1

SSC00-III-5

Thermal Reflector System Design and Testing for the Microwave Anisotropy Probe

Hans D. Neubert1, Wayne Chen2

1Programmed Composites, Inc., Corona, CA2NASA/Goddard Space Flight Center, Greenbelt, MD

Abstract

Scheduled for a June 2001 launch, the Microwave Anisotropy Probe’s (MAP) mission is to studyin detail the cosmic microwave background radiation temperature fluctuations of the universe.The cosmic microwave background is the remnant afterglow of the Big Bang, and the tinytemperature differences from place to place on the sky provides a wealth of information aboutthe basic nature of our universe. The observatory consists of dual back-to-back Gregorian opticsand dual differential pseudo-correlation microwave radiometers. MAP will scan the sky over a22 to 90 Ghz spectrum during a 26 month mission at L2. To achieve the necessary dimensionalstability and stringent total weight limits, the Thermal Reflector System (TRS) consists of dualprimary and secondary lightweight composite reflectors mounted to a lightweight but rigidcomposite truss structure. Design features include the use of lightweight graphite spread fabricand honeycomb core for the reflector shells and stiffening structure, and a bonded mortise andtenon truss structure using two complementary composite materials. Qualification of the TRSincluded in-situ photogrammetry measurements while at 40K and 370K, sine burst and sinesweep dynamic loading, and acoustic excitation. Materials employed, coupon test data, designand construction, and qualification testing will be discussed.

Introduction

The Microwave Anisotropy Probe (MAP),the first of two selected MIDEX missions,is a follow-on to the successful COBEmission. The COBE mission supportedthat the model of the universe is not static,but could be better described by anexpanding one. This discovery earnedPenzias and Wilson the 1978 Nobel Prizein physics, leading to the Big Bang theory.An explosion of such size and temperatureto bring the universe into being must haveleft some mark. The remnant from the BigBang is the cosmic microwave radiationbackground, a measurable quantity.

MAP will seek answers to questions aboutcosmology not answered by COBE. Howdid structures of galaxies form in theuniverse? What are the values of the keyparameters of the universe? When did thefirst galaxies form? Why is the cosmicmicrowave background temperature souniform on scales > 2°K?

The MAP observatory consists of DualBack-to-Back Gregorian optics coupled toa Differential Pseudo-CorrelationRadiometer operating in the 22 to 90 GHzfrequency band. MAP will operate at L2, alocation in space with neutral gravity, spinat 70 rpm, and map the entire universetwice during its 26 month mission. Chuck

2

Bennett of NASA/GSFC is the principalinvestigator.

On April 1, 1997, ProgrammedComposites, Inc. was awarded the contractto provide design engineering, fabricationand qualification testing for the ThermalReflector System (TRS). The TRS consistsof 1.4 x 1.6 meter primary mirrors, .80meter secondary mirrors, dual passivethermal radiators, and a truss structure formechanical alignment of all components.A schematic of the major componentscomprising the TRS is shown in thefollowing figure.

Design Philosophy

At L2, the TRS performs its function whileat a generally uniform 50K (-370F).However, fabrication of the compositeelements occurs at either 250F or 350F.Final assembly and alignment is done atroom temperature. Residual stress betweenfiber and resin matrix at these differential

temperatures can easily lead to warpageand misalignment. Maintenance of criticaldimensions between components requiresthat strict control of Coefficient of ThermalExpansion (CTE) be sustained.

Well known among the spacecraftcomposites community is the fact thathexagonal honeycomb is orthotropic innature, and imparts an orthotropic CTE to aotherwise quasi-isotropic CTE sandwichstructure. As one example, developed byPCI, is the GeoSat Follow-On spacecraftradiometer, where XN-50A fabricfacesheets combined with aluminum

honeycomb produced measured CTE’s of.35 ppm/F and .45 ppm/F in the plate X andY directions. For the MAP application,these CTE’s are too far from zero, and arenot acceptable, but for GFO they are wellwithin established design parameters, sinceGFO operates in low earth orbit withreflector temperatures near ambient.

3

Both the primary and secondary reflectorson TRS are of sandwich construction tomaximize shell rigidity, and incorporate thehigher modulus XN70A spread fabric (75gm/sq. meter) with DuPont’s Korexhoneycomb core. Selection of the Korexwas based on uniformity of coreconstruction, reduced CTE and improvedstrength and stiffness compared to Nomexnon-metallic core. The final CTE’s are .03ppm/F and .15 ppm/F in the reflector X andY plane.

Supporting rib structure for the reflectorsrequired a matching CTE in the directionalong the bonded length with the reflectors.In this case, XN70A tape material togetherwith aluminum core is used. To counteractthe forementioned orthotropic nature of thecore, a slightly non-quasi-isotropic plyorientation was used, which resulted infinal CTE’s of -.02 ppm/F and -.06 ppm/Fin the plate X and Y directions. The ribplate X direction is parallel with thereflector surface (bonded edge).

The truss structure supports allcomponents, and attaches to an aluminumring attached to a gamma-alumina thermalisolation cylinder within the spacecraftbody. When integrated with the spacecraftpayload, four lift points permit the entireTRS and focal plane assembly to bemated/de-mated from the buss structure.With primary reflectors located high abovethe interface plane, the dual requirement ofhigh structural stiffness and low CTE ledPCI to choose a mortise-tenon designapproach. Two styles of corner joints,slotted and nested, are used to form thetruss elements. The TRS XZ plane is mostcritical for dynamic loading during launch.Given our weight, geometric size and

configuration limitations, typically, onewould incorporate increasingly highermodulus graphite for improvement to thefirst mode.However, increasing the fiber moduluscauses the CTE to become increasinglymore negative. In the TRS YZ plane, thelateral frequency requirements are lessstringent, and the primary concern ismechanical load. PCI’s design of the truss,taking in all factors, was to useXN70A/cyanate ester for all faces normalto the XZ plane, and M46J/cyanate ester forall faces normal to the YZ plane. Theresult is a hybrid truss using two graphitecomposite materials, each having differentstiffness, and with one having positive CTEand the other negative CTE.

Two large radiators are used to removeinternal electronic component heat andradiate it to space. These panels employ1100 series H14 aluminum facesheets and 2inch thick aluminum core. To save weight,the facesheets were chem-milled in fourthickness steps having a pattern equivalentto the heat flow distribution. The tworadiators represent over 30% of the totalTRS weight.

Materials Employed

The PCI approach to the TRS design was tomanage the thermal deformations throughmaterials selection and fiber orientation,manage the overall stiffness through shapecontrol of elements and stiffeners, andoverall weight control through optimizationof detail parts.

The list of materials used on the TRS isshown in Table 1.

4

Design Details

The TRS design is composed of four majorsub-assemblies. Primary reflectors,secondary reflectors, truss structure, andradiators.

Primary reflector shape is determined froma software package furnished by Princeton.Given a local x-y coordinate, the programgenerates the local z, resulting in a shapedsurface having no unique focal point. Frompoints generated over a .50 inch squaregrid, a surface was created using SDRC’sIdeas software. Accuracy of the generatedsurface to the theoretical was checked at alllocations, and was found to be within 5E-5inches. One tool was created for theprimary reflector, and another for thesecondary. Two quasi-isotropic facesheets,balanced and mirrored from the prepregstock were fabricated. The core is bondedwhile the cured facesheet is held to the toolwith vacuum. Subsequently, the core ismachined to the same shape as the tool, but

offset. Finally, the rear facesheet is bondedto the core.

The purpose of this design/manufacturingapproach is to negate any mechanicalstresses conceivably imparted by themanufacturing processes. Core bondingoccurs at the stress free temperature of thefacesheets. In spite of our best efforts tomitigate reflector warpage at its operationaltemperature, a temperature differential of –356C (–640F) from the bondingtemperature, the unsupported outboardreflector edges elastically deformed beyondspecification, and a hat shaped stiffenerwas subsequently added. The secondaryreflectors followed an identicaldesign/manufacturing path. There also, ahoop ring/radial rib structure was added tomitigate the operational thermaldeformations.

The truss structure presented someinteresting challenges. The TRS interfaceto the MAP structure is to the upper surfaceof a gamm-alumina cylinder having an

5

aluminum T shaped ring. The TRS isbolted to this T ring. At 70K, the ringshrinks .080 inches radially, which the PCItruss needed to accommodate. The designapproach selected is the use of four I beamshaped Titanium flexures. Being a hardinterface, all axial loads and bendingmoments are carried through the flexures atlaunch, while still retaining sufficientflexibility to accommodate relativedisplacements when at L2.

A mortise-tenon design approach for thetruss was selected due to thermaldeformation concerns and the need toestablish high structural rigidity. At thebeginning, PCI considered about 15 trussconfiguration possibilities. A popularchoice is to fabricate square tubes, and jointhem using bonded gussets and clips.Using multiple tooling, layups and curecycles, our fear was that all tubes would notbe identical, leading to grading, CTE

measurement and selection from multiplecandidates. Similarly, other competitivedesign approaches were discarded due touncertainties and manufacturing cost risk.

Ultimately, the mortise-tenon approachallowed PCI to fabricate one very large flatsheet of material, and CNC machine thetruss elements in a very predictablemanner. This approach also allowed theuse of two complementary materials andfiber orientations to solve the CTE andstiffness issues. Any truss element faceparallel to the XZ plane is quasi-isotropicM46J/RS3 with a slightly positive CTE,while any truss element face parallel to YZplane is XN70A/RS3 with a slightlynegative CTE. The XN70 faces are notquasi-isotropic, but a layup maximizing in-plane stiffness within the allowable CTErange. This methodology is commonlyreferred to as “athermalizing”. A break-away view of the truss is shown below.

6

Radiators are employed to maximize theefficiency of the Low Noise Amplifiers atthe feed horns. These large panels arefabricated using Series 1100 H14 purealuminum for the facesheets and standardhoneycomb core. To achieve the weightbudget, including the radiators, PCIselectively chem-milled the facesheets infour steps based on heat flow contour plotsprovided by Goddard. At all load/supportpoints, the .090 inch thickness wasretained, with .063, .040 and .020 thicknessin the outboard regions. A design

requirement was to thermally isolate theradiators from the remaining structure. Thelower support point uses G-10 flat stock asfitting tabs, aluminum lateral support tubesuse G-10 fitting inserts, and the uppersupport flexure is a folded back Titaniumsheet stock design. A maximum allowable.75 watts heat leak requirement was metanalytically and experimentally verified.

The main features of the TRS design, and anumber of the details are found in theisometric view shown below.

7

Mechanical Testing

The TRS assembly has been subjected tothese tests, representing all major phases offlight and data collection.

• Hot Soak to 100C (Vacuum)• Cold Soak to 40K (Vacuum)• Sine Burst (Ambient)• Sine Sweep (Ambient)• Acoustic Excitation (Ambient)

Hot Soak Test

The hot soak test was performed tosimulate solar heating effects during the‘barbecue’ roll mode during earth-moonphasing loops. This test also provides thedata to calculate a Cold Thermal TimeConstant, provide model correlation for onesteady state sun on the primary reflector,and support Bi-Reflectance DistributionFunction test results. Although the time atpeak temperature (100C) is a relativelyshort 10 minutes, concern was for visco-elastic realignment of the reflectors. At100C, the film adhesive used in thesandwich is at its glass transitiontemperature.

A large bank of heating lamps wasassembled at Goddard, and temperatureswere monitored so that the time-temperature profile would not be exceeded.Displacement measurement of the truss,primary and secondary reflectors, and theradiators was carried out usingphotogrammetry. The TRS met allrequirements, and no permanentdeformation or realignment of the structurewas measured.

Cold Soak Test

The cold soak test represents the simulationof temperature while in data collection

mode at L2. A large vacuum chamber atGoddard was used, with LHe as the coolingmedium within the shroud. Within thechamber, a rotating gantry assembly heldthe photogrammetry equipment, used tovalidate the surface accuracy and alignmentof the entire TRS. As a result of this test,the addition of the hat stiffeners at theperimeter of the primary reflectors wasdeemed necessary. The ‘sweet spot’ of theboth primary and secondary reflectors wereshown to meet specification of .0015inches rms when at this steady coldcondition (40K). A subsequent test at LN2temperatures proved the surface accuracyrequirements had been met. Asupplemental requirement at the 40Kcondition was that no point anywhere onthe TRS could deviate outside a .020 inchpositional location.Subsequent to the coldtesting, extensive visual inspection wasperformed. There were no anomalies ofany kind found.

Acoustic Test

Goddard has a very large acoustic chamber.Upon completion of the vibration testsequence, two acoustic runs wereperformed (-6db and protoflight). Testduration is 60 seconds for both levels. It isinteresting to be standing outside thechamber with the large steel doors closed.The impression is that one is at the launchpad.

This test was uneventful except oneaccelerometer became debonded. A photoof the TRS and test personnel illustrates thesize of the chamber.

The TRS is a stiff structure, and therefore,acoustic loading was not expected to excitelarge accelerations. RMS accelerations andforces from the test are captured in thefollowing table.

8

Sine Burst/Sine Sweep Tests

Sine burst and sine sweep testing wereperformed at Goddard. The TRS wasinstrumented with 19 accelerometers and 6

force gages, as shown in the followingtable.

In terms of strength qualification, theenvironment during the sine burst test ismore severe than during the sine sweep

9

vibrations. The levels were derived fromcoupled loads analysis, which involvedintegrating the TRS model with otherinstrument and spacecraft models to a forma full-up MAP observatory model. Loadsfrom the Delta 7425-10 launch vehicleflight environment were applied to thismodel, with force and acceleration outputscaptured throughout the TRS.

The levels are different in each axis. Usingaxis dependent loads was justified since theTRS load paths depend on load direction.Loading in the Y-axis qualified theinterface flexures, radiator flexures andstrut connections. The X-axis burstqualified the lateral strut/radiators. The Z-axis qualified the overall configuration atlift-off.

A picture of the TRS while on the shakerprovides some insight of the setup.

The typical test sequence consisted of a lowlevel sine sweep from 5 to 50 Hz at 4octaves/minute. Then, sine burst testingwas initiated at –6db and –3db levels, with

the final run level increased to reach therequired accelerations. A low level sinesweep was performed between each burstlevel. A final sine sweep at full level,followed by a low level completed thevibration sequence. This was done in eachaxis. These tests produced voluminousoutput data. A typical summary of theforce gage output is seen in the table.

10

Summary of Testing

The TRS has been subjected to testssimulating all phases of flight, and has beendeemed suitable for integration to the MAPstructure. As a consequence of these tests,minor changes to the design were required,which were incorporated by PCI while thestructure was at NASA/Goddard. Atriangular stiffener was necessary at thejunction of the truss apex and the primaryreflectors to increase a Y axis mode.Mechanical fasteners were added to thebrackets supporting the lateral struts,elevating peel stress concerns at theattachment bracket. Subcomponent testsconfirmed fatigue strength of the G-10radiator supports, flexure buckling strength,and flexure housing static ultimate strength.

PCI expects the TRS to perform itsintended function during its missionlifetime, and perform its function inproviding answers to the origins of theuniverse.