A Novel Near-field Gregorian Reflectarray Antenna ...

A Novel Near-field Gregorian Reflectarray Antenna Design with a Compact Deployment Strategy for High Performance CubeSats Yahya Rahmat-Samii, Joshua M. Kovitz, Jordan Budhu, and Vignesh Manohar Electrical Engineering Department University of California Los Angeles Los Angeles, CA, USA [email protected], [email protected], [email protected], [email protected]

Abstract—We propose a novel antenna system which fits within the cubesat small satellite paradigm is presented which contains a Series-Fed Patch Array Feed Antenna, a parabolic subreflector, and a Reflectarray main reflector all folded into one of the most compact and conformal dual reflector designs to date. This antenna geometry is seen to be unique in that it produces a very compact configuration suitable for the small satellite platform. The subreflector is deployable and is identified as the only moving part eliminating cable movement allowing simplified deployment. The patch array feed antenna generates a plane wave in the near field which illuminates a parabolic subreflector. The scattered field from the parabolic subreflector illuminates a reflectarray main reflector which is designed to emulate an offset parabolic main reflector which shares the same focus as the parabolic subreflector. This paper presents the design and characterization of the antenna system and provides measurements and/or simulation results for each of the antenna components.

Spring-loaded Hinge points

10cm

30cm

20-22cm

CubeSat Chassis

(a)

Parabolic subreflector

Main Reflectarray (acting as a parabolic reflector)

I. I NTRODUCTION CubeSats represent a major paradigm shift for the satellite community and access to space in general. Traditionally, the design trend for satellites focused on durability, long lifetimes, and high performance, which translates to a high cost and lengthy time to deployment. CubeSats, on the other hand, compromise performance and lifetime with the goal to dramatically reduce costs and development time. Their small size allows launching as a secondary payload using a standardized launcher known as the Poly Picosatellite Orbital Deployer (PPOD) [1]. Having a standardized chassis and launching system is a key factor in cost reduction and helps to reduce development time; using a well-developed chassis is one less item to design for satellite engineers. The chassis can be built up using modules having a size of 10 × 10 × 10cm3 , which is often referred to as 1U (meaning 1 CubeSat-unitvolume). These are incredibly small dimensions compared to the typically large satellites. Typical missions use a size of 3U to 6U, making it difficult to integrate high performance wireless systems (radar or communications) into the CubeSat chassis [2]. A major challenge is the volume required for a high-gain antenna. Deployable, high-gain antennas are attractive for CubeSats, especially ones that can be stowed in a compact footprint.

CubeSat Chassis

Feed array (b)

Fig. 1. The near-field gregorian reflectarray with a compact deployment strategy for CubeSats. (a) Stowed configuration with subreflector nearly flush with the CubeSat chassis. (b) Deployed configuration, where the subreflector is deployed using spring-loaded hinges.

Deployment strategy is a major consideration for designers, and current attention has focused on developing large apertures that deploy outside the CubeSat chassis (e.g. umbrella or truss-net reflectors) [3]–[6]. Inflatable reflectors [7] and memory composites [8] have also been investigated for future developments. Another interesting design is a folded-panel reflectarray, which can be compactly stowed on the periphery of the CubeSat chassis [9]. This folded-panel reflectarray

Planar wavefront leading to collimated beam

Parabolic subreflector 40mm

Parabolic subreflector

Spherical wavefront 90mm Planar wavefront

Main Reflectarray 209mm

Feed Array

Equivalent Main Parabolic Reflector

110mm

(a)

Feed Array

(b)

Fig. 2. (a) Ray tracing diagram illustrating the operation of the near-field gregorian reflectarray antenna. (b) Ray tracing diagram illustrating the operation of the near-field gregorian reflector antenna using an equivalent main parabolic reflector.

design acted as the main reflector aperture, and a small antenna array is deployed to feed the reflectarray once the panels have unfolded. Little work, however, has been done to develop ultra-compact, deployable antennas with moderate aperture sizes that seamlessly integrate with the chassis. We propose a novel CubeSat antenna concept that utilizes a near-field Gregorian reflectarray, as illustrated in Fig. 1. The feed is a planar array conformal to the CubeSat surface that radiates an effective plane wave in its near-field zone. A parabolic subreflector focuses the plane wave towards the focus of the main reflectarray aperture, which has been designed to emulate an offset parabolic main reflector. This reflectarray is also conformal to the CubeSat chassis. The novelty of this design lies in the deployment strategy, where only the small subreflector is reoriented for deployment. This overall design avoids cable movement and maintains a compact volume when stowed, while achieving desirable efficiencies. Our compact antenna arrangement for Ka-band operation with an aperture of 209mm directly integrates with a 6U CubeSat chassis. There are many challenges in such a near-field Gregorian reflectarray design. Our focus in this paper will be in developing the feed array for this system, where we must maintain a plane-wave above the array aperture in order to properly illuminate the reflectarray. We present the design methodology, simulated response for each component, integrated response for the system, and the measurements of the fabricated array demonstrating the desired aperture fields. II. A NTENNA D EPLOYMENT AND O PTICAL C ONFIGURATION Near-field Gregorian reflector antenna arrangements have been well known for their scanning capabilities in imaging applications [10]–[12]. The classical configuration uses two reflectors, where both are paraboloidal and share the same focus. A feed array is placed nearby the subreflector, illuminating the subreflector with a plane wave. As illustrated in Fig. 2b, the plane wave is converted to a spherical wave emanating

from the focus of the subreflector. Geometrical optics (GO) predicts that this spherical wave is then converted into a plane wave over a larger aperture area, leading to a collimated beam with a higher directivity than that in the feed array. In our design, we replace the main (larger) parabolic reflector with a reflectarray that operates as a parabolic reflector, as illustrated in Fig. 2a. This allows our structure to be completely conformal to the CubeSat chassis, as indicated in Fig. 1. The two effective paraboloidal structures have an Fm /Dm = 0.431 and Fs /Ds = 0.364. The feed array has a diameter size of 110mm, and the reflectarray has a diameter of 209mm. The parabolic subreflector has an identical diameter to the feed array. The feed array and the main reflectarray are both coplanar, i.e. they are located in the same plane since they are both mounted to the outer surface of the CubeSat chassis. The focus for both reflecting surfaces is located at 90mm above the reflectarray surface, and the subreflector’s apex extends to 130mm above these two surfaces.

Our interest in this near-field Gregorian reflector antenna arrangement was primarily in the simplified deployment mechanism. It is attractive for low-cost space applications to avoid any RF cable movement, which is a significant source of issues for deployment. The only part of the antenna that deploys is the subreflector. Our plan is to implement deployment in a similar fashion to the deployable folded-panel reflectarray antenna in [9]. The subreflector would use two spring-loaded hinges with stops strategically placed in order to tilt the subreflector properly. During launch, the subreflector would be held flush to the chassis surface using a Vectran tie-down. A nichrome burn wire can be used to initiate the deployment sequence and break the tie-down in space [9], [13]. Such a deployment sequence is both simple and only requires a single deployment initiator. This simplification makes a more reliable deployment process.

III. I NITIAL S YSTEM E VALUATION USING A RRAY T HEORY AND AN E QUIVALENT PARABOLA To initially validate our system concept, we sought to use a simplified array and reflector model. For the feed array, we simulated the field distributions using an ideal array of cosq sources, where the nth element’s far-field patterns have the form [14] i e−jkr h (1) En (r) = θˆ cosqx (θ) cos(φ) − φˆ cosqy (θ) sin(φ) r In our model, we assumed that each element had a qx = 1.7 and qy = 1.6 based on previous work with arrays [15]. Our primary interest with the feed array was to evaluate the nearfield characteristics. To simulate this array in the near-field, we assume that we are in the far-field distance to each of the elements individually, and the near field distributions can be then be simulated by using the superposition of all the sources by N X Etot (r) = En (r) (2)

(a)

(b)

(c)

(d)

n=1

0 Normalized far fields (dB)

0 Normalized far fields (dB)

where we implicitly assume that the vectors are decomposed into their Cartesian components and added together. Our array used a circular aperture, where the elements were arranged in a rectilinear grid having element spacings of ∆x = ∆y = 0.65λ0 . The total number of elements was 112, and the elements were excited with equal amplitude and equal phase. The near-field distributions from this simulation are shown in Fig. 3 for the co-polar component. The results are plotted for two different distances above the feed for d = 8cm and d = 12cm, which is near the bottom and top of the subreflector. For both cases, the magnitude appears to taper slightly, while the phase shows a variation ≤ ±25◦ within the radiating aperture of the feed array for both distances. This indicates that we are achieving a quasi-plane wave that is illuminating the subreflector, which is need for the near-field Gregorian reflector system. Next, we wanted to analyze the entire system as a whole. As a starting point, one can use the equivalent paraboloid as an initial evaluation of the entire system, as illustrated in Fig. 2. This is important because a reflector can be simulated using physical optics (PO) diffraction analysis, which runs with reasonable memory and CPU speeds. When a reflectarray is simulated, a significant amount of computing resources are needed for such large apertures (Dm = 18.1λ0 ) Largescale resources are needed to properly mesh the resonant patch structures within the reflectarray. Therefore, running the simulation would not happen in a reasonable amount of time, especially for many different variations. We setup the simulations for the equivalent paraboloid system, where we used the PO on the subreflector and PO on the main equivalent paraboloid. The radiated near-fields from the array were used as the incident source for the subreflector. The PO currents from the subreflector were then integrated in the near-field region to determine the final PO currents on the main. The resulting patterns were computed, and overall

Fig. 3. Simulated aperture field distributions for a uniform phase, uniform amplitude circular array at different distances d from the aperture. (a) Copolar magnitude for d = 8cm. (b) Co-polar phase for d = 8cm. (c) Co-polar magnitude for d = 12cm. (d) Co-polar phase for d = 12cm.

−10 −20 −30 −40 −50 −60 −20

−10

0 θ (deg)

(a)

10

20

−10 −20 −30 −40 −50 −60 −20

−10

0 θ (deg)

10

20

(b)

Fig. 4. Performance of the equivalent near-field gregorian system evaluated using the equivalent main parabola (using the 112 element cos-q array). (a) Eplane performance. (b) H-plane performance.

decent performance was achieved. This can be observed in Fig. 4, where we achieved D0 = 33.80dB at boresight, which was located at θ = 0. Despite the large offset angles, we were able to achieve a low cross-polarization performance for this system, which would be nearly -20 dB if a singlereflector were used. The half-power beamwidths (HPBW’s) in each plane were HP BWE = 3.64◦ and HP BWH = 3.56◦ . The important outcome of this assessment was that the concept was theoretically validated with several simulations. Moreover, the near-fields and far-field reflector evaluation serve as good benchmarks to determine if our array is suitable for this application. IV. R EFLECTARRAY D ESIGN AND E VALUATION After using the equivalent paraboloid to quantify the performance of the subreflector and feed array, we now turn to the reflectarray. The reflectarray uses square patch elements whose size determines the reflection phase of the scattered fields.

Focus

−20 −30 −40 −50 −60 −20

Fig. 5. Reflectarray design coordinate system.

0

Equivalent Paraboloid Reflectarray

Normalized fields (dB)


0 −10

−10

0 θ (deg)

10

20

Equivalent Paraboloid Reflectarray

−10 −20 −30 −40 −50 −60 −20

(a)

−10

0 θ (deg)

10

20

(b)

Fig. 7. Comparison between the performance of the reflectarray using a cosine-Q feed with the equivalent parabola system (with the designed feed array). (a) E-plane performance. (b) H-plane performance.

(a)

(b)

Fig. 6. Reflectarray design process. (a) Phase plot for reflectarray design and (b) Synthesized reflectarray artwork.

For the elements, we used a unit cell size of λ0 /2. Assuming that we have a spherical wave emanating from the focus of the subreflector (and also the reflectarray), we can compute the necessary phase needed to have equal phase across the aperture. Referring to Fig. 5, we have φm = k0 |Ri | + φ0

(3)

where |Ri | is the distance from the focus to the mth element and φ0 is a constant phase offset which can be arbitrarily set for certain purposes. Note that if φm is outside of the range [0, 360◦ ], then a constant ±(360◦ )n phase wrap is added to make the given phase fall in this range, where n is an integer. The computed design phase for each element are shown in Fig. 6a, where each square refers to a certain unit cell. The reflectarray aperture was also chosen to be a circular aperture whose perimeter is marked by the white squares, which represent a null block. As expected, a spherical pattern emerges from this phase plot, and this is the phase needed to compensate for the spherical wavefront emanating from the focus. The design of this reflectarray was computed assuming a cosq source fed at the focus of the reflectarray using a custom spectral domain Method of Moments (SDMoM) code [16]. The reflectarray is designed using the needed phases of Fig. 6a, and the resulting reflectarray artwork is shown in Fig. 6b. The resulting patterns were then calculated using the full-wave simulation tool CST MWS assuming cosq source placed at the focus of the reflectarray. The patterns were then compared to those of the reflector geometry with parabola as the main reflector. The Eplane and Hplane results are shown in Fig. 7, where a directivity of 31.55 dB is obtained. The

beamwidths of this pattern produced HP BWE = 3.7◦ and HP BWH = 4.1◦ for the Eplane and the Hplane, respectively. A reduction in directivity is expected when the reflectarray replaces the paraboloidal main reflector, and thus a reduction in the 3 − dB beamwidths is observed. The H-plane pattern of the reflectarray shows improved Sidelobe levels as expected as the feed used in this simulation utilized a cosq feed which produced a 10 dB taper at the reflectarray edges. The nearfield Gregorian configuration will not likely produce a 10 dB taper at the edges unless a tapered array is implemented. This is a subtle difference, but it can change features such as the sidelobe levels and the cross-polarization performance. In any case, this simulation provides a starting point evaluation to demonstrate the performance that should be expected for this configuration. V. F EED A RRAY A. Design and Development Our goal in this section is to develop and realize a feed array that would serve the near-field Gregorian reflector antenna. Two priorities in this design are low-loss and integration simplicity, and microstrip planar arrays can be very nicely integrated within monolithic circuits or RF modules. A major challenge of microstrip arrays at these frequencies and aperture sizes is minimizing losses. It is well known that series-fed arrays can noticeably reduce losses in a microstrip array [17], [18], and thus we opted for a design with an integral feed network. In other words, the feed network was on the same metallization layer as the micostrip patch radiators. Developing a series-fed, circular aperture is not an easy task, and after several design iterations we opted for the array geometry shown in Fig. 8. A coax input is split in two ways to feed two sets of differential-microstrip lines. In isolation, each of these microstrip lines were roughly 50Ω, but this impedance changes slightly when placed nearby another microstrip line fed with 180◦ phase difference. The 180◦ phase difference between each line is needed because a null will appear at broadside if each 50Ω line was fed with the same phase. From each 50Ω line are subarray branches of various lengths, which were sized in order to achieve a circular aperture. Each of the subarray branches are fed with a 100Ω line, and the series-fed patch radiators are sized to resonate at 26 GHz. Each branch

0 −5

differentiallyfed lines

S11 (dB)

−10 −15 −20 −25 −30 Coax Input

Measurement Simulation

−35 −40 24

50Ω line

25

26 27 Frequency (GHz)

28

100Ω line

Fig. 9. Measured and simulated impedance matching performance of the designed series-fed array.

Fig. 8. A compact, single-layer, single-input, series-fed array providing a uniform phase aperture distribution with low losses.

is spaced by λg , which ensures that the feed network input sees each branch in parallel (enabling nearly equal amplitude excitation). The patch elements are also spaced at λg , which ensure both equal phase and allows the patch elements to be seen as effectively parallel loads at the input. B. Measurement and Near-Fields The design shown in Fig. 8 was chemically etched onto a Rogers Duroid 5880 (εr = 2.2) board with a substrate thickness of 0.508mm. The smallest gap size was 0.5mm and the thinnest trace width was also 0.5mm. Both sizes could be reasonably fabricated within UCLA facilities, but several prototypes had to be iterated before a good fabrication recipe was realized. For the coax input, a SMPM connector was connected behind the ground plane, using a 0.3mm hole to connect the inner conductor to the feed network. We then soldered the connector ground to antenna ground plane. The impedance matching performance for this array showed excellent agreement with that of our simulated model (using HFSS), which helped confirm that our fabrication was successful. The results are shown in Fig. 9. It should be noted that this measurement includes a 6in cable that converts a 1.85mm coax to SMPM coax connector. Potentially the most important step was to measure the patterns of the array. Two brackets to position the array were fabricated using 3D printing. A FormLabs stereolithography (SLA) printer is available on campus that can achieve highly accurate structures. We fabricated these brackets with resin and attached the array, as shown in Fig. 10. We measured the near-fields at 5.5cm above the array using our UCLA Bipolar Near-field chamber. To obtain the near-fields at 8cm and 12cm distances, we applied probe compensation and the near-field to far-field (FFT) transforms to obtain the far-fields. With the far-

(a)

(b)

Fig. 10. (a) Array and brackets in the UCLA bipolar near-field chamber. (b) Magnified view of the fabricated prototype.

fields known, we then used back-projection holography to find the field distributions at 8cm and 12 cm. The results are shown in Fig. 11, where similar amplitude distributions and nearly constant phase distributions are observed. This reaffirms our design of this feed array for the near-field Gregorian reflector antenna. C. Integated Evaluation We also integrated the measured near-fields with the equivalent paraboloid system in order to benchmark how well this array worked with the previous ideal array when used with the equivalent paraboloid system. The results shown in Fig. 12 indicate that good performance was achieved, although higher sidelobes were observed compared with the ideal array results. Higher sidelobes from spurious feed network radiation are likely the cause of these sidelobes, along with a low amplitude taper at the aperture edges. The directivity for this system was D0 = 32.54 dB at boresight, with beamwidths of HP BWE = 2.6◦ and HP BWH = 3.6◦ . VI. C ONCLUSION Our goal was to provide a proof-of-concept for the CubeSat near-field Gregorian system along with an array measurement. This array measurement was critical to determine if the feed array approach was manageable for this application, which is difficult to quantifying in terms of a few performance metrics. By examining the measured near-field distributions and integrated system performance based on the array measurement,

(a)

(b)

(c)

(d)

0

0

−10

−10 Normalized fields (dB)


Fig. 11. Measured aperture field distributions for the series-fed array in FIg. 8. (a) Co-polar magnitude for d = 8cm. (b) Co-polar phase for d = 8cm. (c) Copolar magnitude for d = 12cm. (d) Co-polar phase for d = 12cm.

−20 −30 −40 −50 −60 −20

−20 −30 −40 −50

−10

0 θ (deg)

10

20

−60 −20

(a)

−10

0 θ (deg)

10

20

(b)

Fig. 12. Near-field gregorian system evaluated using the equivalent main parabola and the measured near-field array patterns. (a) E-plane performance. (b) H-plane performance.

we were able to determine that the fabricated array component is completed. This is significant progress towards developing the full near-field Gregorian reflector antenna, and we plan to fabricate the subreflector and the reflectarray in the near future. R EFERENCES [1] H. Heidt, J. Puig-Suari, A. Moore, S. Nakasuka, and R. Twiggs, “Cubesat: A new generation of picosatellite for education and industry low-cost space experimentation,” in Annual AIAA/USU Conference on Small Satellites, 2000. [2] Y. Rahmat-Samii, V. Manohar, and J. M. Kovitz, “For satellites, think small, dream big: A review of recent antenna developments for cubesats.” IEEE Antennas and Propagation Magazine, vol. 59, no. 2, pp. 22–30, 2017. [3] N. Chahat, R. E. Hodges, J. Sauder, M. Thomson, E. Peral, and Y. Rahmat-Samii, “Cubesat deployable ka-band mesh reflector antenna development for earth science missions,” IEEE Transactions on Antennas and Propagation, vol. 64, no. 6, pp. 2083–2093, 2016.

[4] V. Manohar, J. M. Kovitz, and Y. Rahmat-Samii, “A novel customized spline-profiled mm-wave horn antenna for emerging high performance cubesats,” in Proceedings of the Antenna Measurements and Techniques Association Meeting, 2016. [5] J. M. Kovitz, V. Manohar, and Y. Rahmat-Samii, “A spline-profiled conical horn antenna assembly optimized for deployable ka-band offset reflector antennas in cubesats,” in IEEE International Symposium on Antennas and Propagation, July 2016. [6] V. Manohar, J. M. Kovitz, and Y. Rahmat-Samii, “Ka band umbrella reflectors for cubesats: Revisiting optimal feed location and gain loss,” in International Conference on Electromagnetics in Advanced Applications. IEEE, 2016, pp. 800–803. [7] A. Babuscia, B. Corbin, M. Knapp, R. Jensen-Clem, M. Van de Loo, and S. Seager, “Inflatable antenna for cubesats: Motivation for development and antenna design,” Acta Astronautica, vol. 91, pp. 322–332, 2013. [8] R. Barrett, R. Taylor, P. Keller, D. Codell, and L. Adams, “Deployable reflectors for small satellites,” in Annual AIAA/USU Conference on Small Satellites, 2007. [9] R. E. Hodges, N. Chahat, D. J. Hoppe, and J. D. Vacchione, “A deployable high-gain antenna bound for mars: Developing a new foldedpanel reflectarray for the first cubesat mission to mars.” IEEE Antennas and Propagation Magazine, vol. 59, no. 2, 2017. [10] C. Dragone and M. Gans, “Imaging reflector arrangements to form a scanning beam using a small array,” Bell System Technical Journal, vol. 58, no. 2, pp. 501–515, 1979. [11] Y. Hwang and C. Han, “A scanning offset-fed, near-field gregorian reflector antenna,” in IEEE International Symposium on Antennas and Propagation, vol. 19. IEEE, 1981, pp. 716–719. [12] J. A. Martinez-Lorenzo, A. Garcia-Pino, B. Gonzalez-Valdes, and C. M. Rappaport, “Zooming and scanning gregorian confocal dual reflector antennas,” IEEE Transactions on Antennas and Propagation, vol. 56, no. 9, pp. 2910–2919, 2008. [13] A. Thurn, S. Huynh, S. Koss, P. Oppenheimer, S. Butcher, J. Schlater, and P. Hagan, “A nichrome burn wire release mechanism for cubesats,” in The 41st Aerospace Mechanisms Symposium, 2012, pp. 479–488. [14] Y. Rahmat-Samii and S.-W. Lee, “Directivity of planar array feeds for satellite reflector applications,” IEEE Transactions on Antennas and Propagation, vol. 31, no. 3, pp. 463–470, 1983. [15] J. M. Kovitz, J. P. Santos, Y. Rahmat-Samii, N. Chamberlain, and R. Hodges, “Enhancing communications for future mars rovers: Using high-performance circularly polarized patch subarrays for a dual-band direct-to-earth link.” IEEE Antennas and Propagation Magazine, 2017. [16] J. Budhu and Y. Rahmat-Samii, “Accelerating the spectral domain moment method for reflectarray’s by two-orders of magnitude,” in IEEE Antennas and Propagation Society International Symposium (APSURSI), July 2013, pp. 1340–1341. [17] B. Jones, F. Chow, and A. Seeto, “The synthesis of shaped patterns with series-fed microstrip patch arrays,” IEEE Transactions on Antennas and Propagation, vol. 30, no. 6, pp. 1206–1212, 1982. [18] T. Metzler, “Microstrip series arrays,” IEEE transactions on Antennas and Propagation, vol. 29, pp. 174–178, 1981.