0 évaluation0% ont trouvé ce document utile (0 vote)
19 vues4 pages
Four different models for the input impedance of a probe in a parallel plate waveguide are compared. These models include a uniformcurrent model, a cosine current model, and a gap voltage source model. A simple CAlD formula for the probe inductance for a coaxial feed probe for a rectangular microstrip patch antenna is also introduced.
Four different models for the input impedance of a probe in a parallel plate waveguide are compared. These models include a uniformcurrent model, a cosine current model, and a gap voltage source model. A simple CAlD formula for the probe inductance for a coaxial feed probe for a rectangular microstrip patch antenna is also introduced.
Four different models for the input impedance of a probe in a parallel plate waveguide are compared. These models include a uniformcurrent model, a cosine current model, and a gap voltage source model. A simple CAlD formula for the probe inductance for a coaxial feed probe for a rectangular microstrip patch antenna is also introduced.
in a Parallel Plate Waveguide and a Microstrip Patch
Hao Xu, David R. J ackson, and J effery T. Williams Department of Electrical and Computer Engineering, University of Houston, Houston, Texas, 77204-4005, USA Introduction Four different models for the input impedance of a cylindrical coaxial-fed probe of radius a in an infinite parallel plate waveguide of height h (Fig.1) are compared. These models include a uniformcurrent model, a cosine current model, a gap voltage source model, and a coaxial frill model. These different models treat the vertical variation of the fields within the substrate and the field with varying degrees of rigor. The simplest one assumes a uniformcurrent on the probe. The second one assumes that the probe current has a cosine function variation along the length of the probe. The third model assumes that a voltage excites the probe across a finite vertical gap between the probe and the ground plane, in the region 0 <z <A. The fourth model assumes a radial l l p distribution of impressed field Ep across the aperture feed, which corresponds to the TEM mode of the feeding coaxial cable (inner radius a, outer radius b). Numerical HFFS simulation data is also compared with the results of these models in order to draw reliable conclusions about the accuracy of these models as a function of the substrate thickness. As an extension of these results, a simple CAlD formula for the probe inductance of a coaxial feed probe for a rectangular microstrip patch antenna (Fig. 2) is also introduced. Summary of Models for a ]Probe in Parallel-Plate Waveguide A. Uniform Current Model. If a uniform z-directed surface current on the probe is assumed, the probe impedance can be readily obtained as [l], [2] where kb and r ] l are the wavenumber and intrinsic impedance of the substrate, respectively. 0-7803-7846-6/03/517.00 62003 IEEE 124 B. Cosine Current Model An improved model assumes that the surface current density is a cosine function, I ( z ) =I , cos@, ( z - h)) . Using this current, the field within the waveguide is found by a Fourier series analysis. The complex power radiated into the waveguide is then found, and fromthis the input impedance is obtained as [3] C. Gap feed Model In this model the feed is assumed to be a region of impressed uniformvertical electric field in the region O< z < A on the surface of the metal probe, corresponding to a one Volt feed. After matching the boundary condition at p =a [3], the input admittance is found fromthe current at z =0 as L D. Magnetic Frill Model In this model the coax aperture is closed off with conductor by the equivalence principle, and a magnetic surface current is placed on the bottomconducting plate in the region a <p <b . For a one-volt feed, the current I ( z ) on the metal probe is then determined exactly by using reciprocity [3], and the input admittance is found fromthe current at z =0. Omitting the details, the final result is Probe Feed for a Rectangular Patch Antenna The probe inductance of a rectangular microstrip antenna (Fig. 2) varies with the feed position, and this is not accounted for by the infinite parallel-plate waveguide model. To address this problem, it is noted that the patch has approximately a PMC boundary at the four edges. Using image theory, the field inside the patch cavity is the sumof the contributions fromthe original probe and an infinite two- 125 dimensional set of images. If the probe is close to an edge, there will be one image current with a primary influence. If we just take only one image current into account, and use the uniformcurrent model, the following formula for the probe impedance is obtained, where s is the distance of the probe center to the nearest edge Results Figure 3 shows that the uniform, cosine, and frill models agree well when the waveguide height h is small. But when the height becomes large, the three models begin to diverge significantly fromone another. The HFFS simulation data agrees with the frill model very well, even when the height of the waveguide reaches a quarter of a wavelength. An interesting observation, based on numerical experimentation, is that that gap provided the ratio of the probe radius a (the i.nner radius of the frill) to the outer radius of the frill b is chosen to correspond to a 50 P coax. The probe reactance for a cylindrical probe feed in a rectangular patch antenna (Fig. 2) is calculated in three different ways: Z;, fromEq. (l), Z,? fromEq. (5), and an accurate calculation fromthe cavity model. Fromthis comparison (not shown here), it appears that a good approxi.mate CAD formula for the probe reactance is Z, , =max( Z , , , Z p ) . With this formula, the maximum error occurs at the intersection point of the two curves, and is about 552. model agrees very well with the frill model if a gap height A =(b - 0) / 3 is used, Conclusions All four models agree (uniform, cosine, gap, and frill) agree well for small waveguide heights, but the differences become quite large as the waveguide height increases. The fiill model is the most accurate of the four, and agrees well with numerical HFFS results, even for large heights. The gap source model gives results that are almost identical to the frill model, provided the size of the gap source is chosen as A =( b - U ) / 3 . The case of a coaxial probe feed for a rectangular microstrip patch antenna was also considered. The finite dimensions of the patch cause the probe impedance to deviate quite a bit fromthe infinite parallel-plate waveguide results. However, a simple one-termimage theory correction to the infinite parallel-plate waveguide impedance was shown to give satisfactory results for the probe impedance of the patch antenna. 126 References [ I ] D.C. Chang, On the Electrically Thick Monopole, Part 1 - Theoretical Solution IEEE [2] W.F. Richards, J .R. Zinecker, R. D. Clark, and S. A. Long. Experimental and Theoretical Investigation of the Inductance Associated with the Microstrip Antenna Feed. Electromugnetics, Vol. 3, no. 3-4, J uly-Dec. 1983. Trans. Antennas undPropugut., Vol. AP-16, J an 1968. [3] R. F. Hamngton, Time-Hurmonic Electromagnetic Fields, McGraw-Hill, Inc., New York, 1961. 5 C a Fig. 1. Coaxial probe feed in an Fig. 2. Coaxial probe feed for a infinite parallel-plate waveguide. rectangular microstrip patch antenna. . . . . . . . . Uniform current model 1 1 500 I 1 450 400 350 300 250 200 150 100 50 0 Cosine current model _ _ _ _ 0 HFFS simulation 0 0.005 0.01 0.01 5 0.02 0.025 Fig. 3. Probe resistance versus the height of the waveguide h in meters. E, =2.2, U =0.635 mm, f= 2 GHz, Z, =50 Q (corresponding to b =2.19 mm). 127