Investigations on Multifunctional Microstrip Radiators in Association with Optimum Feeding Techniques

Abstract

Traditional feeding techniques for microstrip antennas use direct-contact of patch and the preceding transmission-line – either a planar microstrip/stripline or a coaxial line with probe-excitation. The former radiate spurious; consuming precious substrate space while in the latter probe-inductance causes detuning and need drilling/soldering procedures; both becoming progressively impractical at higher frequencies. Proximity-coupled feeds employ fringing fields to link the patch; avoiding a physical contact e.g. end- or sidecoupling. Tighter couplings may be obtained by embedding microstrip between patch and ground-plane but requires a second substrate. Aperture-coupling is a significant improvement; a coupling slot appearing in the ground-plane with microstrip and patch on opposite sides. The two substrates may be independently optimized; the upper for radiation and the lower to confine fields. At higher frequencies, substrates must be considerably thinner which led researchers to consider waveguide feeding. This offers reduced line losses and a robust, mechanically rigid antenna assembly. Excitation through an end-wall aperture is reported with excellent performance but is not easily adapted for array realization. Transverse or longitudinal slots that run along one of the long walls of the waveguide are more amenable. In this doctoral thesis, a waveguide longitudinal- or shunt-slot feed for a microstrip antenna has been investigated and presented. A theoretical analysis using the Method of Moments has been developed and implemented as a computer program that is used for parametric studies. An FEM-based commercial e.m. simulator is used for validating the M-o-M results. Application to linear arrays is also investigated theoretically followed by experimental investigation of the single antenna element. An M-o-M formulation has been developed for the problem using entire-domain sinusoidal basis functions; in conjunction with a Galerkin procedure. Green’s functions for the waveguide and the grounded dielectric slab are used to obtain the M-o-M matrix. The latter being in the Fourier transform domain, a combination of analytical and numerical integration is used to obtain the terms. Symmetry properties of the matrix are used to optimize the computational burden. Solution of the M-o-M equation yields the unknown coefficients; thereby the unknown slot and patch currents. Both input characteristics and radiated fields are derived in terms of the determined coefficients, the latter using reciprocity principle. Algorithms and code have been implemented in FORTRAN based on the developed analytical expressions. An important consideration is the circumvention of integrand poles in the expressions for the substrate-related matrix terms. These correspond to the TE- and TM-poles of the surface waves. The integration path is modified to describe a small semi-circle in the complex plane around each precisely computed pole location. The formulation, thus, explicitly considers surface waves as opposed to more approximate transmission line- or cavity-models. The computer program has been developed in a modular form, organized into subroutines. Functions are defined for each integrand and a common integration loop computes all terms. Coding resembles analytical expressions to aid in verification and troubleshooting. A prototype antenna for the satellite communication uplink band has been designed to resonate at 6 GHz nominally; detailing the design procedure. The developed M-o-M program is then used to analyze the performance of this antenna. Criteria for convergence of matrix terms are 60/60 mode-indices in the waveguide section; discretization of 3 in azimuthal direction and radial truncation at 63k0. The number of basis functions across the slot and patch are 15 & 11 respectively for converged current distributions and input parameters. A series of parametric studies are performed with regard to the prominent design variables of the proposed radiator. Of these, a slot width of 0.01 yields smoothest predicted responses. A slot length of 0.08 yields a clear shunt resonance. Slots shorter than this show capacitive behaviour and longer inductive. Slot transverse offset offers a useful monotonic control range for coupled power upto ~ a/4 from waveguide centre-line. Tolerance of centering of the patch over slot is not critical though offset detunes the radiator slightly due to asymmetrical slot current distribution. Radiation pattern contributions of each expansion function on the slot and patch are determined – the oddorders typically contributing a peak on boresight and the even-orders resembling difference patterns. The aggregate pattern considering all expansion functions resembles the fundamental TM01-mode across patch. The H-plane pattern exhibits nulls in the plane of the substrate while the E-plane doesn’t due to direct slot radiation. A validation of the developed analysis is carried out on a commercial e.m. solver, the Ansoft® HFSS® using seeded meshing for efficient convergence. Input characteristics and patterns broadly validate the M-o-M results – a slight tweaking of the slot length being necessary to obtain resonance. Additional effects of varying ground plane size were assessed and are seen to introduce slow undulations in the pattern due to edge scattering. Foldover currents to the rear of the ground-plane significantly alter the backlobe. An explicit analysis is essential in some cases like active arrays where backlobe may induce spurious oscillations. For the same problem size, the developed M-o-M program is eight times faster than FEM even on a slower machine. Thus, the developed program may be used for rapid optimization during antenna design phase followed by verification / inclusion of second-order factors like finite ground. Subsequently, the proposed element has been cascaded to obtain linear arrays. Two representative cases are studied: uniformly-excited and Dolph-Chebyshev for reduced sidelobes with five elements each. The parametric studies obtained previously were effectively used to obtain the slot positions for the prescribed excitation amplitudes. The performance of the two modules was analyzed using an array factor calculation for arbitrary element positions together with the MOM-computed patterns; and subsequently validated on HFSS®. The uniformly-excited array gave the expected maximum directivity with sidelobe levels of -14.43 & -18.18 respectively. The Dolph-Chebyshev array; targeted for -20-dB equalized sidelobes, gave closely matching levels of -18.62 & -19.38 respectively. Sample hardware for the proposed WGMPA is realized on RO®-3003 Duroid® substrate and a machined integral aluminium waveguide fixture. Impedance measurements using a VNA show good return-loss behaviour. Radiation pattern measurements taken in an automated Anechoic Chamber, superimposed on the analyzed results indicate a broad match to theoretical predictions. However, leakage from the substrate-waveguide assembly and low coupled power results in undulations in the present hardware that may be overcome in the future by a bolt pattern or conductive adhesive. Theoretical investigations have established the efficacy of the proposed feeding mechanism. The analytical procedure stands validated through an alternate analysis on a proven e.m. solver. The use of the proposed element to obtain linear arrays with prescribed amplitude distributions has been successfully demonstrated through analysis. Such modules may be stacked to obtain a full-fledged planar array, though this has not been presently attempted. Hardware results are also seen to confirm the theoretical predictions within limits of measurement uncertainties; thereby experimentally validating the proposed waveguide feeding technique.