Circularly Polarized Reflectarray Antenna for Satellite Broadcasting Reception Shigeru MAKINO, Yukihiro YOSHIDA, Kosuke OKADA, Shusuke SASAKI Shin-ichi BETSUDAN, Kenji ITOH, Keisuke NOGUCHI, Tetsuo HIROTA Kanazawa Institute of Technology 7-1 Ohgigaoka, Nonoichi, Ishikawa, 921-8501, Japan Email: makino@neptune.kanazawa-it.ac.jp Toru TAKAHASHI Mitsubishi Electric Corp. 5-1-1 Ofuna Kamakura Kanagawa 247-8520, Japan Abstract In design of a reflectarray antenna, we considered the incident angle and the polarization on the refectarray surface. In the result, the reflection phases can be controlled only by varying the radius of ring elements, and it was found that the thickness of a dielectric substrate had an optimal value that minimizes the phase errors. To verify the design method, we manufactured and measured a circularly polarized reflectarray antenna for satellite broadcasting reception. I. INTRODUCTION Reflectarray antennas[1] are very attractive for the application to the large deployable reflector antennas because of their highly reliable deployment mechanism. And also they will achieve low cost small reflector antennas by using printing technique with the conductive ink. A lot of researches on reflectarray antennas are performed, however in designing the size of each element, the incident angle and the polarization depending on its position is not necessarily considered. In this paper, a design method of the reflectarray antenna considering the incident angle and the polarization on the refectarray surface will be shown, where a simple ring type is used as the element. Further it will be shown that the thickness of the dielectric substrate between the element plane and the ground plane has the optimum value on the view point of the phase errors of the elements. To verify the validity of the design method, a circularly polarized reflectarray antenna model for satellite broadcasting reception was designed and manufactured. II. DESIGN OF REFLECTARRAY A. Design parameters The important design parameters of the refectarray shown in Fig.1 are as follows. Fig. 1. Reflectarray antenna. The reflection phase is controlled by the radius r of the element ring, so r depends on the position of the element. Other parameters independent of elements are used commonly, so should be determined to minimize the phase errors. B. Evaluation function for design parameters d : period of elements l : thickness of dielectric substrate ɛ r : relative permittivity of dielectric substrate r : radius of element ring w : width of element ring Fig. 2. Reflection phase Φ vs. ring radius r.
Fig. 3. Δφ 1,Δφ 2 vs. thickness l. φ = tan 1 1 B (2) ɛ r cot βl where B is a normalized susceptance of the FSS and β = 2π ɛ r /λ is a propagation constance of the dielectric substrate. At first, think about ΔΦ. A ring type FSS works as a band rejection filter and B = ± at the resonance. The ring radius r has a limitation less than 6.1mm not to overlap each other, so B has an unrealizable value range ΔB from 0.9 to 0 as shown in Fig.4. The unrealizable reflection phase region ΔΦ can be approximated as follows. ΔΦ dφ ΔB (3) db where dφ db = 2 1+(B ɛ r cot βl) 2 (4) using Eq.(1) and (2). From Eq.(3) and (4), the condition for l which maximize ΔΦ is as follows. ɛr cot βl = B (5) Fig. 4. Susceptance B vs. ring radius r. Fig.2 shows an example of the calculated reflection phase Φ changing r from 0.4mm to 6.1mm at 12.0GHz, where d =13.0mm, l =3.0mm, ɛ r =2.65, w =1.0mm and the incident angle θ=0deg. Using the figure, two types of phase errors Δφ 1 and Δφ 2 can be defined. At first, as shown in the figure, an unrealizable reflection phase region ΔΦ exists between 60 deg and 120 deg. In this case, the maximum phase error Δφ 1 caused by this phenomenon is ΔΦ/2. Second phase error Δφ 2 is the fabrication error caused by the etching accuracy Δr. The phase error for r = r 0 is the half of the difference between the reflection phases of r = r 0 Δr/2 and r = r 0 +Δr/2. So, the phase error Δφ 2 is the half of the maximum phase difference δφ when r is changed. Fig.3 shows an example of Δφ 1 and Δφ 2 (Δr =0.1mm is assumed) at 12.0GHz changing the thickness l from 1.0mm to 20.0mm of dielectric substrate (glass fluorine substrate, ɛ r =2.65), where d =13.0mm, w =1.0mm and the incident angle θ=0deg. As shown in the figure, Δφ 1 and Δφ 2 are conflicting relation, and the thickness l has an optimum value (cross points of the Δφ 1 curve and the Δφ 2 curve) which minimizes the maximum value of Δφ 1 and Δφ 2. The relation can be explained as follows. The reflection phase Φ of the frequency selective surface (FSS) with a metal plate can be expressed as follows using an equivalent circuit[2]. Φ = 2φ π (1) In this case, the unrealizable value B is around 0.45, soδφ becomes maximum at l =4.5mm and 12.2mm using Eq.(5). On the other hand, at cot βl = ±, in other words l/λ = n/2 ɛ r, dφ/db and also ΔΦ become 0, where l =7.7mm and 15.4mm. Next, think about δφ. δφ can be approximated as follows. δφ dφ dr = dφ db db dr (6) dφ/db has the maximum value under the condition that l satisfies Eq.(5). And db/dr has the maximum value ± at the resonance and B becomes ± at this time as shown in Fig.4. So, δφ has the maximum value at B = ±,in other words l/λ = n/2 ɛ r using Eq.(5), where l =7.7mm and 15.4mm. On the other hand, db/dr 0 at B 0, soδφ has the minimum value at B = 0, in other words l/λ = (2n 1)/4 ɛ r using Eq.(5), where l =3.8mm and 11.5mm. These results coincide very well with the calculated results of Δφ 1 and Δφ 2 shown in Fig.3. C. Determination of design parameters Fig.5 shows the Δφ 1 and Δφ 2 changing the thickness l of dielectric substrate considering incident angle θ from 0 to 60 deg and the both polarization TM and TE incident. Frequency is 12.0 GHz assuming the satellite broadcasting reception application. As shown in the figure, thickness l of dielectric substrate has the optimum value 2.8 mm which minimizes the phase errors Δφ 1 and Δφ 2. In consideration of obtaining the two sided copper-lined substrate, the 3.2 mm thickness dielectric substrate was selected. Fig.6 shows the Δφ 1 and Δφ 2 changing the period d of the element in the case of l = 2.8mm and l = 3.2mm. By changing the thickness of dielectric substrate l from the optimum value 2.8 mm to the selected value 3.2 mm, the maximum phase error becames about 10 deg worse. As the period d be larger, the phase error be smaller, however the
period d was selected to be 13.0 mm to avoid the grating lobe generation. Fig.7 shows the Δφ 1 and Δφ 2 changing the width w of the element in the case of l =3.2mm and d =13.0mm. The width w of the element has the optimum value which minimizes the phase errors and was selected to be 0.4 mm. Fig.8 shows the arrangement of designed resonant elements under the condition of the determined parameters, where the incident angle and the polarization on the refectarray surface are considered. Fig. 8. Arrangement of designed resonant elements. Fig. 5. Phase errors Δφ 1 and Δφ 2 vs. l. III. REFLECTARRAY ANTENNA MODEL To verify the design method, a circularly polarized reflectarray antenna model for satellite broadcasting reception was manufactured and measured.two marketed parabola antennas with the aperture diameter 450mm were prepared. The palabolic reflector of one antenna was replaced to the manufactured reflectarray with an paerture size 450mm 430mm and another antenna was used as the reference to compare with the reflectarray antenna. Fig.9 shows the photograph of both antennas. Fig. 9. Manufactured reflectarray and parabola antenna. Fig. 6. Phase errors Δφ 1 and Δφ 2 vs. d. Fig. 7. Phase errors Δφ 1 and Δφ 2 vs. w. IV. MEASURED RESULTS The reflectarray antenna model was measured using a near field measurement system shown in Fig.10. Fig.11 shows the aperture phase distribution at 12.0GHz. The calculated phase error on the aperture was less than 45 deg, however the measured aperture phase distribution has more error. Fig.12 shows the aperture amplitude distribution at 12.0GHz. In the calculated results, the primary radiation pattern was assumed to be cos m θ p where m was determined by the measured amplitude distribution of the parabola antenna. The amplitude decrease seen at the upper region on the aperture was because of the the cross polarization component shown in Fig.13. The cross polarization component was generated by the reflected phase difference between TM incident and TE incident. Fig.14 shows the measured gains of the reflectarray antenna model and the parabola antenna and the catalog gain of the parabola
antenna. The difference between the measured gain and the catalog gain of the parabola antenna may be the loss of the manufactured co-axial cable / waveguide converter replaced by the LNB attached to the original parabola antenna feed. Fig.15 shows the corrected reflectarray antenna gain considering the feed loss. The antenna gain is 1.6dB lower than the calculated gain at 12.0GHz. (a) Calculated Fig. 10. Measurement system. (b) Mesured Fig. 12. Amp. distribution (12GHz). (a) Calculated Fig. 13. Calculated amp. distribution of cross-polarization (12GHz). Fig. 11. (b) Mesured Phase distribution (12GHz). Presumed causes of the differences of the aperture distributions and antenna gains between the measured and the calculated are as follows. The relative position of the primary feed and the reflectarray was not clear at the time of design because the design parameters of the marketed parabola antenna were not opened, so the designed antenna and the measured antenna may have some differences. At the time of design and analysis of the reflectarray antenna, the primary radiation pattern was assumed to be cos m θ p, because the design parameters of the attached feed horn were not clear. At the calculation of the reflection coefficient at the ele-
Fig. 14. Measured value of gain. V. CONCLUSION In design of a reflectarray antenna, we considered the incident angle and the polarization on the refectarray surface. In the result, even if the shape of the conductive elements are a simple ring type, the reflection phases can be controlled only by varying the radius of ring elements. And it was found that the thickness of the dielectric substrate had an optimal value that minimized the phase errors. The design method was verified to be adequate by the measured results of the antenna model. However, the measured antenna gain was about 1.6dB less compared with the calculated result, so the causes will be clarified. REFERENCES [1] Jhon Huang and Jose A. Encinar, Reflectarray Antennas, IEEE Press, Wiley-Interscience, 2007. [2] Seki, Makino, Betsudan1, Hirota, Noguchi, Mizusawa and Nishino, PMC Characteristics of Frequency Selective Reflector with Metal Plate and Its Application to Thin and Metal Attachable RFID Tag, 2009 International Symposium on Antennas and Propagation, 2009. Fig. 15. Corrected value of gain. Fig. 16. radiation patterns (12GHz). ment, the infinite periodic array of elements is assumed. So the reflection coefficients differ between the designed and actual value. And the effect of the surface wave is not included in the analysis. The actual loss of the manufactured co-axial cable / waveguide converter is not clear. Fig.16 shows the measured and calculated radiation patterns of co-polarization and cross polarization at 12GHz. As shown in the figure, the reflectarray antenna model realized the low sidelobe radiation pattern and the measured and the calculated radiation patterns coincide very well. However the peak level of the cross polarization component differs about 12dB between the measured and the calculated.