1 Introduction

The mission of the ETS-VIII is to establishsatellite-based mobile communication technol-ogy and multimedia broadcasting technologyat an affordable price [1]. Specifically, thesatellite’s mission is to enable communicationbetween handheld terminals through anonboard satellite switch and to permit recep-tion of broadcast waves by small terminalswithin S-band (2.5-GHz) frequencies.Achievement of these objectives will require amultibeam antenna with a large-diameterdeployable reflecting surface, a high-outputtransponder, an onboard switch, and smallportable terminals. In this paper, we will dis-cuss the technology relating to the onboardmultibeam antenna.

Requirements for the antenna includehigh-gain multibeam capability, low sidelobesfor the redundant use of frequencies, controlof beam directionality, flexible control ofbeam shape, arbitrary power allotment among

transmitted beams, a high-output amplifierwithin a redundant configuration, and efficientradiation of RF power. To satisfy theserequirements, a reflector antenna employingan array configuration for the primary radiatorwas adopted. This antenna functions in amanner similar to an antenna featuring anactive phased array [2]; however, the design ofthe current configuration has yet to be finallyestablished. We have determined the diameterof the antenna aperture, the setting position forthe feeder array, and the spacing of array ele-ments through calculation of the desired com-munications area of the ETS-VIII [3].

To construct an antenna-feed system thatcan withstand the space environment, we havefocused on developing an array element, asolid state power amplifier, and a beam-form-ing network in the context of overall systemdesign. We constructed the feed system bycombining manufactured devices, confirmingelectrical operations, and experimentally veri-fying the antenna multibeam pattern. A num-

UENO Kenji 57

3-6 Feed System3-6-1 Configuration of the Feed SystemUENO Kenji

Japanese Engineering Test Satellite VIII (ETS-VIII) is being developed and is scheduled tobe launched into a geostationary Earth orbit to experimentally establish mobile communica-tions and multimedia broadcasting services. Of the key technologies in ETS-VIII, we focuson an array-feed system for making multiple beams and illuminating large, deployablereflectors. Assembled array-feed system have been tested, and their ability to control RFsignals in amplitude and phase has been confirmed satisfactorily. Based on measuredarray-feed radiation patterns, we have derived reasonable secondary radiation patterns ofETS-VIII S-band antennas. Environmental tests implemented to the feed units which arebeing placed outside the satellite body have certified sufficient durability in the launchingperiod and space environment.

Keywords Satellite communication, Engineering Test Satellite VIII, Multibeam antenna, Feed sys-tem, Feed unit

58

ber of tests involving heat, acoustics, andvibration were conducted to confirm that thefeed system will function effectively inonboard satellite use, and an analytical modelwas designed to allow for future prediction ofantenna performance.

The function and configuration of the feedsystem of the large deployable reflector on theETS-VIII will be discussed in Section 2, theperformance of the individual devices consti-tuting the feed system will be described inSection 3, and Section 4 will address the elec-trical characteristics of the feed system, theradiation characteristics of the antenna, andthe results of environmental testing.

2 Antenna and feed system

The appearance of the ETS-VIII is asshown in Fig.1. The satellite features twodishes of array-feeding reflector antennas; onefor transmission and one for reception. Thisreflecting surface forms an offset paraboloidalreflecting surface made of metal mesh, withan aperture interior diameter of 13 m and anouter diameter of 17 m [4]. The primary radia-tor of the active phased array illuminates thereflector, and is arranged not at the focal posi-tion of the reflector, but at a point away fromthe focal position, slightly toward the reflect-ing surface, i.e., in an off-focus position. Thisarrangement is chosen in order to enable oper-ations similar to those performed by a normal

phased-array antenna that does not feature areflecting surface [5], as this arrangementallows for an expanded beamwidth in a sec-ondary radiation pattern (referred to as a com-ponent beam) when the individual array ele-ments direct their beams toward the reflector.Consequently, this arrangement allows forsuperposition of component beams, each ofwhich corresponds to a single array element.

The feed system is composed of primaryarray radiators, amplifiers, beam-forming net-works, and other elements, and is configuredas shown in Fig.2. The dotted lines in the fig-ure show the various subsystems (with theexception of the feed system). In the figure, atransmission feed unit (Tx unit) and a recep-tion feed unit (Rx unit) are shown combinedwithin the feed system; this combination ofdevices must be arranged near the primaryarray radiator. The devices are assembled in asingle enclosure installed on the top of theantenna tower; this equipment must featurerobust environmental resistance. The Tx unitconsists of array elements, filters, and solidstate power amplifiers (SSPAs). The Rx unitconsists of array elements, filters, and low-noise amplifiers (LNAs).

The beam-forming network excites ampli-tude and phase for each array element, shap-ing the RF beam signals into predeterminedradiation patterns. Although each beam portfeatures two beam-forming networks (two fortransmission and two for reception), only one

Journal of the National Institute of Information and Communications Technology Vol.50 Nos.3/4 2003

Appearance of the ETS-VIIIFig.1 Configuration of feed systemFig.2

transmission network and one receiving net-work are employed in actual operation. More-over, the input terminals and output terminalsof the two beam-forming networks are con-nected to hybrids, resulting in continuous in-parallel connection. The beam-forming net-works and the hybrids are incorporated intothe satellite structure.

3 Devices constituting feed sys-tem

3.1 Array feed elementThe transmission feed unit (Tx unit) is as

shown in Fig.3. A planar array consisting ofthirty-one micro-strip antennas [6] equippedwith cups (metal cylinders) is disposed acrossa single surface. Each micro-strip antenna,acting as an array element, is affixed with ahybrid substrate layer (on the reverse side ofthe antenna), and radiates a circularly polar-ized wave by a dual-point feeding method.Moreover, output feed is performed via a non-contact feeding method employing slot con-nection, designed to suppress passive inter-modulation (PIM) and to improve overallfunctional efficiency. Experimental verifica-tion of the metal cylinder led to a diameterspecification of 120 mm, with height set to 40mm. The transmitting frequency was set at2,500 to 2,505 MHz for communications and2,535 to 2,540 MHz for broadcasting, and thereceiving frequency was set at 2,655 to 2,660

MHz. Characteristics of this micro-strip cup

antenna are as follows: bandwidth is 4%, gainis 8.9 to 9.8 dBi (for transmission) and 10.3 to10.8 dBi (for reception), half powerbeamwidth is 60 degrees, axial ratio is 1.5 dBor less, mutual coupling with the adjacentarray element is －40 dB or less, and weight is180 g.

3.2 FilterA transmission filter is inserted between

the solid-state power amplifier and the arrayelement to suppress spurious and intermodula-tion waves. In particular, its main purpose isto limit leakage to the received frequencyband. The filter is of the dielectric resonatortype, features insertion loss of 0.3 dB or less,attenuation of 70 dB or more in the receivedfrequency band, power endurance of 40 W ormore, installation area of 100×100 mm2, anda weight of 300 g. A reception filter is provid-ed to prevent high-power transmitted wavesfrom entering the low-noise amplifier via thearray elements and causing interference. Fig.4shows a group of reception filters arranged inthe reception feed unit (Rx unit). Theemployed filter is of the coax resonator type,featuring insertion loss of 0.4 dB or less,attenuation of 70 dB or more in the transmit-ting frequency band (for communications), aninstallation area of 180×60 mm2, and aweight of 190 g.

UENO Kenji 59

Appearance of Tx unitFig.3Filter arrangement in Rx unitFig.4

60

3.3 Solid-state power amplifierThe primary array radiator of the transmit-

ting antenna was designed with a series offixed excitation values for amplitude; onlyphase is freely controlled in beam formation.This specification was selected based on theconcept of allotting an appropriate poweramplifier to each array element in order toavoid unnecessary power consumption and toreduce weight. This option was selected overthe allotment of high-output power amplifiersto all of the array elements, which would haveprovided more flexible control of excitationamplitude.

The excitation amplitude of the array ele-ment is illustrated in Fig.5; this value can beset to 0, －3, or －6 dB. Two types of solid-state power amplifier are employed, featuring10-W and 20-W output, respectively. Twenty-three of the 10-W amplifiers and eight of the20-W amplifiers are employed. The threeexcitation amplitude values and the input levelof a given amplifier to one of two levels areset for this configuration. These amplifiersare designed to ensure gain and phase stabilityat different input levels and variable environ-mental temperatures [7].

Fig.6 shows the 10-W solid state poweramplifier, featuring gain of 55.0 ±0.5 dB, tem-perature-dependent gain deviation of 0.8 dBp-p or less (for a temperature variation from 0℃to 50℃), gain flatness of 0.15 dBp-p/5 MHzor less, temperature-dependent phase devia-

tion of 19 degp-p or less, phase flatness of 13degp-p/5 MHz, a noise/power ratio of 17.4 dBor more, an installation area of 260×60 mm2,and a weight of 660 g.

The 20-W solid state power amplifier fea-tures gain of 57.9 ±0.4 dB, gain flatness of0.14 dBp-p/5 MHz or less, phase flatness of 2degp-p/5 MHz, a noise/power ratio of 17.2 dBor more, an installation area of 220×210mm2, and a weight of 2,500 g.

3.4 Low-noise amplifierFig.7 shows the low-noise amplifier, a

low-power component, arranged within the Rxunit. For the primary array radiator of thereceiving antenna, independent excitationamplitude is given to each beam, as fixing theexcitation amplitude (as in the transmittingantenna) would not reduce power consump-tion. Only one type of low-noise amplifier isemployed, featuring gain of 44.8 ±0.5 dB,gain flatness of 0.5 dBp-p/5 MHz or less,phase flatness of 0.6 degp-p/5 MHz, phasetracking of 5 degp-p, a noise factor of 1.4 dB

Journal of the National Institute of Information and Communications Technology Vol.50 Nos.3/4 2003

Excitation amplitude of array elementFig.5

Low-noise amplifier within Rx unitFig.7

10-W solid-state power amplifierFig.6

or less, an installation area of 42×40 mm2,and a weight of 50 g.

3.5 HybridSince this feed system includes two beam-

forming networks, three hybrids are arrangedon the beam port side (the side to be connect-ed to the transponder) and thirty-one hybridsare arranged on the element port side (the sideto be connected to the array elements). On theelement port side, eight hybrids are housed intwo tiers within a single housing in order tosave installation space. The combined hybridsfor transmission are referred to as “combin-ers” and those for reception are referred to as“dividers.” Fig.8 shows the appearance of thehybrid. The hybrid features insertion loss of3.5 dB or less, passing amplitude error of 0.2dBp-p or less, passing phase error of 5 degp-p,an installation area of 200×70 mm2, and aweight of 360 g.

4 Overall characteristics of feedsystem

4.1 Electrical characteristicsThe devices described in Section 3 have

been assembled within the feed system. Apartial view of this system within the satellitestructure is shown in Fig.9. Electrical testingof the feed system confirmed that the formedbeam is of the required amplitude and phasewith the given excitation values; i.e., the for-mation of a predetermined beam was con-firmed within the beam-forming network andwithin the field of radiation in the vicinity ofeach array element under measurement [8].Accumulated errors in gain, loss, and phasecaused by the array elements, the amplifiers,and connection cables are all corrected by the

beam-forming network, thus satisfying therequirements applicable to amplitude andphase of the individual beams. Actual finalresidual error will depend on the precision ofcontrol and measurement errors within thebeam-forming network.

Fig.10 shows both the amplitude andphase measured in the vicinity of the arrayelements and the required values. Fig.10(a)corresponds to transmission beam #1 (directedto the Kyushu area), and Fig.10(b) corre-sponds to reception beam #1. Deviations fromrequired values for all beams were 0.5 dB rmsand 4.2 deg rms, both for transmission andreception. These results satisfy the initialrequired values of 0.6 dB and 7.3 deg. All ofthe power amplifiers are to be used in the lin-ear region, and the total of their RF outputs inthis configuration reached 355 W. The overallfeed system weighs 235 kg.

4.2 Radiation characteristicsThe radiation pattern of the primary array

UENO Kenji 61

Appearance of hybridFig.8

Feed system within the satellite struc-ture

Fig.9

62

radiator was measured with a planar near-fieldmeasuring instrument. This radiation patternis used to evaluate the radiation pattern of theoverall antenna, including the reflector.Fig.11 shows the radiation pattern of the pri-

mary array radiator of the beam #1 from thetransmitting antenna, drawn from the meas-ured results for individual beams and radiationpatterns. The axial ratio of the antenna wasalso measured together with pattern measure-ment; the value was determined to be 1.5 dBor less.

The procedure for obtaining the radiationpattern of the overall antenna (including thereflector) is as follows: the aperture distribu-tion on the primary array radiator is obtainedby a reverse projection of near-field data pro-cessing; the RF current induced on the reflect-ing surface is then calculated based on thisaperture distribution, and the radiation patternis determined through integration of the radia-tion field. The reflecting surface is composedof fourteen hexagonal modules featuring metalmesh surfaces. The radiation pattern is ana-lyzed under the assumption that the reflectingplane is a conducting metallic plane, takinggaps between modules into consideration.

Fig.12 shows a comparison between thesecondary radiation pattern obtained throughthe application of this method to the measuredresults of the primary radiator and the second-ary pattern obtained by theoretical analysis ofthe primary radiator. The figure indicates thatthese patterns agree for most elements of themain lobe and the sidelobes; however, a dis-crepancy appears in the －25 dB contour(transmission) or the －30 dB contour (recep-tion). This discrepancy is attributed to resid-ual errors in the excitation amplitude andphase of the primary array radiator and to ana-lytical error in near-field measurement.

Summarizing the conclusions drawn fromthese secondary antenna radiation patterns, wecan confirm that the isolation between a beamand the nearest beam using the same frequen-cy is 19 dB or more in transmission and 25 dBor more in reception. Moreover, the axialratios of all beams were 2 dB or less for bothtransmission and reception.

Measurement of the near-field radiationcharacteristics of the antenna using a part ofthe deployed reflecting surface and the feedsystem was also conducted, confirming the

Journal of the National Institute of Information and Communications Technology Vol.50 Nos.3/4 2003

Measured values and required val-ues of excitation amplitude andphase

Fig.10

Feed system radiation pattern intransmission (beam #1)

Fig.11

validity of the analytical models of thedeployed reflecting surface [9].

4.3 Thermal characteristicsSince the Tx unit and the Rx unit are

installed on the east/west panels of the antennatower on the satellite structure, these unitsmust be subject to thermal control independ-ent of the satellite system. In conventionalgeostationary satellites, heat-generating

devices such as SSPAs were fixed on thenorth/south mission panels to allow heat todissipate directly from the radiating surfaces.However, in a phased-array feed system,SSPAs must be installed in the vicinity of thearray elements in order to reduce SSPA outputloss. The SSPA installation area is further lim-ited due to the structure of the antenna fairingand the antenna tower. Thus, a total of 31SSPAs must be installed in a high-density con-figuration. As a result, up to 1.4 kW of heatenergy is generated within the Tx unit.

It was determined that a heat pipe wouldbe used to transport heat to a radiating panelmounted on the north/south panels of theantenna tower for dissipation. Fig.13 showsthe SSPA arrangement and the disposition ofthe heat pipes in the Tx unit.

Fig.14 is a photograph of the antennatower with the feed-unit thermal modelinstalled (the thermal model will be modifiedto a flight model following completion of the

UENO Kenji 63

Comparison of secondary patterns(feed-system measurement results vs.analysis)

Fig.12

Arrangement of SSPAs and heatpipes in the Tx unit

Fig.13

Appearance of antenna tower andthermal model of feed system

Fig.14

Transmitting antenna secondary pattern (beam #1)

Receiving antenna secondary pattern (beam #1)

Calculation based on feed system measurement resultsCalculation based on feed system analysis

heat pipes

Transmitting antenna element

64

thermal test). The Tx unit is affixed to the eastpanel of the antenna tower (right-hand side ofthe figure); heat generated within the feed unitis dissipated from a radiating panel (center inthe figure) of the north/south panels of theantenna tower. The primary array radiator isinsulated from the interior of the feed unit, andis thermally controlled independently, with theouter surface of the radiator covered withTedra Kapton film.

The Rx unit is installed on the west panelof the antenna tower (left-hand side of the fig-ure); heat generated within the feed unit is dis-sipated from a radiating surface on the upperside of the unit. The primary array radiator isinsulated from the interior of the unit as withthe Tx unit, and is thermally controlled inde-pendently, with the outer surface of the radia-tor covered with Tedra Kapton film.

Thermal balance testing was conducted ina space chamber after combining the antennatower, an antenna for a highly precise timereference apparatus, an antenna for feederlink, and an earth head panel dummy [10].Including the earth head panel in the testingconfiguration enables verification of the ther-mal design of the feed system, at the sametime confirming the performance of the ther-mal interface between the feed system and theoverall system.

For the test, four stationary cases (wintersolstice, summer solstice, equinox-1, and equi-nox-2) and two nonstationary cases (eclipseand orbital transfer) were evaluated. Externalthermal input was generated by a solar simula-tor. Evaluation of the stationary cases wasintended to provide data for correlation withthe results of the thermal mathematical model,while experimentation involving anticipatedthermal environments in orbit was designed toyield thermal balance data. The nonstationaryexperiments were structured to acquire tem-perature data in order to evaluate expectedthermal behavior during eclipse or orbitaltransfer.

Since the measured test results differedfrom the predicted temperatures for the arrayelement, the LNA, and the reception filter, the

thermal mathematical model has been correct-ed accordingly. Consequently, in post-testanalysis using a model subject to such correc-tion, the difference between the predicted tem-perature and the measured temperature wasreduced to within 10℃ for all test cases. Thecorrected model was thus shown to providesufficient accuracy in the prediction of orbitaltemperatures. In particular, for the key SSPAcomponents, temperatures can be predictedwithin 5℃. We then performed further orbitaltemperature analysis using the correctedmodel, in the process confirming that allonboard devices can be maintained within thepredetermined ranges from launch through thefinal mission stages. Table 1 shows the resultsof analysis.

4.4 Acoustic test characteristicsThe acoustic test is conducted to verify

that the feed units can withstand acousticvibration during launch. Fig.15 is a photo-graph of the acoustic testing apparatus for thestructural model of the feed system. Theresults of these tests clarify the following threepoints:

(1) Visual inspection of the Tx unit and the Rxunit revealed no abnormalities such asdeformation of the test specimen. As noproblems were seen with the units in thiscase, we have successfully demonstratedthat the structural model of the feed systemmounted on the tower is capable of with-standing acoustic qualification test levels.

(2) Although the PSD (Power Spectrum Den-sity) levels of the array elements, the fil-ters, the 10-W SSPAs, and LNAs revealedthat the effects of random vibration weregreater than anticipated, examination of

Journal of the National Institute of Information and Communications Technology Vol.50 Nos.3/4 2003

Results of orbital temperature pre-diction

Table 1

the individual devices confirmed that thisexcess vibration presented no problems inthe given frequency bands.

(3) The test jig presented no problems in use,with an effective acoustic response as lowas 2 Grms or less, confirming that this testjig may be successfully applied to proto-flight testing (PFT).

4.5 Vibration test characteristicsThe vibration test is conducted to verify

that the feed units can withstand the vibrationenvironment at the time of launch. Fig.16 is aphotograph of the experimental configurationof vibration testing of the structural feed-unitmodel. Results of these tests have indicatedthe following.

(1) The structural feed-unit model can with-stand the sinusoidal vibration environment.

(2) The feed unit structural model satisfies thenatural frequency requirement of 50 Hz orhigher.

(3) The sinusoidal vibration characteristics ofthe devices comprising the onboard feedunit structural model were confirmed.

(4) Based on the results of the sinusoidalvibration test, the structural analysis modelwas modified so that the natural frequencyof the model agreed with test results withinseveral Hz.

(5) There is no resonance point at 150 Hz orless for the Z-axis of the test jig. Althoughresonance points are noted at 100 Hz orhigher for the X-axis and the Y-axis, accel-eration response is high even at the risingslope of the resonance point. These resultshave established that controlling the testjig response acceleration using a limiterwill allow for the use of the test jig (aswell as the vibration test facility) in con-nection with PFT.

5 Concluding remarks

In the development of the ETS-VIII, wehave focused on and provided a detailedanalysis of a feed system featuring a primaryarray radiator. This system consists of thirty-one array elements feeding a large deployableantenna reflecting surface, permitting effectivemultibeam formation. We have completedmanufacture of array elements, filters, solid-state power amplifiers, beam-forming net-works, hybrids, and the remaining componentdevices of the feed system, and have con-firmed that these devices can withstand use inthe satellite in terms of electrical characteris-

UENO Kenji 65

Experimental configuration ofacoustic testing of structural feed-unit model

Fig.15

Configuration of vibration testing ofthe feed-unit structural model

Fig.16

66

tics and environmental resistance. We con-ducted testing of the overall feed system, con-firming that RF signal control in multibeamformation can be performed with sufficientaccuracy. Measurement of the radiation pat-tern of the feed system revealed that the multi-beam pattern of the overall antenna was inaccordance with design. Moreover, environ-

mental testing (thermal, acoustic, and vibra-tion testing) has confirmed that the feed sys-tem will be able to function aboard the satel-lite, in the process permitting modification ofthe analytic model. Data thus obtained will inturn facilitate future testing and improve theprediction of system performance.

Journal of the National Institute of Information and Communications Technology Vol.50 Nos.3/4 2003

References1 Y. Kawakami, S. Yoshimoto, Y. Matsumoto, T. Ohira, and N. Hamamoto, "S-band mobile satellite communi-

cations and sound broadcasting systems in ETS-VIII", 21st Int. Symp. on Space Tech. and Science, 98-h-

02, pp.1362-1367, May 1998.

2 R. Lo. Forti, "Advanced imaging antenna for satellite application at L-band", IEEE AP-S Int. Symp., pp.1618-

1621, 1989.

3 K. Tokunaga, H. Tsunoda, H. Shoki, and M. Okumura, "Design optimization of phased-array-fed reflector

antennas for mobile communication satellites", AIAA Journal of Spacecraft and Rockets, Vol.36, No.1,

pp.105-108, Jan.-Feb. 1999.

4 K. Yonezawa, M. Homma, S. Yoshimoto, S. Hama, T. Ohira, N. Natori, and Y. Tsutsumi, "Outline of Engi-

neering Test Satellite-VIII (ETS-VIII)", 21st Int. Symp. on Space Tech. and Science, 98-h-01V, pp.1356-

1361, May 1998.

5 K. Ueno, "Multibeam antenna using a phased array fed reflector", IEEE AP-S Int. Symp., pp.840-843, 1997.

6 M. Tanaka, "Effect of the height and diameter of the cup on cup microstrip antennas", IEICE Trans. Com-

mun., Vol.E81-B, No.8, Aug. 1998.

7 H. Ishida, Y. Kawakami, H. Hirose, and M. Shigaki, "Solid-state power amplifiers for the Japanese Engineer-

ing Test Satellite-VIII", Sixth Int. Mobile Satellite Conf., pp.258-261, June 1999.

8 M. Atobe, K. Ueno, M. Okumura, K. Tokunaga, and Y. Yamasa, "Development test results of phased array

antenna feed for the Engineering Test Satellite-VIII", IEICE National Conv., B-3-29, p.285, Mar. 2000.

9 K.Ueno, M. Atobe, A. Miyasaka, T. Orikasa, and M. Okumura, "Ground test of array-fed reflector antennas

for ETS-VIII", IEICE Tech. Rept., AP2000-155, Jan. 2001.

10 A. Tanaka, A. Saito, M. Okumura, J. Asaga, M. Atobe, K. Ueno, and A. Tsujihata, "Thermal Balance test for

large deployable antenna feed of the Engineering Test Satellite-VIII", 44th Space Science Tech. Conv., 00-

1F15, Oct. 2000.

UENO Kenji, Dr. Eng.

Professor, Department of InformationNetwork Engineering, Hokkaido Insti-tute of Technology

Satellite Communication and AntennaEngineering