
29.21
C 187 R 48 3.95–5.85 47.55
22.15
C 159 R 58 4.9–7.05 40.39
20.19
137 R 70 5.85–8.2 34.85
15.80
112 R 84 7.05–10.0 28.50
12.62
X 90 R 100 8.2–12.4 22.86
10.16
Ku 62 R 140 12.4–18.0 15.80
7.899
51 R 180 15.0–22.0 12.95
6.477
K 42 R 220 17.6–26.7 10.668
4.318
Ka 28 R 320 26.5–40.0 7.112
3.556
Q 22 R 400 33.0–50.0 5.690
2.845
U 19 R 500 40–60 4.775
2.388
V 15 R 620 50–75 3.759
1.880
E 12 R 740 60–90 3.099
1.549
W 10 R 900 75–110 2.54
1.27
F 8 R 1200 90–140 2.032
1.016
6 R 1400 110–170 1.651
0.8255
G 5 R 1800 140–220 1.295
0.6477
3 R 2600 220–330 0.8636
0.4318
*
Several systems for band letter designations exist [1,2]. This table is adapted from [3].
**
EIA ¼ Electronic Industries Association, New York; IEC ¼ International Electrotechnical
Commission, Geneva.
340 Slotted waveguide array antennas: theory, analysis and design

References
[1] Brady M. M. ‘A plea for clarity in waveguide designations’. IEEE Transac-
tions on Microwave Theory and Techniques. 1965;13(2):247–249.
[2] Lewin L. ‘Letter symbols to designate microwave bands’. IEEE Transactions
on Microwave Theory and Techniques. 1964;12(5):551.
[3] https://en.wikipedia.org/wiki/waveguide_(electromagnetism)#References.
Retrieved 25 Oct 2017.
 Index

active impedance 44, 195, 197 numerical examples 256

Additive Manufacturing 326 scattering by a single slot in a
AESA systems (Active Electronically planar ground plane 250–2
Steered Arrays) 331 scattering from an array of slots
airborne early warning (AEW) system 252–5
223–4 spatial filters 256–8
aluminium/aluminium alloys 322–3 second-order lobes 258
antenna systems 221 analysis 258–9
conformal slotted waveguide arrays reducing the secondary lobes
239–42 259–65
dual polarised slotted waveguide slotted waveguide array antennas,
arrays 232–6 bandwidth of 272
frequency scanned slotted linear resonant array antenna
waveguide arrays 223 275–7
centre-fed travelling array 231–2 mutual coupling 278–9
phase reversal travelling array overloading 277–8
226–30 planar array and feeding
resonant series feed array 230–1 arrangements 279–81
travelling wave linear array slot antenna 273
225–6 travelling wave arrays 281
monopulse slotted waveguide waveguide-fed longitudinal slot
arrays 242–5 273–5
multiple beam slotted waveguide stripline slotted arrays 245
arrays 237–9 independent excitations 247–8
omnidirectional slotted waveguide aperture admittance 29–30,
array antennas 266 108–9, 111
pattern ripple problem 269–72 aperture in an infinite ground
slotted circular waveguide array plane 25
266–8 aperture admittance 29–30
slotted rectangular waveguide far field 26–7
array 266 general field expressions 27–8
phase scanned slotted waveguide problem formulation 25
arrays 221–3 radiated complex power 29
scattering from slotted waveguide array coupling 44, 47, 51
arrays 248 asymmetric ridge waveguide 223
feeding network 255–6 AWACS 332
342 Slotted waveguide array antennas: theory, analysis and design

Babinet’s principle 2 radial line antennas for monopulse

baffles 88, 186, 233–4, 262, 264–5 applications 303–4
bandwidth limitation single-layer radial line slot arrays
beam split 232 301–3
mutual coupling 278–9 slot coupling and attenuation
overloading 149–50, 277–8 coefficient of propagating
return loss 277 wave 296–7
sub-arraying 151–2, 155–6, circular waveguide array 266–8, 291
279, 281 coaxial line scanner 19
barline transmission lines 281 commercial computational
beacon antenna 268 electromagnetic (CEM) codes 96
beam squint effect 116, 122, 124 communication 333
bifurcated waveguide 232–3 complex power 7–8, 29–30
bistatic scattering cross section 252 complex waveguide feed network 245
Blass matrix 237–8 compound slot 128, 130, 184–5
breadboard panel 235–6 Compton solution 107
breadboard sub-array 233, 235 computational electromagnetics
Butler matrix 237 (CEM) code 178, 189, 334
butterfly lobes 152. 160, 186, 222, computed H-plane pattern 98–9, 115
258–9, 263 conductor backed coplanar waveguide
(CBCPW) 308
carbon fibre-reinforced plastic (CFRP) conformal slotted waveguide arrays
materials 246, 324–6 239–42
centred-inclined coupling conventional array architecture 160–1
slot 176–9 cosine integral function 32
higher-order mode coupling coupling slot models 143, 156, 164–6,
between 193–7 169, 175–6
and longitudinal radiating slots centred-inclined coupling slot 176–9
197–9 shunt-series coupling slot 179–81
theory for 195–6 transverse/transverse coupling slot
centre-fed array 95 181–2
centre-fed travelling array 231–2 CST Microwave Studio 334
Chebyshev pattern 119 curved slotted waveguide array
circularly polarised radial line slot antenna 241
antenna 291–6, 304 cylindrical slot array 269
circular parallel plate waveguides, slot cylindrical slotted waveguide array
arrays in 291 antenna 239–40
accurate analysis and design
techniques 304 defence systems 331–2
circularly polarised radial line slot DESA X-band SAR subpanel
antenna 291–6 demonstrator 236
design of slot coupling coefficients die-casting 322–3
299–300 dielectric cover, slot arrays with 191–3
linearly polarised radial line slot dielectric-filled rectangular
antenna 297–9 waveguides 307–8
 Index 343

dielectric filled waveguides 43, 95, vector and scalar potentials 9

101, 131 electric vector potential 10
dip brazing 323 electroplating 326
Dirac delta function concept 69 on dielectric materials 2
direct broadcast satellite (DBS) 291 element by element method 183
Dolph–Chebyshev array 113–16, 120, element pattern 115, 141, 270
259, 277 embedded/isolated 43
Doppler navigation 239 element spacing 116–17, 312
double-ridge waveguide 190 Elliott design equations 95, 99, 101,
dual plane monopulse antenna system 123, 209
243 first design equation 99–101
dual polarisation 183, 234 iterative design algorithm for
dual polarised slotted waveguide standing wave arrays 102–3
arrays 232–6 second design equation 101–2
dumbbell slot 273–4 Elliott design procedure 96, 126,
dyadic Green’s function 5, 13–15, 196 168–9, 209
improvement to 130–1
Eagle Scanner radiating dipoles 20 for iris-excited slotted waveguide
Earth Resources Satellite 20 arrays 189
edge condition 65, 69, 75, 203 for planar standing wave arrays 136–9
edge diffraction coefficient 57 6  6 planar array, example of
edge diffraction effects 199 141–5
edge slot array 126, 182–3, 237 design algorithm 139–41
edge wall slot 127, 182–3, 264 EMI Central Research Laboratories 17
electric field distribution 63, 72–6 E-plane element pattern 43
electric vector potential 10, 26, 48, 252 equivalence principle 12, 202
electroforming 322 equivalent shunt admittance 60, 62,
electromagnetic models 62–6, 76 93, 95
boundary problem 63–4 equivalent slot conductance 25, 40–4
choice of basis and testing ERIEYE 332
functions 65–6 error tolerance problem 320
zero wall thickness 64–5 ERS satellites 332
electromagnetic theory 5 European Earth Resource Satellites 325
boundary conditions 6–7 excitation errors 317
energy and power 7–8 external aperture interface 47
field equivalence principle 12 external mutual coupling 95, 99, 101,
Green’s functions 13 104–6, 193
dyadic 14–15
free-space 13 fabrication techniques 2
image principle 11 far field 26–7
Maxwell’s equations 5–6 fast wave antenna 226
reciprocity theorem 8 feed waveguides 128, 135, 137, 139
infinite region or a region bounded total slot conductance in radiating
by perfect conductors 9 waveguides and resistance in
source-free region 8 149–51
344 Slotted waveguide array antennas: theory, analysis and design

field equivalence principle 12 Hamadallah’s analysis equations 124,

finite difference time domain technique 210
(FDTD) 334 Hankel function 104, 270, 294
finite element method (FEM) 111–12, hard coupling 197–8
171, 334 Helmholtz equation 10, 13, 27
finite ground plane effects 55–7, 175, HFSS (high-frequency structural
199–200 simulator) code 126, 168–70,
finite wall thickness 64, 66–8, 74–5, 106 320, 334
flat plate antenna 20, 247, 331 higher-order mode coupling between
flat plate array 135 centred-inclined coupling
flat plate waveguide array antenna 232 slots 193
Fourier series 269–70 in the design of centred-inclined
Fourier transform 26–7, 104 coupling slots 196–7
free-space Green’s function 13 and longitudinal radiating slots
frequency scanned slotted waveguide 197–9
arrays 223 theory for 195–6
centre-fed travelling array 231–2 history 17
phase reversal travelling array early years 17
226–30 golden years 147–18
resonant series feed array 230–1 shapes of slotted waveguide array
travelling wave linear array 225–6 antennas 18–23
frequency selective slotted radome 257 waveguide slot antennas 18
frequency selective surfaces (FSS) H-plane array coupling 53–5
256–7 H-plane coupling for slots 44
full wave analysis techniques 111, 126 H-plane power dividers 155

gap waveguides, slot arrays in 310–13 image principle 11

generalised scattering matrix (GSM) incremental conductance 96, 126–8
model 123, 126, 131, 169–72 independent excitations 247–8
genetic algorithms (GA) 209 infinite array model, large array design
optimisation 169 using 156–60
optimised results 215–16 infinite array mutual coupling model
of planar standing wave array using 136, 145, 169, 307
MoM analysis 211–13 expression 146–8
of travelling wave linear array integral equation solution 68
209–11 external field 70–1
geometrical theory of diffraction internal field 68–70
(GTD) 199 matrix elements 71–2
glacier and land ice surface topography internal aperture interface 47
interferometer (GLISTIN) 151 internal TE20 mode coupling 96, 99,
grating lobe lattice 253 101, 109–11
Green’s function 13–15, 68–70, 109, iris-excited longitudinal slot 186–8
191, 196, 202, 304 Elliott’s design procedure for
ground plane edges 43, 56, 199–200 iris-excited slotted waveguide
Gruenberg lobes 258 arrays 189
 Index 345

iris-excited slot in a rectangular slot array model, accuracy of 97

waveguide 264 ground plane 97
iris-excited slotted waveguide arrays, mutual coupling 97–9
Elliott’s design procedure for network model 97
189 standing wave array, design example
isolated (or self) impedance/ of 112
admittance 44 design of a –30 dB sidelobe-level
isolated coupling 44, 46–7, 51 Dolph–Chebyshev array 113–16
iterative design algorithm, for standing slot data needed for design
wave arrays 102–3 algorithm 112–13
ITT Gilfillan AN/SPS-48 3D radar travelling wave array 116
antenna 229 design algorithm 118–19
design equations for 118
Keller’s geometrical theory of 21-element travelling wave array,
diffraction (GTD) 199 example of 119–26
Kronecker delta functions 67, 125 local multiport distribution systems
(LMDS) 291
laser-welding 323 longitudinal radiating slot 96, 113,
leaky wave antenna 225–6 119, 128, 130
Lewin’s graphs 108 longitudinal slot arrays 93, 169, 175, 186
linearly polarised radial line slot longitudinal slot characteristics 72
antenna 297–9 experimental verification 79–82
linear resonant array antenna 275–7 slot electric field distribution 72–6
linear slot array 94, 98 slot magnetic field distribution 76–9
linear slotted waveguide array antenna 93 weakly excited slots 82–3
design approaches 95–7 longitudinal slot in waveguide wall 36
Elliott design equations 99 equivalent slot conductance 40–3
first design equation 99–101 longitudinal rectangular slot 38–9
second design equation 101–2 preliminaries 36–8
standing wave arrays, iterative rectangular waveguide and TE10
design algorithm for 102–3 mode scattering 39–40
Elliott’s design procedure, Lorentz reciprocity theorem 8
improvement to 130–1 low-temperature co-fired ceramics
incremental conductance, array design (LTCC) technology 308, 326,
using the concept of 126–8 332–3
mutual coupling 104
compensation 111–12 magnetic surface current density 6
external mutual coupling terms, manufacturing 317, 333
efficient computation of 104–6 carbon fibre-reinforced plastic
full wave analysis 111 materials 324–6
internal TE20 mode coupling metallic materials 322–3
109–11 microfabrication 326
self-admittance 106–9 tolerance theory 317–22
scattering parameters, array design Maxwell’s equations 5–6, 10, 111
in terms of 128–30 meandering the waveguide 259, 263
346 Slotted waveguide array antennas: theory, analysis and design

metallic materials 322–3 compensation 111–12

metallisation techniques 324 finite ground plane effects 55–7
method of moments (MoM) 46, 63, full wave analysis 111
97–8, 120–2, 144, 334 infinite array 145–8, 156
formulation of coupled integral internal 109–11
equations 201–2 self-admittance 106–9
improved design procedure using mutual coupling correction matrix,
MoM solution 204–9 insertion of 112
microelectronics packaging 326
microfabrication 326 non-military applications 20
microstrip patch arrays 22 non-resonant series–series coupling
microwave camera 332 slot 178
microwave directional antennas 17 non-tilted edge slots with wire
millimetre waves 317, 323, 332 excitation 265
miniaturisation 326 normalised active impedance 195
minimum scattering antennas (MSA) 45
MIT, Radiation Laboratory at 18 omnidirectional antenna 266
mobile radar system 324 omnidirectional rectangular waveguide
mode matching technique (MM) 212, antenna 267
334 omnidirectional slotted waveguide
modern antenna pattern array antennas 266
measurements 126 pattern ripple problem 269–72
monopulse 242 slotted circular waveguide array 266–8
monopulse applications, radial line slotted rectangular waveguide array
antennas for 303–4 266
monopulse radial line slot array 292 overloading 95, 277–8
monopulse slotted waveguide arrays
242–5 passive electronically steered array
monostatic radar cross section 255 (PESA) 331
Monte Carlo simulation 320 pattern ripple problem 269–72
Montgomery’s expressions 191 perfect electric conductor (PEC) 11
multiple beam slotted waveguide phased array 22, 43, 221, 228, 331
arrays 237–9 phase/frequency scanned C-band
Multiple Input Multiple Output array 83
(MIMO) systems 232 phase/frequency scanned planar array
mutual coupling 43, 97–9, 104, 119, 20–2
278–9 phase reversal travelling array 226–30
array/isolated coupling 43–4 phase scanned slotted waveguide
calculations 44, 48–51, 63, 104–6 arrays 221–3
array coupling 47 phase steered array 222, 232, 253, 259
equivalent circuit approach 45 planar array and feeding
examples 51–5 arrangements 279–81
mutual admittance between two planar array of 160  160 elements
slot apertures 46–7 consisting of 10  10 element
single-slot admittance 46 sub-arrays 151
 Index 347

planar slot array designs, Poisson’s sum formula 147

examples of 156 polynomials 61
8  8 slot array for radiometer PolyStrata“ 326–7, 333
application 160–2 post-excited longitudinal slots 264
large array design using infinite power calculations 63
array model 156–60
planar slotted waveguide array radar 17, 331
antennas 135 airborne early warning (AEW)
analysis and design in literature system 223
168–9 beacon 268
Elliott’s design procedure 136–9 cross section 249
design algorithm 139–41 development 17–18
6  6 planar array, example of SAR 20, 332–3
141–5 radial line antennas 303–4
flow chart for the design of a planar radial line slot antenna 291–3,
array 140 301, 304
generalised scattering matrix model linearly polarised 297–9
for analysis and design 169–72 radiated complex power 29
infinite array mutual coupling radiated far field 26, 204, 269
model 145 radiating compound broad wall slot
expression 146–8 184–5
planar slot array designs, examples radiating slots 64, 96, 102, 130,
of 156 135–8, 142, 156, 166, 168,
8  8 slot array for radiometer 170, 185, 190, 197–8, 204–6,
application 160–2 213–14, 307
large array design using infinite radio link antennas 23
array model 156–60 radiometer application, 8  8 slot array
sub-array architectures 151 for 160–2
planar array of 160  160 random errors 317
elements 151 reciprocity theorem 8, 36
slot array with 8  8 radiating infinite region or a region bounded
elements 151–6 by perfect conductors 9
slot array with four-element source-free region 8
sub-arrays 152–6 rectangular parallel plate waveguides,
total slot conductance in radiating slot arrays in 304–7
waveguides and resistance in rectangular slot antenna 30–3
feed waveguides 149–51 rectangular waveguide 14, 34–6, 191,
travelling wave feed, design of 162 264, 266, 304, 321
computed results 166–8 slots in 1–2
in terms of scattering wave surface currents on the waveguide
representation of coupling slot wall for the TE10 mode in 1
164–6 reflection coefficient 60, 124, 166,
plane wave incident on a waveguide 195, 204, 214
slot 251 bandwidth 94–5
plastic materials 324 remote sensing satellites 332
348 Slotted waveguide array antennas: theory, analysis and design

resonant length 59–60, 73–4, 79–81, wire excited 265

139, 142, 191 slot admittance 59
resonant series feed array 230–1 calculating 42
resonant slots 169 wall thickness effect 67, 73
approximate equivalent networks for 3 slot antenna 25
ridge waveguide 190–1, 222–3, 321–2 aperture in an infinite ground plane
rounded ends 79–82, 144, 201 25–30
longitudinal slot in waveguide wall 36
S-band antenna 163 equivalent slot conductance 40–3
scalar potentials 9–10 longitudinal rectangular slot 38–9
scattering preliminaries 36–8
array of slots 252–5 rectangular waveguide and TE10
cross section 249 mode scattering 39–40
scattering parameters, array design in mutual coupling 43
terms of 128–30 calculations 44–55
Schelkunoff’s field equivalence finite ground plane effects 55–7
principles 25 rectangular slot antenna 30–3
second-order lobes 258 waveguide modes 33
analysis 258–9 definitions 33–4
secondary lobes, reducing 259–65 rectangular waveguide 34–6
self-admittance 106–9 TE10 mode case 36
series–series slot 175 slot apertures and major coordinate
serpentine feed 228 system 49
shapes of slotted waveguide array slot array model 97
antennas 18–23 ground plane 97
shunt conductance 40–1 mutual coupling 97–9
shunt-series coupling slot 175, 179–81 network model 97
single-layer radial line slot arrays 301–3 slot arrays 291
single-mode apertures 45 in circular parallel plate
single-ridge waveguides 190 waveguides 291
single-slot admittance 46 accurate analysis and design
single-slot antenna 25 techniques 304
slot circularly polarised radial line slot
bandwidth 273 antenna 291–6
electric field 72–6 design of slot coupling
equivalent circuit 60–2 coefficients 299–300
impedance 31, 33 linearly polarised radial line slot
iris excited 186–9, 260, 264 antenna 297–9
magnetic field 76–9 radial line antennas for monopulse
parasitic 53 applications 303–4
probe/post excited 264, 266 single-layer radial line slot
stripline fed 232, 241, 245 arrays 301–3
Tee/Pi model 73 slot coupling and attenuation
Tee or Pi network 96–7, 128, 130 coefficient of propagating
wall thickness 66–8 wave 296–7
 Index 349

with dielectric cover 191–3 MoM solution to the integral

with 8  8 radiating elements equations of planar array and
consisting of two sub-arrays applications 201, 203–4
151–6 formulation of coupled integral
with four-element sub-arrays 152–6 equations 201–2
in gap waveguides 310–13 improved design procedure using
in rectangular parallel plate MoM solution 204–9
waveguides 304–7 radiating compound broad wall slot
in ridge waveguides 190–1 184–5
in substrate integrated waveguides slot arrays in ridge waveguides
307–9 190–1
slot arrays, advanced design and slot arrays with dielectric cover
optimisation of 175 191–3
coupling slot models 175 slot conductance 30, 149, 166, 277
centred-inclined coupling slot at resonance 40–3
176–9 slot coupling coefficients 299–300
shunt-series coupling slot 179–81 slot modelling 18, 42, 59, 83
transverse/transverse coupling integral equation solution 68
slot 181–2 external field 70–1
edge wall slot 182–3 internal field 68–70
finite ground plane effects 199–200 matrix elements 71–2
global optimisations for improved longitudinal slot characteristics 72
designs 209 experimental verification 79–82
equivalence between a slot with slot electric field distribution
one basis function and slot with 72–6
many basis functions 213–14 slot magnetic field distribution
genetic algorithms of planar 76–9
standing wave array using weakly excited slots 82–3
MoM analysis 211–13 principles 59
genetic algorithms of travelling electromagnetic models 62–6
wave linear array 209–11 equivalent circuit approach 60–2
genetic algorithms optimised finite wall thickness 66–8
results 215–16 using experimental data 59–60
higher-order mode coupling between transverse slots 83
centred-inclined coupling results 86–8
slots 193 theoretical analysis 84–6
in the design of centred-inclined slot panel 261
coupling slots 196–7 slotted circular waveguide array
and longitudinal radiating slots 266–8
197–9 slotted cylindrical waveguide
theory for 195–6 array 268
iris-excited longitudinal slot 186–8 slotted monopulse stripline array
Elliott’s design procedure for iris- 247, 249
excited slotted waveguide slotted rectangular waveguide array
arrays 189 266, 310
350 Slotted waveguide array antennas: theory, analysis and design

slotted waveguide array antennas, Stegen’s curve 320

bandwidth of 272 stripline-fed tapered slots 241
linear resonant array antenna 275–7 stripline slotted arrays 245
mutual coupling 278–9 independent excitations 247–8
overloading 277–8 sub-array architectures 151
planar array and feeding planar array of 160  160 elements
arrangements 279–81 151
slot antenna 273 slot array with 8  8 radiating
travelling wave arrays 281 elements 151–6
waveguide-fed longitudinal slot slot array with four-element
273–5 sub-arrays 152–6
slotted waveguide arrays 332 substrate integrated waveguides
conformal 239–42 (SIW) 307–9
dual polarised 232–6 SWANTM 335
frequency scanned 223 SWASS (Slotted Waveguide Antenna
centre-fed travelling array 231–2 Stiffened Structure) 325–6
phase reversal travelling array symmetric array architecture 162
226–30 systematic errors 317
resonant series feed array 230–1
travelling wave linear array 225–6 Taylor series expansion 104
monopulse 242–5 TerraSAR-X Radar satellite 325
multiple beam 237–9 theoretical calculations 57, 62, 80
phase scanned 221–3 Toeplitz matrix 45
scattering from 248 Toeplitz simplification 51
feeding network 255–6 tolerance theory 317–22
numerical examples 256 transmission line model 96–7
in a planar ground plane 250–2 for travelling wave array 116
scattering from an array of slots transverse slots 83, 231, 233–4, 245
252–5 definition of offset for 84
spatial filters 256–8 in rectangular waveguide 83
soft coupling 197 results 86–8
Sommerfeld radiation 9 theoretical analysis 84–6
spaceborne synthetic aperture radar transverse/transverse coupling slot 181–2
(SAR) 1, 20, 332 travelling wave array 102, 116, 281
spaceborne systems 332–3 design algorithm 118–19
spatial filters 234, 256–8 design equations for 118
squintless antenna 231 transmission line model for 116
stainless steel 323 21-element travelling wave array,
standard waveguide frequency bands 339 example of 119–26
standing wave array 112, 151 travelling wave feed design for planar
design of a –30 dB sidelobe-level array 162–4
Dolph–Chebyshev array 113–16 computed results 166–8
slot data needed for design in terms of scattering wave
algorithm 112–13 representation of coupling slot
steerable beams 333 164–6
 Index 351

travelling wave linear array 225–6 waveguide-fed longitudinal slot 273–5

two-layer radial line slot array 293 waveguide modes 33
two-slot geometry for calculating the definitions 33–4
magnetic field 50 rectangular waveguide 34–6
TE10 mode case 36
vector potential 9–10, 13, 26, 252 waveguide slot antennas 18
visible and invisible space 28 WiFi 333
voltage standing wave ratio (VSWR) WiGig 333
151, 191 wire antennas 17
wireless communication system 17
WASP-NET 334 wireless local area networks
waveguide (WLAN) 291
aperture 249–50
feeding system 18 X-band Doppler radar 238
feed network 245 X-band flat plate antenna in stripline
standard dimensions 339 technology 247
and stripline technologies 245–6 XFdtd 334