Improving Compact Short Backfire Antenna Gain and Cross-Polarization using Choke and Ring Cavity Loading

We present two designs that improve the gain and cross-polarization performance of the waveguide-fed short backfire (SBF) antenna by introducing a choke (SBF-2) and by loading the cavity with a metallic ring feature (SBF-2-Ring). A series of parametric simulation studies on antenna dimensions provide information on how to improve the antenna gain and cross-polarization performance, while simultaneously extending the impedance bandwidth. For SBF-2, the peak gain was 16.6 dBi, the minimum cross-polarization ratio was -23.8 dB, and the maximum impedance bandwidth was 27.3%, with gain bandwidth of 19.2%. For SBF-2-Ring, the peak gain was 15.8 dBi, the minimum cross-polarization ratio was -29.1 dB, and the maximum impedance bandwidth was 43.5%, with gain bandwidth of 31.8%. The concepts were verified by designing, fabricating, and testing two prototypes in the microwave C-band. Excellent agreement between simulation and measurement was achieved. Measured gain for SBF-2-Ring was >14 dBi for 4.7 GHz to 6.2 GHz and worst-case cross-polarization in the diagonal plane was <-22 dB for 5.3 GHz to 5.8 GHz. Cross-polarization in the principal planes has significantly greater bandwidth and the worst-case analysis is presented to give limitations on the performance.


I. INTRODUCTION
HE short backfire antenna (SBF) is an established antenna design that is used in medium gain applications where a compact structure is desired. The first designs were reported in publications by Ehrenspeck [1][2]. Since then, prototypes with varying feed mechanisms and design optimizations have been presented in the literature [3][4][5][6][7][8][9]. The directional aspects of the antenna make it a good candidate for point-to-point communications, remote sensing, and array feeds for reflector antennas. SBF designs have been reported in the low end of the microwave spectrum (e.g., S-band), mid-range as Wi-Fi antennas, X-band antennas, and even up to millimeter wave designs [10][11][12]. The early designs were fundamentally narrowband, and therefore, research has been conducted to extend the bandwidth of operation [6], [13]. The designs scale with frequency and can be adapted to a variety of functions [14].
Recently, antennas and radiating systems have been mounted on unmanned aerial vehicles (UAVs), with applications ranging from remote sensing to search and rescue operations. Such antennas need to be compact, low profile, low mass, and highly efficient. The SBF is a strong candidate since it can be designed and fabricated to meet these needs.
Many contemporary remote sensing systems operate in the microwave C-band. For example, Radarsat-2, the Radarsat Constellation Mission (RCM), and Sentinel 1A/B operate in the C-band. Operations range from marine environmental monitoring to improving agricultural practices, all for the benefit of the world's population. Consequently, we seek to advance the design of antennas in the C-band to provide new options and possibilities for future remote sensing systems.
The overarching objective of this manuscript is to develop and verify SBF designs that meet the operation specifications of a C-band remote sensing system. To do this, we address the following three objectives. First, the antenna design must meet several specific physical requirements, namely, low profile, low mass, and able to be constructed of stable materials that will handle space requirements. Second, the SBF must meet the desired antenna performance that is required for a radar or passive microwave sensor. These requirements are described in Section II. Third, the antenna must be relatively simple to construct and assemble. In this design, we are aiming for a prototype with a low number of parts.
In Section II, we present the design concepts, including a discussion of several candidate prototypes. In Section III, we provide measurement results from two fabricated devices that were used to verify the concepts. Section IV gives a conclusion and recommendations for next steps in the design flow.

II. DESIGN CONCEPTS
The conventional SBF design requires a feed element (e.g., dipole) or a waveguide feed. In our design, we used standard waveguide components that would be easily assembled into our antenna prototypes. All designs in this manuscript used WR159 (dimensions a = 40.386 mm, b = 20.193mm). We selected a T This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TAP.2022.3161463, IEEE Transactions on Antennas and Propagation waveguide feed partly based on simplicity in fabrication and partly based on the greater power handling capabilities. We encountered many design trade-offs in our development, such as configurations with reduced gain with the benefit of wide impedance bandwidth. These aspects are discussed in the presented prototypes of this section. For our design, we established several performance metrics that were based on design requirements from existing C-band remote sensing systems and from comparison with existing designs in the literature. These are summarized in Table I. In this manuscript, all reported wavelength scales are given with respect to 5.5 GHz in free space (i.e., λ = 0.05454 m), as this is our primary operation frequency. All simulations were performed using Ansoft HFSS ©.
Typically, authors report on the impedance bandwidth of the antenna. However, this is not the only metric of importance. The gain bandwidth (where gain G ≥ Gmax -3 dBi, where Gmax is the maximum (peak) gain within the gain bandwidth) and the cross-polarization bandwidth are defining metrics for the antenna operation. In many designs (including the ones presented here), the bandwidth is less than the impedance bandwidth, and is the true operating limitation of the antenna.
Furthermore, in this manuscript we emphasize the worst-case cross-polarization as a limiting performance metric. In many manuscripts, authors will often refer to the cross-polarization levels in the principal planes (φ = 0° and 90°), where the crosspolarization is theoretically zero. Measurement results in these planes are often indicative of the limitations of the measurement accuracy. We have presented cross-polarization results in the principal planes for reference and direct comparison with other works in the literature, but also present the worst-case crosspolarization (in this design, φ = 46°).

A. SBF-1 Design
We developed several prototype designs. The first design, SBF-1, which was based on the conventional waveguide-fed SBF, gave reasonable performance with respect to our design specifications. In our design process, we started with a smooth main reflector to establish a benchmark for a C-band SBF design [15]. The geometry of the SBF-1 design is shown in  Fig. 1. By reviewing our simulation results, we established that for all designs with a main reflector diameter (Dm) that was less than 1.8λ, the realized gain did not exceed 11 dBi. This meant that the benefit of using an SBF structure would be negated, as a similar gain could be achieved through the use of a simple open-ended waveguide. For our SBF-1 design, the highest realized gain was 14.88 dBi, and this was found for Dm = 2.2λ. Consequently, our remaining designs in this manuscript used Dm = 2.2λ. The rim height was set to 0.6λ, based on recommendations in the published literature [16]. The best configuration that we could derive had an impedance bandwidth of 1.45 GHz (26.4%). When we evaluated the crosspolarization, we found that while the levels in the principal planes (i.e., φ = 0° and φ = 90°) easily met the specification, the peak cross-polarization (which was in φ = 46°) was significantly worse, giving a cross-polarization ratio of only -16.5 dB.
The surface current distribution of SBF-1 at 5.5 GHz is shown in Fig. 2. Current maxima are observed near the wide edge of the waveguide aperture (region 1) and on the subreflector. Region (2) has a relatively low surface current and in region (3), current maxima are seen near the edges of the main reflector aligning with the y-axis. The aperture field distribution for SBF-1 at 5.5 GHz (on a fictitious circular surface with a diameter of 6λ, located 1λ above the sub-reflector, as shown in Fig. 3) exhibits a non-uniform characteristic, with large peak electric field magnitudes in the areas along the x-axis and immediately above the waveguide opening.
We investigated ways to modify the surface current distribution in order to simultaneously improve the impedance bandwidth and the gain. The approach considers a variety of cavity loading methods, each with trade-offs for fabrication simplicity and performance metrics. The second design, SBF-2, was an improvement of SBF-1, in which we increased the gain by modifying the main reflector of the antenna with a choke. In a third design, SBF-2-Ring, we extended our results from SBF-2 by loading the cavity with a ring, which enhanced the impedance bandwidth, improved the aperture field distribution uniformity, and improved the cross-polarization performance. Our novel designs retain a compact structure, while simultaneously improving gain, impedance bandwidth, and cross-polarization performance.

B. SBF-2 Design
Building on the past research on conventional SBF antennas, we sought methods to increase the gain of the antenna by modifying the main reflector geometry. Simulations of the SBF-1 antenna showed that the surface current distribution on the main reflector surface was primarily focused near the wide edges of the waveguide feed. We modified the surface to have a single corrugation, which is also known as a "choke." The idea of adding a choke to the main reflector beside the rim was first introduced in [17]; however, to the authors' knowledge, there have been no further reported studies on it. The geometry of the design is shown in Fig. 4. A choke, with width Wchoke, depth dchoke, and distance from the outer rim, Rchoke, has been introduced into the bottom of the main reflector.
We performed parametric studies on the choke dimension to find configurations that met the gain requirement at our desired frequency of operation 5.5 GHz. From these results, we analyzed the S11 simulations and selected design configurations that met both the bandwidth and frequency metrics (see Table I). We also considered the worst-case cross-polarization ratio in our analyses. Later, in the fabrication section of this manuscript, we show the construction of one of the designs that has the highest gain and widest impedance bandwidth for the SBF-2 configuration.
In our parametric studies, we varied the gap between the rim and the outer diameter of the choke (Rchoke), the depth of the choke (dchoke), and the choke width (Wchoke). All other  parameters were kept constant. Several candidate configurations met the requirements and we determined that we needed to constrain the values, as follows: As a demonstration of one of the parametric studies, Fig. 5 shows the realized gain and worst-case cross-polarization ratio, while Fig. 6 shows the S11 plot for fixed values of Wchoke and dchoke, while Rchoke is varied. The results in Fig. 5 demonstrate that Rchoke = 0.1λ gives the highest gain, with a value of 16.6 dBi at 5.5 GHz. The configuration with Rchoke = 0.1λ affords the widest impedance bandwidth of operation (800 MHz, from 4.9 GHz to 5.7 GHz, or 15.1%) that meets the gain specification. Using the 3 dB gain definition, the bandwidth of operation extends from 4.7 GHz to 5.7 GHz, or 19.2%. Simultaneously, the cross-polarization ratio is less than -14 dB for the same frequency range, with a minimum value of -23.8 dB at 5.6 GHz.
The S11 plots in Fig. 6 exhibit two resonances. One is at 5.5 GHz, the primary design frequency, and a second resonance can be seen at a lower frequency. Reducing Rchoke corresponds to increasing the choke diameter. This results in a downward frequency shift and a resultant increase in impedance bandwidth. Further reduction of Rchoke pushed the curve above -10 dB for 5 GHz to 5.2 GHz, which was not an acceptable design. The impedance bandwidth of operation for the configuration with Rchoke = 0.1λ was 1.3 GHz (from 4.4 GHz to 5.7 GHz, or 25.7%); however, as stated for the gain specification, the actual operation is constrained by the gain Fig. 6. SBF-2 -S11 variation as a function of spacing between choke and outer rim edge (Rchoke). Wchoke = 0.3λ and dchoke = 0.3λ.
performance. It should be noted that in our design study we used the antenna parameters to broaden its impedance bandwidth, rather than using a separate impedance matching section that could reduce antenna gain and efficiency. Fig. 7 shows the surface current distribution for SBF-2. By virtue of the current distribution, each of the antenna components (i.e., the main reflector, sub reflector and rim) contribute to the radiation. The main reflector surface current was primarily concentrated in the region that was in closest proximity to the waveguide (region 1). The choke contribution (region 2) to the radiation is the smallest amongst the three areas of the main reflector. The surface current magnitude of the section of main reflector that is beside the rim (region 3) is approximately 1/6 of region (1). The surface current was partially constrained to region (2), due to the presence of the choke. As a result, the concentration of the current yields a more focused pattern, and consequently, a higher realized gain than designs without the choke.  The aperture field distribution at 5.5 GHz for SBF-2 is given in Fig. 8. In comparison to SBF-1 (Fig. 3), we note that while the peak field magnitudes are immediately above the waveguide aperture, the field magnitudes are more equally distributed across the main reflector aperture. This results in a higher gain and lower cross-polarization than SBF-1.

C. SBF-2-Ring Design
In this section, we present a method for enhancing the crosspolarization performance and antenna impedance bandwidth through a modification of the SBF-2 structure. There are some applications in which antenna impedance bandwidth is more important than the gain, and a small sacrifice in gain performance can be traded for a much broader impedance bandwidth. The technique was a simple, elegant, and effective modification of the SBF-2 design. It involved adding a conductive ring within the rim. The ring diameter was slightly smaller than the main outer diameter, and as a result, a mode of operation at a higher frequency was added. Additionally, the presence of the ring changed the current distribution of SBF-2 on the rim, resulting in a better cross-polarization result (as will be seen in the experimental verification). SBF-2-Ring geometries are shown in Fig. 9.
In our parametric studies, we varied the ring height (Hring), the distance between the ring and the main reflector (D1-ring), and the distance between the ring and the rim (D2-ring). The ring thickness was set to a constant (2mm) in the simulations. During the parametric study, all other dimensions of SBF-2-Ring were held constant.
The variation in gain as a function of ring height at the design frequency of 5.5 GHz is shown in Fig. 10. An increase in ring  height results in a gradual decrease of gain. At the same time, the impedance bandwidth shows a marked increase between 0.400λ and 0.525λ. The peak impedance bandwidth was at 0.450λ; however, the S11 curve was equal to -10 dB for several frequencies, meaning there was no margin for error in the fabrication process. The maximum gain was found with ring height (Hring) at 0.275λ. Therefore, a design trade-off between impedance bandwidth and gain performance must be made, and will be application specific. Fig. 11 shows the gain variation as a function of the ring height for D1-ring = 0, D2-ring = 0.05λ. We present these particular ring heights based on the results in Fig. 10, where Hring = 0.275λ had the highest gain and the impedance bandwidth for ring heights between 0.400λ and 0.525λ were significantly larger than other heights. The gain results in Fig. 11 illustrate that the gain bandwidth of operation was 1.5 GHz (e.g., from 4.7 GHz to 6.2 GHz, or 27.5%, for Hring = 0.4λ) and the ring height did not significantly affect the realized gain magnitude (it was 16.2 dBi maximum for each of the cases). Using the 3 dB gain definition, the bandwidth of operation extends from 4.5 GHz to 6.2 GHz, or 31.8%. Simultaneously, the crosspolarization ratio was <-11 dB for the same frequency range, with a minimum value of -29.0 dB at 5.5 GHz. For comparison with SBF-2, we examined the same bandwidth (4.9 GHz to 5.7 GHz) and the cross-polarization ratio was <-15.2 dB. On average, the ratio was -18.7 dB for SBF-2 and -19.0 dB for SBF-2-Ring. Therefore, the SBF-2-Ring configuration had a better cross-polarization performance in all respects. Since the initial design goal of SBF-2-Ring was to increase the impedance bandwidth, the approach was successful. Fig. 12 shows S11 plots, in which we varied the ring height (Hring) and kept D1-ring = 0, D2-ring = 0.05λ. Note that in this figure, the frequency axis extends to 8 GHz, whereas the axis for SBF-2 only went up to 6 GHz. Similar to SBF-2, the antenna exhibits a resonance at around 4.7 GHz; however, the addition of the ring has made a significant change at frequencies greater than 5.5 GHz. A dual resonance behavior appears, with a notch between 6 to 6.5 GHz and another at 7.3 GHz. This behavior is more pronounced for ring heights ≥0.400λ, but it is still evident for the Hring = 0.275λ (maximum gain) case. For substantially shorter and taller rings, the impedance bandwidth was similar to SBF-2, and so, these results are not shown.
We determined that the maximum impedance bandwidth (4.5 GHz to 7 GHz, or 43.5%) and gain bandwidth (1.5 GHz, or 27.5%) required us to constrain the values as follows: 0.4λ ≤ Hring ≤ 0.475λ, D1-ring ≤ 0.1λ, and 0.05λ ≤ D2-ring ≤ 0.1λ. In other words, the ring needed to be shorter than the outer rim and could not be in contact with it. Contacting the bottom of the main reflector had no detrimental effect. The surface current distribution of SBF-2-Ring is shown in Fig.  13. In comparison with SBF-2, adding the ring beside the rim altered the current distribution on the rim. In this design, similar to SBF-2, the majority of radiation came from the main reflector. The addition of the ring resulted in a uniform distribution of the current in the area between the ring and the top of the rim, and a new, higher frequency mode of operation, was introduced. This extended the impedance bandwidth of operation significantly.  The aperture field distribution at 5.5 GHz for SBF-2-Ring is given in Fig. 14. In comparison to SBF-2 (Fig. 8), the field magnitudes do not have the drastically higher field magnitudes immediately above and beside the waveguide aperture. In fact, the field appears to exhibit a radial symmetry. The symmetry results a significant improvement in cross-polarization when compared with SBF-2.

III. ANTENNA PROTOTYPE FABRICATION AND TEST
In order to verify the antenna simulations, we built prototypes of the designs. Throughout our design process, we considered practical fabrication aspects. Components were constructed in our machine shop using traditional metalworking methods. We chose aluminum for its low mass and ease in fabrication processes. Our machinist did not optimize the mass of the antenna at this stage, as the primary goal was performance verification. Consequently, the SBF-2 and SBF-2-Ring designs could be significantly reduced by trimming additional material in a future build. The mass of SBF-2 was 462.5 g and SBF-2-Ring was 482.0 g. To position the sub- reflector at the desired height, we used a piece of foam material (εr ≈ 1.05) that was cut to fit the interior of the cavity. We adapted our simulation geometry to include the segment of foam for appropriate comparison. All simulation results presented in this section incorporate the foam material. The main reflector of the SBF-2 prototype is shown in Fig. 15 a), the SBF-2-Ring prototype is shown in Fig. 15 b), and the fully assembled antenna (SBF-2) is shown in Fig. 15 c). Dimensions for the fabricated prototypes are given in Table II.
Through further simulation studies we established that the diameter of the main reflector (Dm), the height of the subreflector (hs), and the sub-reflector tilt angle are the critical design parameters. For the diameter of the main reflector and the height of the sub-reflector, we recommend the tolerance of 1 mm and 0.5 mm for the fabrication, respectively. This should be relatively easy to achieve for a typical machine shop. Ideally, the sub-reflector should be perfectly parallel to the main reflector (i.e., a tilt angle of 0°). In actual fabrication, this can be challenging to attain. The error margin for the sub-reflector tilt angle should be limited to 5°.
We used an Agilent VNA (Model N5224B) to measure S11 for the prototype antennas. To connect the antennas to VNA, we used a low-cost commercial waveguide-to-coaxial adaptor, which fed its waveguide with a probe. It was therefore necessary to evaluate the performance of the adaptor first. This was done using two identical adaptors, connected back-to-back using a waveguide and terminated in a matched load. The set up and the measurement results are shown in the Appendix. The results show that the adaptor S11 is satisfactory, as a quality test hardware, only for frequencies above 5.2 GHz. Below that, it represents a complex impedance, with multiple oscillations, which interact with the antenna input impedance, affecting its measured S11, as shown in Fig 16(a). To improve S11 of the adaptor-antenna combination, so that the antenna gain and radiation patterns can be measured accurately, we added two strips of copper tape with conductive adhesive to the waveguide aperture, to create an inductive iris, and lower the simulated antenna S11 to below -14dB. The resulting S11 of the combined adaptor and antenna with the matching iris, is compared with simulation in Fig. 16(b). It clearly shows the effects of the waveguide adaptor, but the combined results now remain below -10dB, necessary for conducting the measurements. The details of the results are shown in the Appendix. Experimentally, we found the width of ~4.5 mm was effective at improving the match. We added the strips in HFSS simulations to find the effect of the width and determined that values between 4 mm and 5 mm were all effective. The impedance bandwidth is 1.35 GHz (4.27 to 5.62 GHz, or 27.3%); however, the operational bandwidth of the antenna is determined by the gain performance as well, so radiation pattern results were evaluated. We examined the effect of the iris on the far field patterns and gain for both SBF-2 and SBF-2-Ring. The results showed negligible difference.
The radiation patterns were measured in the University of Manitoba's Compact Antenna Test Range (CATR). Radiation patterns for SBF-2 at 5.5 GHz are shown in Fig. 17. The measured peak gain was 15.7 dBi and the simulated gain was 16.3 dBi. If we account for the 0.2 dB insertion loss from the waveguide to coaxial cable connector, the simulated and measured co-polarized radiation patterns are in excellent agreement (within 0.4 dB). The remaining difference can be attributed to the imperfect bonding between the waveguide flange and the bottom of the main reflector. Cross-polarization was calculated using Ludwig's 3 rd definition [18]. The peak value was at φ = 46° and θ = 38°, giving a worst-case of -20.9 dB cross-polarization ratio.
We constructed the ring for SBF-2-Ring using a piece of flexible dielectric substrate (εr ≈ 2.5) with copper cladding. A section of substrate material was trimmed, shaped into a ring, and installed in the SBF-2 main reflector. The ring height was 0.4λ (21.8 mm), the gap, D2-ring, was 2mm and the copper ring thickness was 0.035 mm (formed by the copper cladding on the substrate). This ring dimension was chosen as a compromise between gain performance and impedance bandwidth.
Measured S11 results for SBF-2-Ring are shown in Fig. 18. Measurements with the inductive iris showed good performance, so we left it in place. Later, after radiation tests, we removed the iris to see the effect and the results shown in Fig. 18 indicate that the S11 profile degrades since the values increase substantially. We investigated the width of the strips in further HFSS simulations to find the effect of the width and determined that values between 4 mm and 5 mm were all effective at improving the impedance match. Notches in the S11 results occur at 4.84 and 5.56 GHz. The impedance bandwidth is 2.5 GHz (4.5 GHz to 7.0 GHz, or 43.5%). We note that for frequencies below 4.7 GHz, the S11 measurement is influenced by performance of the waveguide-to-coaxial adaptor.   Cross-polarization was calculated using Ludwig's 3 rd definition and the peak value was at φ = 46° and θ = -45°, giving a worstcase of -29.1 dB cross-polarization ratio.
Radiation patterns were measured in a range of frequencies and we present the results from 4.9 GHz to 5.8 GHz in Fig. 20. SBF-2 gain and worst-case cross-polarization measurement results were in good agreement with simulation, including a drop-off for frequencies above 5.6 GHz (see Fig. 5, Rchoke = 0.1λ). SBF-2-Ring gain performance was similar in magnitude to SBF-2, but there was a near-constant gain value across the entire measurement bandwidth. Worst-case crosspolarization was similar to simulation for frequencies below 5.2 GHz. The cross-polarization measurements were excellent (<-22 dB) between 5.3 and 5.8 GHz. The SBF-2-Ring design has met all of the proposed specifications.
For comparison with other published designs, we have provided a summary in Table III. Column 1 gives a description of the antenna construction/excitation. The impedance bandwidth is given as a percentage, following the standard definition. The maximum gain, Gmax in dBi is given next. The antenna volume, which is expressed in electrical units, relative to the reported design frequency, is given. Finally, the reference for each design is reported in the last column. Note that in this table, we presented the impedance bandwidth using regularly reported metric of |S11|<-10dB. We are unable to provide a good comparison for cross-polarization, as reported values and levels are somewhat inconsistent in the literature From this summary table, we note that small volume antennas can achieve high impedance bandwidth, but at the cost of low peak gain (e.g., [6] and [13]). Some of the designs with the highest gain had the lowest reported bandwidth (e.g., [16], [19][20]). In comparison, our designs show a good compromise between achieved bandwidth and gain, while still maintaining a moderate volume (e.g., [12] has higher gain and bandwidth, but it is more than twice the volume of our antenna).    [20] IV. CONCLUSION We presented innovative designs, which improve the gain and cross-polarization performance of the waveguide-fed short backfire (SBF) antenna. Our emphasis was on improving design aspects for compact, low profile, and low mass antennas that are simple to construct and can be used for operation in the microwave C-band. We have succeeded in meeting design specifications for such an antenna.
The first design, called the SBF-1, had a flat main reflector. We presented these results as a benchmark for the improvements proposed and presented here. The second design, called the SBF-2 antenna, had a choke feature in the main reflector that concentrated the current distribution and resulted in an increase in realized gain when compared to the typical flat main reflector. The bandwidth of operation that met the gain specification of 14 dBi was 800 MHz (from 4.9 GHz to 5.7 GHz, or 15.1%), with a peak value of 16.6 dBi. Using the 3 dB gain definition, the bandwidth of operation extends from 4.7 GHz to 5.7 GHz, or 19.2%. Cross-polarization ratios were moderate throughout the design bandwidth (<-14 dB) and reached a simulated minimum of -23.8 dB at 5.6 GHz. Principal plane ratios were significantly better, as expected theoretically, and shown in the measurement results. We examined the input impedance and S11 results showed that the impedance bandwidth was significantly larger than the gain bandwidth, and was therefore not a limiting factor on the performance. For SBF-2, the peak gain was 16.6 dBi, the minimum crosspolarization ratio was -23.8 dB, and the maximum impedance bandwidth was 27.3%.
The second innovation, called the SBF-2-Ring, added a ring feature to the SBF-2 design. The ring diameter was slightly smaller than the main rim and this extended operation to a higher frequency. The bandwidth of operation that met the gain specification of 14 dBi was 1.5 GHz (from 4.7 GHz to 6.2 GHz, or 27.5%). Using the 3 dB gain definition, the bandwidth of operation extends from 4.5 GHz to 6.2 GHz, or 31.8%. Crosspolarization ratios were moderate throughout the design bandwidth and were better than SBF-2 across the same band (<-15.2 dB). They reached a simulated minimum of -29.0 dB and measured minimum of -29.1 dB at 5.5 GHz. In general, the SBF-2-Ring configuration has a better cross-polarization performance than SBF-2. We examined the input impedance and S11 results showed that the impedance bandwidth was significantly larger than the gain bandwidth (approximately 300% larger), and the addition of the ring resulted in a uniform distribution of current in the region between the ring and the waveguide. For SBF-2-Ring, the peak gain was 15.8 dBi, the minimum cross-polarization ratio was -29.1 dB, and the maximum impedance bandwidth was 43.5%.
Prototypes of each design were fabricated and tested. Excellent agreement between simulation and measurement results were achieved for both SBF-2 and SBF-2-Ring. The optimized SBF-2-Ring met all of the performance specifications. It operated with moderate gain (>14 dBi) and had a good cross-polarization ratio (<-22 dB), and over 500 MHz bandwidth. With appropriate selection of the SBF geometry and cavity loading, as described in this manuscript, many usable configurations can be obtained.
From these results, we plan to use the antenna in remote sensing systems. Future work will include adding dualpolarization capability, reducing mass, and enhancing gain.

APPENDIX: ANALYSIS OF WAVEGUIDE ADAPTORS
In our design, we proposed to use WR159 as the main waveguide feed. Our design specifications have a primary operating frequency at 5.5 GHz and a bandwidth of 500 MHz. We chose this standard waveguide size because its recommended operating frequency is 4.90 GHz to 7.05 GHz and would ideally meet these specifications. In our design, we were able to show by simulation that the antenna would operate at lower frequencies (i.e., down to 4.4 GHz). In HFSS, the simulation assumes an ideal TE10 mode operation for the waveguide when it is fed with a port excitation.
We performed additional measurements on the waveguide and adaptors, as shown in Fig. 21. The results of the measurements are given in Fig. 22, in comparison with the antenna simulations. Referring to Fig. 16, when we looked at the S11 results for the SBF-2 antenna measurement, we noticed that there were notches at 4.125 GHz, 4.43 GHz, 4.82 GHz. At ~4.6 GHz, there was a local maximum in the measurement, which was not present in the simulation results. To find the discrepancy, we measured S11 for the waveguide and adaptors (set-up as depicted in Fig. 21) and the results are shown as the "WG only" curve in Fig. 22. The notches at 4.125GHz, 4.43 GHz, and 4.82 GHz are visible in the WG measurement. Furthermore, we can see that at 4.6 GHz, the waveguide has a local maximum in the S11 at 4.6 GHz. The antenna measurement follows this. So, it appears that the WG (and namely the WG adaptor), is the cause of this discrepancy. It is clear that the antenna measurement is influenced by the waveguide adaptor for frequencies less than 4.7 GHz.