Inter-University Electronics Series, Vol. 7 ANTENNA THEORY part 2 Robert E. Collin Division of Electrical Sciences and Applied Physics Case Western Reserve University Cleveland, Ohio Francis J. Zucker Air Force Cambridge Research Laboratories L. G. Hanscom Field Bedford, Massachusetts McGraw-Hill Book Company New York St. Louis San Francisco London Sydney Toronto Mexico Panama PROPERTY OF THE tinpsaneANTENNA THEORY, PART 2 Copyright © 1969 by McGraw-Hill, Inc. All rights re- served. Printed in the United States of America. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or other- wise, without the prior written permission of the publisher. Library of Congress Catalog Card Number 68-8031 11800 1234567890 MAMM 7654321069INTER-UNIVERSITY ELECTRONICS SERIES Series Purpose The explosive rate at which knowledge in electronics has expanded in recent years has produced the need for unified state-of-the-art presentations that give authoritative pictures of individual fields of electronics. The Inter-University Electronics Series is designed to meet this need by providing volumes that deal with particular areas of electronics where up-to- date reference material is either inadequate or is not conveniently organized. Each volume covers an individual area, or a series of related areas. Emphasis is upon providing timely and comprehensive coverage that stresses general principles, and integrates the newer developments into the overall picture. Each volume is edited by an authority in the field and is written by several coauthors, who are active participants in research or in educational programs dealing with the subject matter involved. The volumes are written with a viewpoint and at a level that makes them suitable for reference use by research and development engineers and scientists in industry and by workers in governmental and university laboratories. They are also suitable for use as textbooks in specialized courses at graduate levels. The complete series of volumes will provide a reference library that should serve a wide spectrum of electronics engineers and scientists. The organization and planning of the Series is being carried out with the aid of a Steering Committee, which operates with the counsel of an Advisory Committee. The Steering Committee concerns itself with the scope of the individual volumes and aids in the selection of editors for the different volumes. Each editor is in turn responsible for selecting his coauthors and deciding upon the detailed scope and content of his particular volume. Overall manage- ment of the Series is in the hands of the Consulting Editor. Frederick Emmons TermanPREFACE When we first agreed to undertake the development of a book on antennas for the Inter-University Electronics Series, the plan called for a rather modest effort devoted mostly to recent advances in this field. Subsequent discussions led to the conclusion that there was a real need for a more complete treatise dealing with all aspects of antenna theory. We therefore agreed to join forces as coeditors to carry out this rather ambitious plan. Twenty years have passed since the publication of Silver’s classical work on antennas in the M.I.T. Radiation Laboratory Series.} Considering the impor- tance of antennas in modern communications technology, the number of antenna books published since then is surprisingly small (see Bibliography). Also, most of these books were written more than a decade ago and do not reflect the present state of development in the antenna field. Consequently, we made plans for a book that would develop the underlying theory from basic principles, would apply it to the many varieties of radiating struc- tures that we call antennas, and would emphasize the more recent developments in this field. We also recognized the intimate relationships between the performance of an antenna and its environment, and we have therefore included chapters on the behavior of an antenna immersed in lossy media or plasmas or located on a lossy earth. The demands on antenna systems in the current era can often be met only by overall system optimization, and this made it essential to also include chapters dealing with the signal-processing and spatial-filtering aspects of antennas. The first seven chapters cover the general theory of antennas without reference to any particular type of structure except by way of illustration. Much of this material may be regarded as classical antenna theory. However, the development has been carried far enough to include topics that have become important in more recent years. For example, in the chapter on receiving antennas, quasi-monochromatic and partially polarized waves are dealt with. A whole chapter is devoted to large non-uniformly spaced arrays, and another chapter brings together and assesses the large number of pattern synthesis schemes that have been proposed over the years. The next three chapters (Chapters 8 to 10) present the latest developments in the theory of cylindrical antennas. Significant portions of these chapters are new contributions to the field and have not appeared elsewhere before. The remaining five chapters of Part 1 cover the older classical radiating struc- tures such as loop antennas, conical and spheroidal antennas, slot antennas, 48. Silver (ed.), “Microwave Antenna Theory and Design,” M.I.T. Radiation Laboratory Series, vol. 12, McGraw-Hill Book Company, New York, 1949. viiviii PREFACE and open waveguides and horns. The reader will find that these chapters bring together in a single place a rather complete overall view and the relevant, references to the literature on these types of radiators. The first three chapters of Part 2 are a coordinated sequence covering large-aperture antennas such as reflectors and lenses, preceded by an account of ray optics fundamentals that are basic in their analysis and design. These are followed by four extensive chapters on traveling-wave antennas. Leaky- wave and surface-wave antennas are of more recent origin, and the underlying theory given here is the most extensive and up-to-date treatment available. The fourth chapter in this sequence deals with the very important “frequency- independent” structures known as log periodic antennas; their theory has many features in common with other traveling-wave antennas, and it was therefore included here. The remaining six chapters of Part 2 are devoted to environmental and systems aspects; their contents were briefly mentioned earlier. We originally planned to devote a chapter to mutual impedance coupling effects in large arrays and to other problems associated with the electronic scanning of such arrays, but a three-volume series devoted to microwave scanning antennas was recently published.} To avoid duplication, we decided to omit this topic; it is the only major aspect of antenna theory not included in these volumes. As editors, we strove for a reasonably uniform notation, but because of the wide diversity of topics, complete uniformity was not possible. That the same symbol often means different things in different chapters should not, when taken in context, be confusing. A major point of departure from convention is the use of both e and e—‘*! (in separate chapters) to represent the time dependence of steady-state monochromatic fields. The former is standard among electrical engineers; the latter is preferred by physicists and applied mathematicians. We did not take a firm stand on this point, but instead let individual authors follow their own preferences. The use of j or 7 as the imaginary unit ¥—1 will alert the reader to which convention is adopted in each chapter. Since the contemporary literature is about evenly divided in the use of the two conventions, and since it is as impossible to talk an electrical engineer into writing an inductive impedance in the form Z = R — iwl as it is to convince an applied mathematician an outgoing attenuating wave should be represented in the fourth quadrant of the complex-wave-number plane, we did not feel it desirable to legislate against either. The two parts offer a number of possibilities as texts for antenna courses. A complete course on antennas cannot be offered in a single semester on the basis of three lectures per week, so there must be a selection of topics, usually according to the interests of the instructor and the class. For a basic short {R. C. Hansen (ed.), “Microwave Scanning Antennas,” Academie Press Inc., vol. I, 1964, vols. II and III, 1966.PREFACE ix course Chapters 1 to 5, selected parts of Chapters 6 and 7, Chapter 9, and topics from Chapters 12 to 15 would provide a good coverage of classical antenna theory. A course in which aperture antennas are emphasized would include Chapters 16 to 18, while Chapters 19 to 22 covering traveling-wave antennas, Chapters 23 to 25 for environmental effects, and Chapters 26 and 27 for systems aspects would be included in courses designed to emphasize those topics. A number of problems and exercises, dispersed throughout the various chapters, have been included to enhance the usefulness of the books in antenna courses. We have been very fortunate in obtaining chapter contributions from truly outstanding antenna specialists. To the contributing authors we also owe a great deal for their wise counsel and enduring patience with a project that has extended over several years. Any remaining shortcomings must be blamed on the editors. The work described in these two volumes is that of a great many engineers and physicists. A reasonable effort was made to ensure proper references and credits. To those whose work is not properly referenced, we offer our apologies. The lack of credit is not intentional; it was caused by the difficulty of tracing the history of any particular topic back to its originators. Finally, as editors, we would appreciate having misprints and errors called to our attention so that future printings may be corrected. The editors express their sincere appreciation to the following organizations for their kind permission to use a number of figures from their various publica- tions: The Institute of Electrical and Electronic Engineers, Inc. 345 East 47th Street New York, N.Y. 10017 American Institute of Physics 335 East 45th Street New York, N.Y. 10017 The Institution of Electrical Engineers Savoy Place London W.C.2, England Institution of Electronic and Radio Engineers 8-9 Bedford Square London W.C.1, England National Research Council of Canada Ottawa 7, Ontario, Canada Robert E. Collin Francis J. ZuckerCONTENTS Foreword =v Preface vii CHAPTER 16 WAVE FRONTS, RAYS, AND FOCAL SURFACES F. Sheppard Holt 16.1 Introduction 1 16.2 The Etkonal and the Eikonal Equation 2 16.3 Geometrical Optics as a Zero-wavelength Approximation 4 16.4 Fermat’s Principle and Snell’s Laws 6 16.5 Phase Analysis and Phase Synthesis 8 16.6 Power Flow in Ray Tubes 10 16.7 Power Distribution in the Aperture i 16.8 Power Distribution in the Focal Region 13 16.9 Power Distribution in the Far Field 15 16.10 Reflection from a Conducting Surface 17 16.11 Ray Collimation and Off-focus Feeding 19 16.12 Congruences and the Equal-path-length Law 21 16.13 Focal Surfaces 23 16.14 General Comments on Focal Surfaces 28 Appendix A: Principal Normal Radii of Curvature, Principal Directions, Principal Planes, and Lines of Curvature 30 Appendix B: Geometrical Optics Power Flow in a Source-free, Nonconducting, Isotropic, Homogeneous Medium 31 Problems 33 References 34 CHAPTER 17 REFLECTOR ANTENNAS C. J. Sletten 17.1 Importance of Reflector Antennas 36 17.2 Types of Reflector Antennas and Their Primary Uses 36 17.3 The Paraboloid with Source at the Focus 37. 17.4 The Wide-angle or Off-focus Characteristics of Paraboloids 55 17.5 Techniques for Compound Primary-feed Designs 64 17.6 Spherical Reflector 69 17.7 Cylindrical Reflector Antennas — 83 17.8 Parabolic Torus Antennas 86 17.9 Stepped Reflector Antennas 89 17.10 Multiplate Antenna 92 17.11 The Corner Reflector Antenna 96 Problems 101 References 101xii CONTENTS CHAPTER 18 LENS ANTENNAS John Brown 18.1 Introduction 104 18.2 Artificial Dielectrics 104 18.3 Design of Homogeneous Lenses 7 18.4 Lenses for Beam-scanning Applications 121 18.5 Extension of Scanning Analysis 126 18.6 Nonhomogeneous Lenses 131 18.7 Configuration Lenses 136 18.8 Discussion 144 Problems 145 References 149 CHAPTER 19 GENERAL CHARACTERISTICS OF TRAVELING-WAVE ANTENNAS Alexander Hessel 19.1 Introduction 151 19.2 Modal Properties of Uniform Traveling-wave Structures 155 19.3 The Dispersion Relation for Shielded Waveguides 156 19.4 Guided Waves on Planar Open Structures 162 19.5 The Excitation Problem in Traveling-wave Structures 168 19.6 Characteristics of Specific Guided Waves 177 19.7 Guided Waves on Periodically Loaded Traveling-wave Structures 184 19.8 The Brillowin Diagram and Mode Coupling in Closed Periodic Structures 191 19.9 Guided Waves on Radiating, Periodically Loaded, Traveling-wave Structures — Kinematic Properties 195 19.10 Spectral Properties of Spatial Harmonics 203 19.11 Guided Waves in Radiating, Periodically Loaded, Traveling-wave Structures — Dynamic Properties 209 19.12 Examples of Brillouin Diagrams of Periodically Loaded, Radiating, Traveling- wave Structures 225 Appendix A: Review of Modal Formalism 235 Appendix B: The Square Root Function in the Complex kz Plane 238 Appendix C: The Transformation kz = ko sinw 241 Appendix D: The Spectral Representation of the Field of a Magnetic Line Source above an Inductive Surface Reactance Plane 242 Appendix E: Some Properties of the Dispersion Relation of Lossless, Planar, Periodic, Open, Traveling-wave Structures 246 Appendix F: The principal Branch of Solutions of the Dispersion Relation for SMRS at Bd = 2" 248 Problems 250 References 255CONTENTS xiii CHAPTER 20 LEAKY-WAVE ANTENNAS T. Tamir 20.1 Introduction 259 20.2 The Field of a Leaky-wave Distribution 262 20.3 Radiation-pattern Calculations 268 20.4 The Complex-angle Plane w and Its Relation to the Leaky-wave Field 273 20.5 Control of Aperture Distribution 277 20.6 Determination of the Complex Wave Number ks 279 20.7 Specific Antenna Structures 289 Problems 295, References 295 CHAPTER 21 SURFACE-WAVE ANTENNAS Francis J. Zucker 21.1 Introduction to Surface Waves 298 21.2 How Does a Surface-wave Antenna Radiate? 304 21.3 Surface-wave Excitation 313 21.4 Discontinuities and Tapers 325 21.5 Surface-wave Structures 331 Appendix: Saddle-point Integration 338 Problems 341 References 343, CHAPTER 22 LOG-PERIODIC ANTENNAS R. Mittra 22.1 Historical Introduction 349 22.2 Principle of Scaling and Application to Log-periodic Design 350. 22.3 Application of k-8 Diagrams to Log-periodic Antenna Analysis 357 22.4 Other Approaches to the Analysis of Log-periodic Structures 361 22.5 Transient Response of Log-periodic Antennas 378 Appendi: Solution of Equations for P(a) and Q(a) Appearing in Sec. 22.4d 380 Appendix B: The Expressions for the Far Field of a Conical Spiral Antenna 382 Problems 383 References 384 CHAPTER 23 CHARACTERISTICS OF ANTENNAS OVER LOSSY EARTH James R. Wait 23.1 Fields of Elementary Dipoles over a Homogeneous Half Space 386 23.2 Radiation Resistance and Impedance Increments Resulting from the Presence of the Homogeneous Half Space 392xiv CONTENTS 23.3 Characteristics of a Thin Linear Antenna over the Conducting Half Space 401 23.4 Linear Antenna over a Solid Metal Disk Which Is Lying on the Half Space 405 23.5 Radial Wire Ground Systems 414 23.6 Influence of the Ground Plane on Antenna Patterns 424 23.7 Conclusions 435. References 435 CHAPTER 24 ELECTROMAGNETIC FIELDS OF SOURCES IN LOSSY MEDIA James R. Wait 24.1 Introduction 438 24.2 The Infinitesimal Radiating Sources in a Lossy Medium 442 24.3 The Finite Current Element 458 24.4 The Horizontally Stratified Earth 461 24.5 Electric Dipole inside a Two-layer Earth 476 24.6 The Infinite Wire over the Three-layer Earth 482 24.7 Grounded Wires on the Homogeneous Earth 492 24.8 Self-impedance of Circuits in a Dissipative Medium 495 Appendix A: Plane Wave Transient Propagation 502 Appendix B: Evaluation of Integrals 505 References 506 CHAPTER 25 ANTENNAS IN PLASMA James R. Wait 25.1 Introduction and Basic Considerations 515 25.2 Radiation from a Current Distribution in an Isotropic Electron Plasma 518, 25.3 Application to a Cylindrical Antenna 520 25.4 The Cylindrical Shell Model 526 25.5 Collected Results 529 25.6 Boundary-value Approaches to Antennas in Plasma Media 532 25.7 Influence of the DC Magnetic Field 548 25.8 Concluding Remarks 551 References 553 CHAPTER 26 THE ANTENNA AS A SPATIAL FILTER A. C. Schell 26.1 Introduction 557 26.2 Spatial Frequency Analysis 557CONTENTS xv 26.3 The Distinction between Coherence and Incoherence 560 26.4 The Incoherent Source Antenna Transfer Function 561 26.5 Different Antenna Types and Their Spatial Frequency Responses 563 26.6 Optimizing System Performance 565 26.7 Optimizing Incoherent Source Transfer Functions 567 26.8 Data Processing 570 26.9 Angular Resolution Enhancement 574 Problems 577 References 578 CHAPTER 27 SIGNAL-PROCESSING ANTENNAS A. A. Ksienski 27.1 Introduction 580 27.2 Definition and Description of Multiplicative Arrays 583 27.3 Response of Linear and Nonlinear Multiplicative Antennas to General Source Distributions 586 27.4 Performance of Multiplicative Arrays in the Presence of Noise 596 27.5 Multiple Correlation Arrays 608 27.6 Aperture Synthesis 608 27.7 Applications of Multiplicative Arrays 611 27.8 Pattern Multiplication by Two-way Mode of Operation 616 27.9 Definition and Description of Synthetic Array Radar 620 27.10 Synthetic Array Pattern 622 27.11 Parametric Design Considerations of Synthetic Array Radar 624 27.12 Unfocused Synthetic Arrays 627 27.13 Focused Synthetic Arrays 630 27.14 Recording and Optical Processing in Synthetic Array Radar 631 27.15 The Optical Properties of the Records 634 27.16 Performance Degrading Factors of Synthetic Array Radar 635 27.17 Detection Methods 643 Problems 650 References 651 CHAPTER 28 LARGE ANTENNA SYSTEMS Merrill I. Skolnik 28.1 Introduction 655 28.2 Limitations to Antenna Size 657 28.3 Large Antennas — Examples 665 Problems 673 References 673 Bibliography 675 Index 677CHAPTER 16 WAVE FRONTS, RAYS, AND FOCAL SURFACES F. Sheppard Holt 16.1 Introduction This chapter is concerned with the laws, principles, and procedures of geometrical optics that are applicable to the design and analysis of microwave antennas. Considered as a zero-wavelength approximation to exact electro- magnetic wave theory, geometrical optics is very accurate in the design and analysis of optical focusing devices, since optical wavelengths are indeed ex- tremely small compared to the aperture dimensions of optical systems. At microwave frequencies the wavelength is not always relatively small compared to the aperture dimensions of microwave systems. However, geometrical optics, although certainly an approximation, is still sufficiently accurate to produce meaningful and useful results, even for antennas with aperture dimen- sions as small as five wavelengths. The advantages of certain microwave components over existing optical components and the relaxation in mechanical tolerance requirements due to the finite wavelength allow the exploitation of geometrical optics analysis and design at microwave frequencies in certain cases considerably beyond that achievable in optics. For example, low-loss isotropic artificial dielectrics in a wide range of index of refraction, phase and amplitude control of sources and receivers, and aspheric as well as non- rotationally symmetric reflecting and refracting surfaces are all available to the antenna designer. In general, geometrical optics is concerned with the analysis and synthesis of optical systems to the approximation that diffraction and interference can be neglected. In an isotropic medium classical geometrical optics assumes that the power flows along paths called rays at a velocity characteristic of the medium and that there exists a family of surfaces known as wave fronts that are everywhere normal to the rays. Point-to-point correlation between wave fronts can be established by the rays, and no power is assumed to be present in regions where there are no rays. It is evident that if all the wave fronts are given, then all the rays are determined, and vice versa. Classical geometrical optics, therefore, neglects wavelength, phase, and the vector nature of electro- magnetic wave motion. For microwave applications it is most useful to extend the classical theory to include effects of the above neglected factors. The form of the extension is justified by the asymptotic solution as w > © (A — 0) of the exact electromagnetic field equations to be discussed in Sec. 16.3. The exten- 12 ANTENNA THEORY sion consists in introducing wavelength as a small but finite quantity, identi- fying the wave fronts with equiphase surfaces, and at each point on a ray in a homogeneous medium introducing the electromagnetic field vectors E and H and relating them as in a plane wave propagating along the ray. We shall refer to this extension as the geometrical optics approximation. 16.2 The Eikonal and the Eikonal Equation Using the geometrical optics approximation, assume for a particular wave motion in a source-free isotropic medium that the equiphase wave fronts are given by the level surfaces of the function L = L(y) and that the phase ¢ on a general wave front W is given by $= of — 2 Leyz) (16.1) where w is the angular frequency, c is the velocity in free space, and (z,y,2) is any point on W. The function L(z,y,z) is known as the eikonal, and it, to- gether with the wave-front velocity in the medium, completely describes the given wave motion from the standpoint of classical geometrical optics. Consider a general ray path C given by the equations x = x(s) y = y(s) z= a(s) (16.2) where s is arc length measured along C from a fixed reference wave front Wo, given by L(z,y,z) = 0, to a propagating wave front W (see Fig. 16.1). Let the position of W along C as a function of time be given by the formula 8 = s(t) (16.3) General wave front W—" Ray path © Reference wave front Wo Fig. 16.1 Ray path and wave fronts.WAVE FRONTS, RAYS, AND FOCAL SURFACES 3 where s increases with t. The wave-front propagation velocity » along C is then ats dt The value of L (x,y,z) on the wave front W will change as W propagates along C. Since time and position along C are related by (16.2) and (16.3), L(z,y,z) may be considered as a function of either s or t. Thus _ dL _ dL/dt El = Gs = as/at Since the phase ¢ on the propagating wave front W is constant, differentiating (16.1) with respect to ¢ yields (16.4) y (16.5) do eth a’ ea or dL iG (16.6) Substituting (16.4) and (16.6) into (16.5), we have \VL] =l =n (16.7) c » where n is the index of refraction of the medium. The partial differential equa- tion (16.7) satisfied by the eikonal L(z,y,z) is known as the eikonal equation. The eikonal equation can be used to determine the curvature of ray paths in an inhomogeneous medium. Assume a wave motion in an inhomogeneous medium prescribed by the eikonal L(x,y,z) and let C be a ray path with are length s in the direction of the wave motion. At each point on C let § be the unit tangent vector in the direction of s, let A, be the unit principal normal vector, and let p be the radius of curvature. The vector VZ evaluated at any point P is normal to the wave front passing through the point P. Hence, at each point on C the vector VL is along C, and in view of the eikonal equa- tion (16.7) VL» zs (16.8) The first Frenet formula! of differential geometry can be written ® _ yas a) = Be a7 & v)s = -SX (VX8) = > (16.9) Substituting (16.8) into (16.9) we have = -8x [v(2) x ve] = 8x Ivan) xa 8 fy-Vn (16.10) 1_. s 1 and a7 fe 7, = f,-V(Inn) = n4 ANTENNA THEORY Exercise 16.1 Show the details of the derivation of (16.10) by carrying out the indi- cated substitutions. Equation (16.10) shows that the ray path curvature 1/p is related to the rate of change of the index of refraction n normal to the ray path. In particular, for a homogeneous medium (n = constant) the curvature is zero and the ray paths are straight lines. 16.3 Geometrical Optics as a Zero-wavelength Approximation For a source-free, nonconducting, isotropic, and homogeneous medium the Maxwell equations for the electric field E and the magnetic field H are VX E = —jouH (16.114) VXH= jock (16.116) V-E=0 (16.11¢) V-H=0 (16.114) where w is the angular frequency, ¢ is the permittivity of the medium, y is the permeability of the medium, and an e** time dependence is assumed. Equa- tions (16.1la-c) combine to produce VE + PE = 0 (16.12) where k = wyeu is the phase constant of the medium. To obtain a zero-wavelength or high-frequency approximation, the electric field is assumed in the form of an asymptotic series’! in descending powers of w as follows: E(x,y,2,0) = eobewn2) x, eae! (16.13) where E(z,y,2) is real and ky = w-Veouo is the phase constant of free space. Sub- stituting (16.13) into (16.12) and independently equating to zero the coeffi- cients of w? and w, the highest powers of w present, yields |VLP = n2 (16.14) WL and (VL-V)Ey + a E,=0 (16.15) where n = k/ko is the index of refraction of the medium. Note that (16.14) is the eikonal equation and hence L(z,y,z) is the eikonal. As w becomes large, the leading term of (16.13) predominates and becomes Enr, the high-frequency approximation to the electric field, as follows: Eur = Ep(z,y,z)eol eu) (16.16)WAVE FRONTS, RAYS, AND FOCAL SURFACES 5 Since Ep(z,y,z) is real, the equiphase surfaces, L(x,y,z) = constant, of Eur are identical with the wave fronts of geometrical optics. Again, the rays can be defined as the family of curves normal to the wave fronts. In the present case of an isotropic homogeneous medium the rays are straight lines (see Sec. 16.2). It has been shown that at each point on a ray VL/n = § is a unit vector tangent to the ray [see (16.8)]. Thus VL 2 d VL-V = nav = nV = na (16.17) where d/ds is the directional derivative in the direction of a ray with respect to are length s along the ray. Equation (16.15) can now be written vt Meg, = 0 (16.18) The solution to (16.18) along a ray through the point (.ro,yo,20) can be expressed in the form (16.19) Eo(s) = Eo(s0) exp ( lpn ) “3h. where % is the distance from a reference wave front to the point (20,70,20).. Thus E)(s) at any point on a ray is completely determined once its value at one point on the ray is known. This is a very clear statement of the geometrical optics property of point-to-point correlation from wave front to wave front along a ray. It is also evident from (16.19) that the direction of E)(s) is the same for all points on a ray (except possibly for sense). Substitution of (16.13) into (16.11c) and equating to zero the coefficient of w’, the highest power of w, leads to the result VL-Eur = 0 (16.20) and hence Eyr at each point on a ray is normal to the ray. Substitution of (16.18) into (16.11a) and retaining only the highest-order terms in w yields Hur, the high-frequency approximation to the magnetic field, as follows: =ahe Hur = =F VL X Eye eh = vis X Enr (16.21) Equations (16.20) and (16.21) show that Enr and Hur along a ray are related precisely as in a plane wave propagating in the direction of the ray. Simplification of (16.19) is best accomplished by considering the vector field 2 1 F= Kns K= RRs (16.22) where at each point 2; and FR; are the principal normal radii of curvature of the wave front passing through that point (see Appendix A at the end of this chapter). The quantity K = 1/RR, evaluated at a point on a surface is known6 ANTENNA THEORY as the gaussian curvature of the surface at that point. It can be shown that the surface integral f F-dA =0 (16.23) where 2 is any closed surface lying in the isotropic homogeneous medium being considered (see Prob. 16.1). The divergence theorem then requires that V-F=0 throughout the medium, and hence (n8-V)K = —KV-n8 = —K VL In view of (16.17) it follows that dK , (VL\, _ at (ee =0 K(s) _ _ pk and hence Kl) > exp f. 7 ds Equation (16.19) can now be written = K (s) Ey(s) = Eo(s0). Ks) (16.24) Applying (16.17) to the eikonal L(z,y,z) and using (16.14), we have VIVE = t= nh or dL=nds For a homogeneous medium n is constant and L=ns+Iy (16.25) Substituting (16.24) and (16.25) into (16.16) yields Eus = Eo(s) Kyo (16.26) Note that the amplitude factor \K(s)/K(s) is precisely equivalent to the geometrical optics power factor K (d)/K (0) shown in (16-B.3) of Appendix B. 16.4 Fermat’s Principle and Snell’s Laws The optical path length (OPL) along a ray path C is defined as the line integral J. n ds, where n is the index of refraction of the medium and s is are length along C. Fermat’s principle states that electromagnetic energy travel- ing between two points will follow any ray path that makes the OPL integral stationary. Clearly then, in a homogeneous medium, ray paths will be straight lines.WAVE FRONTS, RAYS, AND FOCAL SURFACES 7 Snell’s laws of reflection and refraction of rays at a boundary surface between two different media can be derived directly from Fermat’s principle (see Prob. 16.2), but they will be merely stated here. The law of reflection states that the incident ray and the reflected ray lie on the same side of the boundary surface, are coplanar with the normal to the boundary surface at the point of reflection, and make equal angles with the normal. Thus 6; = 8,, where 6; is the angle of incidence and 6, is the angle of reflection, as shown in Fig. 16.2. This reflection law can be expressed in any of the equivalent vector forms (16.274) (16.27) (16.27¢) or (16.27d) where [8] = [8-| = [a] = 1 (16.27e) and 8; is in the direction of the incident ray, 8, is in the direction of the reflected ray, and fi is normal to the boundary surface (see Fig. 16.2). Exercise 16.2 Perform the vector product of & + &; with (16.27a) and use (16.27e) to deduce (16.276). Derive also (16.27c,d). Reflected ray Transmitted roy N >» Incident rey Index of refraction nj Index of refraction np Fig. 16.2 Reflection and refraction at a boundary. The law of refraction states that the incident ray and the transmitted or refracted ray lie on opposite sides of the boundary surface, are coplanar with the normal to the surface at the point of refraction, and satisfy the condition n, sin 0; = 4 sin 6 (16.28) where n; and n, are respectively the indices of refraction in the incident and transmitted regions and 6; and 6, are respectively the angles of incidence and8 ANTENNA THEORY refraction as shown in Fig. 16.2. This refraction law can be expressed in the vector form AX (nS; — n8,) = 0 (16.29) where Ai, §,, 7;, and n, are as defined above and 8; is a unit vector in the direction of the transmitted ray. Combining (16.27a) and (16.29), we have as a statement of both of Snell’s laws axXns = aX 28, = 0X nd (16.30) where n, = 7,, that is, the incident and reflected regions have the same index of refraction. 16.5 Phase Analysis and Phase Synthesis The geometrical optics approximation can be usefully applied to certain problems in phase analysis and phase synthesis. An application to phase syn- thesis is given in this section. An example of phase analysis based on the geometrical optics approximation is described in Sec. 17.6 in the determination of the phase along an axial line in a spherical reflector receiving a plane wave. If L(a,y,z) is the eikonal function, then according to (16.1) the phase ¢ at a point P with coordinates (z,y,z) is @ = of — 2 L(@y,2) If Wo is the wave front L(x,y,z) = 0, then the phase ¢, over the whole wave front W is do = ot The phase at P relative to the phase on Wo, that is, the phase difference $4 between P and Wo, will be $3 = 6 ~ bo = —2L(ey,2) (16.81) Assuming that the ray through P intersects W» at the point Po, then according to Sec. 16.4 the OPL from P to P» is defined to be P OPL = nds Ir where n is the index of refraction of the medium and the path of integration is along the ray from P, to P. The phase difference ¢3 between P and Py can be expressed in terms of OPL as 2r ¢P oa = “yl, nds (16.32)WAVE FRONTS, RAYS, AND FOCAL SURFACES 9 where \y is the free-space wavelength. Noting that w/e = 27/X) and compar- ing (16.31) with (16.32) indicates that the eikonal function L(z,y,z) evaluated at P is equal to the OPL from the reference wave front to P. The geometrical optics determination of phase will clearly fail in any region where two or more rays pass through each point. Consider now the problem of designing a point-source-fed reflector in two dimensions in a medium of index 7 that will synthesize, to within the geometri- cal optics approximation, a given phase distribution along a line.’ Let the point source be located at the point F with coordinates (0,a), and let the x axis be the line along which it is desired to achieve the phase function f(x). Ifaray reflected from the reflector at the point P with coordinates (x,y) intersects the x axis at the point Py with coordinates («,0) (see Fig. 16.3), then it follows from (16.32) that (Fev FU ai+ (2)eve= Sw FP = —fles) + Cr (16.38) yr Reflector x Al%,0) Prescribed phase f(r) Fig. 16.3 Geometry of reflector to produce desired phase along a line. where C; is a constant that can be adjusted to control the position of the re- flector and the extent of usable aperture. For a» fixed, (16.33) as a function of & and y represents an ellipse (see Prob. 16.7b) with foci at F and Py. With 2o variable, (16.33) represents a one-parameter family of ellipses that in general hasan envelope. To each ellipse in the family there corresponds a unique point on the envelope at which the ellipse is tangent to the envelope. Using the envelope as the reflector surface, it is clear that at each point on the envelope the incident ray is reflected precisely as from the ellipse corresponding to that point and hence arrives at the x axis in proper phase. The envelope of the family of ellipses is given by the simultaneous solution of (16.33) with the partial derivative of (16.33) with respect to xo, that is, with 2mm __2—%0__ _ (q) (16.34)10 ANTENNA THEORY Carrying out the simultaneous solution of (16.33) and (16.34) leads to the result, cent e_Atwn-@ 2 Bey +A+ayl — B _1-B At+ae-a@ YS Ba +A+ayl—B do where A= pele: —f(e)| and B= 5%’) (16.35) An experimental line source feed designed by the above procedure for off- focus feeding of a paraboloid was very successful in producing a well-focused off-axis fan beam. 16.6 Power Flow in Ray Tubes One of the important assumptions of geometrical optics is that the power flows along the ray paths. To within the accuracy of this approximation, a ray path diagram therefore presents an overall picture of the power flow for a loss- less source-free medium. The totality of rays that pass through any given closed curve constitutes a ray tube; and under steady-state conditions the total power flowing across any cross section of a ray tube must be constant, since no power can flow across the lateral surfaces of the tube. Thus, as a ray tube cross section decreases, the power density increases, and conversely, as a ray tube cross section increases, the power density decreases. These two cases correspond respectively to converging and diverging rays. In particular, if dA is a differential area on a wave front W and if the ray tube passing through dA intersects the wave front W’ in the differential area dA’ (see Fig. 16.4), then the total power flow through dA must equal the total power flow through dA’. Thus PdA = P'dA’ (16.36) wt dA Ray tube w dA Fig. 16.4 A ray tube between wave fronts.WAVE FRONTS, RAYS, AND FOCAL SURFACES ll where P is the power density at dA and P’ is the power density at dA’. It is clear that the ray tube concept of power flow will break down at focal points, since at such points the ray tube cross section vanishes and (16.36) predicts infinite power density. The concept of power flowing in ray tubes as applied to the region between. the primary feed and the aperture is particularly useful in the design and analysis of antennas. It is also useful in far-field considerations in the design and/or analysis of certain shaped-beam and off-focus-fed antenna systems. The ray tube concept finds little application in far-field considerations relative to well-focused pencil-beam systems, since in such cases the pattern is deter- mined entirely by diffraction. 16.7 Power Distribution in the Aperture In using geometrical optics to investigate antenna aperture plane power distributions or to design primary feeds for optimum illumination, the im- portant quantity furnished by the ray tube concept is an approximation to the relative power distribution. To determine the relative power distribution it is necessary to know the absolute power distribution only to within an arbitrary multiplicative con- stant. Hence, if either the absolute or relative power distribution is known on a wave front W (see Fig. 16.4), then in using (16.36) in the form _dA ~ dA’ to determine the geometrical optics approximation to the relative power distri- bution on W’ it is necessary to know dA/dA’ only to within a multiplicative constant. For a well-focused point-source-fed antenna system the usual procedure for determining the geometrical optics approximation to the relative power dis- tribution on the antenna aperture plane for a given primary feed power pattern is to first analyze the system when fed by an isotropic point source. For the isotropic feed all wave fronts will be spheres and over any fixed wave front W the power density will be constant. Also, any differential element of area dA on the wave front W will be proportional to the differential solid angle dQ sub- tended by dA as measured at the feed. A general ray trace of the antenna sys- tem will determine the aperture coordinates of the exit rays as functions of the ray direction coordinates as measured at the feed. These relations will deter- mine the area dA’ in the aperture plane W’ (assumed to coincide with a wave front) that corresponds to the differential element of area dA in the wave front W. Since dQ = K dA, where K is a constant, the quantity dQ/dA’ is propor- tional to dA/dA’ and therefore can be used in the calculation of the relative power distribution on W’. If the primary feed has a relative power pattern P(6,6), where 6 and ¢ are the ray direction coordinates, then the geometrical P’ P (16.37)12 ANTENNA THEORY optics approximation to the relative power distribution on the aperture plane is proportional to P(6,¢) d2/dA’, where (in theory) @ and ¢ can be expressed in terms of the aperture coordinates. The geometrical optics approximation to the relative power distribution on the aperture of a well-focused line-source-fed cylindrical antenna system can be determined by procedures analogous to those used for the point source case. It will be assumed that the relative power pattern of the line source feed is of the form P(6,z) = g:(6)go(z), where z is the linear coordinate parallel to the feed line and @ is the angular coordinate around the feed. Since the antenna system is cylindrical, all rays from any one point on the feed line remain in a plane normal to the feed line and ray trace diagrams in all such planes are identical. Thus, power will only flow parallel to these planes and the identity of the ray trace diagrams guarantees that the relative power distribution on the aperture plane in the z direction will be proportional to ga(z). This reduces the problem by one dimension and leads to the consideration of the power flow in a typical plane normal to the feed line. For a feed isotropic in @ the wave fronts are circles and the power density on these circles is uniform. If ds is a differential element of arc length on a fixed wave front C and if dé is the differential angle at the feed subtended by ds, then since C is a circle, it follows that ds is propor- tional to dé. A general ray trace will relate 6 to h, where h is the exit ray aper- ture coordinate normal to z, and this relation will determine the differential element dh in the aperture plane (assumed coincident with a wave front) that corresponds to ds. Since ds = K’ d6, where K’ is a constant, it follows that d@/dh is proportional to ds/dh and therefore can be used in the calculation of the relative power distribution on the aperture plane with respect to the h co- YY F Focal point-~ Parabolic cylinder section Aperture plane -} Fig. 16.5 Central plane section of a parabolic cylinder antenna.WAVE FRONTS, RAYS, AND FOCAL SURFACES 13 ordinate. For the given feed with relative power pattern P(6,z) = 9:(0)go(z) the geometrical optics approximation to the relative power distribution on the aperture plane is proportional to g:(0)go(z) d@/dh, where @ (in theory) can be expressed in terms of h. As a simple example of this procedure, applied to a cylindrical system, con- sider a parabolic cylinder reflector of focal length f with its focal line coincident with the z axis. Figure 16.5 shows a central plane section of this reflector with polar coordinates r, @ describing the reflector surface and the linear coordinate h designating position in the aperture plane x = 0. The equation of the re- flector surface in r, 6 coordinates is -_ 2 ~ 1+ cos6 The primary feed located at F (see Fig. 16.5) is assumed to radiate uniformly in 6, and all rays originating at F are assumed to be reflected parallel to and in the direction of the negative x axis. Thus r = f soe? § 6 h=rsin@ = 2f tan 5 dh = fscet§ do =rdé ad dh If the given primary feed has a power pattern P(@,z) = g(6)go(z), then P,, the geometrical optics approximation to the relative power distribution on the aperture normalized at the center of the aperture, is given by and — f gr(6)ge(2) += 5 gu(O)gx0) (1688) This is certainly a correct result from the geometrical optics standpoint, since it clearly demonstrates that the power spreads radially, i-e., in inverse proportion to distance, only in planes normal to the feed line and only over the distance from the feed to the reflector, ie., over the distance r. After reflection the rays are collimated and no more spreading takes place out to the aperture plane. 16.8 Power Distribution in the Focal Region In the analysis of a non-perfectly focused antenna such as a spherical re- flector or an off-axis-fed paraboloid, it is frequently desired to obtain an esti- mate of the relative power distribution along some focal surface or curve in the focal region. In this case the ratio of a differential element of surface area of the incoming plane wave front to the corresponding differential element of surface area or arc length in the focal region is an estimate of the relative power distribution on the appropriate focal surface or curve.14 ANTENNA THEORY As an example, consider the problem of determining the power distribution along the axis of a spherical reflector of radius a illuminated by a plane wave incident along the axial direction (see Fig. 16.6). By symmetry, all rays after reflection pass through the reflector axis. The height h of an incident ray and the coordinate z of the intersection of the corresponding reflected ray with the reflector axis are related to the angle @ as follows: . a h=asin@ and 2= d0088 2 Henee t= (1 = ) (16.39) |, Incoming plane wave Spherical reflector Reflector center of curvature Fig. 16.6 Central section of a spherical reflector. The differential area dA generated by revolving the line segment dh about the reflector axis is dA = 2nh dh All incident rays passing through dA will, after reflection, pass through the differential line element dz. Thus the power distribution P along the reflector axis is = pA a’, be the final focus point. The condition a’ < a/2 guarantees that the corrector will lie in a region where one and only one ray of the incident ray congruence and one and only one ray of the re- flected ray congruence will pass through each point. If the restriction b > a’ is violated, then the resultant solution may suffer depolarization cancelation. Exercise 16.4 Make a ray trace drawing of the reflection of parallel rays by a spheri- cal reflector and determine the allowable regions for location of a corrector. Equation (16.41) is applicable here.WAVE FRONTS, RAYS, AND FOCAL SURFACES 23 General roy Plx,y) Center of Reflector oxis curvature — vio) 0 Reference wove front Spherical reflector Fig. 16.13 Spherical reflector section in the zy plane. Applying the equal-path-length law to the general ray QPP’F and the axial ray OVV'F we have P+ PP +PR=OV+0V + or a(cos 6) + ¥@’ — x) + (y — y)? + VQ@" — bP + y? = 2a +b — 2a’ where x = a cos 6, y = asin 6, and y’ — y = (x’ — 2x) tan 26. Solving for 2’ and y’, we obtain x’ = acosd [a? cos? @ — 4a(a — a’) cos 6 — (a? + b*) + (2a + b — 2a')*](2 cos? @ — 1) ~ 4(b cos? 6 — acos@ — a’ + a) (16.55a) + e088 — x 22! 0088 — a) Sin g (16.556) y 2 cos?@ — 1 These are the coordinates of the corrector surface in terms of the parameter 6. Provided a’ and b lie in the indicated ranges, the answer given above is valid and unique. (The + sign for the y’ value merely indicates symmetry with respect to the « axis.) 16.13 Focal Surfaces Let us consider a wave front W and its associated ray congruence. Let P be a point on W and let PN be a line segment normal to W (see Fig. 16.14). If Cp, is a line of curvature of W with p = R, at P (see Appendix A), then in the near vicinity of P the normals to W along Cr, appear to converge to a point Qi on PN on the concave side of Cp, at a distance R; from W. Similarly, if Cr, isa line of curvature with p = R; at P, then in the near vicinity of P the normals to W along Cr, appear to converge to a point Q. on PN on the concave side of Cr, at a distance R; from W. Thus to each point P on the wave front W there24 ANTENNA THEORY \\ Envelope 5, Fig. 16.14 Ray paths in the principal planes. is associated a unique pair of points Q; and Q, on the normal line PN. These points are called focal points, and the totality of all these points in general constitutes two surfaces which are called focal surfaces or caustics. These surfaces may degenerate into such forms as a single surface, a surface and a curve, a single curve, or a single point. It is important to note here that the focal surfaces are the same for all wave fronts of a particular ray congruence. The focal surfaces can also be described in terms of the envelopes of certain families of rays. A family in this case consists of the totality of rays passing through a particular line of curvature of W, and the ruled surface so con- stituted is known as a principal surface. Two typical families are shown in Fig. 16.14. It is characteristic of each family that it envelopes a curve in space (S: and S, in Fig. 16.14), and it can be shown that each of these space curves lies on a focal surface. Thus the rays of a congruence are tangent to the focalWAVE FRONTS, RAYS, AND FOCAL SURFACES 25 surfaces along these envelope curves. The totality of all the envelope curves constitutes the focal surfaces. In any specific problem regarding the determination of focal surfaces, either a ray congruence or a wave front will generally be given. Since all wave fronts can be determined from the ray congruence and vice versa, these two forms of the statement of the problem are entirely equivalent. However, in actually carrying out the computations to obtain the focal surfaces, the form of the statement of the problem and the functions involved will determine the best solution procedure. Two procedures will be given here, one particularly appli- cable when the wave front is given and the other when the ray congruence is given. Consider the problem of determining the focal surfaces given a wave front W in the form P= x(uv)k + yup)y + 2(uv)z where P is a vector from the origin to a point P on W and u and v are curvilinear coordinates on W. The normal vector to W at P is given by N=P.XP, where the subscripts denote partial differentiation and the expression is evaluated at P. To each of the two lines of curvature Cp, and Cr, through P there corresponds a principal untt normal vector. These unit vectors will be in the direction of N or —N, whichever points to the concave side of the cor- responding line of curvature. Let fi; be the principal unit normal for Cz, and let fi: be the principal unit normal for Cz, The principal normal radii of curvature R; and R, of W at P are given by the solutions to the quadratic equation (eg — Pet — (Ey — 2Ff + Geo + (EG — FY) = 0 where B=P.P, = Pah PN r=p.p, f= Pa PN G=PeP, g =e and Dt = EG — P= NN The subscripts in the above expressions denote partial differentiation. The focal points of W at P will then be given by the formulas Q=P+Ri and Q=P+ Rh As P ranges over W, the above formulas determine the focal surfaces. Consider now the problem of determining the focal surfaces given a ray congruence. In general, a ray congruence is specified by giving the ray direc-26 ANTENNA THEORY tion at each point on a reference surface, the reference surface not necessarily being a wave front. Let the reference surface V be given in the form R= cup) + y(up)¥ + z(up)z (16.56) where R is a vector from the origin to a point # on V, and let the ray directions at each point on V be specified by the unit vector function w = W(up) If Lis the ray through R, then the distances a; and o from R along L to the focal surfaces are given by the solutions to the quadratic equation (EG — F*)B* + (Eg — 2Ff + Ge)B + (eg — f?) =0 (16.57) where E=Wew, ¢ = Ru Wy F= WW, f=Ruw, = RW, (16.58) G=%%, 9 = Rew, with the subscripts denoting partial differentiation. The focal points on L will be Q=R+ Aw (16.59a) and Q=R+ Bw (16.59) As R ranges over V, the above formulas determine the focal surfaces. As an example consider the determination of the focal surfaces or caustics produced when a paraboloid of revolution receives off axis.’ Let the paraboloid of focal length f be located with its vertex at (0,0,f) and its focus point at the origin and let the incoming rays be in the direction 8; = —sin v¥ + cos yz as shown in Fig. 16.15. The paraboloid surface will be considered the reference surface, and in terms of the parameters r,¢ its equation is 2p t=reos@ y=rsing TS (16.60) Applying Snell’s law, (16.27c), at the reflector surface leads to the reflected ray directions 8, = ak + a + ad (16.61) _ 2r* sin y sin ¢ cos ¢ — 4rf cos y cos ¢ = P+ ap __ 2r* sin y sin? ¢ — 4rf cos y sin ¢ oy = Pye _ 4rf sin y sin ¢ — 8f? cos y a where a: —siny (16.62) a, + cos yWAVE FRONTS, RAYS, AND FOCAL SURFACES 27 ™~ Paraboloid x?+y? =-4/(2-F) Fig. 16.15 Geometry of paraboloid receiving a plane wave off axis. With —, . 4 F Tt od wW=§&, the equations (16.60) to (16.62) constitute a normal congruence in the usual form of a reference surface with ray directions specified at each point on the reference surface as a function of curvilinear coordinates. The second pro- cedure for determining the focal surfaces is therefore the most applicable in this case. The fundamental quantities become R=rcos¢%+rsin gy + B= 16f?(cos* y + sin? y sin? ¢) rr P= 8rf sin y cos ¢ (2f sin y sin @ + r cos y) aa—rs G- gp CAP = (2f sin y sin ¢ +r cos y)? a and e = 2tsiny sin @ — 2f cos f=0 9= ape TSin ysin 6 — 2f e084 where A = r? + 4f%. Substituting into (16.57) and solving for @ (with much labor), we obtain 8 woe teal — 4f2) sin? y + 8f? — 4rfsin y cosy sin = 16f?(2f cos y — r sin y sin ¢) & (? + 4f?) sin ya/L — [= = 47) os a sin y sin a728 ANTENNA THEORY Equations (16.59a,b) then give the focal surfaces in terms of the curvilinear coordinates r, ¢. In terms of cartesian coordinates 2, y, z on the paraboloid surface the equa- tions for the focal surfaces become _ _ Br\. _ y (2f — z)siny |i a= (oa) ly [ht apemy as _ _ _@f = 2) cosy |}? where + |. aa Feosy —ysiny Z (16.63) B = 2f —z+ cosy (zcos y — ysin y) _ . _ (ze087 — ysiny) & (2f — z) siny qi (Epa usr =a ) (16.64) Calculations for the particular case y = 20° have been carried out, and the focal surfaces have been calculated and constructed.” Figure 16.16a to ¢ shows the focal surfaces individually and superimposed. 16.14 General Comments on Focal Surfaces In general, the normals to a wave front through a line of curvature tend to focus on one or the other of the two focal surfaces. The physical extent of the focal surfaces is therefore a measure of the focusing ability of the ray con- gruence associated with the wave front. If the focal surfaces are extensive and far apart, then the focusing is poor. If the focal surfaces are small and closely spaced, then the focusing is good. If the focal surfaces degenerate to a line or to a point, then the focusing may be considered as perfect, depending on the application. If it is desired to receive the energy of a converging wave front by means of a point source, then, in general, the optimum position for the point source will be either on one of the two focal surfaces or between them. Regions where the focal surfaces are close together or intersect generally have high energy density. Degenerate focal surfaces are also regions of high energy density. Since every ray is tangent to each of the fwo focal surfaces, the total energy in the wave front can be collected by locating properly phased receivers on one or on the other or on parts of both of the focal surfaces. For purposes of low side lobes in a focusing system there generally is a power taper across the wave front whether considered in the focal region or in the aperture plane. This taper places greater importance on the rays through the center of a focusing system (i.e., rays near the chief ray) than on rays near the edge of the system. The existence in general of two distinct focal surfaces in the focal region of a focusing system indicates the presence of astigmatism. For each different direc- tion of incoming energy the separation of the focal points Q: and Q, (see Fig. 16.14) along the chief ray is particularly important because it is a measure ofWAVE FRONTS, RAYS, AND FOCAL SURFACES 29 Fig. 16.16 Focal surfaces of a paraboloid receiving a plane wave 20° off axis. (a) Paraboloid and one focal surface; (b) paraboloid and a second focal surface; (c) paraboloid and com- posite of focal surfaces. the astigmatism in a region of high power density. In most one-dimensional scanning systems the plane of scan is a plane of symmetry of the system and also a principal surface of all ray congruences in the focal region. The other principal surfaces for all chief rays in the plane of scan will then be normal to30 ANTENNA THEORY the plane of scan. As the direction of the incoming energy changes in the plane of scan, the two focal points associated with the chief ray describe two curves. Rays in the plane of scan near each chief ray will focus on one of these curves known as the 7’ (tangential) curve, while rays near the chief ray but ina plane normal to the plane of sean will focus on the other curve, known as the S (sagittal) curve. The separation of the S and 7' curves is thus a measure of the astigmatism along the chief ray for directions of incoming energy in the plane of scan and hence is an indication of the scanning capability of the system. If the S and T curves are close together, then beam scanning can be accomplished by point source feeding along a mean curve between the two. If multiple beams broad in the plane of scan but narrow in the other dimension are de- sired, then point source feeding along the S curve will produce the desired result. If multiple beams narrow in the plane of scan but broad in the other dimension are desired, then point source feeding along the 7’ curve will produce the desired result. APPENDIX A Principal Normal Radii of Curvature, Principal Directions, Principal Planes, and Lines of Curvature Let P be a point on a surface W and let PN be a line segment normal to W at P (see Fig. 16-A.1). Each plane through PN is a normal plane to W and intersects W in a curve C, each curve C having a unique radius of curvature p at P. In general, for each point P there is also a unique position of the normal plane such that the intersection curve ( = Cy has a radius of curvature p = Ry at P that is maximum (see Fig. 16-A.2). Similarly, there is a unique position of the normal plane such that the intersection curve C = C2 has a radius of curva- Qo N Fig. 16A.1_ Surface with normal line and normal plane.WAVE FRONTS, RAYS, AND FOCAL SURFACES 31 Fig. 16A.2 Principal normal radii of curvature and principal directions at a point on a surface. ture p = FR, that is minimum. These extreme values, 2, and Rz, are called the principal normal radii of curvature of W at P; the directions of C, and C2 at P are known as the principal directions of W at P; and the normal planes whose intersections with W produce C, and ( are called the principal planes of W at P. Acurve on W that at each point has a direction corresponding to a principal direction of W at that point is known as a line of curvature of W. It can be shown that the principal planes at each point are mutually orthogonal and hence the two families of lines of curvature form an orthogonal curvilinear co- ordinate system on W. APPENDIX B Geometrical Optics Power Flow in a Source-free, Nonconducting, Isotropic, Homogeneous Medium The ray paths for this case are straight lines (see Sec. 16.2). In Fig. 16-B.1, P is a point on a wave front W, P’ is the corresponding point on the wave front W’, and d is the distance from P to P’. The differential element of area dA on W is centered at P, and the corresponding element dA’ is determined by the intersection of the ray tube through dA with the wave front W’. The32 ANTENNA THEORY Principal normal centers of curvature R Fig. 16B.1 Geometrical optics power flow. curves C; and C2 through P are lines of curvature of W and hence are mutually orthogonal (see Appendix A). The corresponding curves C and Cin W’ are also orthogonal lines of curvature. If the principal normal radii of curvature of W at P are R, and R, and the centers of curvature are located as shown in Fig. 16-B.1, then the principal normal radii of curvature of W’ at P’ are Ri = Ri +d and R, = R, +d. Thus dA is proportional to RiRy and dA’ is proportional to R;R; = (Ri + d)(R2 + d). If Sis the power density at P and if S’ is the power density at P’, then (16.36) becomes SdA = S'dA’ SdA s Riks or S =a ~S@ toda (16-B.1)WAVE FRONTS, RAYS, AND FOCAL SURFACES 33 The gaussian curvature K of W’ at P’ is defined to be eto K = Fe TF OC +o) (16-B.2) and hence is a function of d. Since K(0) = 1/RiRe, we can write - gk@ ~ "KO This equation relates the power density S’ at the point P’ on a ray to the power density S at a reference point P. The power flow so expressed is equivalent to the power flow implied by the field intensity relation (16.26). Ss (16-B.3) PROBLEMS 16.1 In a source-free, nonconducting, isotropic, homogeneous medium use the rela- tionship of the differential areas dA and dA’ expressed by (16-B.1) to (16-B.3) of Appen- dix B (see also Fig. 16-B.1) to show that J fevansan: F-4A = 0 where F is given by (16.22) and 2 is the lateral surface of a ray tube whose end faces are dA and dA’ as shown in Fig. 16-B.1. 16.2 Derive Snell’s law of reflection for a plane reflector directly from Fermat’s principle by minimizing the sum of the path lengths from a general point on the plane reflector (0,y,z) to two fixed points (z1,y1,0) and (22,y2,0), 21 > 0, zz > 0. 16.3 By using the procedure of Sec. 16.9, given P() = csc? d, o: = 5°, ¢2 = 30°, and P,(8) = cos* 6, determine ¢ = $(6). 16.4 Assume a linearly polarized plane wave incident on the concave face of a spheri- cal reflector. Use the geometrical optics approximation for polarization (Sec. 16.10) to determine the direction cosines of the polarization vector on a general reflected ray. 16.5 Let (2,yo,20) be the location of a line source feed in the focal region of the parabolic cylinder y? = 4f(f — z). Consider the rays that originate at the feed and are reflected by the parabolic cylinder at the point (x,y,z). If B is the direction angle relative to the y axis of the reflected rays, set 98 = Oalongy =0 oy and show that, for rays reflected from the reflector in the neighborhood of the vertex line, y = 0, the values of yo and 29 for which the ray divergence in the yz plane is mini- mum are given by the relation yor = 2(f — 20) 16.6 Given two wave fronts W and W’ and their associated families of rays in three dimensions, let R be a reflecting surface designed according to the equal-path-length law, i.e., so that path lengths are constant along all ray paths from W to R to W’. At each point on R assume there is incident a unique ray from W and a unique ray from W’. By means of these rays any curvilinear coordinate system p, q on R will determine corresponding curvilinear coordinate systems on W and W’ with p and q as parameters34 ANTENNA THEORY along the coordinate curves. If the vectors W, R, and W’ designate points on the sur- faces W, R, and WW”, respectively, then the equal-path-length requirement can be ex- pressed as follows: !R(p.q) — W(p.9)| + |W'(p,) — R(p,q)| = constant Differentiate this expression with respect to p and q and use the results to show that Snell’s law of reflection is satisfied at R. Thus for the reflective case, satisfaction of the equal-path-length law implies satisfaction of Snell’s law. (A similar proof is possible for the refractive case.) 16.7 a. A parabola can be defined mathematically as the locus of points in a plane each of which is equidistant from a fixed point called the focus and a fixed line called the directrix. Use this definition together with the equal-path-length law to prove the fundamental focusing property of a parabola, i.e., rays originating at the focus point will after reflection be collimated parallel to the axis of the parabola. 6, Anellipse can be defined mathematically as the locus of points in a plane the sum of whose distances from two fixed points, F, and F> (the focus points), is a constant. Use this definition along with the equal-path-length law to prove the characteristic optical property of the ellipse, i.e., all rays originating at a focus point are reflected so that they pass through the other focus point. c. A hyperbola can be defined mathematically as the locus of points in a plane the difference of whose distances from two fixed points, /’; and F2 (the focus points), is a constant. Use this definition along with the equal-path-length law to prove the char- acteristic optical property of a hyperbola, i.e., all rays originating at a focus point are reflected by either branch of a hyperbola so that they appear to have originated,at the other focus point. 16.8 Assume a plane wave incident on a spherical reflector along the reflector axis. Write an equation for the one-parameter family of reflected rays lying in a plane through the reflector axis. Find the envelope of this family by differentiating the equa- tion of the family with respect to the parameter and solving simultaneously with the family equation. This envelope is a section of the focal (or caustic) surface char- acteristic of a wave front suffering spherical aberration. Rotation of this envelope about the reflector axis generates part of the focal surface. The rest of the focal surface is degenerate. What is it? REFERENCES 1, Hildebrand, F. B.: “Advanced Calculus for Engineers,” p. 294, Prentice-Hall, Inc., Englewood Cliffs, N.J., 1949. 2. Luneburg, R. K.: Mathematical Theory of Optics, Brown University notes, Providence, RL, 1944. 3, Kline, M., and I. Kay: “Electromagnetic Theory and Geometrical Optics,” Inter- science Publishers, Inc., New York, 1965. 4. Kouyoumjian, R. G.: Asymptotic High Frequency Methods, Proc. IEEE, vol. 53, pp. 864-876, August, 1965. 5. Sletten, C. J., et al.: Corrective Line Sources for Paraboloids, IRE Trans. Antennas Propagation, vol. AP-6, pp. 250-251, July, 1958. 6. Spencer, R. C.: Synthesis of Microwave Diffraction Patterns with Applications to ese? @ Patterns, Mass. Inst. Technol. Radiation Lab. Rept. 54-24, June 23, 1943. Pro- cedure described here is due to L. J. Chu. 7. Silver, 8.: “Microwave Antenna Theory and Design,” M.LT. Radiation Laboratory Series, vol. 12, pp. 132-134, McGraw-Hill Book Company, 1949.WAVE FRONTS, RAYS, AND FOCAL SURFACES 35 Eisenhart, L. P.: “A Treatise on the Differential Geometry of Curves and Surfaces,” p. 393, Dover Publications, Inc., New York, 1960. Eisenhart, L. P.: Ref. 8, p. 403. Holt, F.S., and E. L. Bouche: A Gregorian Corrector for Spherical Reflectors, IEEE Trans. Antennas Propagation, vol. AP-12, pp. 44-47, January, 1964. Calculation of the Caustic (Focal) Surface When the Reflecting Surface Is a Parab- oloid of Revolution and the Incoming Rays Are Parallel, Parke Math. Lab. Study 3, Contract No. AF19(122)-484 for AFCRL, Concord, Mass., Feb., 1952. Calculations of the Caustic Surface of a Paraboloid of Revolution for an Incoming Plane Wave of Twenty Degrees Incidence, Parke Math. Lab. Rept. 1, Contract No. AF19(604)-263 for AFCRL, Concord, Mass., May, 1952.CHAPTER 17 REFLECTOR ANTENNAS C, J. Sletten 17.1 Importance of Reflector Antennas Reflecting metal surfaces have important advantages in the design and construction of large aperture antennas. For instance, the analysis and con- struction of reflector surfaces are simpler than those of lens and array antennas of comparable performance, and most reflector antennas perform well over a wide frequency spectrum. Although reflector systems lack the degrees of freedom of lenses, which have two surfaces and the index of refraction avail- able, or of arrays with many feeding coefficients to be specified, these deficien- cies can be largely corrected by the use of extended primary feed systems properly located and excited in the focal region. Reflector surfaces composed of many movable surfaces can achieve a wide variety of focusing and beam- forming characteristics. Such combinations of reflector and focal-region feed designs can lead to giant, low-cost-per-unit-area antennas with capability for beam scanning and pattern synthesis. We shall develop the theory and design principles needed to form metal surfaces for high gain and for shaped antenna patterns. 17.2 Types of Reflector Antennas and Their Primary Uses It is possible to classify reflector antennas according to the geometric form of the surface, Surfaces of revolution including paraboloids, ellipsoids, hyper- boloids, and spheres generated from conic sections, as well as cylindrical and toroidal surfaces also generated from conic sections, are used. In addition, special reflector surfaces are used for synthesis of special antenna patterns. Because paraboloids of revolution and spheres are most frequently used, their characteristics, together with their appropriate feeding structures, will be described in greatest detail. It is well to have in mind the primary uses for reflector antennas, the func- tions they perform, and their principal advantages in order to appreciate the significance of the analytical emphasis. Reflectors are used chiefly as high-gain collecting or collimating antennas. Reflector designs are usually based on geometric optical or zero-wavelength approximations. A precisely constructed metal reflector collects and focuses electromagnetic waves from optical wave- 36REFLECTOR ANTENNAS 37 lengths to wavelengths about one-third the aperture dimension. In this high-gain collecting function certain design objectives are usually sought. We often wish to control the side-lobe radiation outside the high-gain principal lobe or pattern caused by aperture diffraction, primary-feed spillover, feed block- ing, and other sources of spurious lobing. As with an optical telescope we would also like to produce many high-gain beams from a single reflector aperture, or equivalently to change or scan the direction of high-gain antenna pattern(s) by motion of the feed in the focal plane of the reflecting collector. These objectives require a wide-angle focal region for the reflector antenna. It will be shown that the spherical reflector, selected portions of a paraboloid, a parabolic torus, and certain other reflector surfaces have wide-angle focal regions allowing the generation of multiple antenna patterns of high gain with suitable feeding methods. In addition to these primary uses, reflector surfaces are also used to produce special shaped patterns such as the esc? @ pattern used in search radar systems. In this pattern synthesis function, complicated reflector contours are usually fed or illuminated by a small point or line source feed to generate a prescribed antenna power pattern. Alternatively, a wide-angle focusing reflector antenna of the type mentioned above can produce similar wide-angle search patterns by construction of suitable extended sources in the broad focal regions. Less frequently, reflectors are used as primary feeds for other high-gain antennas. The most familiar example is the Cassegrain antenna system, which employs a small hyperboloidal reflector to feed a large paraboloidal reflector. 17.3. The Paraboloid with Source at the Focus The concave surface of the paraboloidal reflector, or ubiquitous “dish,” is by far the most frequently used reflector antenna. Because in a geometrical optic sense a mirror of this form focuses an incident electromagnetic plane wave propagating parallel to its axis to the focal point of the paraboloid of revolution, the antenna is considered “unaberrated” and its radiation patterns depend largely on diffraction effects. When plane waves are incident from directions other than along the axis or, conversely, when transmitting sources are located at positions other than at the focus of the mirror, the circularly symmetrical paraboloid is highly aberrated and radiation patterns degrade sharply in gain and side-lobe levels from those obtained by a small feed lo- cated at the focus. We shall first analyze the at-the-focus characteristics of the reflector, treat- ing primarily the influence of the primary-feed patterns on the secondary or far-field pattern of the paraboloid,! and discuss how the amplitude taper or power distribution on the aperture depends on the feed pattern, focal length, and the section and edge contour of the paraboloid surface selected. Polariza- tion components introduced by reflector curvature are discussed with methods38 ANTENNA THEORY for summing the surface currents on the reflector. The effect of phase varia- tions on the aperture field due to feed displacement from the focus will be treated by using optical principles developed in Chap. 16. Finally, methods are developed to show how primary feeds may be located and designed for the short-focal-length paraboloid to synthesize antenna patterns of complex shape and to scan antenna beams by motion of the primary source. Certain approximations are made in finding the far-field patterns for a circularly symmetric paraboloid (see Fig. 17.1) functioning as a transmitting antenna with a source located at the focus F. For the analysis of most micro- wave situations we can separate the radiation fields of the paraboloid into several regions: (1) a primary-feed region near the focus containing a primary feed giving rise to a directional primary or feed pattern P;(0,¢), which is assumed not to be influenced by interactions with the reflector; (2) the re- flector region with surface currents due to the fields of the primary feed flowing only on the concaved side of the reflector, which is usually considered a perfect conductor; (3) an aperture region usually taken as a plane surface through the focus F (see dotted circle in aperture plane in Fig. 17.1 normal to the parab- oloid axis with the boundary shape a projection of the outer edge of the paraboloid; this plane is located in the “near zone” directly in front of the reflector) ; and (4) the exterior region containing the diffraction fields of interest including the Fresnel field and Fraunhofer field (far-field) regions. Ge \ | a Poraboloid of revolution Aperture =< plone La Fig. 17.1. Polar, cylindrical, and rectangular coordinates defining paraboloid of revolution reflector and its aperture plane.REFLECTOR ANTENNAS 39 The Aperture Phase The paraboloid surface shown in Fig. 17.1 may be expressed in rectangular or polar coordinate systems which have their origin at the focus F. In the 2, y, 2 system e+ y = 4f(f—2z) with x? + 7? < a@ (17.1) and in spherical coordinates p, 0, @ =F = goer f p= T+ os07 f sec’ 5 a<% (17.2) It is convenient also to represent positions in the aperture plane in terms of polar coordinates r, w. For the paraboloid of revolution all path lengths measured from the focus along rays to the reflecting surface and then along reflected rays parallel to the axis out to the aperture plane are equal to 2f. Thus p+ pcos6 = 2f which is merely a rewriting of (17.2). The assertion that the reflected rays are parallel to the z axis implies that Snell’s law is valid when applied to each differential element of surface dS. We shall use the ray approximations of geometric optics applied to the parabo- loid geometry to describe the field distribution on the aperture plane due to a small source at F. Using (17.2), a normal to the reflector surface will be 8 @ ~ 6 6. — a 2 = = VF — pcos?s cos? 6 + sin 5 cos 56 The unit normal i is then i = —cos 56 + sin 9% (17.3) 2 2 Since fi-6 = fi-2 = —cos (6/2), the projections of the unit normal fi on the incident ray s, and reflected ray s, are equal, satisfying the Snell’s ray relation- ships on a reflecting plane surface. Thus the electrical phase on the aperture is constant for electromagnetic waves diverging from a point source at the focus within the validity of Snell’s law applied to each differential area of the reflector surface. When aperture diameter and focal length contain several wavelengths, the aperture phase is found to be very nearly uniform (when a small source is located at the focus). The Aperture Amplitude Distribution The amplitude distribution or power density on the aperture is not uniform when the reflector is illuminated by a point source at the focus. This can be seen by recalling that energy diverging from a point source decreases in in- tensity as 1/p* in three dimensions (and as 1/r in two-dimensional space; see40 ANTENNA THEORY Sec. 16.7). After reflection from the paraboloid, all rays are parallel and thus maintain the same energy density distribution for any aperture plane con- structed normal to the axis in the near zone of the paraboloid. An instructive method for showing the natural amplitude taper on the aperture of a paraboloid fed by an isotropic radiating source is to consider the angular divergence of power. Note that, from Fig. 17.2, r= psing or, using (17.2), r= 2%tang (7.4) Incident roy 8 Fig. 17.2 Cross section of paraboloid showing optical rays. For the paraboloid we need to consider the power radiated by an omnidirec- tional source in the solid angle formed by rotating d@ about the z axis: dp = 2nC sin 6 do where C is constant. This power must also appear in the aperture ring dA = 2erdr. Thus dp _ 2rCsin@d6_ ,, do dA” Qardr_——s pdr _ C _ Ccost (6/2) =30 dp 1 “r a4 © + CPE in any plane through the paraboloid axis.REFLECTOR ANTENNAS 41 If the source is not omnidirectional, but radiates a rotationally symmetric pattern G(@), dp 8 a G@) cos’ 5 (17.5) Thus this ideal symmetric paraboloid shown in Fig. 17.1 fed by a small source at F produces a “patch” of aperture field distribution with equiphase fronts normal to the axis and an amplitude distribution which is nonuniform unless G(6) = sec (6/2). Polarization Components for Focus-fed Paraboloid To study the polarization of the aperture fields we must introduce a descrip- tion of the polarization of the primary source and also boundary conditions for fields on the conducting paraboloidal surface. If we treat each differential area on the reflector as a large plane conducting mirror, then from the optical methods given in Sec. 16.4 and 16.10 we can write for rays in the direction 8; incident on the paraboloid surface (see Fig. 17.2) with inner normal fi that the reflected ray will be in the direction 8, such that ax &—8 > (17.6) or The electromagnetic boundary conditions that the tangential E fields vanish on a perfect conductor and the E; and E, are continuous mean that ax (+E) =0 a-E, = 2-E, (17.7) or E, = 2(8-E,)a — E; [E,| = [Ei| By assuming also that both the incident and reflected fields are TEM waves in vacuum, or H; = ./* @ x E) (17.82) Ho H,= ve @, x E,) (17.88) then the surface current density is J.=aixX Gi+ 4H) (17.9)42 ANTENNA THEORY Substituting (17.6), (17.7), and (17.8), (17.8b) into (17.9), we obtain J. = 2X Hi) = 2,/@ [a x @ x E,)] (17.10a) J. = 28 XH) = 2/8 i x @ xX ED] (17.106) Exercise 17.1 Show that fi X H; = i X H, on a perfectly conducting reflector. We can now proceed to find the surface currents on the reflector surface and the field produced by these currents on the aperture plane for an idealized point source feed linearly polarized with gain G;(6,¢). We shall study the interesting case of a small electric dipole at the focus oriented parallel to the y axis. The radiated power at a large distance p surrounding a source can be found from the gain function of the feed G;, or P P0,8) = 7°6,(0,9) where P, is the total radiated power of the source and P; is the radiated power per unit solid angle. We can express the incident E, vector on a sphere of radius p about a primary feed located at F with pattern P;(6,¢) as E, = va("2)" Pio" awe, (17.11) assuming we are in the far field of a primary feed linearly polarized in the y direction and 6, is a unit vector in the direction of the component of § that is normal to 6. We can for convenience let C; = V2 (uo/eo)“(P./40) 4. Note that at the aperture plane the factor e~*/p produces the constant phase and the cos! (6/2) amplitude taper discussed earlier. To study polarization effects we can write enter E; = CG; (0,6) ]"é& > (17.12) To examine polarization components for a paraboloid fed by a point source at focus we compute the currents on the reflector by substituting (17.12) into (17.102) and evaluating J, on the paraboloidal surface as follows: L= 2/2 Cit X (6 X &) ee)” onthe (17.18) A source polarized parallel to ¥ can be expressed in spherical coordinates by using the relationship ¥ = sin ¢sin 66 + sin ¢ cos 46 + cos ohREFLECTOR ANTENNAS 43 The unit vector é; is a, - 2X G Xs) _ ¥= GH 1X Gxal ly- @Hel hence &= sin ¢ cos 06 + cos oh _ sin ¢ cos 06 + cos $$ * sin? ¢ cos? 6 + cos? ¢ V1 — sin? ¢ sin? 6 Finally, in (17.13), we can evaluate the vector i X (6 X @,) by using (17.3): us fix (@X &) = -8)9 — G-a)é _ sin (6/2) sin ¢ cos 86 + cos (6/2) sin $ cos 8 6, cos (6/2) cos o$ V1 — sin® ¢ sin? 6 v1 — sin? ¢ sin? 6 Converting to rectangular coordinates, we obtain the following components of current on the reflector: u=aX (6X &) = [ -fsin 6 cos # sin # sin 5 + § cos 5 (sin* # cos # + cos ¢) 1 —Zsin ¢ cos @sin 5] Vrosn'gsnte (17.14) or = Cw HOT a0 n= 2 (17.15) To find the aperture field E,, due to the reflector currents (17.15) we write E, in the form E,= (C1 aes e008, (17.16) Now remembering that the rays reflected from the surface are in the direction &, = —2 (see Fig. 17.2) and using (17.10b), ic, ool Pp J. = 20 aX (—2 X & enter (17.17a) where from (17.10), (17.13), and (17.15) u= AX @X &) = fx (-2 x &) = —(4-6)2 + G28, = —@-6)2 — cos 5 8 (17.178) _ Xsin ¢ cos $ (1 — cos #) — F(sin? ¢ cos 6 + cos? $) V1 — sin? ¢ sin? 6 (17.184) Ey = C1 CE") -—rassne, (17.18b)44 ANTENNA THEORY Exercise 17.2 Show that &-€, = —u, on the paraboloidal surface [which implies, together with (17.17b), that uz + u,¥ + cos (6/2) é, = 0]. Use (17.7) together with the definitions of é; and @, as implied by (17.12) and (17.16) to write 5B _o@E. Boy... 2 & = 2 = 277 a a = 28-4) — 6 fel? Tey fei 2 and evaluate to obtain (17.18a) and (17.18). It is interesting to study the aperture fields when a short electric dipole of length l, « oriented along the y axis is located at the focus. Usually the far field about a short electric dipole is written in spherical coordinates (R,7,8) with the polar axis y = 0 along the axis of the dipole. If the polar axis y = 0 is the y axis, then in terms of the coordinates of Fig. 17.1 sin y = V1 — sin? $ sin? 6 Thus for an electric dipole feed we have only to substitute SkoDoly PVT — sin? Osin®§ for — CHG (0,6)* in (17.18b) to obtain a somewhat simpler expression for the aperture field as the expression 1 — sin? ¢ sin? 6 is eliminated. Note that the x or cross component of aperture field vanishes as 1 — cos @ in (17.182) for @—> 0, which means that for large f/D ratios the reflector section is relatively flat and hence generates weak cross-polarized components. There is a technique? for eliminating cross-polarized aperture field com- ponents on the reflector section of small f/D ratios by primary-feed design. When the transmitting source at focus consists of a small area of y-polarized plane wave propagating toward the reflector, the undesirable z components of aperture field disappear. A small or differential sector of linearly polarized plane wave can be produced by an electric and a magnetic dipole oriented at right angles and excited in phase with relative magnitude {. Such a radiator is called a Huygen’s source. For the short magnetic dipole of moment — MX we can write for the far field with E components in the yz plane hehe 4nR When this source is located at the focus and the illumination pattern [P(6,¢)]** is calculated from E,,, we can again calculate the surface currents on the reflector and from them find the fields on the aperture. The x component of the aperture E field has the same dependence on the position coordinates (p,8,¢) as the x components for the y-polarized electric dipole except that the sign is reversed. When the ratio of the electric to the magnetic dipole moments at F is made proportional to \0/é,, the cross-polarized components of aperture fields vanish. (cos 6¥ — sin 6 sin $2)REFLECTOR ANTENNAS 45 Exercise 17.3 Carry out the necessary steps to show that the cross-polarized com- ponents vanish. In practical feed design a small electromagnetic horn with aperture impedance nearly equal to {» will aid in reducing cross-polarized components in the radia- tion field. By studying the symmetry of the field components on a circularly symmetric aperture for simple dipoles or horn feeds, one can show that in the principal planes zz and yz the sums of the positive and negative components of cross- polarized fields over the aperture are equal and therefore radiation patterns taken in these planes are free from cross polarization. The 45° or skew patterns contain most of the cross-polarized energy introduced by reflector curvature under most conditions. Diffraction Patterns for the Focus-fed Paraboloid General procedures for calculating radiation patterns from aperture field distributions or from surface current distributions on metal surfaces have been developed in Chap. 3. Both of these methods can be applied with advantage to the paraboloid to obtain the antenna gain, beamwidth, and side-lobe and polarization characteristics of the far-field patterns. Diffraction methods”? have also been extended to treat the paraboloid with simple feeds displaced. laterally off the geometric focus. We shall discuss these results after examining the geometric optics structure of the focal region for an arbitrarily located feed point. Our purpose in this section is to show how pattern characteristics and aperture efficiency relate to available design parameters, which are primary-feed pattern Gp(6,4), f/D ratio, and size and shape of the paraboloid section. The simplifying assumptions mentioned previously are used in order to make computations tractable, and they consist in making use of the optical ray-tracing technique we have developed thus far to calculate the phase, amplitude, and polarization of surface currents or aperture fields. Radiation fields are then calculated by a superposition of the direct radiation from the primary source in directions greater than ® and the radiation from the teflected-field distributions based on the expressions for J, and E,,. Whether using the surface current method or the aperture field method, it is generally assumed that the currents on the shadow region of the reflector have negligible effect on the far-field pattern and hence may be set equal to zero. Most analy- ses have been carried out for axially centered circular reflectors and primary feeds with no ¢ variations in pattern. Edge effects are usually neglected, and the pattern on the shadow side of the reflector is not obtained. To represent the far-field radiation patterns we shall use (R,O,®) as co- ordinates for a far-field point as illustrated in Fig. 17.3. For consistency with the equation of continuity the current distribution method requires the presence of line charge distributions along the edge of the46 ANTENNA THEORY y Fig. 17.3. Coordinates used to express far fields of a paraboloid. reflector. The contribution to the far-field pattern of these line charge dis- tributions is just sufficient to cancel the radial far-field components contributed by the surface currents and surface charges on the face of the paraboloid. Neglecting direct radiation from the source, the far-field pattern’ is [see Eq. (3.47)] E= Teme ff Ue - ¢ reflector surface R) exp (jh R) dS Taking components, we can write Ee - es eciheR Sf -J. exp (jhog-R) dS (17.194) Ey =~ eine ff b-J, exp (jheg R) dS (17.190) To interpret this, we evaluate the integral, using (17.17a,b), , 4 2x 70 (G;(6,6)]% [f,Seexp Ghe®) as = aoc, [* [°° eo BX (6 X &,)emserll-c0s 6 cos G-sin # sin @ cos (-#)]p? sin 8 sec $a dg (17.19e) where fi X (6 X 6) =.cos (6/2) &, — (@-6,)Z from (17.17b). The z term will not contribute to Zs and its contribution to Eg is proportional to — 6-2 = sin 0, which is nearly zero for the narrow beams we are concerned with, so this term may be omitted in (17.19c). Principal E- and H-plane patterns are patterns measured or computed in the yz and zz planes, respectively, if we take the principal polarization vector to be in the y direction or that both the source and aperture field are dominantly jy-polarized. Then the Z-plane pattern is defined to lie in the yz or 6 = +1/2 plane. With asymmetric reflector and feed, i.e., no ¢ dependence, the E-planeREFLECTOR ANTENNAS 47 pattern will have only Eg components. The absence of cross-polarized fields in the E-plane pattern under these conditions of symmetry is seen from the integration of all the cross-polarized current sources on the reflector according to Eq. (17.19c), which yields a zero contribution to the radiation pattern in the xz plane. Likewise, the H plane is thus defined with reference to the dominant polarization component to lie in the zz plane and will under these conditions have only a single Zp component. The z components of surface current radiate like small electric dipoles oriented along the z axis and therefore do not contribute to the far-field pattern along the axis of the paraboloid. They do, however, produce contributions to the diffraction pattern in directions away from the axis. A somewhat more accurate radiation pattern is obtained from the surface cur- rents directly rather than by use of the aperture field; which involves a second application of Snell’s law with its optical approximations. The aperture field and surface current methods therefore result in somewhat different far-field patterns. However, as wavelength vanishes, the diffraction pattern for a given aperture is concentrated near the axis, where the fields produced by the two calculations become identical. Using the aperture field approximation, cos @ = —1, so that by (17.2) p(1 — cos © cos 6) = 2f. Changing variables by r = p sin 6 leads to the follow- ing expressions for the principal patterns. E plane: =_ Jeouo enh cnn x) FO aan 8 lacan Oak e \g3e) 08 ® a % f i Mere enyeitor sin © sin ér dg dr H plane: =— Jono ens R+2f) 1 ry" Bol EO, aR (Ea | . i 2 [G,(6,0)]* enyeiter sin © £08 6 dy dr Pp To evaluate the above expressions we can substitute 4p tr at = @=2tan's p= if 2f and ¢,y is obtained from (17.18a). If we assume that the primary-feed pattern is independent of ¢, a simple formula for the gain on axis can be obtained. The axial far field can be ex- pressed, from (17.19c), as E(R,0,0) = jee eanesan( 5 i y" [2° [5 eels o sin 0a ag48 ANTENNA THEORY Assuming no depolarization of the surface currents, that is, e,, = —1, and expressing the power | per unit solid angle in terms of E, we get for the gain ” @= Poo =* ee ! oF (17.20) Aperture Blocking Blocking is a damaging effect in antenna design where a two or three square wavelength feed may represent 10 percent or more of the reflector aperture. The loss in gain can be computed approximately from the ratio of aperture area to feed blocking area, and the effect on side lobes of the diffraction can be studied by assuming linear superposition of the diffraction pattern from the blocked area uniformly illuminated but 180° out of phase with the total un- perturbed paraboloid aperture distribution F(x) .45 In one dimension the aperture blocked by a feed at the center of the aperture of width 2w would have an aperture distribution approximately as shown in Fig. 17.4. We can compute the loss in gain and change in side-lobe levels by the fourier transform method as follows: S(ko sin €=) = f ~ F(@)eitke sin 8 dx f ° F(a)eittosin # dx where the net field is assumed zero between —w and w. To evaluate we rewrite, where u = sin 0, flu) = f. F(a)e*one da — {- F(x)" dr When w is relatively small, F(x) can be considered constant and also the normalized maximum of the pattern in the region |x| < w. Thus Flu) = [2 Pao de — 2p Bowe) kouw is the modified field pattern. Amplitude F(x) Fig. 17.4 Aperture-blocking effect due to opaque feed.REFLECTOR ANTENNAS 49 Methods for Controlling the Diffraction Pattern Several methods* for controlling diffraction patterns are effective in antenna design. First the off-axis or asymmetric aperture (see Figs. 17.6 and 17.13) with the vertex at one reflector edge is used to reduce aperture blocking by the feed. This configuration also practically eliminates any interaction between the primary feed and the reflector which can cause standing waves (inter- ference) in the primary-feed lines due to the specular reflection when the feed horn is directed toward the vertex. The shape of the reflector contour also influences the diffraction pattern. Because a uniformly illuminated rectangular aperture with an edge parallel to the horizon has —13-db side lobes in the vertical and horizontal planes but —26-db side lobes in diagonal planes, antenna apertures are sometimes cut diamond shape or with an elliptical contour to produce low-diffraction lobes in a plane of scan or to shape slightly the search patterns as required. The spillover lobes caused by direct radiation of the primary feed outside the cone & are also undesirable. These lobes raise the noise temperature of the antenna by permitting warm-background objects to radiate directly into the primary feed. These lobes can be only partly reduced by making the reflector larger or primary feed more directive. One remedy is to surround the aperture with absorbing material.’ This kind of dark shielding is common in optical sys- tems. Its application to antenna systems is best accomplished by studying the Fresnel field near the paraboloid aperture. Figure 17.5 shows how the fields pro- ducing side lobes build up. The illustration shows the contours of equal phase and amplitude in front of the paraboloid before a cylindrical tunnel of micro- wave absorbing material is attached to the edge of the dish. By subsequently adding the tunnel of the length indicated, which terminates in a region of low field intensity, the side lobes for angles greater than a few beam widths from the axis can be reduced substantially. Absorbing material is “warm” and radiates thermally into feeds used to look at “cold” targets like the radio sky. For this reason Cassegrain subreflectors fed from a source located near the vertex of the paraboloid are often used in radio astronomy or other functions employing low-noise (usually maser) receivers because the side lobes produced do not look at warm shields or objects behind the reflector. The entire pattern of the primary feed can be controlled by a metal shield in the form of a horn which attaches directly to the rim of an off-axis section of the paraboloid reflector. This design has been successfully developed by Bell Laboratories to produce low side lobes at wide angles from the axis.® The edge currents on a circular dish aperture will add up in phase to produce a strong back lobe in a direction 180° from the main beam. This effect can be reduced by (1) making the edge irregular, (2) placing chokes or impedances on the edge to suppress the edge currents, or (3) cutting coupling holes properly spaced near the edge of the reflector to allow enough radiation to leak through 180° out of phase to suppress the back radiation. Edge diffraction has received considerable theoretical study in antenna de-50 ANTENNA THEORY Paraboloid tunnel | | - pia’ Me . Ht ho constant intensity, db = haat ipatpernener biter a : ena Nh { ON 7 mal en t A Fig. 17.5 Plot of phase and amplitude distribution in the aperture region of a paraboloid showing best location of absorption tunnel to reduce side-lobe levels. sign since the time when Sommerfeld obtained the half-plane solution. Edge effects were treated by Woonton’ in 1950, and more recently the geometrical theory of diffraction” has been applied to curved, thick reflector edges" and the far- and back-lobe radiation patterns of paraboloids have been computed. Recently antenna workers have recognized the advantage of viewing diffrac- tion as optical workers normally do, in the receiving sense.?_ When a uniform linearly polarized plane wave propagating along the reflector axis is incident on a paraboloid with circular aperture, an Airy ring field distribution approxi- mated by Ji(q)/q, where q = 27ar/Xof and a is the diameter of the circular aperture, is set up in the focal region. This field structure is infinite in extent and will also have cross-polarization components caused by the curvature of the reflector. However, it gives us directly the diameter d of the central in- phase Airy disk, d = fdo/2a, and also the position and phase of nearby rings or side lobes. It is possible at microwave frequencies to fashion a primary feed with dimensions and fields corresponding to this Airy distribution. We must, of course, truncate the feed short of an infinite number of rings in J:(q)/q, but even a few rings approximate the low spillover feed needed to receive only that portion of the plane wave incident on the aperture plane inside the dotted circle shown in Fig. 17.1. Or alternatively, when such a circular feed structureREFLECTOR ANTENNAS 51 is used to transmit from the focal region, very little energy lies outside the cone %. This diffraction view of the focal region distribution can be used as a basis for a concentric ring feed* which has the desired low spillover and high aper- ture efficiency. We can approximate the Airy disk-ring structure with con- centric circles of small horns or slots. Because the spatial distribution of the energy or information in the focal region is given by the Fourier transform of the aperture distribution, we can invoke Shannon’s sampling theorem to determine how many radiating (or pickup) points are needed in the paraboloid focal region and the minimum spacing interval required to approximate the aperture fields. Because the Airy disk structure is nearly the same for all focusing lenses or reflectors with the same f/D, we might use a small-aperture-short-focus sys- tem to more efficiently illuminate a large paraboloid.™ The two off-axis paraboloidal sections are inverted and arranged as shown in Fig. 17.6. The difficulty with this scheme is that of generating the plane wave front over the smaller feeding surface of the subreflector. This can be done with stacks of waveguide arrays at microwave frequencies or with collimating lenses at optical and millimeter wavelengths. Paraboloid Paraboloidal In-phase subreflector feeding surface Fig. 17.6 Method for feeding paraboloid to achieve low spillover losses. In fact, the popular Cassegrain or hyperboloid subreflector accomplishes much the same objective using a point source feed. If f/Du = f’/Ds, the high- efficiency low-spillover conditions are met. Figure 17.7 shows the geometry of such a Cassegrain system.” Of course, the horn does radiate some energy outside the cone 2¢) which is spilled past the subreflector. Also, the sub- reflector blocks the aperture of the symmetric paraboloid. The dashed lines show the equivalent long-focal-length paraboloid. It can be shown that the equivalent long system would have the same illumination taper and size Airy rings and disks on reception of a plane wave as the actual subreflector parabo-52 ANTENNA THEORY ooo Paraboloid ao \ aa \ . \ Equivalent long- focal-length poroboloid | reflector——__,} | | . ~ | Hyperboloid ~~ | reflect > subreflector ss | ~~} Fig. 17.7 Cassegrain fed paraboloid showing equivalent long-focal-length antenna. loid arrangement. It is clear that a large feeding horn with a plane wave aperture distribution is needed because of the “magnification” of the long focal length and to reduce spillover around the edges of the subreflector. The essential analysis and design steps for the construction of a horn-fed Cassegrain antenna are given because of the aforementioned advantages and the addi- tional desirability of having the primary feed and horn near the vertex to reduce waveguide losses and facilitate feed adjustment. The Cassegrain subreflector is in the form of one sheet from a two-sheet hyperboloid of revolution located as in Fig. 17.8a. From the properties of the hyperboloid of revolution it is clear that a real focus at Q implies a virtual one at P which, in turn, is made to be at the focal point of the paraboloid. The design of a Cassegrain antenna can be accomplished by selecting the appropri- ate set of parameters to produce an aperture field distribution which in turn gives rise to the desired pattern. Assuming a pyramidal horn feed and all dimensions involved much larger than \ so that optical considerations are applicable to the derivation of the aperture fields, we may proceed as follows: 1. Parameters f and D,, are determined mainly from requirements for antenna gain and mechanical consideration. 2. The horn is assumed to have an A/B ratio not very different from unity, a small flare angle, and phase center on its aperture. It therefore has a nor- malized far-field pattern Gu(¢)/G(0) symmetric around the x axis, which does not depart appreciably from Ji(2.50 2 in ¢) a7.21) for (B/N) sin @ < 0.75.REFLECTOR ANTENNAS 53 Edge ray ji 20 Sq pn ah Hperboloid: p?= (22-1) [[e+e1?-24] / Paraboloid: p?= 4F( x +f) (a) (6) Fig. 17.8 Geometric relationships for computing radiation patterns of a Cassegrain fed paraboloid. A ray leaving Q with an angle ¢ will cross the paraboloid aperture plane at a radial distance p, p and ¢ being related by @_pe-—l z2 Auld be tan 3 DET 2 e! , (17.22) or pat ti_sng _yetling (17.23) e—1litcos¢” e-1l where e is the eccentricity of the hyperbola section of the hyperboloid sub- reflector. These equations can be derived from the basic conic section re-54 ANTENNA THEORY lationships for a hyperbola and a parabola positioned in the Cassegrain configuration shown in Fig. 17.8a and b. If the paraboloid aperture field distribution, which is also symmetric around the zx axis, is represented by F'(o), then in any plane perpendicular to the x axis we have F() _ Gul6) FO) ~ Gx(0) For the long equivalent focal lengths encountered in most Cassegrain an- tennas, the normalized aperture distribution F(p)/F (0) is a very good approxi- mation (subject to small space attenuation corrections) to the primary horn pattern G(¢)/G(0). It follows from (17.24), (17.23), and (17.21) that (17.24) F(p) = Peos(2. 50 é se = t) (17.25) The expression giving the Fourier transform of the antenna radiation pattern over the aperture is then GO) = [Pr Fo a dp 5/2 = I Patt PO 2. 50 2 oes Ai (Cee “\p dp (17.26) FO4Dm) _ BDne— and 0) Ji(2. 50 5 OFT +) (17.27) Side-lobe and efficiency requirements usually establish the edge-to-center aperture illumination ratio given by (17.27). For example, a value of Mo is often selected. Trial values of D,, B, and ¢ are tried until a suitable pattern from (17.26) is obtained subject to conditions above and: 1. The values for e are so selected as not to invalidate the approximation cos @ = 1 for ¢ = dmax. 2. For the horn not to cause an aperture blocking over the one caused by the subreflector, its edges must fall inside the rays passing the subreflector edges. According to Fig. 17.8), this condition may be written A 8D.f ae Sip permitting the calculation of ¢. (17.28) Exercise 17.4 By making use of the expansion git cos B — xy (Aa(adein®REFLECTOR ANTENNAS 55 show that the transform (17.26) is simply Dai GO) = 5 2" [°™ F(9) lex (hoop do ag? where 9 = pcos ¢’ & + psin ¢! §. 17.4 The Wide-angle or Off-focus Characteristics of Paraboloids In order to scan an antenna beam by motion of a feed off the focus of a short- focal-length paraboloid or to arrange and excite a group of feeds in the focal region to synthesize a desired far-field pattern, a fresh look at the highly aberrated or distorted focal fields of the short-focal-length paraboloid is necessary."* When the complete caustic surfaces of the paraboloid are examined, certain useful ridge lines or regions of high ray density are apparent. This analytical approach using rectilinear congruences of rays encompasses all the Seidel aberrations and illustrates by the relative location of two distinct caustic surfaces that coma and astigmatism are usually the dominant aberra- tions.” This realization has led to the independent treatment of sagittal and tangential (elevation and azimuth plane) focusing surfaces of the paraboloid. Analysis shows that different sections of the paraboloid surface have distinctly different focusing qualities, and more especially the off-axis section is par- ticularly free from effects of astigmatism and has really two focal regions. The antenna engineer finds that by locating feeds according to the “skeleton” structure of the limiting lines on the caustic surfaces and through proper phasing of the amplitude distribution on his feeds (corresponding to dis- tributed objects or sources in the focal region) he can (1) partially correct the aberrations or optical spreading of rays away from the geometric focus (ridge line corrector), (2) produce a broad-shaped antenna pattern in one plane which is as sharp as the on-focus diffraction pattern in the other plane (midpoint corrector), (3) locate a locus of point feed position where patterns with simul- taneously good azimuth and elevation patterns can be obtained (flute antenna), and (4) obtain accurate angular position data by comparing the phase in the broad pattern produced by an extended phased source in the focal region (dual terminal phase-in-space feed). The correctors mentioned in parentheses refer to special configurations of extended sources which are so located and excited in phase and amplitude as to produce special useful diffraction patterns. Paraboloid Treated as a Transmitting Antenna Some useful information can be obtained by analysis common in both an- tenna and optical practice: study of the aperture phase errors as a power series across the aperture when a transmitting point source is located at arbitrary56 ANTENNA THEORY Parabolodial reflector section Yn qercet >» aay o Focus > To far-field pattern ay SA N A un, Fig. 17.9 Geometry and coordinates used to analyze off-focus characteristics of a parab- oloidal reflector by ray optics (transmitting antenna method). positions (0,y0,20) near the focus F (see Fig. 17.9). Ray analysis allows us to express the direction cosines of the reflected rays & as — (2feo + yyo\& COS a, = (eteye =h (17.29) 2fzo + C08 ay = tee =m, (17.30) 2, 1 — % — 2 cos a, = (Stuy + papal =n (17.31) where f is the focal length and l= V+ U— wy Ft &— a (17.32) To obtain a well-collimated beam in the a, (azimuth) direction we wish to locate the source at a position (yo,20) such that cos a, vanishes. Equation (17.29) indicates that it is not possible to make cos a, = 0 for rays striking the reflector surface at all points x and y. According to (17.29), sources located along the y axis will produce the best azimuth patterns for the usual reflector section with vertex at the center; and when the center of the aperture is at ym, the straight line m%=0 m= aa (17.33)REFLECTOR ANTENNAS 57 is the best locus for azimuth focus. The aperture phase function can be de- veloped in a power series both in the x and y and, subject to approximations involved, locus lines for both elevation and azimuth can be found. Thus v= 11+ 2/nm, — 2f or approximately y = Li + z — 2f. Exercise 17.5 Assuming that the phase function on the aperture of the paraboloid is (x,y) = Li +z — 2f, show that the terms ¥.(0,0), Hzy(0,0), Y222(0,0), and Yyy2(0,0) in the Taylor expansion of y about (0,0) are all zero and that =o OO) = ire =a Given a series approximation to the phase function on the aperture plane, the far field H(az,0,) can be obtained approximately from expressions of the form B= ff Reayretemernectvney dr dy aperture plane We must, of course, include the amplitude and polarization effects which are comparable with the on-focus case.. This analytic work approached with differ- ent insights and emphasis has been reported by Sandler,” who applied the vector surface current method for sources laterally displaced from focus, and by Ruze,” who used a detailed treatment of paraboloid aberrations. Receiving Antenna Analysis of Paraboloid All of the facts needed for locating sources in the focal region of a paraboloid and for the relative phases and intensities of these sources can be found most easily by considering the caustic surfaces of the receiving paraboloid. This analysis is done for a single incident uniform plane wave inclined to the axis of the paraboloid. The analysis is then extended to families of plane waves with the relative phases between them assumed to be known (corresponding to an extended coherent source at infinity). Examination of received diffraction images is, of course, a well-developed technique in optics, and the methods for treating aberrated diffraction patterns reported by Linfoot” are applicable. A general diffraction treatment of the strong aberrations of the short-focus paraboloid has apparently not been published. The antenna designer has thus far used the knowledge of the unaberrated on-focus pattern along with the caustic surface structure to construct feeds for studying combined diffraction- ray optical effects. But even when rigorous diffraction results” are generally available, the caustic surface with its limiting lines or ridge lines will provide simple, indispensable guides for selecting the best focal field regions and feed locations. When a uniform plane wave inclined at angle 6 with respect to the axis of the paraboloid of revolution is incident on the reflector shown in cross section in Fig. 17.10, we can always rotate the x axis so that it is parallel to the incoming58 ANTENNA THEORY (0. yoz0le Fig. 17.10 Coordinate system for analyzing parab- oloid as a receiving antenna. wave front. Assume then that a plane wave, parallel to the x axis but tilted with respect to the y axis, is incident on the paraboloidal reflector. The unit vector along incoming rays sya, so (see Fig. 17.10) is 8 = 0% — sind ¥ + cos dz (17.34) The unit exterior normal to the reflector is ak + yh + 2ft i = (17.35) vat + yt + ap Snell’s law gives 8, = & — 2@8-8)a (17.36) The direction cosines of the reflected rays are _ 2y sin 6 — 2f cos 6)x rr a (78) = _—s 2(y sin 6 — 2f cos 6)y m, = —sin@+ Pipe (17.38) _ 2(y sin 6 — 2f cos 6)z m = cos 6 + Ppt ye (17.39) We can find where these rays pierce any plane; the highest density of inter- secting rays, or in-phase rays, is in the yz plane.REFLECTOR ANTENNAS 59 Since x = 0 is the yz plane, and since the reflected rays are straight lines, we can write YOY _ m7 _ my, m a We can solve for yo and a, thus defining the locus of ray interception in the yz plane, and obtain a parametric set in terms of the xyz coordinates of the dish _sin 6 (v? + y? + 4f?) 4f cos @ — 2y sin @ _ 608 6 (x? + y? + 4f%) = “Feosd— wane 2+? (17.42) In Fig. 17.11 the locus of ray interception for various tilt angles @ has been plotted for the limit when « > 0. All rays do not intercept a line, but the ray (17.40) = (17.41) Paraboloid Line at 8=15° ridge line for o complete porabolic reflector ; others shown are only for the top half of a reflector. foi tif tof of gd fp pt poiij yi i 090 070 0 "O24 020 0.18 0.12 008 O04 OJO -004 -G08 -012 Fractions of a focal length | Feed line to correct strip. | 9= 5°" ot y =0.5f on reflector. -0.48 ~4-0.56 --0.64 Fig. 17.11 Ridge lines produced by plane waves incident on paraboloid section inclined at angles 8.60 ANTENNA THEORY density is highest in the limit as «0 corresponding to the lines shown. These astigmatic lines are the best for azimuth focus at elevation angles @ and are referred to as ridge lines. Relation of Azimuth and Elevation Focusing Loci to the Complete Caustic The focal surfaces for the paraboloid have been developed in Chap. 16. The general equations for the caustic surfaces are given in (16.63) and (16.64), where the angle @ in Fig. 17.10 now corresponds to the angle y in Fig. 16.15. The alternate use of the positive and negative signs in B leads to two focusing surfaces. One of these surfaces is related to azimuth focusing and one to elevation focusing. We shall limit our attention to a narrow strip of the reflector in the yz plane, x = O, and use a unit focal length, f = 1. For mathematical convenience let —zcos@ + ysin@ _ a} sin (5 — 6) (17.43) _ y . Zz and cos 6 = po2 then sin 6 = —— 3 (17.44) Choosing the positive sign in B of (16.50) and (16.51) gives Yo 2 wef 17.4 mY (17-45) which is the straight-line azimuth focus discussed previously. Choosing the negative sign in B of (16.63) and (16.64) leads to the second focusing surface. This is a circle given by _ 38 1 at +t — (PF), +E, = 0 (17.46) with its center at _ 3,0 a=t me (17.47) — yy = 12) Yo = ie (17.48) The complete azimuth caustic and its relation to the reflector are shown in Fig. 16.16a toc. Both caustic surfaces occur simultaneously for every incident plane wave. The loci of azimuth and elevation focus for a number of correction points (y = y,) on the reflector are plotted in Fig. 17.12. For each corrected point (ym) on the reflector, the azimuth and elevation focus loci are represented by a straight line and a circle, respectively. Note that, for all correction points except Ym = 0, the elevation and azimuth loci intersect at two points, one at the true focus and one at a secondary focus.REFLECTOR ANTENNAS 61 ono no25 m2050 [2=60° Paraboloid cross section y?=4f(f-2) oN Ym = 0.75 50° Ym = 0.75 40° Ym=0.50 Yn= 0.25 8 ym=0 F— Elevation and azimuth focal points are shown for plane wove. Inclined 8° -20° Fig. 17.12 Best feed loci for azimuth and elevation focusing for paraboloid section with midpoints ym. On substitution of (17.43) and (17.44) into the quantity B in (16.63) and (16.64) B = 2[1 — sin @ (y cos 6 + zsin 6)] for the negative sign (17.49) B=2 for the positive sign (17.50) The measure of the astigmatism is therefore A = 2sin 0 (y cos 6 + z sin 6) (17.51) and for zero astigmatism defining the second focal point where the circle and line intersect a second time tan O40 = —4 (17.52)62 ANTENNA THEORY Relation (17.52) will give the elevation angle of the secondary focal point. For example, if the correction point on the reflector is chosen to be Y = Ym = 0.50, then z = 0.9375, and 64-0 = 28°4’. Consider, in Fig. 17.12, a typical correction point ym = 0.75. Below focus, the circle and straight line rapidly diverge; above focus, the straight line is near the circle between the true focus and the secondary focus. This means that with an array made to lie along the azimuth focus line (below focus) the beams are sharp in azimuth but badly defocused in elevation. Therefore, extending the array below focus results in loss of gain. If the array is extended above focus, the two loci are closer together between the two focal points and accordingly there is less elevation defocusing over this region. Consequently, in the region between the two focal points, there is less loss of gain. For ex- ample, the secondary focus occurs for a plane wave incident at 38°1' with the axis for a correction point of yn = 0.75. Better gain over this elevation interval can be expected if the primary feed is to lie above focus. Such construction also makes a more compact physical unit. Although the primary feed then extends across the reflector aperture, no serious shadowing has been noted on models built to operate at \) = 3.2 cm and Ay = 1.25 em. The Ridge Line Correctors The particular ridge line of the paraboloid caustic [Eqs. (17.41) and (17.42)] produced by a single inclined plane wave has been studied experimentally for the off-axis section. The ray paths from the reference plane wave surface to the intersection of paraboloid with the yz planes down to the ridge line (see Figs. 17.10 and 17.11) were calculated in order to get the electrical phase along the ridge line locus, ,(yo,20). An extended line feed was constructed to lie along this ridge line with path lengths built in to compensate for the difference of relative phase expressed by the function 4,. Two new methods are de- veloped in Sec. 17.5 to obtain arbitrary phase functions along line apertures. A slotted array method suitable for microwave waveguides was used to design a line feed that produced satisfactory gain and elevation plane beamwidth. The second method for obtaining synthesis on line apertures is based entirely on ray optics. This simple method was developed in Chap. 16. Based on this ray approach a phased line source to lie along the ridge line can be constructed by using a cylindrical shape mirror or pillbox fed by a point source to produce the phase function &, on the aperture. The shape of the cylindrical mirror to produce this phase function is derived by using envelope theory leading to a parametric expression for the coordinates of the special mirror. Rather than the pillbox construction, the mirror curve can be used to generate a surface of revolution about a straight-line approximation to the ridge line. Midpoint Correctors It can be shown by study of Fig. 17.12 that small receiving feeds with feed areas less than \” (and, of course, by reciprocity transmitting sources) dis-REFLECTOR ANTENNAS 63 tributed along lines passing through the foci are best suited for producing sagittally or azimuthally sharp patterns. This result is important for synthesis of wide fan-shaped patterns needed for radar search modes. Equations (17.33) and (17.45) express proper tilt of this feed line when the center of reflector area is at some height (or radius) away from the paraboloid axis. For the vertex- centered paraboloid the best azimuth-focused patterns are produced by feeds along the y axis. Such a feed will receive (or produce on transmission) a family of plane waves inclined at different angles in the plane containing the feed line and the axis. The best patterns will be obtained when the relative phases of sources along this line are such as to focus the energy on the midpoint of the reflector. (Of course, a line source focuses to a line which here follows approximately y = y, on the reflector surface.) To construct such a feed by using ray optics we can feed a cylindrical pillbox ellipse with a point source with the pillbox line aperture on the feed line and the second focus of the ellipse at the midpoint of the reflector surface (Fig. 17.13). The length of the line source will determine approximately the angular width of the wide fan-shaped search beam. The construction is shown for an off-axis paraboloid section with its feed located along a portion of the azimuth focus line lying below the focus —i.e., mounted under the aperture of the paraboloid. This location of the feed is effective for producing fan-shaped patterns without aperture blocking, Paraboloid section reflector contour Fig. 17.13 Geometry for locating and phasing a mid- point corrector using special pillbox reflector contour.64 ANTENNA THEORY but, as noted, there are important advantages in locating the feed above the F in front of the aperture. Best Focusing Surface for Off-axis Section of the Paraboloid A study of Fig. 17.12 shows why designers might (1) prefer an off-axis or asymmetric section of a paraboloid and (2) prefer to locate feeds or photograph- ic films on lines or cones directly in front of the aperture above the focus. In this focal region the circle for best elevation plane or tangential focusing inter- sects the corresponding line for best azimuth or sagittal focusing in two points with a region in between where these loci of best focus curves are close together. Only for the circular, vertex-centered paraboloid section do we find the eleva- tion circle is tangent to the corresponding azimuth focusing line instead of in- tersecting it. The superior focusing properties of the off-axis paraboloid section were demonstrated rather sensationally by feeding with small sources located along this predicted line above the focus. Excellent diffraction patterns were produced in both elevation and azimuth planes for directions up to 60° off the focal axis. The feed actually used was a midpoint corrector composed of slots (approximating small magnetic dipoles) cut in a straight section of waveguide. Slots were opened one at a time, and good diffraction patterns were measured in both the elevation and azimuth planes for each slot alone (much as one would. play a flute — whence the name flute antenna). However, this line source feed when all radiators are connected to one terminal and phased to focus on the midpoint of the off-axis reflector is also the best means for feeding all the point sources simultaneously to synthesize a specified shaped beam. It is instructive to locate the best focal area for an off-axis section of a paraboloid by examining the three-dimensional aspects of the information in Fig. 17.12 for reflector with midpoint y,, displaced f/2 above the axis. The projected view of the focal region shown in Fig. 17.14 represents qualitatively the area around F and the second focus F’ within which the diffraction images have not enlarged significantly from patterns measured at the focus. The second focal point really extends into a nearly circular are. By locating small sources around the flat cone produced by revolving the line, Eqs. (17.33), around the z axis, we might expect to receive the sharpest image of distant extended sources.” 17.5 Techniques for Compound Primary-feed Designs It is often desirable to have several primary feeds in the focal region either interconnected to a single antenna terminal or independently fed to produce multiple patterns. Elemental dipoles or slots can be distributed in the focal region to correct for phase aberrations introduced by the reflector antenna or to synthesize a specified radiation pattern when properly excited by an arrayREFLECTOR ANTENNAS. 65 % Best . focusing Conical surface surface % 2o Conical surface Fig. 17.14 Approximate surface of sharpest images for off-axis sections of paraboloid shown in Fig. 17.9 for case Um = Hf. or coupling matrix connected to a single antenna port. More frequently, in- dependent antenna terminals are available for each antenna feed, and radar beam comparisons on a pulse-by-pulse basis (monopulse) between the ampli- tude and phase of two overlapping beams are made. Wide-angle reflector systems can approach the high data-gathering function of an optical telescope by introducing many separate simultaneous feeds to generate multibeams for mapping, tracking, angular diversity reception in scatter communication, or multiple-frequency operation. In this section are described design techniques for focal region feeds which can produce the wide range of amplitude and phase distributions needed in beam synthesis and aberration correction. Mutual coupling between individual radiators is not treated in these discussions, al- though it can measurably modify the antenna’s patterns and impedance when radiating elements are closely spaced.66 ANTENNA THEORY Half power beom width nw yee ”% L Axis Ph Paraboloidal reflector antenna with two feeds displaced laterally from focus 0 6 from axis Fig. 17.15 Monopulse amplitude comparison of antenna beams to determine accurate angle of arrival information. Monopulse Feed Design A very common method for improving the angular accuracy of radar using an antenna of diameter D which produces an antenna beamwidth of approxi- mately \o/D is to introduce two (or more) feeding horns in the focal region. By comparing the relative amplitudes of signals received on T; and T2, a measure- ment of direction of the incident plane wave or target can be made to a small fraction of a beamwidth. It is sometimes more convenient to attach the two antenna terminals to a hybrid T in order to display simultaneously the sum. and difference patterns. By measuring quantitatively the amplitude differ- ence of the instantaneous signals, for instance, on a pulse-to-pulse or monopulse basis in radar, the direction of an incident plane wave can be determined accurately in a plane containing the axis and the feed horns (see Fig. 17.15). Exercise 17.6 Show that the monopulse function expressed as voltage at the antenna terminals T; and 72, shown in Fig. 17.15, has the form Vi — Va = 2Goite- 7 r+ sinh (4670) assuming the main lobe of the received beams produces a potential Go? exp (—a?6?/2) at a terminal. Waveguide Arrays for Focal Regions When single terminal feeds for synthesis or correction are required, the feed designer must know (1) the shape and location of the feeding locus in the focal region and (2) the proper phase, amplitude, and polarization distributions along the feeding line or surface. The simplest correction results when the required focal distribution can be approximated by a straight line of linearlyREFLECTOR ANTENNAS 67 polarized radiators such as required for midpoint correctors in a paraboloid reflector described in Sec. 17.4. In most cases rather complicated amplitude and phase functions are re- quired along the correcting array, necessitating independent control of ampli- tude and phase at each radiating element. There are several waveguide feeding methods which are able to provide a wide range of feeding coefficients at the individual radiators. To illustrate these procedures we shall show how longitudinal shunt slots can be arranged on a rectangular waveguide to construct a phased line source. As an example we wish to synthesize a csc? a elevation pattern by use of midpoint corrector fed along the line zo/yo = —Yym/2f = —14, where ym = f/2 as shown in Fig. 17.16. The required phasing condition is such as to focus all the energy on the midpoint ym. The power to be radiated per length along the array can be computed also from the geometry shown in Fig. 17.16. This method as- sumes constant phase velocity in the guide and achieves phasing by progres- sively varying the longitudinal phasing of elements, such as longitudinal shunt slots, which can be reversed in phase 180° by alternating them across the centerline of the guide. y? =4f(f-2) sch pottern Fig. 17.16 Geometric considerations for determining the phase and amplitude distribution of a midpoint corrector for a paraboloid forming csc? @ radiation pattern. We can control the phasing on the line source by adjusting the spacing between slots for a traveling-wave array fed by a waveguide with guide wave- length A, = \o/u and input terminal at S = 0. We note from the phasing re-68 ANTENNA THEORY quirements that a linear array of generally broadside to backfire radiation characteristics is needed. In this case an iterative equation for the longitudinal slot spacings is readily found. Focusing to ym requires (see Fig. 17.16) Tn + u(Ss ~ Saas) + (32) = Dass Normalizing all lengths to R, the distance between ym and the focal line at S = 0, gives (+ S42) + uSn + (35) = (1+ Stu) +uSen (17.53) from which expression S,4, can be computed from S,. Exercise 17.7 Define Day1 as Dnit = (1 + Sx2)* + wSn +h + Da th where k = o/2R. Show from the equation that Sax BDng — (v2 — 1 + Diy) nd = When phase distribution or phase lag L as a function of aperture distance S is known but difficult to express analytically, a graphical construction can be used to locate the longitudinal position of slots or dipoles. When the radiating elements allow phase reversals of 180° by, for example, reversing leads to an electric dipole, the graphical construction shown in Fig. 17.17 can be used to fit a variety of phase curves by proper selection of AS. x . ton w = 5° (with no mismatch on guide) 2 Phase lag / So 51 Sa Sn Sn las] waveguide array 5 ml Las Fig. 17.17 Graphical construction for determining the longitudinal spacings of radiating elements to produce a given aperture phase function L(S). Distance along The longitudinal spacings of the slots are thus fixed by the required phase distribution; the transverse displacements can now be used to control the amplitude distribution. As a first approximation to obtaining a csc? a powerREFLECTOR ANTENNAS 69 pattern in elevation, notice that the power radiated in AS along the array will appear in the interval Ae as shown in Fig. 17.16. Thus, if the slots are equally spaced, all that is necessary is for each slot to radiate a fraction of the input power proportional to csc? a,. The longitudinal slot spacing actually de- creases toward the end of the array farthest from the focus. If we do not com- pensate for this, the result is a gradual increase in the power radiated off focus. Let the required powers radiated by each slot be Po, P1, Ps, ..., Pv, where nis the slot number from the feeding terminal. Then the coupling coefficient is (see also discussion on slotted waveguide arrays in Chap. 14) WP, aT 1-WDP; n=0 1- N dX Pp n=0 and r represents the fraction of power fed to the end of the array. The drop per slot is, in decibels, Kn = 4 where We= 1 10 log i-k The actual displacement can now be found from either theoretical or experi- mental slot or dipole coupling-to-waveguide curves. To select the best method for producing the required phase and amplitude distribution, first determine whether the phasing on the line source requires (referred to feed end) end-fire, broadside, or backfire ranges. Then use the constant guide phase, variable-spacing technique above for the general broad- side to backfire requirements. The end-fire region can be obtained by loaded waveguide or coaxial line feeding elements without alternate phase reversals. The pillbox or optical reflector methods (see Chap. 16, Sec. 17.4, and Fig. 17.13) are also useful for wide-band applications, although it is difficult to control the amplitude distribution as well as on slot arrays. 17.6 Spherical Reflector Optical designers have used spherical surfaces almost exclusively in design of lens and reflectors because of the ease with which such surfaces can be shaped and polished to optical tolerances. Unlike the paraboloid surface the metal spherical mirror does not focus a plane wave perpendicular to an axis to a focal point. However, the spherical cap does have the important advantage of focusing plane waves incident over a wide angular sector to nearly identical focal surfaces. This wide-angle feature of the sphere has attracted both optical and antenna designers, who have sought to improve the focusing characteristics70 ANTENNA THEORY in various ways. Optical designers have used long-focal-length systems and specially shaped auxiliary dielectric lenses such as the Mangin lens and Schmidt Jens to reduce the effects of optical aberrations. The Mangin mirror has a lens directly in contact with the spherical mirror, and the Schmidt lens usually takes the form of a radially symmetric lens located in the aperture of the spherical cap such that the focal region is between the reflector surface and the lens. Ray-tracing techniques are available for these correcting lenses, but because of their weight and complexity they have not been widely used by antenna designers. We shall present the correction methods suited to radio and microwave antennas which have as their primary objective correcting the spherical aberration of the spherical mirror for the purpose of scanning the antenna beam by feed motion in a fixed reflector or the development of a simultaneous multibeam antenna with information-gathering characteristics comparable with those of an optical telescope. Focal Region of the Concave Spherical Cap The caustic surface (see Fig. 17.18) generated by the reflection of a parallel bundle of rays by a concave spherical mirror is an epicycloid and is derived by methods described in Chap. 16. There is a degenerate line F'V of this caustic surface lying along a radius parallel to the incident rays (the axial line) which extends from the paraxial focus to the reflector surface through which all the energy over a 120° cone must pass. This line segment F'V of the focal surface can be used for the interception of a plane wave incident in the direction of the axial line.* The electromagnetic energy incident on this line must be rephased by correcting the path lengths between elements to realize maximum antenna gain or aperture efficiency. Such phased line sources located on this line to correct aberrations are referred to as longitudinal correctors. The highest focal field intensity is in the region near the paraxial focus, where the caustic cusps and the focal line are near each other. By placing a suitable primary feed transverse to the radius at the distance about 0.475R from the reflector surface, where F is radius of sphere, effective aperture illumination can be obtained. Feeding structures located in this manner are called trans- verse correctors. We shall first discuss the optical characteristics before in- troducing diffraction effects and feed design techniques for correction of aberrations. Phase Variations along the Axial Line It is convenient to analyze the spherical reflector as a receiving antenna and consider a plane wave AO normal to radius OV incident on spherical cap at a general point B (see Fig. 17.18a). Because of the rotational symmetry about OV, B represents a ring of radius r. The electrical phase caused by differences in path length for portions of the plane wave at different distances r from the radius OV can be calculated from the path length geometry. Let the pathREFLECTOR ANTENNAS 71 (6) Fig. 17.18 Geometry of (a) spherical cap antenna used to compute path lengths and (6) rays that form caustic. difference between axial ray OVF and the general ray ABC be denoted by 5; then 6 = AB + BC — OV — VF R R = Rosé + 5 sec8 — RB — 5 = R(cos @ — 1) +8 (ecco - 1 Since OC = (R/2)sec@ and «= OD = OF = f = R/2, we can write, letting 2 = OC — OF = f(sec @ — 1), o= (1-735)72 ANTENNA THEORY This function is shown in Fig. 17.19. From optical considerations we might expect a correctly phased line of transmitting sources (or, by reciprocity, phased receiving elements) to compensate for these phase differences and, when positioned on the axial line at points z > 0, to produce a circular patch of plane wave on the aperture of the reflector which would in turn give rise to a diffrac- tion pattern such as produced by a paraboloid of the same aperture size. 0.3 8/f -0.1 ° 0.2 0.4 0.6 0.8 1.0 z/t Fig. 17.19 Path difference 6 vs. axial distance z for a spherical cap reflector. This has been experimentally verified, and the aperture efficiency and an- tenna pattern contiol depend on design skill in producing phase and amplitude distribution functions and desired polarization components. Of course, the extended line sources are more difficult to design than the small point source needed to feed a paraboloid. The advantage of the spherical reflector consists mainly in that, because of the symmetry of the sphere, the correcting feed can be rotated about the center of the sphere and produce nearly identical radia- tion patterns in a wide range of look-directions. This feature allows a fixed spherical-cap antenna to be scanned easily by motion of the primary feed. Aperture Amplitude Distribution Produced by Axial Line Source We can relate amplitude or intensity along the line source with distance z on the line source by optical considerations similar to those we used in our treat- ment of amplitude on a paraboloid aperture. Referring to Fig. 17.18a, ot ._f sind = 5 cos 8 = Fre - tY4(fY. Aperture area A = rr? and (5) + (4 3) =2 A fo dA _ _8xft thus apt Gee! ae Gee For uniform power in the aperture, dP /dz on the radius must be proportional to the above expression for dA /dz.REFLECTOR ANTENNAS 73 Polarization Effects The spherical cap is a curved surface which generally introduces cross- polarized components of the electromagnetic field in reflecting linearly po- larized waves. We are interested in feed positions distributed in the focal region and especially along the axial line. It is simpler therefore to consider again the receiving case of a plane-polarized wave incident on the sphere and to examine the distribution of polarization components along the axial line from F to V of Fig. 17.18a. Because of the spherical symmetry any linearly polarized component can be arbitrarily chosen as a reference direction in the incident plane wave. Choosing E; = Eye-*»¥ in the plane wave parallel to AO, we can write down the polarization components by using the ray approxi- mations of (17.7): Erctiectoa = (+E, — (A X E,) X A = 2(-E)A — E; A = xX + yoy + zz = outward unit normal to reflector at point of incidence B, = QWroyo + ys? — LF + Qeoyo as 2% = sin 0 cos ¢ Yo = sin Osin @ 2 = cos6 Then E, = sin? @ sin 26 & + (2 sin? @sin? ¢ — D¥ + singsin 202 (17.54) Now since z on the radius is explicitly related to the angular coordinate 6, Rey = Resin 9 cot 20+ 2+ secd=1 +5 (17.55) we can integrate each polarization component over a circular strip r dé and express the resultant sum as a function of z on the radius. With a uniform plane wave incident on a spherical cap only one polarization component is nonzero on the axial line when terms in (17.54) are integrated over ¢ from 0 to 2x. These integrations are easily carried out as an exercise by the reader. The more difficult case involving polarization off the focal radius including diffrac- tion fields for a general point in the focal region is discussed in the literature.?6" It is clear from examination of (17.54) that cross-polarization components can contain a sizable fraction of the power in the focal region for small f/D, that is, when 6 — 7/2. The ratio of focal length to aperture diameter will always be small, f/D = V, for feed scanning a fixed sphere over a wide angular region. In general, three separate specular regions of the reflector contribute to a point in the focal field between the caustic surface and the reflector. However, on the degenerate focal line rays from a single concentric area of the reflector can be considered to contribute to the field. Fortunately, it is possible to solve the diffraction problem for the amplitude and phase on the axial line with a plane wave incident normal to the axis* and treat the receiving antenna case by using the surface current method. A cross section of a spherical reflector of74 ANTENNA THEORY Fig. 17.20 Coordinates used for calculating diffrac- tion field on axial line of spherical reflector. radius of curvature R is shown in Fig. 17.20. The angle 6 is measured from the center of curvature 0. The distance along the axis measured from the center of curvature is written as £2. A uniform plane wave traveling in the z direc- tion, with the magnetic field intensity in the y direction, is assumed to be incident on the reflector (see Fig. 17.20). A reference aperture plane for the incoming wave is located at {2 = 0. The surface current density on the re- flector is J. = -2@ X H,) which has two components: Jes = —2H;cos@ Jz = 2H: sin 8 cos It is clear that the z-directed current will not produce a field on the axis of the reflector, because equal and opposite current elements are symmetrically disposed on the reflector surface. Considering the current in the x direction, the field produced at a point C on the axis of the reflector is, using only the radiation field of the equivalent electric dipole, kobBC > 1, E,() =A I. °° ritek 008 cos 8 exp (FiteBC) S(8,8)R? sin 6 d8 dé where A « —2H;, and the factor S(6,¢) is the obliquity factor, or pattern, of each current element as viewed from the point C. From the geometry, S@,6) = F UBC x J.) X BCL, = 1 — sin’ 6 eos" 6 which for the ¢ integration yields [* (= sin? 6 cost 4) dé = 2n(1 — 14 sin’ 8) ° At the larger values of 6 the pattern of the elements produces a decrease in the contributions to E,, but since the reduction for most 8 is not large enough toREFLECTOR ANTENNAS 75 warrant the inclusion of this factor, S(6,¢) will be approximated by unity in this section. The distance BC is found from the law of cosines: BC = Ry1 + # — 2& cos 6 = tR The integral to be evaluated is sin 8 cos 6 dé (17.56) and the constant A will be chosen later for appropriate normalization. Chang- ing the variables to EQ = anna [ep Litel (cos + VIF B= 20089) VI + 2 — 2&cos6 =t gives B® = 2eRA [" exp [ ater GE* ze +i- 5) (+e a *\ dt and another change to Weft @— gra Bat or more specifically yields 1+ 2# [sme Se") wise _ fr caso 1 _) eu _ cS “) du Carrying out the integration and using C= [cos (3) du S= fosin (3) du Et) = results in Bale) = A[SE exp (itor: 428) x {(1+ Stee — cru + iste) — 58(ul + 25 (creme Jéua a) 4 a eitelust -j j i creme}76 ANTENNA THEORY where the limits are ta = 2 (VIF B= Best — 8 Pett a — 29 This expression can be evaluated for the field at any value é along the axis. The limits uw; and uw are smooth functions of £, us being zero at the paraxial focus £ = 14 and uw being zero at the crossing of the marginal rays given by cos & = 1/2¢. Tf all current elements were collected by a waveguide in such phase that their fields added in phase at the paraxial focus, the total field at that point would be Ey = RA [" Cpt - B)e = 5 4 0086) [5 _ 2 = arra[(3+ 3 4 COS 6 ra = tRAF (6) Normalizing the field expression by this quantity yields for ko >> 14 and z— (ie., near paraxial focus) Ea®) = [rw veell= exp [-sna(? =) x {1cem) = Clue) + 58(s) — 78(e0)) +3 X (WTF EH Boa, + desorme — ja[ Ee crore} In this expression the first part is the normalization; the second shows that the normalized field decreases as the reflector size is increased; the third is the geometric optics behavior; the fourth is the Fresnel integral behavior; and the rest contributes minor oscillations. The normalized field magnitude for spherical mirrors of half-angle 6 = cos 0.8 = 36.9° and radii of curvature R = 25%, 100A, and 400) are plotted in Fig. 17.21. The curves for R = 400% are slightly smoothed. The geometric optics limit is also shown for comparison. The deviation of the phase from the geometric optics phase oo[ uC 42] is shown for the same cases in Fig. 17.22. To find the far-field pattern of the antenna when the conjugate of the fieldREFLECTOR ANTENNAS 77 10 | r R= 25r 8 == R= 100% J L —-— P= 4009 4 lel F oN 4 Z oS Vl 7 = L / 2 oot AT / Vv . * ~ — ‘ Geometrical optics BSS Les OkerTirisiiisipii tipi tii te 0.45 0.50 0.55 ¢ 0.60 0.65 Fig. 17.21 Magnitude of the field along the axis of a spherical reflector. 3.0 T \ R= 25ro 2.0 | ---- P= 100% \ | —-— RF = 40029 \ é . - - 3 9 oe ea SS 3 Z Ors = = y 2 TT, SS | \% -1.0F, [7 / \ \ ot : af | | \ 0.47 0.50 0.55 t 0.60 0.65 0.67 Fig. 17.22 Deviation from the geometric optical phase (see Fig. 17.19) along the axis of a spherical reflector. distribution of (17.56) is placed on the axis of the reflector we proceed as follows: The current density on the reflector due to source element of strength E%() dé at a point £ on the axis is approximately 0 Jun = S(B,8)EL@) d& 5 cos 078 ANTENNA THEORY The far-field pattern of a thin annulus of radius FR sin 6 and width R cos 6 dé in the transverse plane is R® sin 6 cos 6 d0 f 2 skoR sin 0 sin a cos 0-9S(B,4) de ° where a is the polar angle between the direction of observation and the axis of the reflector and y is the azimuthal angle of the observation direction. The far-field pattern from all currents on the spherical reflector excited by the entire axial distribution is e-HUER+R cos 0) f(a) = constant [[#@ —. R? sin 6 cos 6 x fou sin @ sin a cos (¥-#)S(8.4) dp dO de (17.57) where the effect of the obliquity factor of the current elements is included. Substituting the expression given for H,(f) in (17.56) and using variables u’ = cos @ in (17.56) and u = cos @ in (17.57) yields _ ene (Ht wmu) f(a,p) = constant f ff ff at x few sin 0 sin a 008 W-#6(8.6) dg du du’ dé Evaluating the w’ and £ integration by the method of steepest descent yields S(a,~) = constant f * f & G-itoR sin 0 sin a cos (Y-¢) o Jo X (1 — 4 sin? 6 cos? 6 cos? $) cos 6 sin 6 dé do With the indicated integration performed and a = R sin 6, the pattern in the E plane (y = 0) is given by S(@,0) = constant {Aj (koa sin az) — 4 sin? X [Ar (Koa sin az) — $4A2(Koa sin az)] + 4 sin? [Ai (koa sin a2) — 54Ac(koa sin az) + 542A3(koa sin a:)]} where A,(x) = p!(2/x)*J,(x). The pattern in the H plane (y = 2/2) is given by Has) = constant{Ai(koa sin ay) — sin? OAe(Koa sin ay) + sin! 6[As(oka sin a,) — MYAz(koa sin a,)]} showing the effect of the element patterns of the currents on the reflector and the feed elements. The far-field radiation patterns for @ = 30° are shown in Fig. 17.23.REFLECTOR ANTENNAS 79 1.00 0.75 0.50 0.25 0 -0.25 L 1 1 1 \ 0 v 2 3 4 5 6 7 8 7 10 foo sina Fig. 17.23 Calculated far-field radiation patterns for a spherical reflector with a line source feed. Longitudinal Line Source Correctors The information about phase, amplitude, and polarization calculated accord- ing to the analytical procedure outlined above is sufficient for engineering design of primary feeds to lie along the radial focal line. Additional refine- ments such as amplitude taper for pattern (primarily side lobe) control and circular or variable polarization are also possible. The most important func- tion to achieve along z is the phase or path length condition shown in Figs. 17.18, 17.19, and 17.22 needed in order to realize aperture gain. Fortunately, this phasing condition is-rather insensitive to frequency change and is well predicted by the ray path analysis. Corrective Line Source Arrays for the Spherical Reflector Waveguide array techniques can be used to phase individual radiators to correct for path length differences and, for example, in the receiving sense to cause all the energy intercepted by dipoles along z to arrive in phase at the paraxial focus F. This condition is expressed as =, 2 _ f° [ote of 1| = m2 FC gte le [8 where the integral 4o/A,(z) is the required variation in guide wavelength , to produce 6(z) and \ is the free-space wavelength. Guide width w in a rec- tangular guide, for example, can be varied to achieve this condition:80 ANTENNA THEORY We can also view the problem as one of building an array with variable directivity such that the maxima for each differential length dz are in the direction 20 as shown in Fig. 17.18a. We can then use simple array theory for dipole or slot elements with phase reversals to fulfill the array equation cos 26 = x =- x where S is the element spacing and @ can be related to z by (17.55). Arrays built on these principles have been designed to feed large spheres with diameters of D/No = 400. The 1,000-ft-diam sphere antenna at Arecibo, Puerto Rico, has a 90-ft line source extending from the paraxial focus 435 ft above the fixed reflector surface. The waveguide feed operates at about 430 MHz and is composed of closely spaced transverse slots on a waveguide, where }, is controlled primarily by varying the guide width. A guide with square cross section is used to permit circular polarized waves. On these structures the bandwidth of the correcting feed is restricted by the dispersive character of waveguide, which quickly destroys the in-phase condi- tion between the first and last radiator as frequency is changed. A broadband line source can be designed by feeding each radiator with a separate line equal in length to 4(z). Actual line sources have diameters of finite size, and the problem of satisfy- ing the phasing condition in all ¢ directions is difficult on a thick line source for all values of z. Because different components of E actually travel a different distance to a given point on the spherical surface, it is necessary to study and compensate for path length for field components radiated from the opposite sides of the waveguide.** Amplitude of radiation along the line source guide can be regulated by techniques described for the paraboloid. Because of polarization effects and weight restrictions, line sources extending from the focus to about z/f = 0.5 have been most useful. A short end-fire array has proper phasing near the paraxial region. Sharp cutoff in primary pattern can be achieved at the edge of the reflector, resulting in generally low noise temperature performance for this antenna design. Gregorian Subreflector A subreflector similar to the Cassegrain feed for a paraboloid reflector can be designed by optical techniques to correct the optical aberrations of a spherical- cap antenna completely. The subreflector contour can be derived by making use of the path length condition 4(z) as for the longitudinal phased line source making use of the general relationship given in Chap. 16 for synthesis of an aperture phase with a reflecting curve.* By this method the one-dimensional or pillbox reflector contour is obtained, and the subreflector surface is formed by rotating the curve about the focal line (z axis) (see Prob. 17.1).REFLECTOR ANTENNAS 81 Gregorian subreflectors provide a broadband solution to the problem of aberration correction on a sphere. When the aperture amplitude is calculated for all practical arrangements of horn position and patterns and subreflector diameters and shapes, we find that the aperture amplitude taper is usually inverse; i.e., there is a high power density at the edges of the aperture of the sphere. Such an aperture distribution gives rise to higher near-in side lobes than are desirable for radar, although the resulting narrow beamwidth is ex- cellent for mapping purposes. Aperture blocking is greater than for a line source feed. The general shapes of the Gregorian corrector located with vertex and focus position are shown in Fig. 17.24 for a sphere of radius 1,000 (paraxial focus at 500). Figure 17.24 shows the geometry, and Fig. 17.25 shows the various forms the reflector can take. The general equations for the surfaces are given in Fig. 17.25. Transverse Correctors for the Spherical Reflector The peaks in amplitude shown in Fig. 17.21 correspond to circle of least con- fusion deducible from optical principles. A point source or horn located in this vicinity will intercept the most power from an incident plane wave. When the electromagnetic field is carefully examined in a plane transverse to the principal radius, diffraction rings dominate the field structure. The principle of sta- F = paraxial focus V = reflector vertex EE = reflector aperture x AOJ = reference wave front 2(@-a) = total scon angle P = corrector focus N= corrector vertex KK = corrected aperture Fig. 17.24 Location of Gregorian subreflector in spherical antenna.82 ANTENNA THEORY =200 -160 -120 ~80 40 0 40 80. 120 160___200 400 TT T TT T 440 480 20° 480 AG, O focus 520 560) 600 640 680 A A 8C All-n)e=(1+p2) +(2+0 -2n)?] (2c?-1) 4lpc?-c-n +1) Curve Corrector ve “Vertex Focus A 045. 0.5 where ¢ = cos 8; s=sin 8 é 833 OOS D 00.5 Fig. 17.25 Representative Gregorian corrector shapes and parametric equations for computing them for various positions of feed point P and vertex position n of subreflector. tionary phase can be used to calculate the transverse focal fields, although in general three specular areas (with three components of polarization) con- tribute to each field point.” Special formulations are required in the neighbor- hood of caustics. Using detailed information about the transverse field by calculation or from experimental measurements (the usual antenna pattern obtained by rotating the reflector and feed is very nearly equivalent for the spherical reflector to the angular transverse field distribution measured by the feed when it is rotated about the center of the fixed sphere through the focal region), another kind of distributed feed can be designed to correct for spherical aberration effects and modify the aperture amplitude distribution.REFLECTOR ANTENNAS 83 The basic idea of the transverse corrector is to arrange the primary feed in the form of circular rings as suggested by the Airy ring structure of an un- aberrated focal region. The central maximum produced about the focus F when a plane wave is incident on the sphere can be approximated by a single horn feed with diameter equal to that of the central Airy disk, and additional radiators can be positioned on rings to correspond to locations of the strongest maxima. By coupling the ring radiators and the central horn to a single feed line proportional to the ratios of the total energies in their respective focal regions and by phasing these signals to compensate for approximately 180° diffraction reversals and the secondary phase shifts due to optical errors, a transverse correcting primary feed can be fashioned. This approach to feed design has certain advantages. The radiating ele- ments can be made up of continuous individual horns giving an appearance of an eggcrate or honeycomb. When each of these horns is fully coupled to a receiver, multiple lobes of aberrated beams will be available as with an optical telescope. If, however, one pattern of high quality is desired, light coupling to adjacent horns properly phased will restore this pattern to the gain, side-lobe levels, and efficiency available with a paraboloid of equivalent diameter. It is possible with passive interconnection to restore several patterns simulta- neously. The diameter of the central Airy disk is approximately the size of an open waveguide, because to a good approximation the diameter of the disk is d = \of/D and f/D is effectively about 14 for a large spherical cap. Again the symmetry of the sphere is such that the same matrix of feed couplings will correct the pattern for a wide range of look-directions of the antenna. 17.7 Cylindrical Reflector Antennas Reflector antennas in the form of parabolic and right circular cylinders have been widely used as high-gain apertures illuminated by line sources. A notable example of a large scannable cylindrical antenna is the (400-by 600-ft) parabolic cylinder fed by a linear array located in Illinois and operated by the Uni- versity of Illinois. Analysis of a cylindrical or single-curve reflector antenna is similar and usually considerably simpler than that of paraboloids and spheres. Com- paring in general the principal characteristics of aperture phase, amplitude, and polarization with Sec. 17.3 on the paraboloid, we note that (1) the aperture amplitude taper due to variations in the distance p from the feed line to the reflector is proportional to 1/r rather than 1/p*, (2) no cross-polarized com- ponents are produced in the aperture field when incident and reflected waves are polarized parallel to the axis of the cylinder, and (3) the focal surface formed by an incident plane wave will be a single-sheeted caustic unlike the dual-sheeted caustics of doubly curved reflectors. The percentage of aperture blocked by a line source feed on a cylindrical antenna is usually greater than by a point source feed on a paraboloid. Mechanical construction of singly84 ANTENNA THEORY curved structures is usually simpler than that of doubly curved surfaces. An- tenna beam scanning in one plane can be achieved by feeding the cylinder with a line-source feed with a linear, tiltable phase front. A line source feed for a parabolic cylinder reflector can in effect be obtained by feeding the reflector with another parabolic cylinder. In the double- parabolic cylinder antenna, two parabolic cylinders are arranged at right angles to focus a plane wave incident from the z direction to a focal point. When the two cylindrical surfaces defined by yadfe @t+fP=4feth are constructed in a common coordinate system, their surfaces intersect along the curve when f =} _~@+1 1 ~ SB Some advantages in fabrication and in the independent control of horizontal and vertical beamwidths by adjusting the height of the two cylinders are realized. Line source feeds capable of scanning antenna beams are the most frequently used for feeding cylindrical antennas. Linear arrays for generating linear phase changes between radiating elements are discussed in Chaps. 5 and 7. Certain lens antennas such as the Luneberg lens, the Foster scanner, and the R-2R scanner and other geodesic forms discussed in Chap. 18 can generate the necessary tilted plane waves needed along a linear line source aperture for scanning a cylindrical antenna. Cylindrical antennas in the form of pillboxes are themselves a frequent choice for this function. Pillbox Antennas Pillbox antennas are a form of a cylindrical reflector antenna with the re- flector surface usually very short and terminated at both ends with parallel imaging metal plates.” This construction usually has a parabolic or circular reflecting curve, and the parallel plates are often terminated in a linear radiat- ing aperture flared into the shape of a horn to improve impedance match to free space. The spacing between the parallel plates is usually less than a wave- length and excited only in a single propagating TEM or TE mode. Pillboxes are almost always fed by small point sources, and several interesting forms capable of wide-angle scanning by feed motion have been developed. Figure 17.26 illustrates the basic form of the pillbox antenna which can be fed by a small vertically polarized horn or monopole located near the focus of the parabolic reflector to produce a plane wave along the flared line aperture. In this form the pillbox antenna suffers some pattern degradation and im-REFLECTOR ANTENNAS 85 1.816 in, outside y dimension 14.370in. y?=60« inner surface to probe Feed Co 4.20in— be be 60 in. —_________+| Fig. 17.26 A 3-GHz focus-in-the-face pillbox antenna. Dimensions for an antenna supporting a TEM mode are shown. pedance mismatch due to reflected energy intercepted by the feed. Tilting of the aperture wave front by motion of the feed results in large aperture phase errors and consequent radiation pattern deterioration due to the large off-focus aberrations of the parabolic cylinder. These difficulties can be remedied to some extent by constructing the pillbox in two layers and replacing the back parabolic reflector with a circular cylinder as shown in Fig. 17.27. Exit layer Linear Parallel aperture —> —— plate bend Open-ended . Entrance waveguide layer Semicircular backwall as Fig. 17.27 A double-layer cylindrical antenna using a semicircular backwall.86 ANTENNA THEORY A satisfactory solution to the 180° microwave junction needed to exchange energy without reflection between the entrance layer and the exit layer is available® for a wide range of incident directions and over a considerable fre- quency band. Of course, the semicircular backwall does not focus perfectly to a point, and correction techniques similar to those used on the spherical re- flector can be employed to cause a small source moving on a circular are to generate the tilted plane waves on the line aperture needed for scanning. In particular, the transverse corrector consisting of two auxiliary side horns coupled to an identical central horn is effective in reducing side lobes due to spherical aberration. Again the basic design criteria for transverse correction are (1) to couple the auxiliary horns to the feeding waveguide at the same decibel level as the height of the uncorrected first side lobes received on the central horn above and (2) to space and phase the auxiliary horns so that their peak patterns fall on the first side lobes of the central horn patterns and 180° out of phase with those side lobes. The feed suitable for feeding a pillbox an- tenna with circular back reflector shown in Fig. 17.28 is operated at a wave- length of 1.25 em with parallel plates supporting TE) mode. A correcting symmetric dielectric lens can be introduced in the lower or upper level of the pillbox to improve the linearity of the aperture phase, or the upper and lower levels can be separated by a distance as shown in Fig. 17.29. Geodesic prin- ciples presented in Chap. 16 are used to optimize the dimensions for minimum aperture phase errors. When the right-angle geometry as shown is used, the geodesic pillbox is never entirely free from phase errors. 17.8 Parabolic Torus Antennas The parabolic and elliptic torus reflectors are other doubly curved reflector surfaces which to some degree combine the excellent focusing qualities of a Variable flap attenuators Variable flap c enuators, phase shifters (resistance coated plastic Open (polystyrene) K-band guide 0.25in. x 0.50in. (6) Fig. 17.28 A triple-horn transverse corrector feed for scanning pillbox antenna with circular reflector backwall. do = 1.25 em. (a) Top view; (b) side view.