High gain is not usually associated with a compact antenna. For satellite-communications applications, however, an antenna design must be small and light, yet provide pattern shaping, wide bandwidth, and polarization purity. In developing an antenna for multiband Global Positioning System (GPS) and wireless-local-area-network (WLAN) use, it was possible to create a small, lightweight design with polarization diversity and high gain.

For GPS applications, for example, an antenna may be required to handle both the low band at 1.226 GHz and the high band at 1.575 GHz. For IEEE 802.11a/b/g WLANs, an antenna must operate in both the 2.4- and 5-GHz bands with bandwidth to support data rates at 11 and 54 Mb/s. Additional applications include the planned Air Force satellite systems at 1.8 and 2.25 GHz. For a single antenna to cover multiple wireless applications, coverage at 1.8 through 2.1 GHz should also be considered for third-generation (3G) cellular systems.

Polarization is an important characteristic in a good antenna design. For space applications, it is customary to use circular polarization (CP), such as the right-handed circular polarization (RHCP) or left-handed circular polarization (LHCP), either for transmit and receive or reuse of the same spectrum band for added capacity. Although most WLAN systems require linear polarization, the use of CP will eventually becomes advantageous for mobile systems.

Certain theoretical limits dictate how small an antenna can be made while still providing required gain and bandwidth. For the space-based (satellite) application, the antenna design was required to fit a certain form factor while operating with CP at 1.8 GHz for the uplink (the satellite's receive frequency) and 2.25 GHz for the downlink (the satellite's transmit frequency). Pattern-shaping capability was also a key requirement, to allow the satellite to maintain communication while in different positions and angles. The antenna must be rugged enough to withstand high shock and vibration, environmental/thermal extremes (temperature swings of typically −40 to +70°C), and power cycling. Several options were considered for the design, including a helical antenna, quadrifilar helical antennas (QFHAs), and various microstrip patch configurations. Initial analysis and electromagnetic (EM) software simulations indicated the difficulty of achieving the required performance levels in the small physical size.

After considering several unconventional approaches, a ring radiator technology was selected as the potential solution. This approach uses the resonant structures to effectively create long paths (and high gain) for the radiation currents while reducing antenna size by 25 to 35 percent compared to other approaches. The technique made it possible to meet the form-factor requirements with higher gain than possible with an even larger microstrip patch antenna or a cavity-backed helical antenna.

Design and analysis of ring antennas requires very intuitive engineering (and educated guessing) compared to the better-understood design and analysis approaches used for microstrip patch antennas. Fortunately, by performing a detailed initial design and analysis procedure, and carefully reviewing the EM simulation results, it was possible to reduce the design risk for the ring antenna in spite of its complexity.

In a simple rectangular patch antenna, the source of radiation can be thought of as two slots at the two ends of the patch, approximately one-half wavelength apart. If each of these slots is about one-half wavelength in length, 2.1 dBi gain should result. Any two such antennas working as a two element array should theoretically provide an additional 3 dB gain. Therefore, it should be possible to achieve about 5.1-dBi gain from a simple patch antenna. With some refinement, it may be possible to get even better gain or pattern shaping depending on the ground plane type or resonance mode.

For a ring antenna, it may be possible to design a structure with multiple resonances that could be spaced and coupled for multiple-frequency or broadband-frequency use.

High gain and pattern shaping can be achieved by properly phasing the various modes to operate in a predetermined manner so as to combine constructively or destructively in the far field and in appropriate directions. In most cases, it was possible to achieve as much as 9 dBic gain (theoretically) from these structures with as much as 17 percent bandwidth. In theory, it was possible to achieve bandwidths of 15, 20, and 30 percent for VSWRs of 1.50:1, 2.0:1, and 3.0:1, respectively. Unfortunately, it was not possible to develop a systematic design approach that would meet all required physical and electrical parameters at all frequencies. With some effort, it may be possible to develop an approach that would meet the requirements for certain modes of operation, however.

Figure 1 shows a swept-frequency plot as it was predicted by EM simulation for an optimized antenna design. The plot indicates multiple resonances, although not all would be used for the satellite antenna The lowest resonance at 1.8 GHz provides better than 13-dB return loss, while at the high end, at 2.25 GHz, the return loss can be better than 17 dB. For a certain combination of parameters, it is possible to achieve a 10-dB return-loss bandwidth of about 15 percent. This would be an excellent broadband antenna for many applications. The return loss for the resonance at 2.1 GHz is even better, at nearly 20 dB. The antenna design's multiple resonances make it suitable for use as a single wideband antenna or for three discrete-frequency applications.

The predicted radiation pattern for the RHCP antenna at the lower resonance of 1.8 GHz indicates nearly 5.5 dBic gain (Fig. 2, top left). The axial ratio is nearly 13 dB at zenith (Fig. 2, bottom left). At the upper resonance of 2.25 GHz, the predicted radiation pattern indicates nearly 8-dBic gain (Fig. 2, top right). At this frequency, the axial ratio appears to be nearly 12 dB (Fig. 2, bottom right).

Page Title

Figure 3 shows simulations of the surface current densities on the rings. As expected, the highest current densities (in red)—indicating the radiation mechanism for this configuration—can be found along the edges. The top illustration simulates the upper ring while the bottom illustration shows a simulation of the lower ring for the upper resonance at 2.25 GHz. The radiation mechanism changes slightly at the lower resonance, where the gain is lower, although this was optimized according to the requirements of the satellite link budget.

From a side view (Fig. 4), the radiation mechanism can be seen with the coaxial input connector. The larger box surrounding the antenna is a convention of the EM simulation program in which the device to be modeled is enclosed within finite boundaries (the box). The enclosure's boundaries were chosen to have minimal effects on the antenna's performance.

Based on these analyses and simulations, several antennas were manufactured, with two such designs shown in Fig. 5 (antenna A on the left and antenna B on the right). The antennas met all the electrical requirements and passed all the space qualification requirements the first time around due largely to a well-controlled design process, extensive use of simulation and verification, and good mechanical design and fabrication practices.

Figure 6 shows the measured return loss of antennas A and B, with response shapes remarkably similar to the desired simulated performance of Fig. 1. Differences between the simulations and actual hardware can be traced to some tuning performed in the laboratory, albeit minor. The measured radiation patterns and gains for the two antennas are shown in Fig. 7, with measurements at 1.8 and 2.25 GHz for antenna A in Fig. 7 left top and left bottom, respectively, and at 1.8 and 2.25 GHz for antenna B in Fig. 7 right top and right bottom, respectively. Each pattern includes 0, 45, 90, and 135 deg. azimuth pattern cuts. Note the similarity of these measured patterns to the simulations of Fig. 2. The measured backlobe performance was similar to the simulations but not measured for all the antennas.

In addition to the "conventional" antenna requirements, prelaunch ground testing of the satellite payload required a way to test the communications link in the satellite's high bay and to provide communication with the satellite payload without radiating into the high bay. As a result, the antenna design was required to work effectively in close proximity to the satellite's various other subsystems, including solar arrays. To provide a means for the antenna not to radiate into the high bay and a way to communicate with the antenna required some thought. Waveguide approaches were considered, but the form factors were impractical. EM simulations of various antenna boxes and "hats" were performed to identify cutoff behavior and hotspots, and ultimately, a design that combined aspects of a filter and antenna, called a "filtenna," was developed.

Part of the challenge in developing this new design involved having an antenna resonate in a cavity or waveguide. After some fruitless experimentation, filter and antenna theory were combined and the coupling resonator models were carefully optimized to develop the filtenna. The design includes a hat-like cover with minimal effect on return loss (Fig. 8) which is added for testing purposes (but not needed in the satellite application). The EM simulations showed that the position of the resonances, particularly lower resonances, was very sensitive and shifted with the position of the hat/cover. Simulated results of return loss and insertion loss are shown in Fig. 9, while Fig. 10 shows measured two-port insertion loss (top) and two-port return loss (bottom). The simulated and measured data agree closely, with the exception that the measurements including some tuning in the laboratory to improve the lower bandedge return loss. Figure 11 shows a side view of the simulated EM fields of the filtenna, indicating the mechanism of coupling from one port to the other.

This design was applied to two other applications, one for a dual-band Wi-Fi antenna suitable for IEEE 802.11a/b/g WLAN "hot-spot" applications at 2.4 and 5 GHz and the other for dual-band GPS use. Figure 12 shows simulations of the Wi-Fi antenna, indicating high gain for the linearly polarized design, although the design requires additional bandwidth at the lower end to meet the requirements of IEEE 802.11g at 2.4 GHz. Simulated performance of the antenna for dual-band GPS (not shown) matches test data.

The design involved the development of a degenerate mode structure, which would support two very near modes with 90-deg. phase shift. The design, in fact, was optimized by means of this. Still, it would have been helpful to have been able to map the antenna's fields after prototyping to check magnitudes and phases. By optically mapping the field vectors and comparing them with simulated results, it would then be possible to tune the phases of the modes. Such a tool would further reduce the guesswork from the engineering of antennas; it is currently available but as of yet too expensive for practical design.

ACKNOWLEDGMENTS
The author would like to thank Mr. Thierry Guichon and his team at AeroAstro (Ashburn, VA) for their support, encouragement, and valuable insight. The author would also would like to thank Tom Flynn and his staff at Ansoft (San Jose, CA) for technical assistance with the High-Frequency Structure Simulator (HFSS) software, Dr. Margomenos at EMAG Technologies (Ann Arbor, MI) for testing, Rehan Jaffry for engineering and testing support, John Stolp of Rogers Corp. (San Jose, CA) for materials support, Bob Gray of Triangle Labs (Carson City, NV) for manufacturing support, and Rick Gagelin for mechanical engineering support.