WIXLIB

View metadata, citation and similar papers at core.ac.ukEuCAP 2018 1570406531brought to you byCOREprovided by Infoscience - École polytechnique fédérale de 1525354555657606162636465Wideband Cavity-Backed CubeSat Antenna in S bandMiroslav J. Veljovic, Anja K. SkrivervikMicrowaves and Antennas Group (MAG), EPFL, Lausanne, Switzerland, miroslav.veljovic@epfl.chand Earth Exploration‐Satellite services (EES) in the S band,as allocated by the European Cooperation for SpaceStandardization (ECSS). The initial aperture-coupled patchdesign, selected to acquire a better understanding of theantenna operation, is presented in Section II, including themeasurement results of the first prototype. Section IIIdescribes a detailed set of modifications performed on theinitial model, making it suitable for CubeSat applications. Aprototype of the new cavity-backed stacked-patch antennawas built and validated. The measurement results are alsopresented in Section III. Potential limitations of thisgeometry and instructions for handling such limitations aredescribed in Section IV.Abstract—The Telemetry, Tracking and Command(TT&C) antennas are a crucial component of small satellites,as their in-orbit attitude is not always well-defined. The TT&Cantenna design for a CubeSat is an even more challenging task,considering the volume restrictions imposed by the standard,and the bandwidth requirements for a duplex communication.A detailed design process of a low profile S-band antenna,suitable for CubeSat applications, is presented in this paper.The fabricated prototype exhibits a 10-dB impedancebandwidth of 40 %, and a 3-dB axial-ratio bandwidth of 32 %.Index Terms—Cavity-backed, wideband, stacked patchantenna, CubeSat, TT&C.I.INTRODUCTIONCubeSat is the most popular standard for small modularsatellites. Each CubeSat is comprised of Units (U) - cubeshaving a size of 100 x 100 x 100 mm3 and a mass of up to1.33 kg [1], with the most popular being 1U, 3U and 6Usatellites [2]. The standard attracts many universities andindustries with its low cost of building and orbitaldeployment. The small available volume and weight posestrict constraints on the system design, including the onboard antennas. During the launch, no feature on the satellitesurface should protrude the volume designated by thestandard, and either low-profile or deployable antennastructures are used [3].Several antennas on board CubeSats realize variousfunctions and their frequencies vary from VHF to Kabands [3]. Among others, Telemetry, Tracking andCommand (TT&C) are vital for each satellite mission. Rightafter a satellite is detached from the launch vehicle, theattitude is not defined and wide-coverage antennas arerequired to establish a link with the ground stations andallow for subsequent satellite operations [4]. Therefore, theTT&C antenna reliability is crucial for a satellite missionsuccess. In this sense, low-profile antennas are a favourablecandidate over fail-prone deployable structures. On the otherhand, size restrictions, primarily on the antenna thickness,result in limitations on the achievable impedance and axialratio (AR) bandwidth with this kind of antennas, especiallyin the lower frequency bands. This is a known issue withplanar antennas, and so far, only a few CubeSat antennadesigns that yield narrowband AR performance are proposedfor S-band frequencies [5], [6].In this paper, we present a design of a wideband lowprofile circularly-polarized (CP) cavity-backed patch antennathat fits on a single side of a 1U CubeSat and covers theIndustrial, Scientific and Medical (ISM) 2.4 GHz band andfrequencies for Space Operation (SO), Space Research (SR)II.APERTURE-COUPLED PATCH ANTENNAThe initial antenna requirements are summarized inTable 1.An aperture-coupled patch antenna [7] was selected for apreliminary model, since this geometry provides the mostbandwidth, compared with other methods of exciting patchantennas (coaxial pin, microstrip line). To better understandthe operation of this antenna, a linearly-polarized (LP)microstrip-fed element was first tuned for the desiredfrequency band, having a minimal antenna thickness for thespecified bandwidth. The circular polarization is thenachieved using a crossed slot and a feeding network based onT-junction power dividers, quarter-wavelength transformersand phase-delay lines. The design was carefully analysed inseveral steps, gradually increasing the feeding networkcomplexity. Antenna geometry with intermediate designsteps is shown in Fig 1.The design was verified through a prototype shown inFig. 2(a). RT Duroid 5870 substrate (εr 2.33, tanδ 0.0012,h 0.508 mm) is used for the feeding network and thecoupling slot, the patch is etched on a thin FR4 sheet(εr 4.55, tanδ 0.02, h 0.2 mm) and separated from the slotusing a space-eligible hard foam material (εr 1.08,tanδ 0.0002, h 10 mm). Individual layers are attachedtogether using nylon screws.TABLE I.Frequency band [GHz]Antenna gain [dBi]Beamwidth [ ]S11 [dB]Axial Ratio [dB]Dimensions [mm3]1INITIAL ANTENNA REQUIREMENTS2 – 2.45 ( 20%) 5 50 -10 3100 x 100 x 12

The issue could be overcome if a better matching isachieved for a single element, which would reduce thereflections inside the network, or if Wilkinson dividers wereused instead of T-junctions, in which case the reflectionswould be dissipated in the dividers’ resistors. However, inthis case, the active reflection coefficient of the antenna islimited by the available thickness, and T-junctions wereselected as they allow a simpler manufacturing process.a)b)c)d)a)e)Fig. 1. Aperture-coupled patch antenna model evolution. a) LP model1. b) LP model 2. c) CP 4-port model. d) CP full model. e) Crosssection view. The figures show the feeding lines (black), the patch(dark grey) and the slots (light grey).The prototype is characterized in an anechoic chamber, andthe measured results are compared with the simulations inFig. 2(b).An intermediate 4-port model, that excludes the feedingnetwork, is used to correctly evaluate the coupling to theradiating patch. The response of the 4-port model, shown inFig. 2(b), is a good indicator of the power accepted by thepatch in the desired polarization. The port excitation mimicsthe feeding network performance, and the active Sparameters (1) are calculated in order to evaluate the reflectioncoefficient of the antenna in this configuration.T-junctions are used in the full model as power dividers,thus multiple reflections exist inside the feeding network.The reflected signals ultimately arrive at the coupling slotswith an unwanted phase shift and effectively contribute tothe cross-polarization. This explains the low values of S11 forthe full model, shown in Fig. 2(b). Moreover, the similaritybetween the active-S1 of the 4-port model, and the AR of thefull model seen in Fig. 2(c), confirms the previous analysis.b)1The active S-parameters of an N-port network are obtained usingthe following expression:Nactive - Sm S mnn 1an, m 1,., N ,amwhere Smn are the conventional S-parameters. Parameter active-Smgives the reflected power in port m when all the ports are excited in aspecified way (compared to the conventional S-parameters, whereports are excited one at a time).c)Fig. 2. Aperture-coupled patch antenna. a) A realized prototype. b)The reflection coefficient. c) Simulated gain and axial ratio.2

III.CAVITY-BACKED PATCH ANTENNAA metallized FR4 is used as the second ground plane. Thepatches are etched on two other FR4 sheets (h 0.1 mm) andseparated using two hard foam layers (h 4 mm, h 5 mm).All the layers are attached with nylon screws. Theconductive shield around the stripline was first approximatedwith a pin wall, having the pin separation of 15.6 mm( 0.1 λg, where λg is the wavelength inside the stripline at 2GHz). The pins were manually soldered to the two groundplanes, since metallized vias, or any kind of metallization atall, is not feasible on the foam material. The firstmeasurements were performed in this configuration.Afterwards, the stripline was enclosed with a conductivetape, and the measurements were repeated.Once the antenna is mounted on the CubeSat, it isnecessary to isolate it from the interior satellite components,in order to ensure a proper operation of both the antenna andon-board electronics. This can be accomplished by using astripline-based feeding network. In this case, the electricfield strength is equally distributed between the two groundplanes in the substrate, which results in a poor coupling tothe radiating patch. It is shown that the coupling to the patchis the critical point of this design, considering the bandwidthrequirements and the thickness limitation. Severalmodifications were performed to improve it: A second, smaller patch was added below the originalone, to improve the coupling at the higherfrequencies. By adding the second patch, the totalantenna thickness remains unchanged. Instead of a stripline, a layer of low-permittivity foamand a second ground plane were added above themicrostrip lines. The resulting structure can beconsidered as an asymmetric stripline [8] and itfocuses the electric fields in the dielectric layer closeto the slot, ensuring a stronger coupling. Thisapproach alone is, however, ineffective for a smallground plane separation that is required here, as theproximity of the second ground plane deteriorates theperformance. This effect is demonstrated in Fig. 3. Mode-suppression pins were placed around the slot toreduce the power transferred to the parallel-platemode in the stripline. The introduction of pins had anoverall negative effect on the coupling since itperturbed the electric field distribution inside thestripline, in the vicinity of the slot. This effect wasobserved regardless of the number of pins, even if thedensity as high as λg/10 is used [9]. Following the conclusions drawn from the previouspoint, the key modification was to enclose the wholefeeding structure from all sides with a conductivelayer, effectively creating a rectangular cavity in theasymmetric stripline. The electric field distributioninside the cavity has a maximum at the position of theslot, which drastically increases the power deliveredto the radiating patch, and improves the overallantenna performance.Fig. 3. Active S-parameter of the 4-port asymmetric-stripline model,for different separations of the second ground plane (h).Other modifications are also possible, such as changing theshape of the slots, but they were not considered in this work.A modified feeding network, based on Wilkinson powerdividers and phase-delay lines, is used for the full antennamodel, shown in Fig. 4. The result is a substantialimprovement of the AR values, in comparison with the Tjunction-based feeding network, as the reflected signals arenow absorbed by the dividers’ resistors.A prototype of the cavity-backed antenna was built toverify the design. The feeding asymmetric stripline consistsof an RT Duroid 5870 substrate (εr 2.33, tanδ 0.0012,h 0.508 mm) and a low-permittivity hard foam material(εr 1.08, tanδ 0.0002, h 2 mm).a)b)Fig. 4. A cavity-backed stacked-patch antenna model. a) The topview, including the feeding network (black), the patch elements (darkgrey) and the slot (light grey). b) Cross section view.3