IT IS POSSIBLE TO DESIGN MULTISEGMENT DIELECTRIC RESONATOR antennas (DRAs) with significantly higher impedance bandwidths than homogeneous DRAs with the same overall physical dimensions and resonant frequency. For practical purpose, a number of general design rules have been proposed based on numerous simulations involving stacked and embedded DRAs.

Premises of the Project
Dielectric resonator antennas (DRAs) have been the subject of many investigations since they were introduced in the mid-1980s.¹ They are quite useful for high frequency applications where ohmic losses become a serious problem for conventional metallic antennas. In addition, they offer higher bandwidths than the microstrip patch antennas commonly used at the same frequencies.

In 1997, Kishk and his colleagues showed that "stacked" DRAs could be designed to have a larger bandwidth than conventional homogeneous DRAs.² Simple illustrations of a few different DRAs can be found in Fig. 1.

Similarly, it has been shown that DRA bandwidth can be significantly improved by loading a high permittivity, low profile dielectric disk on top of the original structure.³ A similar design, then called a "multisegment" DRA, involves using low profile, high permittivity dielectric inserts below a microstrip-fed DRA to improve coupling between the feed line and the antenna.^4-6 A number of other papers have been published showing similarly advantageous DRA designs.^7-8

Even though stacked DRAs have been proposed, there have been, however, few detailed studies concerning the effects of changing various physical characteristics, including but not limited to the relative permittivities and heights of the stacked sections. While it is known that stacked DRAs have potential advantages over homogeneous DRAs, it is difficult to optimize a practical design because of the lack of illustrative experiments relating to these parameter changes.

Goal of the Project
The primary objective of this project is to investigate the effects of various parameter changes on stacked and embedded DRAs. An "embedded" DRA is defined as a variation of a stacked DRA where at least one section is partially or totally enclosed by another section. Each physical characteristic will be studied individually to show its effects on the global antenna parameters. In general, comparisons will be made between a stacked or embedded DRA and a homogeneous DRA with equal physical dimensions and an equal resonant frequency. This investigation will illustrate the differences between stacked or embedded DRAs and their equivalent homogeneous DRAs and, also, the differences between various designs of stacked and embedded DRAs.

The secondary objective of this project is to provide practical insight into the design of stacked and embedded DRAs, by systematically comparing a number of different designs. A few "most useful" designs will be singled out, which will help define a set of guidelines for designing practical antennas of this type. Because physical size will remain constant for these comparisons, and modifying the relative permittivity of any DRA can easily change its resonant frequency, these "most useful" designs will likely be chosen for either high bandwidth compared to an equivalent homogeneous DRA or desirable impedance characteristics.

Equipment or Special Technology
All of the simulations were done using Ansoft HFSS (High Frequency Structure Simulator).

Results
A few illustrations will help describe the large number of simulations completed in this study. Note that for all cases the physical dimensions are the same, and the resonant frequencies were held as close as reasonably possible. Also note that "bandwidth" refers to the impedance bandwidth (defined as S11 < -10 dB).

The first type of DRA that was tested was the "stacked" DRA, shown in Fig. 2. In general, a stacked DRA had a bandwidth advantage over a homogeneous DRA when d_lower > d_upper, and when e_r.upper > e_r.lower.

The next type of DRA studied was the "simple embedded" DRA, shown in Fig. 3. In general, the bandwidth advantage of this structure over a homogeneous DRA was maximized when e_r.shell > e_r.plug, and when the inner radius b was fairly large compared to the total radius a. Unfortunately, it is very difficult to derive hard conclusions mainly because of the geometry of this DRA. Because the dielectric-air boundary at the primary radiating surface (the top) changes with every modification of any of the physical parameters, it proved nearly impossible to isolate the causes of the various changes to the antenna parameters.

To try to eliminate the changing radiating surface, the next type of antenna under study was the "core plugged embedded" DRA, shown in Fig. 4. Interestingly, the greatest bandwidth advantage compared to a homogeneous DRA came when e_r.plug was as low as possible, meaning that the plug was filled with nothing but air. This is a considerable advantage from a design perspective, given the sensitivity of the DRA to small air gaps caused by imperfect manufacturing. Further, we concluded that the ideal location for the interface b was approximately equal to the probe offset. If the interface were placed too far out, the input impedance of the antenna would tend to be very capacitive, and the walls would probably be too physically weak for a practical antenna.

The final type of DRA we studied was a hybrid embedded/stacked DRA, shown in Fig. 5. This antenna geometry combined worthwhile features from both stacked DRA and core plugged embedded DRA geometries. Because there were too many physical parameters to study rigorously, the design rules that were gleaned from the simpler previous examples were applied to optimize a few illustrative cases. For now, it suffices to say that this geometry provides a small bandwidth advantage when compared to even the stacked DRA geometry. One disadvantage is that some of that bandwidth may not be usable, since the antenna patterns vary somewhat with frequency. Further, this geometry would be quite difficult and, perhaps, prohibitively expensive to fabricate.

At this point, three specific examples will be shown. The first is a homogeneous DRA, the second is a stacked DRA, and the third is a stacked/embedded DRA. The general results are shown in Table 1.

Note that all of these examples have exactly the same physical dimensions and almost exactly the same resonant frequencies. A graph of S₁₁ vs. frequency for these three cases can be found in Fig. 6. Smith Charts for the homogeneous, stacked and embedded/stacked cases can be found in Figs. 7, 8, and 9, respectively. Note that all of these figures were constructed using matched data. As a result, the Smith Charts are not normalized to the same impedance. It is worth noting that HFSS does not model DRA antenna parameters exactly, but the trends are usually correct. It follows that the 68 percent bandwidth that HFSS returned is probably not exact, but the bandwidth of the embedded/stacked DRA is almost surely significantly higher than that of a homogeneous DRA.

Discussion
Three general design rules can apply to any DRA. In this section, each will be addressed individually and applied to the design of multisegment DRAs.

Rule 1: The first rule is that as the relative permittivity of the material gets higher, both the resonant frequency and the bandwidth go down. The opposite is true as the relative permittivity gets lower. For multisegment DRAs, the designer tries to include lower permittivity sections to improve the bandwidth and higher permittivity sections to lower the resonant frequency. Which sections need to be low and which sections need to be high was largely determined by trial and error, but the distinction did turn out to be possible.

Another way to look at this rule as it applies to multisegment DRAs is the following: a smaller, high permittivity section will guarantee that at least some part of the DRA resonates at a low resonant frequency. The presence of another small section with a lower permittivity creates a second resonance; if the two are superimposed, the net result is a relatively large bandwidth DRA. This phenomenon can be seen clearly in Figs. 6 through 9, where a number of resonances prove equal to the number of segments in the DRA. The one segment homogeneous DRA has one resonance, the two-segment stacked DRA has two resonances, and the three-segment embedded/stacked DRA has three resonances within the impedance bandwidth.

Rule 2: The second general rule is that most of the power radiates through the top surface of the DRA. As a result, the design of multisegment DRAs is more complicated if the top surface is made up of more than one DRA segment. This was a problem in analyzing the simple embedded DRA, where the plug was the same height as the shell segment.

Rule 3: The third general rule is that DRAs radiate somewhat like a cavity because a dielectric-air interface behaves like a perfectly conducting magnetic wall when the relative permittivity of the dielectric is fairly high. The higher it is, the better the approximation becomes. This is very important to remember when designing multisegment DRAs, since the low dielectric constant segments often have a direct interface with the air surrounding the antenna. If the dielectric constant is too low, the antenna pattern can change significantly and the DRA will not function as most DRAs do.

As a side note, this is one of the potential advantages of the embedded/stacked DRA over the stacked DRA. If we assume that the upper segment is the same for both antennas, then the air plug in the embedded/stacked DRA allows the shell of the lower section to maintain a higher dielectric constant while keeping the same resonant frequency as the stacked DRA. Thus, it can be assumed that less power is lost through the walls of the lower DRA section.

Conclusions
Clearly, it is possible to design a multisegment DRA with a significantly higher bandwidth than a homogeneous DRA with the same physical dimensions and resonant frequency. A number of general design rules were proposed to provide some practical insight to designers of multisegment DRAs. As long as the designer is aware of the limitations, such as potential changes in the antenna pattern and sensitivity to manufacturing tolerances, the only serious drawback is the increased cost of fabrication due to the complexity of the various multisegment geometries.

References
    ¹S. A. Long, M. W. McAllister, and L. C. Shen. "The Resonant Cylindrical Dielectric Cavity Antenna," IEEE Trans. Antennas Propagat. AP-31.3 (1983): 406-12.
    ²A. A. Kishk, B. Ahn, and D. Kajfez. "Broadband Stacked Dielectric Resonator Antennas," IEE Electronics Lett. 25.18 (1989): 1232-33.
    ³K. W. Leung, K. M. Luk, K. Y. Chow, and E. K. N. Yung. "Bandwidth Enhancement of Dielectric Resonator Antenna by Loading a Low-Profile Dielectric Disk of Very High Permittivity," IEE Electronics Lett. 33.9 (1997): 725-26.
    ⁴A. Petosa, R. Larose, A. Ittipiboon, and M. Cuhaci. "Low Profile Phased Array of Dielectric Resonator Antennas," Proc., International Symposium on Phased Array Systems and Technology, Ottawa, Canada, Oct. 1996. 182-185.
    ⁵B. Henry, A. Petosa, Y. M. M. Antar, G. A. Morin. "Mutual Coupling between Rectangular Multisegment Dielectric Resonator Antennas," Microwave and Optical Technology Lett. 21.1 (1999): 46-48
    ⁶A. Petosa, R. Larose, A. Ittipiboon, and M. Cuhaci. "Microstrip-Fed Array of Multisegment Dielectric Resonator Antennas," IEE Proc., Microw. Antennas Propag. 144.6 (1997): 472-76.
    ⁷S. M. Shum and K. M. Luk. "Stacked Annular Ring Dielectric Resonator Antenna Excited by Axi-Symmetric Coaxial Probe," IEEE Trans. Antennas Propagat. AP-43.3 (1995): 889-92.
    ⁸S. M. Shum and K. M. Luk. "Numerical Study of a Cylindrical Dielectric-Resonator Antenna Coated with a Dielectric Layer," IEE Proc., Microw. Antennas Propag. 142.2 (1995): 189-91.

PDF (367KB)
Table of Contents

Institute for Space Systems Operations - Y2003 Annual Report
Copyright © 2004