Reconfigurable Transmitarray Antennas
by
Jonathan Yun Lau
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Edward S. Rogers Department of Electrical and Computer Engineering University of Toronto
Copyright c 2012 by Jonathan Yun Lau Abstract
Reconfigurable Transmitarray Antennas
Jonathan Yun Lau
Doctor of Philosophy
Graduate Department of Edward S. Rogers Department of Electrical and Computer
Engineering
University of Toronto
2012
Transmitarrays have been shown to be viable architectures for achieving high-directivity reconfigurable apertures. The existing work on reconfigurable transmitarrays is sparse, with only a few experimental demonstrations of reconfigurable implementations. Fur- thermore, of the designs that have been presented, different approaches have been pro- posed, but the advantages and drawbacks of these approaches have not been compared.
Therefore, in this thesis we present a systematic study of the different approaches to designing reconfigurable transmitarrays, and present designs following these approaches with experimental validation.
First, we investigate the distributed-scatterer approach, which is modeled with layers of identical scattering surfaces. We characterize the beamforming capabilities and then present a Method of Moments technique for analyzing and optimizing designs that follow this approach. Then, we present experimental results for a unit cell with varactor-loaded dipoles following this approach. From these results, we demonstrate that the structure thickness following this approach is problematic for beamforming applications.
Taking the coupled-resonator approach, we next present a slot-coupled patch design that is significantly thinner and easier to fabricate than designs that follow the first ap- proach. Implementing this design in a fully reconfigurable transmitarray, we demonstrate two-dimensional beamforming. An advantage of this design is that it can also operate as
ii a reflectarray. Next, following the guided-wave approach, we present a transmitarray design that uses a bridged-T phase shifter and proximity-coupled differentially-fed stacked patches.
Not only does this design not require vias, it is has a large fractional bandwidth of 10 per- cent, which is unprecedented in reconfigurable transmitarrays. Implementing this design in a full transmitarray, we experimentally demonstrate reconfigurable two-dimensional beamsteering, as well as shaped-beam synthesis. The main contributions of this thesis are two-fold. First, we thoroughly and system- atically compare the transmitarray approaches, which has not been previously done in literature. Secondly, we experimentally demonstrate a reconfigurable array design that achieves better bandwidth, scan angle range, and beam-shaping capability, than existing designs, with reduced fabrication complexity and physical profile.
iii To God be the glory Acknowledgements
It was only through the support of many individuals and organizations that this thesis was successful. First, I would like to sincerely thank my supervisor Prof. Sean Hum for his constant support, guidance, and mentorship throughout my studies. When I joined the electromagnetics group, my knowledge of fields and waves was very limited at best.
I am extremely grateful for his patience and effort in guiding and teaching me. Not only have I grown academically, but I have also been greatly enriched and blessed by his regard for my personal development.
I would also like to thank Prof. Sarris, Prof. Eleftheriades, and Prof. Poon from the University of Toronto, and Prof. McNamara from the University of Ottawa for being on my supervisory and examination committees. Through their feedback and questions, I have not only gained valuable insight into the technical challenges, but more importantly, the motivation of the research.
I am also very grateful for the many friendships that I have built with the students and staff in the electromagnetics group. In particular, I would like to thank Marco and Tse for all the time spent answering my many questions about theory, simulation, and fabrication, and also Krishna and Tony for the many insightful conversations and memorable experiences.
I would also like to acknowledge the generous financial support provided by the Nat- ural Sciences and Engineering Research Council of Canada, the Ontario Graduate Schol- arship in Science and Technology, and the Edward S. Rogers Sr. Scholarship, without which this work would not have been possible.
Finally, I would like to thank my parents Edmund and Angela, my parents-in-law
Leonard and Belinda, and my brother and sister-in-law Josiah and Fiona for their un- conditional support and encouragement. Most of all, I would like to thank my beloved wife Melodie for sharing every step of this journey with me. I could not have asked for a better companion through life’s adventures.
v Contents
1 Introduction 1
1.1 ApproachestoTransmitarrays ...... 5
1.2 Motivation...... 10
1.3 Objectives...... 11
1.4 Outline...... 13
2 Background 14
2.1 ArrayAntennas...... 14
2.1.1 PhasedArrays...... 15
2.1.2 Reflectarrays ...... 18
2.1.3 FixedTransmitarrays...... 20
2.1.4 ReconfigurableTransmitarrays...... 25
2.1.5 FrequencySelectiveSurfaces ...... 27
2.2 ScatteringBehaviorofTransmitarrays ...... 28
3 Relevant Background Theory 30
3.1 Preliminaries ...... 31
3.1.1 CoordinateSystems...... 31
3.1.2 VectorPotentialandWaveEquations ...... 33
3.2 Radiation of Finite Structures ...... 36
3.2.1 RadiationfromaPointSource...... 36
vi 3.2.2 Radiation from a Finite Aperture ...... 38
3.2.3 Radiation from a Finite Array ...... 39
3.2.4 Directivity, Beams, and Side-lobes ...... 41
3.2.5 Gain,Losses,andEfficiency ...... 43
3.3 Infinite Arrays and Floquet Analysis ...... 45
3.3.1 Modeling Free-Space Propagation with Waveguides ...... 50
3.3.2 Green’sFunctions...... 51
3.3.3 Periodic Waveguides ...... 52
3.3.4 RectangularWaveguide...... 53
3.3.5 Parallel-Plate Waveguide ...... 57
4 Distributed-Scatterer Approach 61
4.1 Layered Scattering Surfaces for Beamsteering ...... 61
4.1.1 UniformScatteringSurface...... 67
4.1.2 GraduatedScatteringSurface ...... 76
4.1.3 Simulation Results ...... 86
4.2 Analysis of Discrete Layered Designs ...... 91
4.3 Transmitarray Unit Cell of Loaded Dipoles ...... 98
4.3.1 Unit Cell Design ...... 99
4.3.2 SimulatedResults...... 105
4.3.3 ExperimentalResults...... 107
4.4 Conclusions ...... 112
5 Coupled-Resonator Approach 114
5.1 TransmitarraysasFilters...... 115
5.2 Unit Cell with Slot-Coupled Patches ...... 121
5.2.1 Equivalent Circuit Model ...... 122
5.2.2 ImplementationGeometries ...... 123
vii 5.2.3 Simulation Results ...... 124
5.2.4 ExperimentalResults...... 130
5.3 Transmitarray Design and Implementation ...... 132
5.3.1 EffectofMutualCoupling ...... 135
5.3.2 Biasing Circuitry ...... 137
5.3.3 ExperimentalSetup...... 139
5.3.4 ElementCharacterization ...... 142
5.3.5 ExperimentalResults...... 143
5.4 Reflectarray-ModeOperation ...... 151
5.5 Conclusions ...... 155
6 Guided-Wave Approach 158
6.1 UnitCellDesign ...... 160
6.1.1 ReconfigurablePhaseShifter...... 160
6.1.2 AntennaandBalancedFeed ...... 162
6.1.3 Unit Cell Implementation ...... 163
6.1.4 Simulated and Experimental Results ...... 167
6.1.5 Stacked-PatchDesign...... 170
6.1.6 Simulated and Experimental Results ...... 172
6.2 TransmitarrayImplementation...... 178
6.2.1 MutualCoupling ...... 178
6.3 One-DimensionalBeamforming ...... 179
6.3.1 ArrayFabrication...... 181
6.3.2 ElementCharacterization ...... 183
6.3.3 ArrayBeamforming...... 184
6.3.4 Cross-Polarization ...... 193
6.3.5 Shaped-BeamSynthesis ...... 194
6.4 Conclusions ...... 195
viii 7 Comparison of the Approaches 198
7.1 DesignandOptimization...... 198
7.2 PhaseRangeandInsertionLoss ...... 199
7.3 Beamsteering and Side-lobe Levels ...... 200
7.4 Shaped-beamSynthesis...... 201
7.5 Bandwidth...... 202
7.6 ElementControl...... 202
7.7 PhysicalSize ...... 204
7.8 Fabrication ...... 204
7.9 Scalability and Extensibility ...... 206
7.10Summary ...... 206
8 Conclusions 210
8.1 Contributions ...... 213
8.2 ClosingRemarks ...... 215
Appendices 216
A Method of Moments Analysis for Dipole Unit Cell 216
B Multi-mode S-parameters 224
B.1 Definition of Multi-mode S-parameters ...... 224
B.2 Cascading multi-mode S-parameterblocks ...... 229
C Loss Budget Calculations 232
C.1 ArrayLosses...... 232
C.2 SpilloverLoss ...... 233
C.3 TaperEfficiency...... 234
ix D Stacked Patch Antenna and Feed Design 237
D.1 MicrostripTransition ...... 238
D.2 FeedMitre...... 239
D.3 StackedPatches...... 239
E Projection Matrix Algorithm 248
E.1 Source Listing for calc ff mat.m ...... 249
E.2 Source Listing for beamshape.m ...... 250
Bibliography 253
x List of Symbols
A, Ax,Ay,Az Magnetic vector potential AF Array factor
Az Azimuthal angle (measurement coordinate system)
Am MoM constant a Element width
B Susceptance
B Magnetic flux density
Bmn MoM constant b Element height
Cv Varactor capacitance
MN Cmn MoM constant d Layer spacing
δ( ) Dirac delta function
δm Kronecker delta function D Electric flux density D Directivity
Maximum dimension of radiating structure D
E, Ex,Ey,Ez Electric field intensity ǫ Permittivity constant
ε Efficiency
xi El Elevation angle (measurement coordinate system) f0 Operating frequency f˜ Fourier transform of f G¯ Dyadic Green’s function e G¯ Electric dyadic Green’s function
G Gain g Dipole gap size
Γmn Propagation constant in waveguide in z-direction for mode mn H Magnetic field intensity HPBW Half-power beamwidth
I Scalar current
J, Jx,Jy,Jz Current density k Free-space propagation constant k Wave vector kxm, kym, kzmn Floquet wavenumbers
Lv Varactor parasitic inductance l Dipole length
λ Free-space wavelength
M¯ Vector base for TE waveguide modes m Spectral index in x
M¯ Tensor base for TE waveguide modes Permeability constant
N Number of scattering surface layers n Spectral index in y
N¯ Vector base for TM waveguide modes
N¯ Tensor base for TM waveguide modes p Spectral component index
xii P Order of MoM sinusoids used to approximate currents
Prad Radiated power φ Azimuthal angle (beamforming coordinate system)
φ0 Azimuthal beam angle Ψ sin( ) or cos( ) ψ Electric scalar potential
Qo, Qe Multi-mode S-parameter matrices
Rvs, Rvp Varactor parasitic resistance (series/shunt)
RM Reflection coefficient for mode mn Mmn
Ra, Rb Multi-mode S-parameter matrices r Radial distance (beamforming coordinate system) r Observation (field) position vector r′ Source position vector
S11 Reflection coefficient
S21 Transmission coefficient T Transmission matrix
Aperture surface area S SLL Side-lobe level s Dipole position offset
T Total structure thickness
T M Transmission coefficient for mode mn Mmn t Substrate thickness
θ Polar angle (beamforming coordinate system)
θ0 Polar beam angle U Radiation intensity v Window function
W Spatial period of periodic scattering surface
xiii Wn Coefficient of window function
Wrad, Wrad Radiated power density w Dipole width ω Angular frequency (in radians)
Y Admittance
Z Impedance
xiv List of Tables
4.1 Minimum structure thickness ((N 1)d)...... 73 − 4.2 Varactor diode capacitance and biasing voltage ...... 101
4.3 Dipole unit cell design parameters ...... 103
5.1 Specialconfigurations...... 135 5.2 Pencil beam directivity, peak gain, side-lobe sevel, and beam width . . . 146
5.3 Transmitarraylossbudgetforbroadsidebeam ...... 151
6.1 Pencilbeamangles ...... 190
6.2 Pencil beam directivity, peak gain, side-lobe sevel, and beam width . . . 191
6.3 Transmitarray Loss Budget for Broadside Beam at 5 GHz ...... 193
7.1 Summaryofdesigncomparisons ...... 207
7.2 Comparison with existing designs ...... 208
xv List of Figures
1.1 Reflectarrayantenna ...... 4
1.2 Transmittarrayantenna ...... 5
1.3 Thin lens from a refraction perspective ...... 6
1.4 Thinlensfromdelayperspective...... 7
1.5 Thin lens from a Huygen’s wavelet perspective ...... 7
1.6 The lens as a collection of scattering surfaces ...... 8
1.7 The lens as a collection of resonators ...... 8
1.8 The lens as a collection of antennas and phase shifters ...... 9
1.9 Transmitarray/lensdesign ...... 11
2.1 Reconfigurablephasedarray ...... 15
2.2 Phasegradientofaphasedarray ...... 16
2.3 Rotmanlens...... 17
2.4 Equivalent Norton circuit for a receiving antenna ...... 29
3.1 Arraycoordinatesystem ...... 31
3.2 Beamformingcoordinatesystem ...... 32
3.3 Measurementcoordinatesystem ...... 33
3.4 Free-spaceradiation...... 35
3.5 Pointsourceinfree-space...... 38
3.6 Radiatingapertureinfree-space ...... 40
xvi 3.7 Radiatingarrayinfree-space...... 41
3.8 Side-lobes ...... 43
3.9 Transmitarraylosses ...... 44
3.10 Angled plane-waves due to different phase and magnitude periodicities . 45
3.11Infiniteperiodicarray...... 46
3.12 Radiated plane waves from an infinite periodic aperture ...... 49
3.13 Identical field solutions for freespace and periodic waveguide ...... 50
3.14 Rectangular waveguide T E10 mode and equivalent array and plane waves 54 3.15 Waveguide ports forscattering parameters ...... 56
3.16 Parallel-plate waveguide T EM mode and equivalent array and plane waves 58
4.1 Lensingwithscatteringsurface(s) ...... 62
4.2 Surface inside a waveguide with periodic boundary conditions ...... 63
4.3 Arrangement of multiple surfaces and points xm ...... 64 4.4 Powernormalization ...... 67
λ 4.5 The effects of the number of layers for d = 4 ...... 69 4.6 Theeffectsoflayerspacing...... 70
4.7 The effects of layer spacing (continued) ...... 71
4.8 S21 remapping...... 72 4.9 Maximumaveragetransmissionpower ...... 73
4.10 Susceptance ranges (solid - max, dashed - min) ...... 74
λ 4.11 S-parameters for N = 4 and d = 4 ...... 75 4.12 Computed real-part (arbitrary units) of normalized electric fields for dif-
ferent susceptance values (and phase shifts) ...... 76
4.13Graduatedphaseshift ...... 78
4.14 Computedelectricfields ...... 79
4.15 Fields in each mode for W =2.4λ ...... 80
4.16 Fields in each mode for W =7λ ...... 81
xvii 4.17Lateralpropagation...... 82
4.18 Beam powers of different modes p for scan angles θ1 ...... 83 4.19 Beam powers for different N and d ...... 84
4.20 Beam powers for different N and d (continued)...... 85
4.21 SEMCAD scattering surface implementation with lumped inductors and
capacitors ...... 87
4.22 Infinite scattering surface modes at different beamsteering angles..... 88
4.23 Implementation of finite scattering surface structure in SEMCAD .... 88
4.24 Radiated fields from a finite layered scattering surface ...... 89
λ 4.25 Simulated far-field pattern for a 180 mm surface (N =4,d = 4 ) ..... 90 4.26 Discretelayeredstructure ...... 92
4.27 Point current source a waveguide with a dielectric slab ...... 93
4.28Waveguidesetup ...... 98
4.29 Equivalent circuit models for dipoles ...... 99
4.30Unitcellgeometry ...... 100
4.31 Varactor diode series model L =0.4 nH, R = 2 Ω,C =0.15 pF 2.0 pF 100 v vs v − 4.32 Best S-parametersfromMoMoptimization...... 102
4.33 Effectofcenterdipolelength...... 104
4.34 Effectofdipolespacing...... 104
4.35Effectoflayerspacing ...... 105
4.36 MoM and FDTD S21 ...... 106
4.37 MoM and FDTD S11 ...... 106 4.38Fabricatedunitcell ...... 107
4.39Fabricatedunitcell ...... 108
4.40 Measured unit cell S-parameters...... 109
4.41 Summary of simulated (FDTD), numerical (MOM), and experimental
(EXP) and results at 6.4GHz ...... 110
xviii 4.42 Simulated S21 withlossandbiasingstructure ...... 111
5.1 Poleangleanddistance...... 117
5.2 Exampletwo-polecircuit...... 118
5.3 Example two-resonator complex poles and response, with s1 and s2 loci as
C is varied, with L = Lc = 1 nH and Z0 =50Ω...... 118 5.4 Example three-resonator complex poles and response ...... 120
5.5 Reconfigurable slot-coupled patches transmitarray element structure . . . 121
5.6 Circuitmodel ...... 123
5.7 Ideal simulated S21 for different configurations (Cp,Cs) ...... 126 5.8 Ideal transmission response at 5.7GHz ...... 127
5.9 Theeffectofvaractordiodeloss ...... 128
5.10 Simulated S21 in rectangular waveguide, with varactor losses ...... 129 5.11 Transmission response at 5.5GHz,withvaractorlosses ...... 130
5.12 Experimental waveguide test harness ...... 130
5.13 Measured S21 (magnitude) ...... 131
5.14 Measured S21 (angle)...... 132 5.15 Measured S-parametersforoptimalconfigurations ...... 133
5.16 Transmitarrayperformance ...... 134
5.17 Mutual coupling simulation waveguides ...... 136
5.18 Simulated mutual coupling for pairs of configurations ...... 137
5.19 Array element and bias network design (vertically exaggerated) ..... 140
5.20 Fabricated 6 6array ...... 141 × 5.21 Transmitarray experimental setup in the near-field scanner ...... 141
5.22 Phase range histograms for fabricated elements ...... 143
5.23 Co- and cross-polarizations in the two principal planes ...... 144
5.24 Measured far-field pattern as the elevation angle is scanned (E-plane) . . 145
5.25 Measured far-field pattern as the azimuth angle is scanned (H-plane) . . 147
xix 5.26 Measured aperture fields (array bounds denoted by rectangle) ...... 148
5.27 Measuredfar-fieldpattern(E-plane)...... 150
5.28 Measuredfar-fieldpattern(H-plane) ...... 150
5.29 Measured S11 for 1 V slot varactor bias and varied patch capacitance . . 152
5.30 S11 of unit cell at 5.5GHz ...... 152
5.31Reflectarraysetup...... 153
5.32 Reflectarray measured far-field pattern (E-plane) ...... 154
5.33 Reflectarray measured far-field pattern (H-plane) ...... 155
6.1 Bridged-T circuit topology ...... 161
6.2 Return loss and transmission phase a single bridged-T filter with L1 = 0.75 nH, L =1.25 nH, and C =0.3 pF 1.2pF...... 162 2 − 6.3 Balanced bridged-T circuit topology ...... 164
6.4 Biasingscheme ...... 164
6.5 Dimensions of the bridged-T phase shifter in microstrip ...... 165
6.6 Overallunitcelldesign ...... 166
6.7 Varactor diode model (C = 0.15 pF 2.0 pF, R = 3 Ω, R = 1.5 kΩ, − vs vp
Lv =0.4nH) ...... 167
6.8 Simulated results for the single-patch element ...... 168
6.9 Fabricated single-patch element ...... 169
6.10 Experimental results for the single-patch design ...... 169
6.11 Experimental S21 at 4.87GHz...... 170
6.12 Designofthestacked-patchelement ...... 171
6.13 Fabricatedstacked-patchelement ...... 172
6.14 Simulated results for the stacked-patch design ...... 173
6.15 Experimental results for the stacked-patch design ...... 174
6.16 Experimental S21 at various frequencies (solid - magnitude, dashed - phase)175
xx 6.17 Summary of simulated and experimental results (*denotes the results from thesingle-patchdesign)...... 175
6.18 Simulated far-field radiation pattern of the stacked-patch antenna . . . . 177
6.19 Setup of the side-by-side (x) mutual coupling simulation ...... 179
6.20 Mutual coupling between adjacent array elements ...... 179
6.21 Radiated fields from a finite layered scattering surface ...... 180
6.22 2D simulated results for a six-element stacked-patch bridged-T transmitarray181
6.23 Internalviewofthefabricatedarray ...... 182
6.24 Assembled 6 6transmitarray...... 183 × 6.25 Element classification for the 6 6array ...... 184 × 6.26 Element characterization 6 6array ...... 185 × 6.27 Measured far-field beam patterns at 5.0GHz...... 187
6.28 Measured far-field beam patterns at 4.7GHz...... 188
6.29 Measured far-field beam patterns at 5.2GHz...... 189
6.30 Comparison of expected (AF) and measured (EXP) far-field patterns at 5GHz ...... 192
6.31 Measured cross-polarization for φ0 = 45◦ ...... 194 6.32 Shaped-beam synthesis measurements ...... 196
7.1 Comparison between 2D 180 mm bridged-T transmitarray and four-layer
scatteringsurfaceapertures ...... 201
7.2 Moving pass band and resulting bandwidth ...... 203
A.1 Dipolegeometries...... 216
B.1 Waveguide port definitions (only two modes shown) ...... 224
B.2 S-parameter characterization for a single layer, in a waveguide section of
length 2d ...... 225
B.3 S-parametercases...... 226
xxi B.4 Scatterercascading ...... 229
C.1 Transmitarraysetup ...... 233
D.1 Return loss for the microstrip transition ...... 238
D.2 Return loss for the microstrip transition ...... 239
D.3 Mitresimulationsetup ...... 240
D.4 Returnlossfordifferentmitres...... 240
D.5 Stackedpatchlayoutwithfeed...... 241
D.6 Qualitative comparison of input impedance trajectories for single patch (loop-less) and when a stacked patch is added (with loop), for different
lowersubstratethicknesses ...... 243
D.7 Theeffectoflowerpatchlength ...... 244
D.8 Theeffectofupperpatchlength...... 244
D.9 Theeffectoflowerpatchwidth ...... 245
D.10Theeffectofupperpatchwidth ...... 246 D.11Theeffectofthefeedgap ...... 246
D.12Theeffectofuppersubstrategap ...... 247
xxii Chapter 1
Introduction
The antenna is an intrinsic part of any wireless communication or sensing system, tran- sitioning electromagnetic energy from its guided form within a structure to radiation in free-space, and vice versa. Antennas have evolved into all types of shapes and sizes, embedding themselves ubiquitously into everyday applications, such as personal commu- nications, home electronics, and transportation.
One of the major trends in antenna design has been in making the antenna less intrusive and improving aesthetics, while at the same time pushing the performance and increasing the number of supported frequency bands. For many years, antennas pervaded our view in the form of long monopole antennas on automobiles and large backyard satellite dishes. However, while some antennas have remained overtly visible, most are disappearing from view. For example, very few modern cellular handsets, GPS navigation devices, or laptop computers have protruding antennas. Antennas on automobiles and aircraft for the most part are integrated into the bodies of the vehicles. Antennas are becoming more discrete, conforming themselves to the form factors of the devices on which they reside. Consequently, due to proximity, device components have increased interaction with the radiating characteristics of the antenna, often making it difficult to clearly define the boundaries of the antenna on a device. The notion of radiation from a
1 Chapter 1. Introduction 2 single dedicated structure is changing.
In many instances, such as satellite and point-to-point communications or radar, the radiated power needs to be concentrated in a specific direction to maximize the power transmitted to the receiver, or to reduce interference to other systems operating within the same frequency band. In such cases, antennas with high gain and directivity are required, and two general approaches have been used to design such antennas. Aperture antennas, the first, achieve high directivity by creating large illuminated apertures. The larger the aperture and the more uniform the field phases and amplitudes on the aperture, the higher the directivity will be. Examples include the horn antenna and the widely recognizable parabolic reflector antenna. The other approach is to use an array of small antennas, each excited with a specific phase, to effectively create a large aperture of uniform fields.
However, directivity and gain can be achieved without the driven radiator being highly directive. Passive scatterers can also be used to modify the fields from a low directivity feed antenna and achieve high directivity. For example, Yagi-Uda antennas, which are commonly seen on rooftops for television signal reception, consist of a dipole surrounded by parallel parasitic elements. As the feed antenna radiates, currents are excited on the parasitic elements such that the fields radiating from the parasitic currents add constructively in a specific direction. In fact, the parabolic reflector of a parabolic reflector antenna can be viewed as a large passive scatterer illuminated by a small source.
Nevertheless, regardless of how directivity is achieved, a large effective aperture with uniform fields needs to be synthesized, which necessitates a structure with a large physi- cal size. Large antennas are not only cumbersome and unsightly, but they are also costly to use in many cases. For example, rigid reflector antennas, which are commonly used for high-gain satellite communications, are bulky and heavy, making them extremely expen- sive to launch into orbit. For this reason, antenna arrays have been used to synthesize large apertures. Chapter 1. Introduction 3
Antenna reconfigurability is also crucial in many applications. For instance, in a radar system where a high-directivity aperture needs to be scanned, typically a large reflector or antenna array is mechanically rotated rapidly, requiring a large amount of space and mechanical hardware. Reconfigurability is also required in communications systems where the transmitter or receiver are in motion such as satellite tracking, or for adapting to changing environments where there are moving scatterers and multi-path.
The ability to electronically reconfigure an antenna without mechanical movement is very desirable, because much less physical space is required and the reconfiguration speed is many orders of magnitude faster. Moreover, mechanical systems require significantly more maintenance and are more prone to failure over time.
Phased arrays, which are simply arrays of antennas driven with specific phases, have been proposed as elegant solutions for achieving high-directivity and electronic reconfig- urability. Unlike large reflector antennas or horn antennas which must be mechanically moved to redirect the beam, the beams produced by a phased array can be electroni- cally scanned very rapidly. Since the elements of the array can take planar or conformal arrangements, phased arrays are also generally less bulky than aperture antennas. How- ever, one major challenge with phased arrays is scalability. Since each element in the array needs to be individually excited, the length of the transmission lines needed to drive each element in the array increases quadratically with the size of the array, leading to quadratically increasing transmission line losses and space requirements. The losses and complexity associated with the feed network are major disadvantages.
For this reason, we can turn to spatial techniques for exciting an array, using a single feed antenna and an array of passive scatterers to synthesize a large effective aperture. A prominent example of this is the reflectarray antenna [1], shown in Figure 1.1, where an array of passive elements is spatially excited by a feed antenna and resonates, radiating in such a way that an aperture with uniform phase is created. A reflectarray can be reconfigured by electronically modifying the resonances of the elements. Chapter 1. Introduction 4
Figure 1.1: Reflectarray antenna
The transmitarray, also known as an array lens1, is a discretized lens, and is the dual structure of the reflectarray, as shown in Figure 1.2. From an optics perspective, a transmitarray is to a lens, what a reflectarray is to a mirror. A transmitarray has several advantages over a reflectarray. While the scanning range of a reflectarray is fun- damentally limited to one side of a plane, or 180◦, conceptually a spherical or cylindrical transmitarray surrounding an isotropic feed antenna can achieve a 360◦ scanning range. Furthermore, while feed blockage is a challenge for reflectarrays because the feeding an- tenna is on the same side of the array as the radiated fields, such is not the case for a transmitarray. In fact, with the feed out of the way, transmitarrays can not only produce beams in the far-field, but they can also be used for focusing fields from the feed into a spot in the near-field, which is not possible with reflectors due to feed blockage. Trans- mitarrays and reflectarrays can also be combined into more elaborate structures, as has been done with lenses and mirrors in catadioptric telescopes2.
Compared to traditional reflectors and lenses, transmitarrays and reflectarrays have the major advantage of being planar. Not only are planar antennas easier to integrate into everyday structures such as walls and roofs, they are more portable because they can
1The terms transmitarray and array lens will be used interchangeably in this thesis 2A catadioptric system is one where both refraction and reflection are used Chapter 1. Introduction 5
Figure 1.2: Transmittarray antenna be rolled up or dismantled in panels and easily reassembled. Unlike parabolic reflectors whose shape needs to be rigidly supported, the shape of a large planar transmitarray can simply be supported using stretched cables or membranes. Moreover, because the properties of transmitarrays can be arbitrarily designed, they can be made to take any form factor. If made to be reconfigurable, distortions in the shape of a transmitarray can be electronically corrected, which is a significant advantage when the antenna is operating in a remote location.
Although a criticism of transmitarrays and reflectarrays is that the feed significantly increases the profile of the antenna, this is in fact not a problem in many applications. For example, for space applications, physical profile is only a constraint when the antenna is being sent into orbit. Therefore, the feed structure can easily be collapsed and moved into place when the antenna is actually deployed. Alternatively, other feeding mechanisms such as Cassegrain reflection can be used to reduce the physical profile.
1.1 Approaches to Transmitarrays
In this thesis, we will explore different approaches to understanding reconfigurable trans- mitarrays. We begin with an illustration using thin lenses from optics. The most basic Chapter 1. Introduction 6 way to understand a thin lens is to assume that the media through which the waves prop- agate is homogeneous. In this case, the boundary conditions, described by Snell’s law, relate the material properties with the angles of incidence and refraction. In a properly shaped lens, all refracted waves converge on a focal point, as shown in Figure 1.3.
Figure 1.3: Thin lens from a refraction perspective
Alternatively, we can look at the paths on which waves propagate. Because different media have different associated phase velocities, waves propagating through different distances in the media experience different phase shifts and time delays. The focus of a lens is the point at which all paths to that point incur the same phase shift or delay. As illustrated in Figure 1.4, the two paths have different lengths, but the delay τ experienced on both paths is the same.
It is also possible to look at the lens from the Huygen’s principle, which views each point in the lens as a scatterer that produces waves in every direction, as illustrated in
Figure 1.5. Each scatterer will be excited by the incident wave according to the properties of the lens medium. In this case, the focal point is a result of the interference pattern created by an infinite number of tiny scatterers in the lens.
We can glean from these simple examples three approaches with which we can design a lens. If we perceive the lens as a collection of passive scatterers distributed in space, Chapter 1. Introduction 7
Figure 1.4: Thin lens from delay perspective
Figure 1.5: Thin lens from a Huygen’s wavelet perspective Chapter 1. Introduction 8 then we will seek to manipulate the properties and positions of the scatterers to control the focusing abilities of the lens. When the scatterers are small, then the scatterers can be perceived as layers of closely-spaced scattering surfaces, as shown in Figure 1.6. We will call this approach the distributed-scatterer approach.
Figure 1.6: The lens as a collection of scattering surfaces
For a lens, we are most interested in controlling the phase through the structure.
Since phase of the scattered fields has the greatest variation when the scatterers are near resonance, we can also see the transmitarray as a collection of coupled resonators, as illustrated in Figure 1.7, and take the coupled-resonator approach.
Figure 1.7: The lens as a collection of resonators
Finally, if we look at the delay or phase shift incurred over different paths, then we may be inclined to discretize the lens and use time delaying or phase shifting circuits to control the properties of the lens. In order to use such circuits, the free-space wave must first be coupled into transmission lines with antennas, as shown in Figure 1.8. This Chapter 1. Introduction 9 approach will be called the guided-wave approach.
Figure 1.8: The lens as a collection of antennas and phase shifters
These approaches are classified and named in this thesis for the purpose of comparing them systematically, and we note that the terminology is not commonly used in the community. Also, these three approaches are not mutually exclusive, since a particular structure can belong to multiple classes depending on how it is perceived. For example, a guided-wave transmitarray design may use a coupled-resonator phase shifter, or the guiding mechanism may be so short that one could argue that the waves are not in fact guided. Alternatively, the surfaces of a distributed-scatterer design may be resonant and tightly coupled. Nevertheless, for the purpose of this thesis, we will focus on the following concepts for each approach. For the distributed-scatterer approach, we will focus on scattered fields and surface impedances and lateral interaction of adjacent array cells. Although equivalent circuit models can be used to model a single cell, they cannot model the interaction between adjacent cells, which is an important consideration as we will see later. For the coupled-resonator approach, we will focus on the presence of resonant structures and how they interact. Unlike the distributed-scatterer approach, we will focus on longitudinal coupling of the resonators. Finally, for the guided-wave approach, we will focus on the structure as an assembly of microwave circuits connected by transmission lines. Chapter 1. Introduction 10
1.2 Motivation
This thesis demonstrates how transmitarrays can be used to effectively address three major challenges that have emerged in antenna design: directivity and gain, reconfig- urability, and hardware cost (in terms of monetary cost, complexity, or physical space).
Many solutions for each of the above challenges have been proposed, but typically ad- vances in one goal have led to trade-offs in another. The difficulty is thus in addressing the three challenges simultaneously.
Firstly, antenna directivity and gain are crucial. As the demand for wireless band- width continues to increase, higher frequencies will be needed. However, with increasing frequency, atmospheric attenuation also increases, resulting in the need for high-gain an- tennas. That is, transmitters and receivers must direct their power towards each other, rather than radiating in every direction. Such spatial directivity can also help alleviate the effects of multipath and co-channel interference that reduce the available bandwidth.
Furthermore, directivity is crucial for achieving high-resolution imagery in radar or re- mote sensing applications.
Secondly, the radiation characteristics from highly directive antennas often need to be reconfigurable when terminals are in motion or when the channel conditions change.
As frequency increases, channel variations in time increase also, resulting in the need for adaptive communication systems, for which real-time reconfigurable antennas are important. In radar applications, electronic reconfigurability can significantly reduce the space and hardware costs associated with the mechanical rotation of antennas, and improve reliability since no moving parts are required.
Thirdly, as wireless devices are becoming ubiquitous, more and more constraints are being placed on antenna shape, size, complexity, and cost. While single antennas have low associated hardware costs, they are typically not able to achieve the desired directivity and reconfigurability discussed previously. On the other hand, while phased- array antennas are able to achieve greater directivity and reconfigurability, they require Chapter 1. Introduction 11 large bulky feeding networks that result in significant transmission line losses. This underscores the importance of finding practical low-cost techniques for implementing arrays.
1.3 Objectives
Given the issues outlined above, the goal of this thesis is to demonstrate how the trans- mitarray paradigm can be used to achieve high-directivity reconfigurable apertures, while at the same time minimizing hardware cost and complexity. To achieve this goal, we first need to understand the fundamental problems, issues, and challenges. We can succinctly define the transmitarray or lens design problem as follows, as illustrated in Figure 1.9:
For given incident fields on one side of the transmitarray, produce outgoing
fields with specific phase and maximal power on the opposite side.
Figure 1.9: Transmitarray/lens design
For instance, the incident waves may be spherical waves produced by a nearby source or quasi-plane waves from a distant transmitter. The desired outgoing waves may be quasi-plane waves in one or more directions, or waves that converge to a nearby receiver.
We want the outgoing waves to carry as much power as possible, meaning that loss in the transmitarray and the reflections on the incident side must be minimized. Chapter 1. Introduction 12
As we will see in Chapter 2, reconfigurable transmitarrays have not been extensively studied in literature. Different approaches have been proposed for transmitarray design, but their relative strengths and limitations are not well understood. So, to systematically investigate transmitarray approaches, we will take three perspectives: the distributed- scatterer approach, the coupled-resonator approach, and the guided-wave approach. For each approach, we will identify the advantages and the limitations.
Following, detailed design methods and experimental implementation are crucial for demonstrating viability and how practical issues such as element control and fabrication complexity can be addressed. Only a few experimental array implementations have been presented and actual reconfigurable beamforming results have been very limited. Thus, drawing from the insight gained from studying the approaches, we will design planar reconfigurable transmitarrays for beamforming. In order to assert the practicality of transmitarrays, the experimentally tested arrays must be fully reconfigurable in two- dimensions, performing beamscanning over both azimuth and elevation angles.
To summarize, the three objectives and contributions of this thesis are:
1. Systematic examination and comparison of the different approaches to transmi-
tarrays to identify the strengths and limitations of each approach
2. Detailed description of design methods for reconfigurable transmitarray antennas
3. Design and experimental demonstration of reconfigurable transmitarray beamform-
ing to show that transmitarrays can be used to produce practical reconfigurable
high-gain apertures Chapter 1. Introduction 13
1.4 Outline
The outline of this thesis is as follows. In Chapter 2, we first review existing work on transmitarrays and related work in literature. Following, in Chapter 3, we provide relevant definitions and theoretical background for understanding the thesis. In Chap- ters 4, 5, and 6, we present three approaches and designs following these approaches: the distributed-scatterer approach, the coupled-resonator approach, and the guided-wave ap- proach. We compare the different approaches in Chapter 7 and present our conclusions in Chapter 8. Chapter 2
Background
In this chapter, we will review results and designs that have been proposed in literature that are relevant to this thesis. Because a transmitarray is essentially a discretized lens, it belongs to a class of antennas called quasi-optical or spatially-fed antenna arrays. For conciseness, we will not review dielectric lenses or Fresnel zone plates that are often used for focusing, because they are neither reconfigurable nor discretized. Therefore, we will first review the history of antenna arrays, focusing on spatially-fed arrays, from their origins in reflectarray design to the state-of-the-art in reconfigurable transmitarrays. Next, we will discuss the concept of maximum receivable power for an antenna and how it relates to transmitarrays. Finally, we will review beamforming techniques that will be used later in this thesis.
2.1 Array Antennas
Fundamentally, the maximum directivity of an antenna is proportional to its size. Rather than using a single large antenna, large radiating apertures can also be synthesized using an array of smaller antennas.
14 Chapter 2. Background 15
2.1.1 Phased Arrays
Phased arrays, which are arrays of antennas where each array element is directly fed by a transmission line with a specific phase and amplitude, have existed for many years. Be- cause of their conceptual simplicity, they have been popular for achieving high-directivity
fixed and reconfigurable radiating apertures.
To briefly review the governing principle of a phased array illustrated in Figure 2.1, consider the linear array with element spacing d shown in Figure 2.2. Suppose the elements of the phased array are excited with phase-shifted signals, where the phase difference between adjacent elements is φ. Consider parallel rays traced at an angle of
θ0 from the array axis. The additional distance of each subsequent ray to an imaginary aperture plane denoted by the solid line is d cos θ0. If k is the free-space phase constant, or the rate at which the phase changes in space for a propagating wave, then the rays arriving at the imaginary aperture will have all have uniform phases if
φ = kd cos θ0. (2.1)
In this case, the radiation from each element will add constructively, resulting in a beam directed at an angle θ0. From this, we can see that a phase gradient in the array excitation produces a narrow coherent beam, or a pencil beam, in a particular direction. A similar principle applies when the array is used to receive a signal from a particular direction.
Figure 2.1: Reconfigurable phased array
To drive each element, traditional phased arrays split the input signal into multi- ple feed lines using power dividers, and then phase shifters are used to manipulate the Chapter 2. Background 16
Figure 2.2: Phase gradient of a phased array radiated phase of each array element. Non-reconfigurable phased arrays typically use delay line phase shifters to create pencil beams in a specific direction. The amplitude of each element can also be manipulated by tailoring the power-splitting ratio of the power dividers to gain additional control of the radiated waves.
The body of research on phased arrays is vast and so it is not possible to thoroughly review the literature on phased arrays. The theory and design techniques of phased arrays have been well studied and many comprehensive books are available [2, 3]. Therefore, in the following sections, we will only highlight examples showing techniques that have been used to implement reconfigurability.
Because phased arrays require a phase shifter for each element, it is challenging to make each phase shifter reconfigurable for larger arrays. As a result, Rotman lenses were proposed as a way to introduce reconfigurability into linear phased arrays [4] without reconfigurable phase shifters. A typical Rotman lens, illustrated in Figure 2.3, consists of a convex-shaped parallel-plate waveguide, with input and output transmission lines connected by cone-shaped impedance matchers. In transmitting mode, by exciting a given input port, the different delays caused by the path distance to the output feed in the lens create different phase gradients across the antenna array, which result in a beam in a particular direction. Superposition by exciting multiple ports allows for multi- beam synthesis. While Rotman lenses provide an easy way to achieve reconfigurable Chapter 2. Background 17 beamsteering, the beam directions are discrete and limited to the number of inputs. Moreover, they can only be used for linear arrays and electronic input switching with a large number of inputs is non-trivial, particularly at high frequencies.
Figure 2.3: Rotman lens
To achieve independent phase control, each element requires an independently con- trollable reconfigurable phase shifter. A great deal of work has been done on monolithic- microwave integrated-circuit (MMIC) phase shifters, but complex fabrication processes are involved and the insertion losses are significant (more than 4 dB at 5 6 GHz) [5]. − More recently, with advances in semiconductor technology, researchers have studied electronic tuning techniques by integrating diodes in phased arrays. Varactor diodes, when reverse biased, provide a continuously tunable capacitance in the depletion region of the diode. They have been used to implement the extended-resonance power-dividing method for reconfigurable phased arrays without the need for separate power dividers and phase shifters [6]. At 2 GHz, a four-element linear array achieved a 20◦ reconfigurable scanning range with only 1 dB of insertion loss.
Alternatively, PIN diodes can be used as switches for changing signal paths. An as- sembly of Rotman lenses and PIN diode switches has been used to create an electronically reconfigurable phased array [7]. Operating at 35 GHz, a 10-element linear array achieved a scanning range of 49◦ experimentally with a gain of 15.6 dB. The scanning range was covered by seven beams spaced 7◦ apart, since Rotman lenses can only produce beams Chapter 2. Background 18 in discrete directions.
As mentioned in the introduction, one of the challenges with phased arrays is that the complexity, size, and losses of the feed network increase substantially with array size.
For this reason, researchers have turned to spatial feeding techniques for antenna arrays.
2.1.2 Reflectarrays
Spatial feeding of antenna arrays originates from optics, where a feed antenna illumi- nates the array elements, which in turn re-radiate the power with specific phases. Since reflectarrays have been a prominent example of discretized optics for microwaves [1], we will briefly review milestones in their development.
The reflectarray is analogous to a parabolic reflector, where the guiding principle in its design is that the phase delay from the source to the reflector, to the aperture plane must be equal for every possible radiation path. Thus, the primary challenge in reflectarray de- sign is the control of individual element phases. The first reflectarray proposed consisted of an array of open-ended waveguides, where the depth of each waveguide controlled the phase delay of the element [8]. As microstrip technology increased in popularity due to its ease of analysis and fabrication, the microstrip patch became the most popular reflectarray element. Many aspects of microstrip reflectarrays were studied, including control of element phases using microstrip stubs attached to each patch [9, 10], control of element phases using varied patch length [11, 12], and reflectarray feeding configurations
[13].
Compared to the parabolic reflector, the main drawback of the reflectarray is that the bandwidth tends to be narrow. As a result, work has focused on increasing the bandwidth of reflectarrays using stacked patches, achieving a fractional bandwidth of
10% [14]. Another technique for achieving large bandwidth is the use of true-time-delay
(TTD) lines. Reflectarrays have also been able to achieve fractional bandwidths of 10% by aperture-coupling microstrip delay lines to patch elements [15]. Chapter 2. Background 19
A number of techniques have been presented for electronically tuning reflectarrays [16, 17]. Most techniques either involve tuning a capacitance by manipulating of the
DC voltage across a varactor diode [18, 19, 20, 21, 22], micro-electromechanical systems
(MEMS) capacitor [23, 24], or by shorting and connecting components using PIN diodes
[20] or MEMS switches [25]. Other techniques also include mechanically raising and lowering patches [26], or changing the dielectric properties of the substrate using liquid crystal substrates [27, 28]. Amplifiers to augment the reflected power have also been proposed in fixed [29] and reconfigurable reflectarrays [22].
Generally, two approaches for reconfigurable reflectarrays have emerged. One ap- proach is to have the antenna of each element couple power into a transmission line, and then to use a reflection-type microwave phase-shifter to manipulate the phase of the signal that is re-radiated. To tune the reflection phase of each element, PIN diodes have been used to discretely change the length of the transmission line [20, 30]. Alternatively, varactor diodes have been used to tune capacitances, allowing continuous tuning of the reflection phase [20, 21, 22]. The other approach is to integrate the tuning component directly into the antenna, so that the phase of the re-radiated waves is manipulated by changing the resonance of the antenna. Varactor diodes have been directly connected to microstrip patches [18, 19], and MEMS switches have been used to connect or dis- connect structures that reactively load patches [31, 25]. These two approaches parallel the guided-wave and distributed-scatterer approaches to transmitarrays introduced in
Chapter 1.
Each tuning technology has its respective advantages and disadvantages. PIN diodes are inexpensive and readily available, but because they act as switches, multiple control lines are required for each element to achieve higher-resolution phase control. Varactor diodes are also readily available and allow for continuous phase tuning with a single control voltage, but are non-linear when the input power is large. Because the capacitance produced by the varactor diode is controlled by a DC bias voltage, a large input power Chapter 2. Background 20 affects the bias voltage, modulating the capacitance and producing distortion. MEMS technologies are capable of high power handling with low losses (3.5 dB for varactor diode vs. 0.5 dB for MEMS at 5.8 GHz [32]) and with virtually no distortion [32], but they are more challenging to fabricate. In terms of operating frequency, discrete components such as PIN and varactor diodes cannot operate at higher frequencies because of the parasitic effects caused by the component size and packaging. While tunable liquid crystal substrates are more suitable at millimeter-wave frequencies, they suffer from significant losses.
Regardless, given the dual nature of transmitarrays and reflectarrays, tuning tech- nologies that have been shown to be effective for reflectarrays can generally be applied in similar manners to transmitarrays. The advantages and disadvantages such as distortion and parasitic effects are all very similar. In the following sections, we will first review fixed transmitarrays, followed by reconfigurable transmitarrays, which are tuned using techniques similar to those that we have just discussed.
2.1.3 Fixed Transmitarrays
The transmitarray is similar to the reflectarray in that the phase delay from source to transmitarray element to the aperture plane must be equal for every possible radiation path, in order to produce a pencil beam. However, the most crucial difference between a reflectarray and a transmitarray is that in a reflectarray, all of the power will be reflected regardless of the frequency and unit cell design. In the extreme case with frequencies well below the resonant frequency of the cell, a reflectarray will reflect with 180◦ phase-shift, or a coefficient of 1, because the elements will have little effect and the reflecting ground − plane will dominate the response. Similarly, in the other extreme case with frequencies well above resonance, a reflectarray will also reflect with 180◦ (or 180◦) phase-shift. For − frequencies in between, a reflectarray will reflect with a phase shift between 180◦ and − 180◦. However, with a transmitarray, if the structure is not properly matched to free- Chapter 2. Background 21 space, all of the power will be reflected from the input of the transmitarray, resulting in zero transmission through the structure. Thus, the magnitude of transmission is an additional design consideration for transmitarrays.
An interesting observation is that the bandwidth limitation of reflectarrays can in part be attributed to the presence of the ground plane. To increase the bandwidth, the elements need to be less resonant. However, if the resonance of the elements is reduced, then the magnitude of the fields radiated by the elements is also reduced, and the response of the reflectarray becomes dominated by reflection off the ground plane. Because of this, it may be easier to achieve larger bandwidths using the transmitarray paradigm.
Fixed transmitarrays have been studied for many years, with most of the work taking the guided-wave approach, connecting two antennas together with a phase shifter. Two early microstrip-based transmitarray designs based on coupled patch antennas were pro- posed in 2006 [33]. The usefulness of the first design, which used slot coupling to connect the antennas, was deemed to be limited due to the limited bandwidth. The other design, which consisted of patches slot-coupled to a common transmission line, had an improved bandwidth, but at the cost of more substrate and metallization layers.
The concept of transmitarrays became more popular in the early 2000’s as integrated active antennas emerged. Transmitarray designs based on transmission line-coupled patch antennas were extended to include active elements [34]. A series of studies were presented on active transmitarrays [35, 36, 37], where a 4 2 transmitarray was pro- × totyped and measured. The elements consisted of an amplifier, with aperture-coupled patch antennas connected to the input and output of the amplifier. This structure was also used as power amplifier [38].
Development of transmitarray principles also progressed in the design of spatial power amplifiers, as this last example shows. Using a large number of low-power elements, a spatial power amplifier uses the principles of spatial power combining to achieve large power handling at high frequencies, while avoiding transmission line loss problems. A Chapter 2. Background 22 spatial power amplifier is very similar to a fixed transmitarray in that an array of elements is used as a lens. Fixed array lenses with amplifiers have been demonstrated for use as spatial power combiners [34, 39, 40, 41, 42]. The major difference between spatial power amplifiers and transmitarrays is that the design goal of a spatial power amplifier is to maximize power, with the collimating element very close to the array. On the other hand, for transmitarrays, the design goal is beam synthesis for a receiver that is very far away.
A5 5 transmitarray design was first experimentally demonstrated at 12 GHz in 2007 × [43, 44, 45]. The array elements each consisted of a stripline delay line between transmit and receive patch antennas, where the patches were connected to the stripline using coaxial probes. The phase delay of each element was varied by controlling the length of the delay line. While simple in principle, the implementation of such a structure was difficult, since the design required five conducting microstrip layers (one for each patch, and three for the stripline), all connected through vias.
To avoid the use of vias, an asymmetric fixed transmitarray design involving one microstrip-fed patch connected directly to a slot-coupled patch was proposed at 9.6 GHz
[46, 47]. In this design, additional stacked patches were also added to increase the bandwidth of the structure. A circular array with 208 elements experimentally achieved a directivity of 28 dBi with side-lobe levels of about 20 dB and 1.5 dB of insertion loss − through the structure.
In the transmitarray designs discussed so far, the phase shift through each element is manipulated by adjusting the length of the transmission line connecting the receive and transmit antennas. In this way, element phase shifts can be selected from a continuous range of values. Phase-quantized transmitarrays, where elements can only have discrete phase shifts have also been proposed, with 1-bit (0◦/180◦) and 2-bit (0◦/90◦/180◦/270◦) designs [48, 49]. It was suggested that phase quantization of the elements only has a minor impact on the array directivity, with only a 1 dB difference between the continuous phase and 2-bit phase designs in the configurations studied. Chapter 2. Background 23
The proposed 2-bit design [48] used two patch antennas connected by co-planar waveg- uide (CPW) transmission lines, where the lengths of the transmission lines were quantized to four lengths. A 20 11 array achieved a peak gain of 14.3dB at 9.6 GHz with a 3 dB × fractional gain bandwidth of 10.4%. While the simplicity of this CPW implementation was an advantage, requiring only two layers of substrate and three layers of metallization with no vias, higher than expected side-lobe levels of 6.6 dB were observed. Multi- − ple beam operation was also proposed, by changing the angle of the transmitarray feed antenna, where at 30◦ offset the array produced a peak gain of 11.5 dBi.
In the 1-bit design [49], two square patch antennas were connected using a single via, with one of the two patches rotated by 90◦ or 90◦ to achieve the 0◦ and 180◦ phase shifts. − Consequently, the input and output fields have orthogonal polarizations. A 20 20 array × was experimentally demonstrated at 9.8 GHz, achieving a gain of 23.6 dBi and side-lobe levels of 15 dB with a very large 3 dB fractional gain bandwidth of 18.4%. − Similar designs with probe-connected square patches were presented at V-band [50] and X-band [51, 52]. The 1-bit linearly polarized design was extended to a 2-bit circularly polarized design by rotating the patches by 0◦, 90◦, 180◦, or 270◦. A20 20 array × achieved a gain of 24.3 dBi with 13.7 dB side-lobe levels, with a large 3 dB fractional − gain bandwidth of 13% at 9.8 GHz [51, 52]. A similar 20 20 array achieved a gain × of 22 dBi with 16 dB side-lobe levels, with a large 3 dB fractional gain bandwidth of − 13.3% at 60 GHz [53]. The insertion loss through the structure was 0.46 dB.
A circularly-polarized lens based on rotated patch elements has also been proposed
[54]. Two stacked patches on either side of a ground plane were coupled by a cross-shaped slot. The phase shift through the structure was manipulated by rotating the patches.
At 12.4 GHz, the 349-element array had side-lobe levels of about 25 dB. The array − achieved a bandwidth of 7% where the axial ratio was less than 2 dB, with an insertion loss through the structure of 0.6 dB.
A different approach, the distributed-scatterer approach, has also been used to design Chapter 2. Background 24 transmitarrays, where array elements consist of four layers of coupled rings [55]. Without a ground plane, the waves passed through the structure without being guided. The phase shift through the structure was manipulated by adjusting the width of the rings and the size of the gap between the rings. At 30 GHz, the 21 21 array achieved a gain of × 28 dBi and side-lobe levels of 16.5 dB, with a 3 dB fractional gain bandwidth of 7.5%. − With each ring layer having about 0.25 dB of loss and 0.3 dB of loss from the foam spacers between the layers, the total insertion loss through the structure amounts to about 2 dB. One of the advantages or this coupled-ring design is that it is not restricted to a single polarization, but a drawback is that the layers need to be spaced approximately a quarter-wavelength apart, which makes the structure thick at lower frequencies.
A thinner structure, also employing layers of metallization, has also been proposed
λ using three metallization layers but a structure thickness of only 10 . The structure es- sentially consists of three closely spaced metal layers with large slots. The slots run hor- izontally and are connected end-to-end, effectively creating a long horizontal slot across the entire structure. The phase shift through the structure is manipulated by changing the width of the slots at different points, where the width the slot in each of the three layers may be different. One-dimensional beamforming at broadside is demonstrated at
30 GHz. Although very simple in design and fabrication, the drawback of this design is that it may be difficult to extend to two-dimensional beamforming.
To briefly summarize, many studies have demonstrated high-directivity transmitarray design at many different frequencies, with insertion losses as small as 0.46 dB and frac- tional bandwidths as large as 18%. Most of the designs consist of coupled-patch antennas, using delay lines to achieve phase shifting and stacked patches to increase bandwidth. All of the arrays, with some exceptions [54], are designed to produce a pencil beam when illuminated by a horn at broadside (at prime focus). That is, these transmitarrays are designed to convert the spherical waves produced by the feed antenna to broadside quasi- plane waves. Because these transmitarrays designs are fixed, new arrays would need to Chapter 2. Background 25 be fabricated to produce pencil beams at non-broadside angles. It is for this reason that we rarely see any fixed transmitarray results producing non-broadside beams.
2.1.4 Reconfigurable Transmitarrays
Following the development of tunability in reflectarrays, reconfigurability was introduced to transmitarrays. As is the case with reflectarrays, one of the trade-offs of introducing reconfigurability to transmitarrays is that the insertion loss through the structure will inevitably be increased, due to the additional complexity and resistances of the tuning mechanisms. Moreover, fabrication and experimental verification with reconfigurable ar- rays is more challenging due to the cost and control hardware for the tuning mechanisms.
With reconfigurable transmitarrays, it is more feasible to experimentally explore beamforming capabilities of non-broadside beams. There are certain challenges asso- ciated with beamforming that have often been overlooked with fixed transmitarrays that only produce broadside beams, particularly for beams with larger deviations from broad- side. For example, because the required phase differences between adjacent array ele- ments are larger, the mutual coupling between elements can cause significant element phase error.
Preliminary results for an electronically reconfigurable transmitarray design were first presented in 2008 [56]. Probe-coupled patches were loaded with varactor diodes connected to the ground plane to manipulate the resonance of the patch antenna, but only 90◦ of phase agility was achieved.
Reconfigurable array lenses using MEMS switches to achieve 2-bit phase control, allowing four possible discrete phase tunings, were also proposed for the purpose of an- tenna beamforming [57, 58]. Slot antennas were coupled using short stripline transmis- sion lines, and MEMS switches were used to control which transmission lines were used.
One-dimensional beamforming was demonstrated experimentally witha22 22 array at × 34.8 GHz, producing a gain of 9.2 dB and side-lobe levels of 13 dB at broadside. The − Chapter 2. Background 26 average insertion loss of the structure was 8.0 dB. The reconfigurable array was able to scan a pencil beam in an 80◦ range, producing 3.8◦ of gain at 40◦, and 6.4◦ of gain at − +40◦ off-broadside in the E-plane. This work is one of the first results on reconfigurable lenses that experimentally demonstrated electronic beamscanning.
In 2010, a fully reconfigurable 6 6 transmitarray was presented for beamforming [59]. × Probe-fed stacked patches were connected by three reflection-type phase shifters in series that used varactor diodes to tune the phase shift. Because the phase shifters were larger than the element spacing, the elements were grouped together, with four patch antennas connected to each phase shifter. In this way, there were effectively 9 tunable cells in the array. The reconfigurable transmitarray demonstrated beamforming at 0◦ and 9◦ degrees off-broadside at 12 GHz, producing 16 dBi of gain and side-lobe levels of about 13 dB. − The average insertion loss through the structure was 3 dB. Some of drawbacks of the design, however, include its physical thickness and fabrication complexity, as microstrip layers were soldered together at perpendicular angles.
While the above reconfigurable transmitarray designs follow the typical guided-wave approach, an alternative transmitarray design following the distributed-scatterer ap- proach has also been proposed [60, 61]. In this design, four layers of ground planes with varactor-loaded slots were used, similar in principle to the fixed coupled-ring trans- mitarray discussed previously [55]. Simulation results for a unit cell were presented, demonstrating 400◦ of phase range where the insertion loss was less than 3 dB.
In contrast to the wealth of work that has been presented for reconfigurable re- flectarrays, there have not been many investigations into reconfigurable transmitarrays, especially for beamforming applications. This motivates the need for a systematic un- derstanding of the different design approaches, as well as more experimental results to demonstrate the viability of the reconfigurable transmitarray paradigm. Chapter 2. Background 27
2.1.5 Frequency Selective Surfaces
Transmitarray and reflectarray design is also closely related to the design of frequency selective surfaces (FSS), as many FSSs are designed to behave as lenses or as reflectors.
For example, a surface made of small electrically small patches [62] was shown to improve the gain bandwidth to over 20%. A varactor diode-tuned high-impedance surface has also been proposed as a reconfigurable reflectarray [63]. FSS theory and application has been well-studied [64].
The concept of coupling two radiating structures separated by a ground plane, fol- lowing the guided-wave approach, can be found in many FSS designs, such as the spatial bandpass filter using two hexagonal patches coupled by a CPW resonator [65]. Alterna- tively, FSSs designs have also been proposed following the distributed-scatterer approach, such as the spatial bandpass filter using layers of patches and grids [66]. Because most
FSS designs are uniform, they are more applicable for spatial filtering applications than for beamforming, where the elements need to be varied across the aperture to produce a phase gradient.
Reconfigurable FSSs have also been proposed for lensing applications, such as a tun- able liquid crystal frequency selective surface (FSS) [67], and a varactor diode-based tunable FSS [68]. While these surfaces are reconfigurable, it is the operating frequency of the entire lens that was tuned in these designs. That is, phase tunability was not a de- sign objective, and the elements of the arrays could not be individually tuned. Thus, the surfaces are not suitable for beamforming applications where phase control is required.
Moreover, the structures do not have sufficient order to achieve 360◦ of phase tunability with acceptable insertion loss.
While many of the concepts between reconfigurable transmitarrays and FSSs are similar, the primary difference is that while the array elements of transmitarrays and
λ reflectarrays are typically resonant with sizes on the order of 2 , the element sizes of FSSs λ are electrically small (less than 10 ). The two major consequences are mutual coupling Chapter 2. Background 28 between the elements and element controllability.
Firstly, because FSS elements are very close together, their fields and currents are tightly coupled. Because of this, array beamforming, which is well understood, cannot be directly applied because it assumes that array elements can be individually manipulated.
When elements are highly coupled, it is very difficult to determine the required element tuning to produce a desired far-field pattern. While it may be possible for some FSS structures to produce a pencil beam where a linear phase gradient is desired, accurate beamshaping, or the production an arbitrary far-field pattern is next to impossible with high mutual coupling.
Secondly, due to the high element densities in FSSs, individual element control is very difficult in practice because of the large number of elements, and because there is very little space for control circuitry between the elements. Furthermore, the ratio of achievable far-field tunability versus complexity is much lower than that of antenna arrays, making FSSs less practical for beamforming applications.
2.2 Scattering Behavior of Transmitarrays
At this point we will briefly clarify a notion that has arisen in literature on the maximum proportion of incident power that an antenna can receive. This is pertinent to trans- mitarrays because an ideal transmitarray must receive 100% of the power on one side and re-radiate all of the power on the other side. A maximum receive antenna efficiency would impose a fundamental limit on the efficiency of a transmitarray. Intuitively, one could argue that any transmitarray structure necessarily scatters some power because it is a physical structure in the presence of incident waves, and thus the efficiency cannot be 100%. Because transmitarray elements are antennas, and antennas are known to have a scattering cross-section, some of the power must be scattered.
Thevenin or Norton equivalent circuits have been used to model receiving antennas Chapter 2. Background 29
[69], as shown in Figure 2.4, where Ya represents the admittance of the antenna, YL is the load, and Ig is the current that is generated when the antenna is excited. By this argument, maximum power is delivered to the load when it is conjugately matched to the antenna impedance, YL = Ya∗, resulting in a maximum receivable power of 50%.
Figure 2.4: Equivalent Norton circuit for a receiving antenna
This misconception was rigorously treated [70] by showing that while electric dipoles have a maximum receive efficiency of 50%, the theoretical maximum receive efficiency is
100% for electric dipoles over a ground plane, or a combination of electric and magnetic dipoles. Essentially the misconception is due to the fact that the equivalent circuit cannot fully capture the behavior the receiving antenna in free-space, especially with antennas containing ground planes. Therefore, following the guided-wave approach, if appropriate ground-backed antennas are used for the receive and transmit antennas, then theoretically it is possible to achieve zero return loss and insertion loss through the structure, as has been demonstrated in many transmitarray designs.
We note that this result does not apply to the distributed-scatterer approach, when the structure does not behave as an antenna. For example, consider an array where each element consists of two passive dipoles of different lengths. If the dipole lengths and spacings are tuned such that the waves that they scatter have equal magnitudes but opposite phases, then there will be zero power reflected. In this way, the transmitarray can achieve 100% transmission efficiency using only electric dipoles. This will be discussed in more detail in Chapter 4. Chapter 3
Relevant Background Theory
In this chapter, we will present the theory used to analyze and design transmitarrays.
In the first section, we will define the coordinate systems that will be used, and then we will introduce the vector potential and wave equations.
In the second section, we will present theory that is used for analyzing a finite radiat- ing structure in free-space. Beginning with a point source and far-field approximations, we will develop the radiation characteristics for a finite aperture, followed by the defini- tion of metrics for evaluating antennas such as directivity, gain, and loss. This theory is relevant for the experimental results for arrays that will be presented in later chapters.
Following, the third section presents the theory for analyzing infinite periodic aper- tures, such as an infinite antenna array. Unlike the finite case, far-field approximations cannot be made for infinite structures. Rather, Floquet analysis and waveguide Green’s functions are used. The theory for infinite periodic structures is relevant for unit cell design and characterization.
While in practice, a unit cell is first characterized prior to experimentation with full arrays, we have opted to review the theory of general finite radiating structures first because the theory can be presented with a better logical flow.
30 Chapter 3. Relevant Background Theory 31
3.1 Preliminaries
3.1.1 Coordinate Systems
In this thesis, three coordinate systems will be used. Note that in this thesis, we will only be considering lineary ˆ-polarized array elements.
Array and Element Coordinate System
When discussing an array element or the array locally, the Cartesian coordinate system will be used. We will always assume that both the incident and outgoing waves are propagating in the +ˆz-direction, and that the electric field is always polarized in they ˆ- direction, as shown in Figure 3.1. Unless otherwise specified, the transmitarray is always in the xy-plane and centered at the origin.
Figure 3.1: Array coordinate system
Spherical Coordinate System
The spherical coordinate system is used for describing radiated waves in the far-field, when the distance between the observation point and the source is very large. Following standard notation, r, θ, and φ, will be used for spherical coordinates, as shown in Figure Chapter 3. Relevant Background Theory 32
3.2. The relationship with Cartesian coordinates is given by
z x cos θ = tan φ = . (3.1) x2 + y2 + z2 y
In the spherical coordinate system, the unit vectors are
rˆ =x ˆ sin θ cos φ +ˆy sin θ sin φ +ˆz cos θ (3.2a)
θˆ =x ˆ cos θ cos φ +ˆy cos θ sin φ zˆsin θ (3.2b) − φˆ = xˆ sin φ +ˆy cos φ. (3.2c) −
Figure 3.2: Beamforming coordinate system
Measurement Coordinate System
The horizontal coordinate system, which uses elevation and azimuth angles to specify a direction, is used for measured array beamforming results. This is in following with what is common in the antenna measurement community. While similar to the spherical coordinate system, the azimuth is not to be confused with the azimuthal angle φ, and the elevation is not to be confused with the polar angle θ. In the measurement coordinate system, the array is arranged such that broadside, or the z-axis of the array, aligns with
0◦ in azimuth and elevation, and the y-axis of the array is aligned with 90◦ elevation. As shown in Figure 3.3, the relationship between the azimuth (Az) and elevation (El) and Chapter 3. Relevant Background Theory 33 the two other coordinate systems is given by y sin θ sin φ tan El = = (3.3a) √x2 + z2 sin2 θ cos2 φ + cos2 θ x sin θ cos φ tan Az = = . (3.3b) z cos θ
Figure 3.3: Measurement coordinate system
3.1.2 Vector Potential and Wave Equations
A fundamental approach for determining the electric field due to a current density J is through the use of a magnetic vector potential A, defined as
B = A. (3.4) ∇ × To understand why it is advantageous to use the vector potential, we will develop the wave equation for the vector potential from Maxwell’s equations. Assuming simple media where the electric and magnetic flux densities are linearly related to their respective field intensities (D = ǫE and B = H), Maxwell’s equations are
H = J + jωǫE (3.5) ∇ × E = jω H. (3.6) ∇ × − Chapter 3. Relevant Background Theory 34
Taking the curl of both sides and substituting, we get two wave equations for the electric and magnetic fields
H k2H = J (3.7) ∇×∇× − ∇ × E k2E = jω J (3.8) ∇×∇× − − where k = ω√ ǫ. Because any field can be decomposed into the sum of a curl and a solenoidal (curl-free) field, we can express the electric field as
E = ψ jωA (3.9) −∇ − where ψ is the electric scalar potential, producing a solenoidal field ψ. A trivial substi- ∇ tution of (3.4) and (3.9) into (3.6) reveals that Maxwell-Faraday’s equation is satisfied.
Substituting (3.9) into Maxwell-Ampere’s equation (3.5) gives
H = J + jωǫE ∇ × 1 ( A) = J + jωǫ( ψ jωA) ∇ × ∇ × −∇ − J = A jω ǫ ψ + ω2 ǫA − −∇ × ∇ × − ∇ J = ( A) 2A jω ǫ ψ + k2A. (3.10) − − ∇ ∇ −∇ − ∇ Since (3.4) only defines the curl of A, the divergence of A can be arbitrarily assigned.
Defining the divergence using the Lorentz gauge, where A = jω ǫψ, we can simplify ∇ − the wave equation for the magnetic vector potential to
2A + k2A = J. (3.11) ∇ −
With the Lorentz gauge, the resulting expression for the electric field becomes
jω E = k2 + ( ) A. (3.12) − k2 ∇ ∇ In Cartesian coordinates, the vector wave equation (3.11) simplifies into three scalar Chapter 3. Relevant Background Theory 35 equations, one for each component
2A + k2A = J (3.13a) ∇ x x − x 2A + k2A = J (3.13b) ∇ y y − y 2A + k2A = J . (3.13c) ∇ z z − z The vector potential is particularly useful because in free-space and some other scenarios, the components of A correspond directly to the components of J [71] (p. 141)