Datasets:

ai-aerospace

/

ams_data_full_2000-2020

like 0

Table 1. Micro wheel spec overview Power Momentum Torque Life Parameter MRW EMRW/SMRW Mass ~0.75 kg 4.1 kg Volume 1 OO*I OO*I 00 mm' 105*105*105 mm' Voltage 1 1-1 4 V (1 2 nominal) 24-32 V (28 nominal) 0.4 W (constant speed) 3.0 W (peak torque) 0.21 Nms 0.42 Nms 3 mNm lOmNm >I XI 0' revs 1.2 W (constant speed) 5.0 W (peak torque) >2.2x1oy revs SMRW Backqround / Development There were two main design changes to the SMRW from the EMRW: 0 0 The introduction 0; a cove;, deemed necessary due to increased harness packing around the spacecraft which lead to concerns of harness clashing with the unprotected inertia disk. The re-housing of the motor magnetics to increase the size and alignment of the bearings thus striving to increase the life. For reasons explained later in the paper with regards to ground testing the motor was partly reengineered to provide: a) a very low but repeatable bearing preload, thus low contact stress b) a supplementary angular snubber for launch vibrations c) a very high degree of bearing alignment Figure 5 shows a cross section of the SMRW. The motor is supported on a cone that attaches to the housing, which supports the electronics including optical encoder below. The motor houses the bearings above the magnetics, which was a change from the EMRW to increase the alignment accuracy as the EMRW bearings were supported at either end of the shaft in two different housings. Figure 5. Cross section of the SMRW The bearings are arranged in a back-to-back configuration using a deep grove type made from 440C stainless steel. The size was increased size from the EMRW SR4 type to an SR6 size, both of which are standard catalogue parts with relatively short lead times from European suppliers costing about 50 GBP each ($100). This is very affordable when considering that some bespoke bearings and cages can be orders of magnitude more expensive. 191

Document

AMS_2006.pdf

As previously mentioned the lubrication system is dry by means of a transfer cage. The cages are a standard snap on design using PGM-HT, which is a PTFE base using chopped glass fibers for cage strength and it is impregnated with MoS2. As small partials of the cage transfer to the balls and raceways they get squashed out leaving a thin film that provides the lubrication. The increase in bearing size was also necessary to increase the load capability as the launch environments on some new missions are increasing. There was also a change to preload the bearings (the EMRW bearings were shimmed to 50 microns with no preload) using a (standard catalogue part) crest-crest spring to apply a low preload, the spring also simplifies assembly due to the high deflection to low force characteristics of the spring thus designing in simple assembly and keeping the cost down. The motor magnetics were primarily kept the same but a magnetic overlap that provides some preload on the EMRW was removed thus removing the axial preload from the design so a lower preload can achieved to maximize the life. The electronics are nominally unchanged from the MRW consisting of a simple power stage utilizing commercial components and a main PCB which houses the optical encoder and can be seen in Figure 6. All the electrical components are readily available commercial parts, which do go through some increased screening at SSTL upon inspection on critical and delicate components such as the DC-DC converters, which have glass-isolating beads around their legs. Further detailed information on electrical commercial components can be found in [3], [4]. MOTOR E-PROM MOTOR ------- E-PROM HPWER STAGE OPTICAL ~~ ~~ Figure 6. Micro wheel electronics Qualification testinq The testing mentioned above and documented by ESTL clearly shows two effects. Firstly that using this class of self-lubricating material above a critical peak Hertzian contact stress of around 1200MPa results in very poor lubrication and should be avoided, Secondly, that below this stress there is a clear relationship between ball raceway peak Hertzian contact stress and lifetime which suggests that acceptably high lifetimes CAN be achieved with self-lubricating materials IF the contact stress between balls and raceways are maintained at a sufficiently low level. Furthermore, because of the well defined relationship between contact stress and wear rate (i.e. lifetime) it is possible to define a highly accelerated test program in which not only the speed is increased (say by a factor 5), but also an additional acceleration factor is obtained by increasing the bearing preload to some higher value than nominal for the application. In this way life test acceleration factors of order 20-30 can be justifiably achieved. 192

Document

AMS_2006.pdf

One additional benefit of using solid lubricants at low Hertzian contact stress is that the torque noise, which is in pari related to lumpy and uneven transfer of the cage material onto the raceways, is reduced. The disadvantages of using this approach are relatively modest; firstly some extra care must be taken during ground testing (at low preload the rotor mass and 1-g effects can be significant) and reduced angular stiffness of the wheel shaft during launch vibration. Given the improvements to the SMRW over the original unit, and by extrapolation of data from the design guide and other test programs, lifetimes of order 1 -2E10 revs were predicted to be achievable even in the absence of the improved bearing alignment. Two SMRW wheels have successfully passed the following qualification program: 0 Vibration (to 23.5 Grms) 0 Micro vibration 0 . Life test (to a factor of 2 on life in thermal vac performing hot / cold cycles)' 0 Micro vibration Including two different spacecraft operational scenarios, QM#1 nominal four-wheel operation in a 1' tetrahedral configuration biased at constant speed of 300 RPM with a higher preload (6.1 N) to allow accelerated testing, and QM#2 degraded spacecraft mode with three wheels biasing around zero and constantly crossing zero with a nominal flight preload of 2 N. Over a typical mission life of seven years in nominal mode this equates to 2.2 billion revolutions including a factor of 2. Taking into account acceleration philosophy mentioned above the life test only needs to run for a few months. The nominal operational mode also included operational scenarios from orbital maintenance maneuvers, which increase the wear on the cage due to the acceleration of the wheel. During these maneuvers the wheels are spun up from their nominal speed to around 4500 RPM to stabilize the spacecraft during thruster firings. They are then driven slowly back to nominal speed using the spacecraft's magnetorquer rods to dump the momentum. These maneuvers have been scaled to take into account the increased speed of the life test covering a representative number of revolutions at the correct rate of acceleration. This equated to performing approximately 1 slew cycle per minute at 3 RPM per second acceleration, Figure 7 shows the predicted profile. The degraded spacecraft mode with the wheels constantly crossing zero was performed nominally from +/- 100 RPM at approximately 2 crossings per minute for over 156 thousand zero crossing. The wheel was set up in a fully flight configuration including cover and the bearing preload was set at 2 N. During the life test, both wheels also demonstrated hot and cold survival (-30°C / +60"C) / operational 20°C / +50°C) cycles including start ups at the beginning and end of test at the temperature extremes. 1LW 193

Document

AMS_2006.pdf

Figure 7. Typical flight like orbital maneuvers imposed over the accelerated life test profile QM#1 (nominal oDeration) During vibration it is not uncommon to see responses at the inertia disk in the order of 60g. This relates to a peak Hertzian contact stress around 3500 MPa, which is pushing the bearing to its limits with the onset of sub surface plastic deformation at 4000 MPa. No visible damage was evident on any bearing surfaces ball or race as was seen with the MRW. Figure 8 shows a snap shot from the life test of the high slew maneuvers. There is some difference between the predicted acceleration rates, which is due to the way the wheel was controlled during the test. This actually gave the bearings a slightly harder time, as the higher acceleration rates increase wear. The wheel conducted over 1 13 thousand slew cycles at high speed with no anomalies. Figure 8. SMRW QM#1 Snapshot of life test data Micro-vibration was conducted on the wheels pre and post life test to investigate the size of the effect cage degradation has on the micro-vibration, a plot from post life test can be seen in Figure 9. The bearing fundamental frequencies such as cage / ball frequencies can be seen tracking as predicted up to about 350 Hz and showed a small increase in amplitude. The main increase in noise was in two frequency bands, 400-500 Hz and around 1000Hz bands where there is a slight increase in amplitude. This ties in with modes of the inertia disk around 400Hz and 1000 Hz that can be seen from vibration sine sweeps. 194

Document

AMS_2006.pdf

Analysis of the jitter testing is still ongoing with spacecraft system models to gain a better understanding of the effects of wheel micro vibration through the spacecraft with particular interest to imaging missions. BEARING FUN DERMENTAL FREQUENCIES aao 3.5 1 0 'Shp Figure 9. Waterfall plot showing coast down micro vibration post life test QM#2 (Declraded mode) The wheel operated nominally throughout the life test with no anomalies performing over 130 thousand zero crossing. Post life test insDection The bearings were stripped down, inspected and the cages weighed pre and post life test to ascertain the amount and characteristics of the wear. Photos from the strip down can be seen in Figure 10, with even wear around the pocket, clear and even transfer of PTFE material onto the race ways. Both QM1 (6.1-N preload) and QM2 (2.3 N) exhibited very low cage wear on successful completion of the life tests carried out compared to that expected. Typically the cage mass loss was of order 1% or less of the QM1 initial cage mass (with on-loaded bearing cage wear rate being higher than that operated at the nominal preload). For QM2 the measured cage wear rate was still lower, in fact on the threshold of measurement for QM2. According to the lifetime model, and previous experimental data, QM1 bearing cages could have been expected to have lost up to 15% of their cage mass due to wear (0.1g) by the end of the accelerated life test. The excellent condition of the cages in both QM1 and QM2 suggests that the improved bearing alignment achieved has a large impact on the wear rate observed comparing to the EMRW life test. Extrapolation of wear rates is slightly hazardous because typical wear processes follow a so-called "bathtub" curve with high initial and final wear rates but a long period of low wear rate equilibrium performance. Given this a linear extrapolation of mass is likely to under-estimate lifetime, but using such an approach, it may be concluded that the ultimate lifetime of the bearings in the SMRW wheel may be considerably in excess of 1 El 0 revs, that is permitting a lifetime with margin of in excess of 16 years at 300 RPM. Clearly however the ultimate lifetime achievable will need to be determined by further testing/flight experience. The torque performance of the bearings during the life tests was also very good. After the initial run-in, the torque noise generated improved with lifetime. For example in QM1, the Standard deviation torque 195

Document

AMS_2006.pdf

noise during reversals was around 2.5 gcrn initially and slightly improved, to a value of 1.8 gcrn by completion of testing. hearing wiih the shield mrnwed Cage shaving wear fines in the centre of pocket Figure 10. Close up photo of the wear of one of the SMRW QM#1 bearing cage after life test Conclusion Both SMRW life tests were successfully concluded with both wheels meeting and exceeding their requirement; lifetimes substantially in excess of these values may well be achievable. No SSTL wheel has had a mechanical failure on orbit. All wheels are either still working or were working nominally when the spacecraft was decommissioned. To date the MRW and the SMRW (through on ground qualification) have outlived spacecraft life and have still been operating nominally when the spacecraft has been de-commissioned proving self lubrication is fit for this type of wheel application. The EMRW, SMRW and SRW covered in the paper are available commercially from SSTL under the trade names Microwheel and SmallWheel. References 1. Ball bearing tests to evaluate duroid replacements - ESTL 2. Performance guide; self lubrication bearings, NCT Guide, 1976-ESTL 3. 30 Years of Commercial Components in Space: Selection Techniques without Formal Qualif ication-SSTL 4. 25 Years Experience With Commercial Components in Space - SSTL, California Paper 196

Document

AMS_2006.pdf

Design of the ATMS Scan Drive Mechanism Curtis ~llmon' andpave Putnam Abstract The Advanced Technology Microwave Sounder (ATMS) scan drive mechanism is a torque-compensating single-axis dual-mirror gimbal assembly. The scan drive mechanism will fly as part of ATMS on both NOAA's NPOESS (National Polar-orbiting Operational Environmental Satellite System) and NASA's NPP (NPOESS Preparatory Project). The ATMS, a weather monitoring instrument under development by Goddard Space Flight Center, measures microwave energy emitted by the atmosphere which aids weather forecasting. The material covered in this paper will focus on the mechanical design of the scan drive mechanism. The topics covered include the design features of the scan drive mechanism and the methods used to minimize the transmitted torque disturbances from the scan drive mechanism to the spacecraft. Introduction The ATMS scan drive mechanism is a single-axis continuous-rotation gimbal that feeds microwave frequency data into the ATMS instrument. The microwave data generated by the scan drive mechanism is used to develop layered maps of the Earth's atmosphere by measuring the temperature and humidity at different altitudes. Figure 1 shows the orientation of the scan drive mechanism and ATMS instrument in flight and a composite plot of the temperature layers showing a developing tropical storm. Figure 1. ATMS scan drive mechanism flight pattern and data from earlier generation AMSU showing a tropical storm developing in November 2005. The ATMS Scan Drive Mechanism is the 3rd generation scanner to provide atmospheric temperature and humidity data from space for weather prediction. The ATMS scan drive mechanism is the third generation in a family of scan drives that reaches back to 1978. The ATMS scanner offers a three times improvement over the current generation AMSU scanner in number of scans per orbit. For comparison, the first generation MSU scanner rotated one scan or one revolution over 25 seconds. The AMSU second generation scanner rotates one revolution every 8 Lockheed Martin Space Systems Company, Sunnyvale, CA Proceedings of the 38'" Aerospace Mechanisms Symposium, Langley Research Center, May 17-19,2006

Document

AMS_2006.pdf

seconds, and the ATMS scan drive mechanism rotates one revolution once every 8/3 seconds. To achieve the improved scan rate, the ATMS scan drive mechanism inserts three periods of acceleration and deceleration into each revolution as shown in Figure 2. 1 . Nadir Earth scene sector I, AB c I Figure 2. Scan Drive Mechanism Scanning Pattern: When in scanning mode, the Scan Drive Mechanism rotates at a constant velocity through the Earth scan and the Cold and Hot calibration scans. It rotates with constant acceleration or deceleration between these scans As with every new design, there were things done very well in the scan drive mechanism and things that could have been improved upon if time permitted. In the category of things that were done well were the packaging design, the design for manufacturability, and the disturbance torque minimization features. One of the things we wished went better was the EMlhadiated susceptibility design and in a middle category (some good, some bad) was the bearing procurement and testing. Scan Drive Mechanism Design Packaaina and Overall Desian. The overall envelope and packaging of the scan drive is in the category of things that went well. As can be seen in Figure 3 the scan drive mechanism needed to fit in a very compact space on the ATMS instrument. The space between the two reflectors had to house two motors, two resolvers, two pairs of bearings, and a flywheel. The details of packaging the components in this space are illustrated in the Figure 4 cross section. One unique feature that aided in packaging the gimbal in this small space was the use of single-string (non-redundant) motors and resolvers. The single-string main and flywheel motors were driven by redundant wiring and electronics but by making the motors non-redundant we were able to achieve higher torque margin in a smaller package than possible with a redundant winding design of the same size. The motors were also an ironless core construction, which used a significantly thinner stator than a conventional brushless DC motor. As shown in Figure 4, the scan drive mechanism is made up of main and compensating subassemblies. The main subassembly rotates the reflectors while the compensating subassembly limits the disturbance torque into the instrument by driving a flywheel in the opposite direction. The scan drive mechanism can be driven in both a compensated and uncompensated mode. In the uncompensated mode, the flywheel is not powered and torque compensation is eliminated. Weight was another key requirement that went well. The weight and other requirements for the scan drive mechanism are given in Table 1. To achieve the required weight, the reflectors and the main and compensating subassembly housings were made of beryllium. The reflector material selection was important for weight as well as inertia. The spherical shrouds around each reflector were made of 6061 -T6 198

Document

AMS_2006.pdf

aluminum. Other materials were considered for the shrouds but aluminum was selected for both for weight and manufacturability. Figure 3. The Scan Drive Mechanism mounted on the ATMS instrument illustrates the packaging challenge of this design. Two motors, two resolvers, two pairs of bearings, and a flywheel must fit in the center section between the two spherical shrouds. Desian for Manufacturabilitv. The scan drive mechanism design not only met very tight packaging goals and weight requirement but was created as a modular assembly that allowed assembly, disassembly (when required), and test as subassemblies. This approach provided schedule and work load flexibility during assembly and test. The modular design of the scan drive mechanism can be seen in Figure 5 where the main motor subassembly and compensation flywheel subassembly are on the work bench together. The subassembly approach also allowed for the separate balancing of the main assembly with the reflectors and the compensating assembly with the flywheel. Fabrication of the spherical shrouds and a repeatable method of assembling and disassembling the shroud halves with the scan drive mechanism was another important manufacturing challenge. The shrouds attach to rings that have floating nut plates and locating pins for repeatable reattachment of the shrouds. The shrouds are split along a horizontal seam to allow removal of the upper shroud half during reflector alignment verification (see Figure 6). The shrouds are rough machined out of one piece of aluminum and then split along the horizontal seam. The spherical shape of shrouds provided a maximum view factor for each reflector and covered the warm calibration source inside the ATMS during earth scan. 199

Document

AMS_2006.pdf

- 254 mm 7 Beryllium KAV Reflector 155ph Flywheel 397 mm I Green = Fixed Assy Red = Main Rotating Assy Bke = Cornp Rotating Assy I Figure 4. Cross Section of the Scan Drive Mechanism illustrates compact packaging and materials selection used to meet weight and envelope requirements Reflector Alianment The reflectors mounted to the main shaft with titanium spiders that allowed axial and rotational alignment. The spiders attached to the shaft using tapered square holes to ensure repeatability of position during successive assembly and disassembly operations. In order to meet the required alignment, as shown in Table 1, there were shims between the spider and the reflector. The shims were pre-machined in thickness increments of 0.01 mm (0.0005 in). This shim kit allowed for rapid alignment of each reflector to the spin axis on a coordinate measuring machine (CMM). The CMM was used to define the true angle of the reflector relative to the spin axis by defining multiple planes whose normal vectors defined a cone. The upper half of the split shroud was removed during alignment to allow full access across the entire reflector. Radiated Susceptibilitv. The cable harnessing and EM1 design on the scan drive mechanism were things that fell into the category of things-not-done-so-well. Cabling is often one of the last features considered by mechanical designers and the scan drive mechanism was no different. The scan drive mechanism has two harnesses with connectors that extend a short distance out of the gimbal. Inside the gimbal the harness travels on the surface of the gimbal housing inside a cable tray. The harness was originally sheathed with braided shielding and the surfaces of the cable tray and gimbal housing were coated for conductivity. Despite these seemingly normal precautions, the scan drive mechanism dramatically failed the first radiated susceptibility test. Instead of providing a required 60 dB of attenuation in the 1 to 4 GHz frequency range, the scan drive mechanism harness and shielding were effectively acting like an antenna. 200

Document

AMS_2006.pdf

Table 1. Requirements and capability for the ATMS Scan Drive Mechanism - to achieve a 3x increase in the scan rate, the single-axis SDM is accelerated and decelerated between measurement and calibration scans - Parameter Earth View Sector Control Accuracy +f- 35mdeg ~0.003 N-m between 0.01 and 1 .0 Hz c0.1 N-m above 33 Hz Disturbance torque into spacecraft Torque capability Ifunction under 3X worst case drag Life [continuous service) (11 years Mass 14 0.9Kg Average Power 14 3W There were numerous causes for the radiated susceptibility failure: 1. The beryllium housing halves and cable tray were attached with fasteners spaced too far apart and there were no provisions for EM1 gaskets (the gaps were radiation sources). 2. There were no provisions for terminating the harness shield to the cable tray or to the entry into beryllium housing necessary to make a continuous Faraday cage. 3. The beryllium housing mesh air vent was installed with non conductive RTV - an ungrounded air vent can propagate RF energy. 4. The harness inner shield was tied to the outer shield inside the connector backshells - both shields terminated at one end can act as an RF antenna. 5. Motor and resolver harnesses were routed together providing coupling paths inside the harness. Unfortunately these short comings were not uncovered until late in the testing program and schedule did not permit major design changes to fully correct the problem. The fixes as shown in Figure 7 consisted of caulking the cable tray and housing halves with silver-filled epoxy, using silver-filled epoxy to ground the vent screen, and wrapping the harnesses with copper foil over the braided shielding. These fixes fell short of the required 60-dB attenuation, but fortunately the requirement was lowered to 40-dB attenuation and the reworked scan drive mechanism passed the radiated susceptibility test. The painful lesson from this experience is that every wire harness and EM1 task is not necessarily similar to the last design and a quick discussion with a subject area expert can save weeks of time. 20 1

Document

AMS_2006.pdf

.-.. . . ' '..? .h , . 1_, . \r , '., . f. I Figure 5. Torque compensation flywheel. The torque into spacecraft due to accelerating and decelerating the reflectors is compensated for by active control of the counter rotating flywheel. The flywheel inertia is lower than the inertia of the rotating reflectors so the flywheel is accelerated at the inertia ratio times the changing rotation rate of the main rotating assembly. Figure 6. Upper half of Scan Drive Mechanism shroud is removed to verify alignment of reflectors and rotation center before and after each environment 202

Document

AMS_2006.pdf

Gimbal Bearina Handlinq The manufacturing handling of bearina toraue characterization the gimbal bearings was good news, bad news story. On the plus side, tests were implemented as part of the manufacturing plan to detect torque irregularities. On the down side, the bearings did not have sufficient cleanliness controls. The bearings in the scan drive mechanism were designed for a life of 130 million revolutions for the reflector shaft bearings and 230 million revolutions for the flywheel bearings. Cleanliness of the bearings and lubricant was absolutely essential to achieve bearing life and smooth torque performance. Fortunately, we had a bearing torque test after the bearings were installed in the gimbal and on the third scan drive mechanism we found a noisy torque trace on a flywheel bearing duplex pair and on closer inspection we found debris in the bearing. The most likely sources of debris were: 1. Metallic debris was found and was most likely from tooling used for torque testing at the supplier. Inspection of the supplier tooling found metallic debris in threaded holes on the tooling. 2. White fibers were found and were most likely from cleanroom wipes or hair bouffants. Although disassembly and replacement of the bearing was a painful schedule hit, the bearing contamination was caught before entering acceptance test. None of the other gimbal bearings ever showed erratic torque performance and a recommended cleanliness program was implemented by our bearing supplier to eliminate future contamination. Figure 8 shows our bearing torque set-up and samples of the debris found in the bearing. m Air Vent , ”.( Figure 7. During radiated susceptibility testing in the 1 to 4 giga-Hertz frequency range, the scan drive mechanism harnessing acted as an antenna causing interference with the ATMS receivers. Various fixes were implemented to reduce the interference but adding the right EM1 features earlier in the design process would have worked better. 203

Document

AMS_2006.pdf

A Figure 8. Contamination was found in the flywheel bearing of flight unit 3. The in-situ bearing torque test was invaluable at identifying bearing problems during the assembly of the gimbal. Fortunately disassembly features were built into the design of the scan drive mechanism. Disturbance Toraue Reduction. Torque reduction was an aspect of the scan drive mechanism that went well. The level of torque disturbance allowed by the scan drive mechanism as shown in the Table 1 is less than 0.027 in-lb (0.003 N-m) across the 0.01 to 1 .O Hertz frequency range. To achieve this level of quiet operation, a number of disturbance reduction techniques were required. The first was to use a counter-rotating flywheel to compensate for the torque disturbance caused by accelerating and decelerating the reflectors three times each scan rotation. Other techniques used to achieve quiet operation included accurate balancing both the main antenna rotation assembly as well as the flywheel, and careful selection of parts including low run-out bearings and low cogging motors for both main and flywheel assemblies. One of the primary reasons for designing the scan drive mechanism as two subassemblies was to allow access for balancing the main reflector rotation subassembly and flywheel compensation subassembly. In Figure 5, radial and axial threaded holes for balance weights can be seen on the flywheel sFbassemblr The balance weight holes were filled with set screws as required to achieve a 80,935 g-mm (4.4 oz-in ) dynamic balance and 216 g-mm (0.3 oz-in) static balance. Figure 9 illustrates the features on each reflector shroud that allowed balancing of the main shaft and reflector subassembly. The main subassembly was difficult to balance due to the cutout in the shroud for reflector viewing. The cutout was balanced by placing a tungsten weight on the hole side of the reflector assembly. Several other smaller masses were added around the shrouds to account for other inconsistencies in geometry. The main subassembly was balanced to the same level as the flywheel subassembly. 204

Document

AMS_2006.pdf

I Optional balancing set screw locations avsate for 'n top of IT (11 10 Note Reflector rn+ shown Figure 9. To minimize uncompensated torque disturbances, balancing features inside each shroud allowed dynamic and static balancing of the rotating main assembly. One of the unique features of the scan drive mechanism was the counter-rotating flywheel used to minimize the torque generated by accelerating and decelerating the reflector each scan cycle. The scan drive mechanism acceleration/deceleration pattern (Figure 2) shortened the scan period and increased overall gap coverage but generated undesirable torque disturbances. The counter-rotating flywheel cancelled 92% of the torque from the reflector acceleration profile. The success of this approach is shown in Figure 10 where the scan drive mechanism was run both with and without the flywheel compensation. Pointinci Performance The final link in the scan drive performance was the earth scan pointing. Pointing performance success was a combination of mechanical alignment and motor and resolver selection. As noted the brushless DC motor was a low-cogging ironless-core design which resulted in very low torque ripple. The main motor and reflector assemble position was controlled by a 64-speed and single-speed brushless resolver. The resulting pointing performance is shown in Figure 11. 205

Document

AMS_2006.pdf

Compensation Motor Off Compensation Motor On - -- ~ .._ ~ __-- -* Jc 1 -- .o; ...... __ .. Figure 10. Reaction torque cancellation is successful with the compensation flywheel motor on. The Scan Drive Mechanism can operate in reduced mode (greater torque disturbance) using only the main motor. I I I I I I 0 os 1 15 2 2s 3 5.5 4 45 5 -0.8 I €am %en I , I I I I 1 I Figure 11. The Scan Drive Mechanism meets the ultimate test of pointing accuracy for earth scan as well as during hot and cold calibration. Acceleration/deceleration periods offer the ability to provide more earth scans per orbit. 206

Document

AMS_2006.pdf

Acknowledgements Every successful major project results from the combined efforts of many people. It is with great regret we don’t have the space to list all of the many people who contributed time, effort, and talent to the ATMS Scan Drive program. None-the-less, most of those who participated in the Scan Drive program would agree that the following people were key to the success of the mechanism part of the program. Goddard Space Flight Center: Sergey Krimchansky, Robert Lambeck, Rick Schnurr; Northrop Grumman Electronic Systems: Dennis Lord, Terry O’Brien; Lockheed Martin: Ed Boesiger, Caesar Ching, Jeff Fisher, Patrick Herbert, Stu Loewenthal, A.J. Maher, Larry McGovern, Gordon Smith, Nic Mercer and Julie Price. References 1) Advanced Technology Microwave Sounder on NPOESS and NPP, Christina Muth, Paul Lee, Sergey Krimchansky, James Shiue, Allan Webb. NPOES IPO Information center, No. 06125,06- 1 6-2004, http://l40.90.86.6/1POarchive/SCI/sensors/ATMSDraftBriefv4_AlanW ebb. pdf 2) AMSU temperature plots: http://Rm-esiD.msfc.nasa.aov/amsu/ 207

Document

AMS_2006.pdf

Lessons Learned From the Windsat BAPTA Design and On-Orbit Anomalies Steve Koss’ and Scott Woolawaf Abstract The Windsat spin drive assembly is a high precision constant speed drive assembly with challenging design requirements. Many difficulties were encountered during the design and build of the spin drive assembly. Additionally, after 2 years of nominal on-orbit operation the Windsat spin drive assembly spun out-of-control, resulting in a temporary cessation of the W indsat/Coriolis mission. After a lengthy investigation, telemetry analysis revealed that the resolver based tachometer data was likely being corrupted by two distinct kinds of slip ring noise. Workarounds were devised to successfully bring W indsat back on-line, and another similar spin anomaly was recovered from 5 months later. This paper will concentrate on the many lessons learned from the Windsat spin drive development and subsequent on- orbit anomaly. Additionally, the goal of the paper is to convey the “key things that were done well” as well as the “things that should be done different next time” (good vs. bad heritage). introduction Windsat is the primary payload on the Coriolis spacecraft. The Windsat payload size is roughly 2-meters square by 3-meters tall with a mass of 450 kg. Windsat is a demonstration program to evaluate the ability to exploit passive microwave polarimetry to measure the full ocean surface wind field (wind speed and wind direction) from space. This payload is very similar to the NPOESS CMlS payload currently under design. The Windsat payload instrument spins continuously at 31.6 revolutions per minute via a spin drive mechanism called the Bearing And Power Transfer Assembly (BAPTA). For accurate geolocation, the BAPTA is a precision device with a spin rate accuracy of 4.05% and a position feedback accuracy of i Figure 1. WindsatICoriolis (left) and Wind Speed/Direction Data (right) f. U.S. Naval Research Laboratory, Washington, DC Ball Aerospace Corp, Boulder CO Proceedings of the 3@ Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 209

Document

AMS_2006.pdf

20.0055 degree. To provide a zero momentum payload a 189 Nms momentum wheel spins in the opposite direction of the payload. During operation, the Windsat payload and BAPTA speed control is “slaved to track the momentum wheel speed. This system performed nominally for 2 years on-orbit (33 million revolutions), until February, 2005 when the BAPTA abruptly spun out-of-control. BAPTA / Slip Ring Design Overview The BAPTA design (Figure 2) incorporates a “bolt-on” slip ring assembly which provided ease of integration & test and “modular functionality” of the separate units. The BAPTA was designed with a redundantly wound three phase brushless DC torque motor and redundant dual speed (1X & 64X) resolvers. The resolvers provided 18-bit position resolution and better than 16 bit accuracy after conversion via a resolver to digital converter. This also enhanced the servo “stiffness” to maintain the tight 20.05% speed control required. Two angular-contact machine spindle bearings supported the aluminum- beryllium rotating housing, resulting in a very stiff, robust structural design. Figure 2. BAPTNSlip Ring Assembly The slip ring performance was the most challenging aspect of the BAPTA. The design life for Windsat is 3 years at 31.6 rpm, which translates to 50 million revolutions. Many spacecraft slip ring applications are for solar array drives that have rotational life requirements several orders of magnitude lower. The Boeing 376, NGST/Ball DSCS II spun/despun satellites, and a few others, have the only slip rings with life requirements in this class. The Boeing slip rings use silver rings with silver/MoS2 brushes, whereas the DSCSll used a gold-on-gold slip ring design. However, unique to Windsat was the requirement to pass high-rate 1553 digital data and resolver transmitter signals across the slip ring, which are much more sensitive to high-frequency slip ring noise than these previous applications. Much life test data existed for slip rings, but little if any high frequency noise data existed for slip rings at the time of Windsat design. Fortunately, a 1968 vintage DSCS II program slip ring unit was found that had been through a 200 million revolution life test in the 1970’s. This unit was tested in 1999 with a comprehensive slip ring test set, designed to monitor & record high-frequency slip ring noise. After several hours of noisy operation, the brushes pushed the significant amounts of wear debris to the side (Figure 3) and the slip ring noise went away. This test data gave us confidence that this type of slip ring would work, while another life test on a new unit closer to the flight design was run in parallel to the fabrication of the flight slip rings. The Windsat & DSCS II slip ring designs are of the oil-lubricated, gold-brush / gold-ring variety, but differ from any other gold-on-gold designs. The Windsat slip ring design consists of hard gold alloy brushes made from wire that is swaged to a rectangular cross section (Figure 4). The brushes were swaged flat instead of round in an attempt to accommodate more wear debris under the brush. Since Windsat & DSCS II spin in only one direction, a pair of trailing rings sit in a “W” shaped contact groove compared to the more common bidirectional “stubbing” & trailing round wire in a ‘‘V’ groove configuration. The “trailing only” design is less prone to brush bounce compared to a “stubbing” design. Additional details on the DSCSll slip rings can be found in reference 1. Since there was great concern for slip ring noise problems, 100% parallel redundant rings were used for all signal rings resulting in 4-8 contacts per wire (4 vs. 8 contacts 21 0

Document

AMS_2006.pdf

depending on how the brush might "lift off" due to debris). Additionally, unlike the typical gold plated rings, the Windsat rings are solid soft gold with a hard gold flash coating on top (Figure 4). post Lire I ( Figure 3. Windsat / DSCS Slip Rings Figure 4. Windsat / DSCS Slip RinglBrush Contact Cross Section Design & Manufacture Lessons Learned There were numerous lessons learned during the design, build, and test phase of the Windsat BAPTA and slip rings. What will be discussed are the "key things that were done well" as well as the "things that should be done different next time" Test Validated Toraue Marains The performance test program for the Windsat BAPTA required it to meet its rate and position accuracy requirements when operating under a 4X worst-case drag torque. This test was performed by measuring the cold drag torque (worst-case) and driving the BAPTA against an open loop torque motor (acting as a brake) to produce the 4X drag torque. The BAPTA was found to function well under this load but the increased BAPTA motor current caused the resolver based tachometer feedback signal to get noisy. The noise was due to secondary power supply feedback into the resolver-to-digital chip under high load. Adding an RIC filter and a Zener Diode circuit between the secondary power supply and the resolver-to- digital chip eliminated this noise. By performing performance testing at worst-case torque, this problem

Document

AMS_2006.pdf

was fixed on the ground allowing us to fully utilize our torque margin with full closed loop servo performance. It is common to demonstrate margin by analysis only or via friction torque measurements but less common to verify closed-loop servo torque margin performance. Onlv the closed-loop servo toraue marain test performed would have cauaht our resolver noise problem. Additionally, the large margins built into all aspects of the design helped in the decision to fly “as-is” with numerous problems/defects that will be described hereafter. Desian/Test For Indefinite Stall At Hot TVAC The NRL standard test program for the Windsat BAPTA required testing the BAPTA under stalled conditions at worst case hot thermal vacuum until the BAPTA reached thermal equilibrium. Designing the BAPTA to survive these conditions and performing this test gave us the confidence that it was extremely difficult to damage the BAPTA by mistakes or anomalies on the ground or on-orbit. As point of fact, during the Windsat on-orbit anomaly the BAPTA drive motor was hammered back and forth with full stall current at 13 Hz for 15 minutes with no damage. Don’t Put Precision Feedback Device (Resolver) In A Load Path The Windsat BAPTA resolver was mounted at the end of the BAPTA, which supported the cantilever, mounted slip ring assembly. As such all cantilevered slip ring launch loads were transmitted through the resolver mounting structure. While this structure was capable of handling these loads, the resolvers are precision bonded to this structure and some slight settling of the resolver position was seen after the first BAPTA vibration and thermal vacuum testing. After this initial “set” the position did not drift noticeably and was thus more of a “headache” than a real problem. The layout of the Windsat BAPTA does not leave many optional locations for locating the resolver (Le., out of a load path). However, for future designs consideration should be given to placement of precision feedback devices relative to load paths. Blind Holes, Filtered Vents & Labvrinth Seals Prevent Contamination Problems - Both Goina In And Cominq Out - Especially On Slip Rinas The Windsat BAPTA and slip rings had a design requirement for blind holes whenever possible and filtered vents and/or labyrinth seals to prevent contamination problems. Many drives have suffered anomalies due to contamination in bearings or air gaps. This is especially critical on designs with slip rings. Slip rings with the life requirements of Windsat create significant amounts of wear debris. Managing this debris and keeping it out of bearings, air gaps, (and shorting rings) it a crucial part of the design. In the Windsat design the slip ring assembly had labyrinth seals that kept the slip ring debris contained within the slip ring and away from bearings, motors, and resolvers. Slip ring barriers were made with heights large enough to prevent debris buildup ring shorts. Additionally, all power + rings were grouped together and all returns were grouped together to minimize potential shorting paths. On the signal side we had alternating “odd & even” side brush blocks for consecutive signal wiring and ring assignments of shield, +, -, shield (4 rings per twisted shielded pair, Figure 5) to prevent adjacent circuit shorting and minimize noise and crosstalk. A final lesson was learned from inspecting the old DSCSll slip ring life test unit, which was full of wear debris. It was found that the oil impregnated reservoirs used to help replenish the relatively high vapor pressure oil used in the 1970’s had a side benefit of acting as a “fly trap” to collect a thick layer of wear debris away from the critical brushlring interface. Thus while the new low vapor pressure oil used on the Windsat slip rings did not need oil reservoirs, they were maintained for their “wear debris management” function. Figure 5 - Signal Rings - Twisted Shielded Pairs 21 2

Document

AMS_2006.pdf

Motors - Not Challenaina But Acts Like a “Maanet” For “Mundane” Problems For various reasons spaceflight motors seem to have an unusual amount of minor manufacturing problems, which at best impact schedule and at worst impact reliability. As such, special attention should be paid to the motor contract and it is strongly recommended that an extra thorough pre-ship test/inspection/documentation review be conducted at the motor vendor prior to accepting motor delivery. On Windsat, motor manufacturing defects drove us to procure motors from 2 different vendors and neither vendor produced defect-free motors - in the end we flew the motor that had the least defects. A listing of problems that have been encountered with both the Windsat BAPTA motors as well as a few on other programs follows: 1) Windings potted in Stycast thermally conductive epoxy. Stycast cracked (Figure 6) after thermal cycling as a result of CTE mismatch between Stycast & copper windings. Resulted in X-rays of motor to look for crack induced strain on windings. Also resulted in “painting” the Stycast with Urelane in an effort to contain any pieces of Stycast that might pop off and jam the air gap or bearings. On the Windsat BAPTA, we ended up flying a motor that had the least cracks in the Stycast. While Stycast winding encapsulant may not cause a problem on small motors, on large motors like the 147-mm-diameter Windsat BAPTA motor it is probably best to avoid potting windings in Stycast. 4) Figure 6. Fine-Focus X-Rays, Cracked Stycast (left), Nicked Hookup Wires (right) Lamination stack painted with “fluidize”. Problems with fluidize adhesiodflaking off. Riskkoncern that fluidize flakes could get into bearings. Magnets or non-corrosion resistant backiron nickel plated. Problems with nickel plating adhesiodflaking off. Problems with epoxy washcoat bubbling/peeling off. Riskkoncern that platingkoating flakes could get into bearings. Use corrosion-resistant materials to the greatest extent possible to minimize need for plating & epoxy coating. When coating is required perform tape test & detailed inspections. Hookup wire strain relief problems. Various instances of insufficiently designed strain relief in multiple vendorddesigns wherein wire insulation damaged (Figure 6) or wires broken. Strain relief is something that rarely gets enough attention. Problems with air gap tolerances. Tolerance stack analysis required both at room temperature and worst-case temperatures. Analysis should be verified by test/measurement. Problems with motor cleanliness. Particulate contamination, which could get into bearings & reduce lubricant life. In extreme cases debris in tapped/helicoiled holes, which could jam, air gap or bearings. (Watch out for helicoil tangs, small tap pieces, etc) 21 3

Document

AMS_2006.pdf

7) Finally, while not necessarily a problem, a very faint motor ‘’ticking” sound was found wherein the tick corresponded to the commutation frequency of the motor (6-step commutation) and the intensity of the sound was proportional to motor current. This was first noticed during Windsat EM1 in the “soundproof” EM1 chamber. The drive had been listened to with a mechanics stethoscope earlier as standard practice but the faint tick was not noticed. However, the stethoscope test was done in a clean room next to the air handling system so it is quite possible the “tick sound was always there but not noticed until the dead quiet EM1 chamber. The tick might have been just the magnetostrictive contraction of the armature and normal - it is not known for sure. The ticking sound was tracked through system level vibe, TVAC, etc and it never changed and the source was never isolated. A lesson learned is to turn off all noise sources (air handlers, etc) when listening to something and to use a microphone to provide an audio record that is subsequently analyzed by sound /frequency analysis software if possible. Value Of Screenina Toraue Traces On Bearinas - EsDeciallv Thin Section The Windsat slip rings were built as a standalone assembly with their own bearings. This was done for ease of integration and to minimize potential conflicts with the slip ring integration and test schedule. The bearings used were angular contact thin section (50.8-mm (2) ID, 63.5-mm (2.5”) OD, 6.35-mm (0.25) cross section, 42 balls). Since the slip ring vendor had limited bearing design/analysis capabilities, NRL helped them with bearing design and analysis. Additionally NRL provided the bearings and assisted with installation. During bearing screening torque tests conducted at NRL prior to installation, several bearing pairs were rejected as a result of torque spikes at the bearing cage frequency (Figure 7). The bearing had very thin “rubber band like” one-piece phenolic retainers, which were difficult to dimensionally control. Many of the cages were out-of-round and would pinch the riding land of the bearing, resulting in torque spikes at the cage frequency. The torque screening tests ensured that these bearings were rejected and after several iterations acceptable bearings were obtained and flown. I HSG S/N 22 Vettical -30 RPM 20:47 PM 0246 8 10 12 Time (=e) Figure 7. Screening Torque Trace (left) & Thin Section Bearings (right) Analvsis & Life Test Data For Couplinas Cannot Be Overlooked The Windsat BAPTA design used two different bellows shaft couplings. Shaft couplings are often a “necessary evil” when multiple shafts must be connected and small amounts of misalignment would result in large/unacceptable bearing loads. On the Windsat BAPTA the shafts in question were aligned with high precision, however, there was no solid analytical prediction for coupling life. One of the couplings was a “catalog” design with a published fatigue life/misalignment capability. When the vendor was pressed to back up the published data, however, it was found that it was extrapolated off tests of the first few couplings made 50 years ago and that “you really should run a test”. A life test was run on one of the couplings with the flight misalignment duplicated. The life test was stopped after the required design life of 3 years / 50 million revolutions as it was deemed unlikely that a fatigue failure would occur past that point (and because the life test gearbox wore out). There are many different styles/designs of shaft couplings, the more common being Oldham and Bellows. The Oldham can wear and have backlash issues whereas the bellows can fail by fatigue (either in vibration testing or operation). An assessment of the life-limiting 21 4

Document

AMS_2006.pdf

impacts on the design should be done, and if necessary, a life test should be conducted. The program did not have the time or funds to conduct a life test on the second bellows coupling used and the program accepted that risk. This has been borne out by the fact that the BAPTA has so far seen over 40 million revolutions to date without a failure. Manv Commercial Slip Rina Vendors Can Use Help When Desianina For Space. Especiallv When it Comes To Thermal Desian 81 Bearinas There several slip ring vendors being used for space in the USA. All of the slip rings vendors are primarily “commercial” slip ring vendors wherein space is a sideline business. As such, it is recommended that special attention/oversight be given to the slip ring vendor to ensure the unique design and reliability aspects related to space are satisfied. For instance, none of the slip ring vendors have special expertise in bearing design, or detailed thermal analysis that is typically required for long-life high reliability space applications. As such, NRL and Ball assisted the slip ring vendor with the bearing design and purchased, lubricated, screened, and helped install the bearings. NRL also performed a detailed thermal analysis and provided the information to the slip ring vendor to assist in temperature and thermal expansion/brush alignment analyses. The detailed thermal analvsis NRL Performed on the Windsat slip rinas was critical in ensurina adeauate performance from the slip rinas as well as enablina us to understand and recover from the Windsat BAPTA on-orbit anomalv. This will be discussed further in the next section “On-Orbit Anomaly Lessons Learned.” Slip Rinas - “Devil is in the Details” Slip rings are a fairly ‘low tech” device. However, for high reliability, long-life space applications “the devil is in the details” to ensure proper performance and reliability. When passing high-frequency signals across slip rings this problem is multiplied greatly. On the Windsat slip rings many “detail problems” were encountered such as the following: 1) As mentioned previously the Windsat slip rings were solid “soft” gold with a thin hard gold flash top coating. The slip rings were sent to Ball Aerospace for cleaning and lubrication. After cleaning a microscopic inspection was performed wherein it was noticed that several of the slip rings were cracked (Figure 8). After the first cracks were discovered a 100% microscopic “mapping” of the slip rings were performed and other cracks were found. The rings were also eddy current inspected, which also uncovered a few faint cracks. None of the cracks affected the ring resistancehmpedance, and the brush wear actually tended to smear over the cracks in a beneficial manner so it was decided that the cracks were acceptable. Finding cracks on solid “soft” gold slip rings was surprising. It was determined that the cracks were caused by plating with a current density that was “in-spec” but at the high end of the spec. Subsequently the current density tolerance was revised and the second, flight spare slip ring was plated with no cracking. While the crackina was not desirable, no problems were ever caused bv the cracks. Figure 8. Cracked Slip Rings 21 5

Document

AMS_2006.pdf

2) The slip ring wires used Tefzel wire insulation per MIL-W-22749/44 and M27500-28SR2S23. There were numerous instances wherein the wire insulation cracked at bend radiuses (Figure 9). This required a lot of tedious repair work and inspection. Similar problems were also seen on the Tefzel motor hookup wires. While the definitive cause of this was not determined, no cracking was seen in the second, flight spare slip rings where the wire used was procured from Raychem as Raychem wire had been used on the rest of Windsat with no cracking. When digging into other experience with this problem it was found that the Tefzel cracking seemed to occur most often in the presence of elevated temperature epoxy/potting cure operations. However, no definitive causeleff ect relationship was determined. i Figure 9. Cracked Tefzel Wire 3) As mentioned previously, the slip rings were cleaned and lubricated by Ball Aerospace. During cleaning an analysis of the solvent rinse was conducted, which revealed silicone contamination. The slip ring specification prohibited any silicone materials, lotions, etc from being used on or near the slip rings as silicone contamination can cause lubricant dewetting and form compounds, which result in slip ring noise. The source of the contamination was eventually traced to Kapton tape that was mislabeled by the tape vendor and contained silicone adhesive, even though it was labeled as acrylic adhesive. A lengthy cleaning process was undertaken to remove the silicone from the slip rings. If not for the extraordinarily thorough cleaning and lubrication process procedures at Ball, the slip rings would likely have had silicone contamination problems, which would have resulted in disastrous consequences. It is recommended that solvent rinse analvsis be conducted durina slip rina Drocessina of slip rina assemblies. 4) The Windsat slip ring had 137 rings and 274 brushes in a 305-mm (12-inch) length. Slip ring-to- brush alignment was very challenging and key to proper functionality. Alignment features and tight tolerances were necessary to properly control alignment. Additionally, a detailed thermal model/analysis was performed to analyze the effect of thermal gradientdtemperature on slip ring to brush alignment. Finally, displacement during vibration was analyzed to ensure that brushes would not “jump rings” during vibration testing/launch. None of these conditions presented a trivial problem. Furthermore, as the slip rings wear, they create wear debris which tends to get pushed to the sidelout of the contact path with continued rotation. During normal operations the slip rings are maintained within a narrow 25-30°C temperature band. However, occasionally the spacecraft goes into a radiation upset induced safe hold which turns everything off and the slip rings cool to approximately 0°C. Starting up from this temperature results in displacement of the brushes due to differential thermal expansion and presents an opportunity for “running in the debris on the side of the road. In fact, the sliD rina/brush disDlacement at cold temDerature may have Dlaved a Dart in the on-orbit anomalv that will be discussed in detail later as both on-orbit BAPTA anomalies occurred within weeks of a safe hold event. 5) The Windsat slip rings were lubricated with a thin film of oil supplied by Ball Aerospace. During cold temperature acceptance TVAC testing (-20 to 5°C) slip ring noise was found. This noise was due to hydrodynamic effects of the thickened oil at these temperatures. It was determined that 21 6

Document

AMS_2006.pdf

7) 9) this noise could be “burned through” with the 2 Amps of current used on the power rings but not with the 100 mA of current used on the signal rings. Additionally, the rings that exhibited oil exhibited a bead of excess oil that clung to the brush at the ring contact (Figure 10). Rings without this oil bead did not seem to show cold temperature noise. This problem was solved by 2 methods: 1) re-oiling the slip rings with a slightly lower quantity of oil which did not result in the oil beads at the ring contacts, and 2) adding a heater to the slip rings to try to keep the rings above 5°C. After retesting with the lower oil quantity there was no noise found all the way down to -2OC. Figure 10. Original Oil Quantity (Left) Final Reduced Oil Quantity (Right) Since the Windsat slip rings had a high life cycle requirement and slip ring noise & wear are a function of brush force, the brush force was measured on all 274 of the slip ring brushes. This brush force inspection was a difficult. Dainstakina process. but it Daid off in that it revealed that desDite the best of efforts and tolerancina. manv of the brush forces were sianificantlv out of -. It is typically believed that the environment (air vs. vacuum) does not affect gold on gold slip rings but does affect silver/MoSp slip rings. On Windsat, we found that after running at 30 rpm for approximately one week in air we would get high-frequency slip ring noise. This noise would clean up / go away after several days of running at 30 rpm in nitrogen or vacuum. While there were no timelfunds to investigate the mechanism for this, it is postulated that an extreme pressure additive in the oil may be oxidizing in air to form insulating compounds. In N2 or vacuum this material may be wordpushed away and cannot form in the absence of oxygen. This didn’t seem to have any adverse impact on the performance of the life test units. Additionally it had no impact on Windsat as air drag limited the speed of the full up Windsat payload in air to a few rpm and there was never enough revolutions in air to run into this problem. The Windsat slip rings are constructed from a stack of slip ring modules. Prior to the module “stacking” the individual modules were thermal cycled to screen for “bad bond” or rings going open circuit. These screening tests resulted in rejecting a module with multiple opens and finding a module with one bad circuit that was relegated to a spare, prior to wiring and any “pain”. The slip ring brush design was an exact copy of the DSCSll design. However, since DSCS had fewer and smaller diameter rings, the brush placement differed. Unfortunately, the placement used left the tips of the brushes dangerously close to the brushlring contact point. In fact a few of the brushes created large piles of run-in debris, which was traced to contact at the brush tip and “digging in”. Fortunately, this only affected a few brushes and enough spare rings were available to reassign these circuits. SliD rinas are verv process-sensitive devices. It has taken decades to learn all of the Drocesses that need to be controlled/Daid attention to in order to Droduce reliable sliD rinas. Manv of the DeoDle who “learned these lessons” have left and the Droaram should make sure that those details have been carried forward into the current Droduction or the “heritaae” cited mav not be valid. This issue was encountered when building the Windsat slip rings, which traced their heritage back to the DSCSll program, last built in the 1970’s. There were actually still several people at the slip ring vendor and Ball who worked on the DSCSll slip rings. However, undoubtedly there was a lot that was forgotten, especially when it came to “why it was done that way”. The Windsat slip ring design was very similar to DSCSll but not the same. One “heritage” mistake we made on Windsat was to copy the DSCSll signal and power brush design, exactly 217

Document

AMS_2006.pdf

to print. The DSCSll unit had 12-mm (0.5-inch) diameter signal rings with correspondingly short signal brushes and 25-mm (1-inch) diameter power rings with longer power brushes. For various reasons (well thought out) Windsat ended up with both signal and power rings at 38-mm diameter (1.5 inches). The Windsat/DSCSII power brushes were much longer and had a much lower stiffness/spring rate. This made the power brushes much less sensitive to variations in brush force from misalignment and thermal excursions when compared to the shorter signal brushes. In hindsight, this was an obvious error it would have been better to not blindly copy “heritage” and optimize the brush design for brush force sensitivity - i.e., make all brushes like the power brushes. Values of Thorouqh Testinq (vs. Dumb & Hamv) This may seem obvious and not really a lesson learned, however, there is always budget and schedule pressure to reduce testing. On Windsat we did not compromise when it came to thorough testing. This had the extremely beneficial effect of catching many problems and fixing them before they were flown. However, it also had the parasitic effect of “finding lots of blemishes under a microscope” that we might have been just as happy to not know about. A highlighthecap of some of the things that were found by extremely thorough testing that would not have been uncovered with “normal testing” follows. Some of these had a big impact; others were just “interesting”. - NRL test set designed to detect high-frequency noise finds problems not seen in typical slip ring tests Found at least some cold temp noise could be “burned thru” with current Found high-frequency noise in air vs. GN2 & vacuum - - - - Found torque decreases significantly with vacuum level (<pressure=<friction) Found & solved cold temperature noise during acceptance testing which helped diagnose on- orbit anomaly (more on this later) 100% microscopic mapping of rings also helped diagnosis of anomaly & revealed cracks previously unnoticed Value of slip ring module level thermal cycle tests screening “bad bond” On-Orbit Anomaly Lessons Learned After 2 years / 33 million revolutions of nominal operation the Windsat BAPTA abruptly spun out of control (Figure 11). After a lengthy (and confusing) investigation, telemetry analysis revealed that the BAPTA resolver-based speed data was corrupt. Figure 1 1 shows the commanded speed, BAPTA resolver-based speed estimate, and the “true” BAPTA speed derived from spacecraft rates from the spacecraft Inertial Measurement Unit and conservation of momentum. In this figure, you can see the BAPTA speed going unstable and the true speed diverging from the resolver-based speed. Once the instability grows, the three-axis stable spacecraft begins to yaw uncontrollably, which triggers a commanded BAPTA spin- down and spacecraft “safe hold” mode. Since the data shows that the resolver system appeared to be corrupt, and the spin drive had redundant resolvers and electronics, the “B-side” was selected. The redundant B-side, however, was also found to be unstable (Figure 12). After extensive unsuccessful (and very confusing) troubleshooting, some troubleshooting commands were issued wherein the A&B sides of the BAPTA electronics were simultaneously turned on and the stationary 16-bit resolver position was queried. With the BAPTA stationary, both A and B side angles read identical (as they should) and they also read the correct angle (as roughly determined via external means). It was recognized at this point that all post-anomaly spin/troubleshooting was conducted at cold temperatures (around 5°C) as a result of the Windsat payload being off from the BAPTA induced spacecraft safe-hold. From earlier experience gained during cold temperature acceptance testing of the slip rings; it was theorized that cold temperature related slip ring noise was likely corrupting the B-side resolver which had its “transmitter” coil power passed across the slip rings. While the slip rings were able to start up at temperatures around 5°C during the first two years of operation, it was postulated that the effects of two years (33 million revolutions) of operation had resulted in a greater tendency for slip ring lubricant related noise to occur at cold temperature. Additionally, with 20/20 hindsight it would have been better to size the slip ring heater larger, but it was added late in the program when there was no more power left in the budget and no justification for a change that would “break” the power budget as the slip rings worked fine at that time all the way 21 8

Document

AMS_2006.pdf

down to -2OC. Even with a larger slip ring heater care would need to be taken with thermal gradients and brushhing alignment. BAPTA Speed Relative to Bus in SCS Frame I I I I I I I I I I I 4000 4100 4200 4300 4400 4500 4600 4700 4800 4900 Time (t), [sec] Figure 11. Original On-Orbit “A-Side” Spin Anomaly BAPTA Relative Speed Estimates in SCS Frame ..... ..... ..... P ........ , ............ ............... ......... .lJI i.. .... I ..I: ..... ..... ...... ...... ResolverBasedSpeed Tw” Speed (Fm IMU, MA, R WI II m 4 \I V yi 3 8.04 8.05 8.06 8.07 8.08 8.09 x104 Time (t), [sec] Figure 12. Original Anomaly “B-Side” Troubleshooting Low-Speed Spin 21 9

Document

AMS_2006.pdf

Additionally, it was postulated that a “blobs” of slip ring debris could have been the cause of the original A-side instability which occurred at the operational temperature of 25-3OOC. However, with limited on-orbit telemetry there is no easy way to confirm this. One item of note is that while we had full electrical redundancy and multiple redundant (parallel) slip ring contacts, all of the rings have the samekommon wear life (i.e., wear life is not redundant). With these anomaly theories in hand a workaround was devised to power on the payload on for several days to warm up the slip rings prior to BAPTA / momentum wheel spin up. (The payload had to be powered off during spin up to minimize the risk of blowing a fuse on the spacecraft power bus from the “peak” current draw during spin up.) On June 13, 2005 the workaround was successfully implemented and the BAPTA operated nominally on the B-side. On October 27, 2005, however, the BAPTA went unstable again, this time on the B-side. During this second “6-side BAPTA anomaly”, the BAPTA went unstable in a manner similar to, but not quite the same as the original A-side anomaly (Figure 13). B-Side Anomaly 10127105 40 35 30 25 E 15 10 5 0 5.95 IO‘ 6 IO‘ 6.05 IO’ 6.1 IO’ 6.15 IO’ 6.2 IO’ 6.25 IO’ Seconds Past Midnight Figure 13. Second On-Orbit “B-Side” Spin Anomaly Noteworthy during the B-side anomaly, the speed had several unstable “blips” followed by recoveries until it finally went fully unstable and triggered a spacecraft momentum induced safe hold and spin down (Figure 14). The second anomaly on the B-side presented the opportunity to confirm/refute the theory that slip ring debris “ingestion” was the cause of the original A-side anomaly. -If slip ring debris induced noise were indeed the cause of the A-side anomaly then there was a good chance (based on our DSCSll life test unit experience) that subsequent running had cleared the debris and “cleaned up” the noisy A-side resolver rings. Therefore, troubleshooting was conducted at low speed (0.3 rpm) on both the A and 6 sides of the BAPTA after the slip rings were warmed (to 15°C) to avoid the cold temperature slip ring noise phenomenon. Indeed the 6-side remained noisyhnstabie but the A-side was now cleanktable after the 4.5 months of additional run-time since the original A-side anomaly (Figure 15). 220

Document

AMS_2006.pdf

8-Side Anomaly 10127105 I I I I 2 1.5 I K n 1 0.5 0 E-Side 0.3 rpm lest 11/1/05 I I I j. i ........................ C .......____, ........... ;r j :a : j. j i. i i. j ;a j ................................................ ...... I I .................................. j. .j mi ;r ?I .................... :.I ....... fl ; ....& ,I 1 10 jm ................. i ='-""' 7.78 IO4 7.79 IO' 7.8 10' 7.81 10' 7.82 IO' 7.83 IO' 7 84 IO' 6 63 10'6.635 IO* 6 64 10'6;645 IO6 6 65 10'6.655 IO' 6.66 10'6 665 IO' Seconds Past Midnight Seconds Past Midnight Figure 15. Second On-Orbit Anomaly "A & B-Side" Troubleshooting Windsat was then successfully restored to operational status again on the A-side on November 6, 2005. Procedures have been developed to provide a speedy recovery should future anomalies of this nature arise. Additionally, if the condition eventually arises wherein both A&B sides are noisy at the same time it may be possible to run the motor open loop (forwards or backwards) in a troubleshooting mode to clean 221

Document

AMS_2006.pdf

up the slip rings as long as the slip rings are not noisy to the point that the motor cannot be commutated (the resolver is also used to commutate the motor). The ability to run many of these troubleshooting modes has proven invaluable in troubleshooting and recovering from problems both on the ground and on-orbit. Conclusion The Windsat BAPTA is a long-life precision spin drive assembly with challenging requirements and a difficult development program. As of November 2005, it has operated for 2.5 years of its 3 year life goal and it continues to operate well enabling Windsat to generate good wind speed and direction data. However, after two years of operation, evidence of slip ring noise on the resolver rings has appeared which has resulted in speed anomalies requiring safe-hold recovery operations. Looking back on the experiences gained during the development and on-orbit operations some key things were done well and should be done again (good heritage) and some things could be improved next time (bad heritage). A brief recap of the key items follows: What Worked Especiallv Well Redundancv and Robust Desian Marains Without robust margins the BAPTA development schedule would have been “broken” and without redundancy (in ways unanticipated) the BAPTA would not still be operating. Windsat was a low-cost, low reliability experimental mission and a programmatic decision was made to forgo any redundancy to contain cost. However, a cost-benefit analysis was presented to management which convinced management to allow redundancy on the BAPTA (the only redundancy flown on Windsat). Partner Relationshir, Between Government and Slip Rina Vendor NRL and the slip ring vendor combined strengths and areas of expertise to build a better product. NRL provided assistance with bearing design and analysis as well as thermal analysis allowing the slip ring vendor to concentrate on the rest of the slip ring design. Desian For Test & Troubleshooting Designing for test and troubleshooting up front allowed many problems to be solved both on the ground and on-orbit. Some key items were: the ability to spin in both directions, run open loop, and turn on both sides of the electronics simultaneously. What Could Be Done BettedDifferent Next Time Don’t pass critical BAPTA feedback (resolver) over lona-life slir, rinas An optical encoder was dropped early on for budgetary reasons and with hindsight one should trade alternate position feedback devices that do not require slip rings. Alternately, better methods of preventing slip ring noise from affecting resolver signals could be pursued. Noise sensitive lona-life slip rinas for space remain challenainq Based on Windsat experience and recent development and testing on other programs a trade study should be revisited on gold vs. silver/MoS2 vs. non-contacting (i. e., fiber optic, capacitive coupled) “slip rings” for long-life noise sensitive applications. Desian slip rina brushes to accommodate thermal expansion Windsat copied DSCSll heritage brush design and the power brush design was good with a low “spring constant” that was insensitive to wear and thermal expansion but the critical signal brushes had a high ”spring constant”and should have been designed the same as the power brushes. References Phinney, Damon, “Slip Ring Experience in Long Duration Space Applications.” 2dh Aerospace Mechanisms Symposium, (1 986),pp 45-54. 222

Document

AMS_2006.pdf

JWST NlRSpec Cryogenic Light Shield Mechanism Kathleen Hale’ and Rajeev Sharma’ Abstract The focal plane detectors for the Near-Infrared Spectrometer (NIRSpec) instrument on the James Webb Space Telescope (JWST) require a light tight cover for calibration along with an open field-of-view during ground performance testing within a cryogenic dewar. In order to meet the light attenuation requirements and provide open and closed fields of view without breaking vacuum, a light shield mechanism was designed. This paper describes the details of the light shield mechanism design and test results. Included is information on the labyrinth light path design, motor capability and performance, dry film lubrication, mechanism control, and mechanism cryogenic performance results. Background A light tight cover mechanism design and development is discussed for ground testing the Near Infrared Spectrometer Instrument’s focal plane detectors on the James Webb Space Telescope. The NIRSpec focal plane detectors require a light tight cover for calibration along with an open field-of-view during ground performance testing within a cryogenic dewar. These tests include exposing the detectors to infrared light for calibration in “darkness” while under vacuum at approximately 20 Kelvin (-253°C). The darkness requirement is to maintain light levels less than 0.001 electron/sec/pixel at the detectors. In order to provide this low light level during calibration and also allow the detectors to be exposed to the required light sources when needed, the light shield mechanism was developed. Requirements The light shield mechanism is required to attenuate light in the chamber to a level less than 0.001 electronlseclpixel before it reaches the detectors. Volume space allotted to the mechanism is 27.9 cm in diameter by 15.2 cm high (01 1 in x 6 in). The shield must be able to open or close within about 60 seconds, operate in a vacuum Torr) at 20 Kelvin, and in any gravitational orientation. The lifetime requirement for the shield is a few thousand cycles where one cycle is open and close. When open, the shield door is to remain outside the 76-degree cone angle of the detectors’ field of view. The light shield door range of rotation is about 124 degrees from open to close. Shutter Light Path Design A labyrinth light path is created, Figure 1, by the aligning of teeth-like protrusions on the shutter door with teeth cut outs into the plate covering the detector housing, hereafter referred to as the dewar bulkhead. The door and dewar bulkhead protrusions are separated by a gap of 0.64 mm (0.025 in) to prevent any contact debris from being created during the opening and closing operation of the door. This gap allows light entrance to the path, however, the labyrinth design combined with the proper material selection and treatment forces any light entering to bounce many times off the path’s walls and thus be absorbed to levels below the requirement. Aluminum 6061, which is bead blasted and black anodized, makes up both the door and dewar bulkhead. “NASA Goddard Space Flight Center, Greenbelt, MD Proceedings of the 38 Aerospace Mechanisms Symposium, Langley Research Center, May 15- 17,200G 223

Document

AMS_2006.pdf

Door r-O.64 mm ( 2.4mm 0.25 mm 4 t Imm I Bulkhead Figure 1. Light- Path Labyrinth Cross-section View Fabrication of parts was completed on the Micron 600U 5-axis high-speed milling machine as shown in Figure 2. The small sizes and angles of the light path protrusions make it necessary to use this machine. An additional advantage of the high-speed machine is the extremely low amount of stress that is added to the part by the machining as opposed to the level normally added by typical machines. Stress is a concern due to the fact that cryogenic temperatures can cause the part’s stress to increase and warp the movable door. Therefore, as well as using the high-speed machine, Aluminum 6061-T651 is used to minimize stress. Aluminum 6061 -T651 is heat treated to stress relieve the material. Figure 2. Picture of door being cut on Micron machine After rough-cut fabrication, the door is taken and soaked in liquid nitrogen to determine if part warping at cryogenic temperatures is an issue. Key measurements taken before and after this process are compared and it is found that the door’s critical dimensions do not permanently move any significant amount. Therefore, it is determined that the cryogenic environment will not warp the door and affect its operations. 224

Document

AMS_2006.pdf

Next, the dewar bulkhead plate is fabricated. Aluminum 6061- T651 is again used to minimize stress. Figure 3 shows the exploded isometric view of the mechanism. End of Travel End C """ILVII Dewar Bulkhead 0279 mm (1 1 .O in) Figure 3. Exploded Isometric View of Light Shield Mechanism Motor Selection One cryogenic compatible stepper motor is used to drive the light shield door mechanism. The motor was procured from CDA lntercorp with a right angle gear head, a planetary gear train, dry film Molybdenum Di- Sulfide (MoS2) coated parts, a motor shaft mounted resolver for position feedback, and a keyed motor shaft interface to the door. Motor design parameters are discussed below. Toraue Analysis indicates the light shield mechanism requires, with margin, a maximum driving torque of 0.19 Nom (1.67 inolb) in the worst-case gravity orientation. In this orientation, the door is required to be held open or closed against gravity by the motor detent alone as power must be disabled during detector tests. This motor detent torque with margin turns out to be the driving design parameter and calculates to 0.1 Nom (0.88 inolb). A healthy factor of safety of 6 is used to bring the motor output detent torque to a 0.6 Nom (5.3 inolb) value. The procured motor provides a measured detent torque of 0.75 Nom (6.67 in-lb). Gearinq A right angle gear-head is essential for this mechanism design as the volume space available to mount the motor and mechanism to the dewar bulkhead is limited. The total motor gear ratio is 187:l implying one motor step equals the door motion of 0.16 degree. This small step size is required so that if during door motion the motor bounces back or misses a step near the fully closed position, the gap left "open" does not allow a significant amount of light to enter the light path. Analysis confirms the gap is not large enough to degrade the light shield capability of the mechanism. 225

Document

AMS_2006.pdf

Crvoaenic Lubrication Dry lubrication on the motor components is required for proper cryogenic temperature operation. A sputtered Molybdenum Disulfide (MoS2) coating was used on the motor gears, bearings, and moving parts. MoS2 was chosen for this mechanism based on the ability to provide any thickness needed and also its heritage on numerous previous flight programs such as COBE, Cassini, SIRTF, and others. Drawbacks in using many dry film lubricants, including MoS2 are that the coating should be operated in a dry oxygen and water vapor free environment. Operation should not occur at humidity levels above about 40% due to the reactivity of the coating with oxygen and water vapor. The MoS2 dry film lubricant sputtering process was provided by Hohman Plating Inc. In a vacuum, at 149°C (3OO0F), an electron emitter breaks up the lubricant material into atomic size particles that bombard the parts. The coating particles adhere to the parts' surfaces and build up thickness over time. Coatings can be co-sputtered with different elements to improve crystalline structures and lower the friction coefficient. The majority of coatings are co-sputtered with Nickel giving a 0.02 friction coefficient. More advanced co-sputter options are Antimony O3 (p = 0.01) or Antimony O3 with Gold (p c 0.01). This project co-sputtered with nickel as the coefficient of friction meets the required amount and is less expensive than the other co-sputter options. Figure 4 shows the dry film thickness along the tooth of a motor gear part as seen in a cross-section view. Dpto tlme , :I1 .Zr):bl Figure 4. Mo~ coating thickness measured along width of gear tooth near tooth tip of a representative motor gear part. Thickness measured from mounting adhesive to gap. Nominally, pohman plates parts to a thickness of 3000-1 0000 Angstroms (0.3 to 1 pm or 1.1 8 x 1 O5 to 3.93 x 10 in) over 50 minutes. Considering the low life cycle requirements of the mechanism, the associated mechanical loads it would undergo, and margin, a MoS2 coating thickness of 2.5 pm (0.0001 in) was considered. Initially, the relationship of lubricant coating buildup over time with the sputtering process was thought to be linear. However, through the coating of several sample gears, analysis shows 226

Document

AMS_2006.pdf

it to not be so. At first the thickness and time were extrapolated linearly to that for 2.5 Nm (0.0001 in) which resulted in a 3-hour long application. Destructive analysis was performed on a sample gear by the materials group at NASA Goddard, and the average thickness of the coating was measured to be 17.8 pm (0.0007 in). Another gear sample was then coated at a lesser time of 70 minutes and the measured thickness of the coating ranged from 0.5 to 3.8 pm (ZX~O-~ to 1.5 x loe4 in). Based on the gathered data shown in Figure 5, it was decided that an application duration time of 80 minutes would be sufficient for the light shield mechanism motor parts. All parts are coated for the 80-minute duration resulting in a thickness of 1.2 um (0.00005 in). The final dry lube coating applied to the motor parts is based on the sample gear data and the minimum motor assembly tolerances for nominal fit. After the coating process was complete, all motor parts were successfully reassembled and tested. 9.00E-06 - 8.OOE-06 - 7.OOE-06 - 6.00E-06 - E u) 5.00E-06 - Y u c c .- 4.00E-06 - 3.OOE-06 - 2.00E-06 - 1.OOE-06 - O.OOE+OO 7 MoS2 Coating Thickness vs. Time t t v ** A 8 - + * 1 .OOE-05 I The bearings in the motor are also dry lubricated with MoS2. Lubrication thickness is the same as that for all other motor parts, 1.27 pm (0.00005 in). The support bearing on the opposite side from the motor on the bulkhead is a radial bearing and is not sputter coated. Instead, the support bearing is a “BarTemp” bearing from The Barden Corporation. “BarTemp” bearings have a cage made of a Teflon-coated, highly compressed material with very fine glass fibers and molybdenum disulfide impregnation. During operation, the rotation of the bearing causes the balls to rub off small amounts of the cage which coats the raceways with a thin lubrication layer. Mechanism Control The light shield mechanism motor is a two phase DC stepper motor with 30 degree steps and a tots1 right angle gear ratio of 187:l. Nominal operation occurs at 24 VDC. Since the mechanism must operate at room temperature and at 20 Kelvin, a current controlled electrical drive system is used for maintaining proper performance margins under all operating conditions. A Newport 300 Series current control amplifier drives the light shield mechanism motor. Step pulse rate and phase current levels are settable 227

Document

AMS_2006.pdf

quantities through a front panel or macro driver software which controls the mechanism to open or close the light shield door in approximately one minute. Software time out and TTL limit switch interfaces allow electronically disabling motor power redundantly with this controller. Motor resolver analog signal is read and converted to a TTL quadrature signal that the Newport controller can read. Motor power is disabled when contact is made with one of the end of travel switches in either the open or closed position of the light shield door. The motor shaft resolver is used as a method of determining the door position. Over Travel Analysis As the light shield reaches its end of travel in either the open or close direction, it comes into contact with an end-of-travel switch. This switch then cuts off power to the motor and the door stops its motion. In the close direction there is a hard stop that does not allow the door to over travel. Travel in the open direction, however, does not have a hard stop. Therefore, stress analysis is performed on the tab that comes into contact with the switch to determine the maximum amount of force that can be handled. These calculations determine that the load felt on the tab is 20 times smaller than the yield load of the door material, including margin. Therefore, at the slow speed that the motor rotates, the tab contacting the switch will not yield if the motor attempts to continue to rotate, and the shield door will not over travel. Cryogenic Testing Before delivery to the project, the light shield mechanism is tested in a dewar at 20K in the operational configuration. The dewar bulkhead is attached to an interface plate to mate the mechanism to the chamber. Once assembly is complete, the chamber is closed and the then rotated 180" to orient the mechanism to its operational configuration. This places the mechanism so that gravity is acting to open the door from its closed position. Test results are forthcoming and will be presented at the conference. Special thanks goes to Mike Dube and Mark Mclendon of the Materials Group at Goddard for their destructive analysis and insight in evaluating the dry film coating of motor parts. 228

Document

AMS_2006.pdf

Development Tests of a Cryogenic Filter Wheel Assembly for the NlRCam Instrument Sean McCully’, Charles Clark*, Michael Schermerhorn*, Filip Trojanek*, Mark O’Hara*, Jeff Williams* and John Thatcher* Abstract The James Webb Space Telescope is an infrared-optimized space telescope scheduled for launch in 201 3. Its 6.5-m diameter primary mirror will collect light from some of the first galaxies formed after the big bang. The Near Infrared camera (NIRCam) will detect the first light from these galaxies, provide the necessary tools for studying the formation of stars, aid in discovering planets around other stars, and adjust the wave front error on the primary mirror (Fig. 1). The instrument and its complement of mechanisms and optics will operate at a cryogenic temperature of 35 K. This paper describes tests and test results of the NlRCam Filter Wheel assembly prototype. ..- Figure 1. The NlRCam instrument Introduction The Filter Wheel assembly is one of three types of mechanisms on NIRCam. There are a total of four Filter Wheel assemblies on NIRCam, and these assemblies are situated at the pupils of both the longwave and the shortwave beams for both of the NlRCam optical benches. The Filter Wheel assembly mission is to position optical filters and pupil lenses as well as other wavefront sensing elements into respective optic beams. The Filter Wheel assembly prototype was built to retire certain risks and concerns. The end-of-life bearing drag torque is a critical parameter necessary in determining the drive current. The motor torque constant and phase resistance at temperature is critical to meeting operating margins for torque and power. Finally, the position control and stability using an inductive feedback position sensor and cogless DC motor needed to be characterized. * Lockheed Martin Space Systems Company, Palo Alto, CA Proceedings of the 38 Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 229

Document

AMS_2006.pdf

Filter Wheel Assembly Design The Filter Wheel assembly contains two independently rotating wheels with 12 optic positions each (Fig. 2). Each optic is centered on a radius that is 112.5 mm from the axis of rotation. The Pupil Wheel contains a complement of light projectors, wave-front-sensing elements, filters, coronographic wedges, and weak lenses. The Filter Wheel contains 5-mm filters that are tilted four degrees parallel to the plane of the optic bench. The NlRCam prescription requires the Pupil Wheel and Filter Wheels to rotate very close to one another with the final surface of the filters less than 25 mm from the first optic surface in the Pupil Wheel. The performance of the Filter Wheel assembly is driven by the requirement to locate some of the elements in the pupil wheel with a repeatability of less than 310 microns. Additionally, the target location of the prescribed element is to be adjustable on orbit through software. This performance criteria along with a very modest average thermal dissipation requirement that accompanies the 35 K operating condition make the Filter Wheel assembly design a challenge. I" Figure 2. The NlRCam Filter Wheel Assembly The team designed each wheel assembly to be mounted directly onto a custom motor rotor. This design concept is based on heritage designs that have been space qualified and flown before. It is an extremely efficient design concept for both envelope and mass concerns. Each wheel is driven by a 24-pole, three- phase, redundantly wound, cogless DC motor. This configuration minimizes the operating power and space envelope and maximizes the flexibility in commanding positions. A cogless motor has no harmonic content in the drag term, and it has no position detents. This combination of features allows for a very low drag actuator with no mechanically favored wheel positions. A 24-pole motor has the added benefit of simplifying the drive electronics operational scenario since a single 30 degree move from one optic position to another is accomplished by one complete electrical cycle. The motor rotor is supported by a back-to-back mounted duplex bearing pair. The choice of bearing type was driven by heritage and experience. A relatively large bearing was chosen to accommodate a 50-9 launch load and to extend the life expectancy of the dry film lubrication. The lubricant choice was more difficult. The final decision was based in part on an AMS paper that evaluated various lubricants [l]. Additionally, Lockheed Martin has experience with thin film lubrication on a wide variety of applications. These experiences lead to a preference for self-lubricating PTFE and MoS2 retainers. The position feedback was the last aspect to be considered. The team chose a differential inductive sensor system to measure position. Ramp targets on the face of the optical element wheels converts linear sensing devices to a rotation. This choice was driven by previous cry0 experience with the 230

Document

AMS_2006.pdf

technology, envelope, power, mass, and cost concerns. It was made possible by a relatively loose position requirement. Filter Wheel Assembly Prototype Design The prototype design represents a single actuator. The prototype design and test program goals were to confirm a bearing run-in procedure, quantify bearing, motor drag, and motor capability, characterize the inductive sensors, and finally verify the controlled repeatability performance. All tests were conducted at an operating temperature of less than 70 K. Five 35-mm O.D. duplex bearing pairs with TeflonTM and MoS2 impregnated cage material were obtained (Fig. 3). The cage material is identical to the proposed flight. The bearing size and preload is identical to the flight design. However, each prototype bearing is made from 52100 steel rather than 440C as specified for the flight program. Figure 3. Filter Wheel Assembly Prototype Bearing In addition to the bearings, two cogless DC motors were specified and procured and included a customized rotor with a feature that would accept the bearing pair and mate with the surrogate optic wheel (Fig. 4). A motor housing was designed to hold the motor stator with a unique flexure design that accounted for coefficient of thermal expansion differences between the components. A motor shaft and shaft mount was designed and fabricated. The design allowed for adjustment in the motor air-gap. Figure 4. Cogless Motor Installed In Prototype 231

Document

AMS_2006.pdf

A set of inductive sensors were obtained (Fig. 5). The surrogate optic wheel was designed to provide 15- degree differential ramp targets. The sensor array and sensor track were integrated with the prototype design. J Figure 5. Inductive Sensors Prototype Test Configuration In parallel with the design and fabrication of the Filter Wheel assembly prototype, the team designed the necessary setup to test the assembly at 35 K. After selecting a cryogenic test chamber, the team specified the required test equipment. In addition, the team designed the appropriate tooling that would adequately hold the prototype during testing. In order to test the bearing and motor drag, a fixture was designed to hold the prototype within the cryo-chamber, while an external motor and encoder spun the rotor and wheel through a ferrofluidic feedthrough. A vacuum rated torque meter was coupled to the feedthrough inside the chamber to measure torque as close to the prototype as feasible. Figure 4. Prototype Bearing Cold Test Configuration 232

Document

AMS_2006.pdf

Bearing Run-in and Cryogenic Drag Test Results Bearing drag was the first performance characteristic tested. Each bearing was run-in at an ambient operating environment using a run-in test fixture. A ground support equipment (GSE) motor was used to rotate the motor at 40 RPMs for 200 revolutions in one direction then 200 revolutions in the opposite direction. A torque measurement at 1 RPM followed each sequence of 200 revolutions. A total of 10,000 revolutions were accomplished over a period of approximately four hours. Torque telemetry was continuously obtained by an in-line torque transducer. Post-test inspection and tests indicate that an average of 40 to 50 angstroms of Teflon lubrication was successfully transferred to the balls and races. The bearing was then assembled onto the prototype shaft and installed into the cryo-chamber to test bearing drag at temperature. A bearing resistive torque of 5.4 mN-m (0.8 oz-in) was measured. This represents an increase of only 3.2 mN-m from the 2.2 mN-m tested at ambient. While motor and bearing tests continued, additional bearings were run-in using the same 10,000 revolution procedure. These bearings were then inspected and characterized. Periodic inspection and photography indicates that at 20,000 revolutions, the bearing was in “as-new” condition both in terms of torque and physical condition (Fig. 5). I Figure 5. Prototype Bearing Post-Test inspection (20,110 revolutions) It was determined that a bearing with a measured torque in excess of 21 mN-m (3.0 oz-in) was no longer functioning properly. The torque was irregular, and it was accompanied by an audible protest or squeak. A bearing end-of-life torque limit of 14 mN-m (2 oz-in) was established using the “Knee” of the ambient torque/life curve and allowing for measurement uncertainty. This is viewed as a very conservative limit. Motor and Bearing Tests After the bearing run-in tests were completed, the motor stator was assembled into the prototype housing for motor and bearing tests at temperature (Fig. 6). The same thermal vacuum test setup was used. A motor and bearing resistive torque of 9.6 mN-m was measured at temperature. This represents an increase of 4.2 mN-m over the bearing drag test. 233

Document

AMS_2006.pdf

Figure 6. Prototype Bearing, Motor, and Sensor Cryo-Test Configuration The test configuration used to test motor and bearing drag also allowed for the measurement of the motor torque constant. By spinning the assembly at 50 RPM with the GSE motor and measuring the peak back emf voltage, the motor torque constant was determined to be 11 50 mN-m/A. Through both ambient and cry0 test runs, it was also shown that the torque constant was largely insensitive to the temperature and vacuum change. Motor phase resistance dropped from 82 ohms to 2.2 ohms at temperature. This measurement was done with a simple two wire ohm meter in the prototype efforts, and the measurement includes all the GSE wire-some of it still at ambient temperatures. The flight testing will use a four-wire measurement to eliminate this measurement error. Using 2.2 ohms in the power calculation is considered conservative. All mechanical connections to the test shaft were removed, and the assembly was tested cold under its own power. At each of the twelve positions, breakaway current was determined to be less than 10 mA. The motor was commanded to take 7 steps at a commanded current level. The motor was reset to pull it back to the motor zero position, and the commanded current level was increased. The process was repeated. Motion occurred prior to reaching a 10 mA command (Fig. 7). A prototype control board was designed and built. Two power amplifiers (with current control) are used to generate commandable current. Since the net current into the node of a wye-wound motor is zero,-the third phase is generated as a difference between the first two. The electrical cycle was divided into steps using a 12-bit controller effectively giving the ability to micro-step. The motor is driven open-loop 30- degrees (4096 micro-steps) from one position to another. Once in the approximate position, the closed loop mode is activated, and the position sensors are used to “micro-step” to a final optic position. The position signal is differentiated and used in a damping circuit while in the closed loop mode. The result is a nicely damped response. The control loop can be used to drive to any value in the position signal, but the sensor ramp has been designed (and aligned) such that each optic position is nominally represented by a zero according to the position signal. 234

Document

AMS_2006.pdf

Figure 7. Breakaway Current Tests Position Control Tests A 17-bit encoder was connected to the test shaft of the prototype in order to obtain position data (Fig. 8). The sensor track design results in 12 zero-voltage position signals and two higher voltage reference signals in a single revolution of the wheel. The sensor track was aligned to the motor zero during assembly. These data were used to obtain a voltage level of 0.09 V that corresponds to the 0.310 mm position requirement. Figure 8. Typical Voltage Response of Position Signal Control scripts were written to command the motor from position to position switching back and forth from open loop to closed loop modes of operation just as proposed for flight (Fig. 9). A reset was used to pull the motor to the closest motor zero position. Open loop moves were accomplished by commanding a speed, a current, and a number of steps (4096). The scripts then closed the position loop and 235

Document

AMS_2006.pdf

commanded the error to zero. Finally, the motor current was ramped to zero. The whole process was repeated to accomplish 360 degree rotations. Closed-loop error was found to be acceptable at approximately 0.09 mm. However, the results indicate an unacceptable position error “spring-back effect when the holding current is removed. This error is illustrated by the nearly 0.5 V motion near the end of each position hold. Figure 9. Nominal Rotation of Wheel to Twelve Positions Dither and Check Routine Through a series of build-up and breakdown tests of the prototype, it was determined that the powered-off position error is the result of both the bearing and the motor. The bearings are large (for a precision instrument bearing), and they have an outer-land riding cage. It is believed that this cage tends to store some windup energy when in motion. In addition, the DC motor appears to have a restoring energy. Additional tests are planned that will help verify the root cause of the motor restoring forces. It has been shown that increased friction levels reduce the powered-off position error. This is good news as this means that this particular position error will not get worse at the end of bearing life. It also provides a possible solution to the problem. After much deliberation, it was determined that introducing another source of debris into our optic system was not advisable. A common method for improving position results is to dither back and forth about the final position. This tends to relieve friction terms, and it has been shown to dramatically improve final positions of the prototype (Fig. 10). The motor is controlled past the intended final position by 1.4 degrees. This excursion is based on the ball-to-cage pocket clearance and is intended to specifically address bearing cage windup. The wheel is then commanded by half that excursion to the opposite side of the final position. The distance is halved again. The final “dither” is not always possible to see in the position signal. However, this approach does not guarantee a final position. 236

Document

AMS_2006.pdf

Figure 10. Results Showing Dither Improvements The proposed final solution is to dither and check three times before applying a powered-on hold. If the powered off position error is unacceptable, then a dither is commanded. If the position error is still unacceptable after dithering 3 times, then the system leaves power on (8 mA) (Fig. 11). . . .. i dither Figure 11 : Typical Dither Details 237

Document

AMS_2006.pdf

Conclusions A calculated torque margin of 1.8 is now supported by engineering test results. A motor torque constant of 1150 mN-m/amp was measured at ambient and operating temperature and meets the specified minimum of 1100 mN-mlA. A motor drag torque of 4.7 mN-m was measured and is less than the specified maximum of 5.3 mN-m. Cold bearing testing of the WA prototype successfully characterized the bearings at various cold temperatures and established an end-of-life bearing torque limit of 14 mN-m (2 oz-in). Further life tests are planned. A measured motor phase resistance of 2.2 ohms results in an estimated peak power consumption of 89 mW and an average power consumption of 0.44 mW when operated at its maximum duty cycle. Motor control and current margin have been demonstrated at a temperature of 20 K. A powered-on closed loop control error of 0.09 mm was measured at operational temperatures. A breakaway current of less than 10 mA was measured. Acknowledgements Development of the NlRCam instrument at the Lockheed Martin Advanced Technology Center is performed under contract to and teamed with the University of Arizona’s Steward Observatory. The University of Arizona in turn is under contract to the JWST Project at the NASA Goddard Space Flight Center. A great deal of team work was necessary to accomplish these tests. Tom Welsh, Bud Swihart, and Richard Bnrner played critical roles in getting the hardware to the proper test temperature. Their efforts and dedication are greatly appreciated. References [l] Gould S.G., E.W. Roberts, “THE IN-VACUO TORQUE PERFORMANCE OF DRY-LUBRICATED BALL BEARINGS AT CRYOGENIC TEMPERATURES,” 23rd Aerospace Mechanism Symposium, 1989. 238

Document

AMS_2006.pdf

Cryogenic Nano-Actuator for JWST Robert M. Warden' Abstract An extremely precise positioning mechanism has been developed for use in space for optical positioning of large mirrors. The design incorporates traditional mechanical components such as gears, bearings and flexures in a unique configuration covered by two patents. This linear actuator is capable of 10 nanometer position resolution over a range of 20 mm and can operate under cryogenic conditions. The design, assembly, construction and testing of this mechanism are presented. Introduction The James Webb Space Telescope (JWST) is configured to be a large deployable spacecraft as shown in Figure 1. A key component of JWST is the Optical Telescope Element (OTE), which consists of all the components along the optical path including the Primary and Secondary mirrors. The Primary mirror is about 6.5 meters in diameter and is made up of 18 segments. These mirrors are folded up during launch and deployed in space. Once unfolded, the mirrors must be deployed away from the launch restraints and then adjusted very precisely. Figure 2 shows how each primary mirror segment and the secondary mirror is supported and controlled by six linear actuators to obtain six degrees of positioning control. Each mirror can be positioned in tip, tilt, piston, horizontal & vertical decentering and clocking. Two actuators are assembled into a bipod assembly as shown in Figure 3. The final hexapod configuration is made up of three bipods. In addition, each primary mirror segment features a central actuator for adjusting the radius of curvature of the segment. The positioning and focusing of the primary and secondary mirrors require a total of 144 actuators. 7 Actuators segmen\ A / Figure 1. James Webb Space Telescope (JWST) * Ball Aerospace & Technologies, Boulder, Colorado Proceedings of the 3dh Aerospace Mechanisms Symposium, Langley Research Center, May 17-79,2006 239

Document

AMS_2006.pdf

Back side of mirror spacecraft framework and the mirror segments. For the purpose of this paper, the Delta Frame is fixed in space and the mirror moves relative to it. Delta Frame (fixed) Radius of Curvature strut (6 places) Strut to mirror attachment (6 places) Delta Frame to Backplane attachment Wiff le Plate (3 places) I Radius of Curvature Actuator mirror, an additional actuator is used to control the Radius of Curvature of the mirror. Actuator Pdr (3 places) See also Figure 3. Figure 2. Hexapod Mounting 240

Document

AMS_2006.pdf

Athermalization Brackets Hexapod Flexure \ Hexapod Flexure Fixed Interface to Delta Frame Movable Interface to Mirror Wiff le Bracket Figure 3. Actuator Pair 241

Document

AMS_2006.pdf

Actuator Requirements Coarse step size Axial load Axial stiffness The Actuator for JWST has two top level requirements: 1. Accurately position the mirror segments. II Support the mirror segments during ground test and launch These requirements apply to the actuators supporting the 18 segments of the primary mirror as well as to the actuators that support the secondary mirror. e1 .O microns 0.058 microns Yes 1890 N (425 Ib) 2650 N (595 Ib) Yes 24,500 N/mm (140,000 Ib/in) 25,200 N/mm (144,000 Ib/in) Yes Positioninu requirements The first requirement of positioning was used to generate the following derived requirements: Position feedback Cryogenic operation Nominal length 1. Move the mirror from the stowed position to the nominal deployed position. 2. Move the mirror from the nominal deployed position with 6 degrees of motion. 3. Position each segment to nanometer resolution. 4. Support the segments in a hexapod configuration. 5. Operate at cryogenic & ambient conditions 6. Operate over the life of the mission. 20 microns 12 microns Yes 30 K 20 K Yes 138.8 mm (5.5 in) 138.8 mm (5.5 in) Yes Load Reau irements The second requirement of support was used to generate the following derived requirements: 1. Support the mirror segments during launch. 2. Support the mirror segments during ground optical testing. 3. Support the mirror segments during ground transportation. 4. Hold the mirror segments in place with power off. 5. Support the segments in a hexapod configuration. 6. Operate at cryogenic & ambient conditions Mass The derived requirements are summarized in Table 1. 700 gram (1.55 Ib) 665 gram (1.47 Ib) Yes Table 1. Requirements Summary I Coarse range I >20 mm I 21 mm I yes I I Axial holding I 380 N (85 Ib) I 890 N (200 Ib) I yes I 242

Document

AMS_2006.pdf

Early Development The success of the actuator is primarily due to two important inventions: the fine stage flexure and the coarse drive coupling. It is important to understand the operation of these two elements before going on to the overall actuator description. The actuator is comprised of numerous individual design elements that all work together in order to satisfy the demanding requirements of cryogenic nanometer-level positioning. However, these two inventions enable the fine adjustment capability over the long range of motion required for the actuator. Fine Staae Flexure The development of the actuator began at Ball Aerospace in 1997 with the invention of the fine stage flexure. A need was established for amotion reduction'device that could convert a relatively coarse input motion into a well controlled, optical-level, fine motion. It was well known that flexures work well for motion control because they enable motion without backlash or hysteresis. However, simple flexures were not able to achieve the large ratio required for this application. The solution, partially shown in Figure 4, was the "Motion Reducing Flexure Structure", developed by Ball Aerospace and covered by United States Patent number 5,969,892 dated October 19, 1999. This compound flexure operates in two stages. As the middle of the cross bar is moved up and down, the sides are moved out and in, thereby causing a small but controlled change to the overall height. Motion reduction of up to 1OO:l can be achieved using this design. 1042a 1042% Figure 4. Flexure Patent Description Coarse Drive Couplinq The next development was the use of a single motor to operate both coarse and fine motion. Although the fine stage flexure provides the nanometer-range positioning accuracy needed for aligning the mirror segments, it does not accommodate the large range needed for moving the mirrors from the stowed position to the deployed position. For this reason, a coarse motion feature was needed for the actuator. It was desired to have one motor drive both the coarse motion and the fine motion. Figure 5. Coupler To move both the fine stage and coarse stage with a single motor required the use of a coupling that could switch between coarse and fine modes or at least disengage the coarse motion. This was achieved by the invention of a tumbler type coupling, which connects the fine drive to the coarse drive as shown in Figure 5. This coupling consists of two rotating disks, each with a protruding pin. This results in a deadband or backlash of approximately 90% of the rotational input. When the motor is reversed, the drive pin backs away from the driven pin so that the coarse motion is decoupled from the drive train. In this deadband zone only the fine stage is engaged, which enables precise mirror positioning. At the end of this travel, the coarse shaft is again engaged to enable coarse motion. This mechanism was developed Ball Aerospace and is covered by United States Patent 6,478,434 dated November 12,2002. 243

Document

AMS_2006.pdf

Actuator Chronoloay The complete development of the actuator is beyond the scope of this paper but several important - -I .. milestones should be noted 1997 Fine stage flexure developed‘: Large axial input results in small axial output due to compound flexure configuration. 1999 First Actuator built? 1 0-mm motor, dry film lubrication on all gears and bearings, tested at cryogenic temperatures (30 K), demonstrated fine resolution of less than 10 nanometers. 2000 AMSD Actuator built3: %-inch motor, more robust, twin counter-rotating coarse drive screws, three monopods per mirror, first mirror phasing demonstrated. 2002 IR&D Actuator: %-inch motor, more modular to reduce fabrication & assembly costs, large coarse drive screw to accommodate launch loads. Never tested. 2003 Test Bed Telescope Actuator4: 1 0-mm motor, low cost, 150 units built, hexapod configuration, smaller size, flight-quality positioning, ambient conditions, new fine stage flexure design. 2004 Flight Actuator: %-inch motor, ball screw for improved axial load capability, dry lube on all bearings & gears, new fine stage flexure. Figure 7 shows an actual collection of data from the first fine range of motion test. The graph shows absolute height vs. motor steps and . the distorted sine wave is clearly visible. Note that the fine range here is = 10.9 microns (0.0109 mm). Subsequent adjustment of the fine stage flexure resulted ir, a fine range of 10.5 microns. Fine Staae lmwovement Figure 6 shows the simplified fine stage flexure. The shape is generally that of the capital letter “A”. Like the original fine-stage flexure, movement of the cross-beam deflects the side beams, which, in turn change the overall height of the flexure. The cross-beam is attached to an eccentric cam shaft, which is driven by a gearmotor. As the cam shaft rotates, the cross-beam is driven up and down resulting in a sinusoidal displacement pattern. The sinusoid is distorted due to the compound nature of the geometry. The fine range of motion can be easily adjusted by changing out the pair of shims under the feet of the fine stage flexure. A change in shim thickness of about 0.1 mm (.004 in) results in a change of fine motion of about 1 micron. Large movement of cross beam results in small overall Eccentric Bearing drives cross-beam motion Shims under feet are adjustable to modify fine range of motion Figure 6. Simplified Compound Flexure 3 Figure 7. Fine Stage performance 244

Document

AMS_2006.pdf

Actuator Mechanical Description The best way to describe the functioning of this mechanism is to follow the drive train. The major components in the actuator are shown in Figure 8 and in the diagram in Figure 9. All major components attach directly to the main housing. Gearmotor The drive train of the Actuator begins with a stepper motor. This type of motor is often used for positioning mechanisms because the rotation is divided into discrete steps. The motor for this application has 24 steps per revolution or 15 degrees per step. The stepper motor is attached to a 60:l gear head and a resolver to form an integral gearmotor assembly. All bearings and gears are coated with a Ball proprietary dry film lube to enable operation at 30K. First Pass to Cam Shaft The output from the gearmotor uses a simple 3:l spur- gear pass to drive the cam shaft. The cam shaft incorporates an eccentric bearing in the middle that drives the fine stage flexure shown in Figure 8. The shaft is supported by two simple bearings that are preloaded so that the shaft is always pulled away from the gear end. The cantilevered mounting of the shaft enables the use of 1:l right angle bevel gears to change the drive axis from horizontal to vertical. Second Pass to Coupling The bevel gear pass connects to the coarse drive shaft through the coarse drive coupling as shown in Figure 5 described previously. Both sides of the coupling are supported by preloaded bearing pairs. Except for the gears, both sides of the coupling are identical to take advantage of commonality in fabrication. The drive side has a bevel gear and the driven side has a spur gear pinion attached. Third Pass to Coarse Drive The spur gear pinion drives a large ring gear resulting in an 8:l ratio. The coarse drive shaft is an 8-mm ball screw with a 2-mm pitch that is mounted as a cantilever with two preloaded bearings at one end to attach the shaft to the main housing. Other Components The remaining actuator components are discussed later in the paper and include: 0 Friction brake 0 Torsional stabilizer LVDT position sensor 0 Dry film lubrication Figure 8. Actuator on Test Stand Fine Position 0.01 mm Fine Range Compound flexure 1 :3 Spur gear pass I 1 : 1 Bevel gear A Geai Ball screw & nut Torsional Stabilizer 21 mm Coarse Range Figure 9. Schematic Diagram 245

Document

AMS_2006.pdf

Design Discussion Figure 10 is a cut away view of the actuator showing the preloaded bearing pairs and gear passes. Standard recommended fits and tolerances were used. These are all standard mechanical configurations derived from catalog information. DW Lube Operation at 30K prevents the use of liquid lubrication. Instead, dry film lubrication was applied to the moving surfaces of all bearings and gears. This Ball proprietary process as been used successfully on other cryogenic programs such as SIRTF and Hubble. Moving components can therefore be operated at ambient as well as cryogenic temperatures. Gearmotor Life The life requirement for the gearmotor was estimated using expected operational cycles for the life of the unit. A total of 1.7 million motor revolutions were estimated. The dry film lubrication in the gearmotor has been analyzed to last at least 3 million cycles. The gearmotor has been noted as a life limited item and motor revolutions must be recorded. Factor = spec/total = 1.75 Margin = spechotal -1 = 0.75 Toruue Marain The motor is a stepper type that rotates a precise amount for each step command. Torque 'margin calculations were generated to compare the output torque of the motor to all of the resistance torques including friction, inertia, coulomb drag and operational loads. Appropriate safety factors were placed on the various loads including cryogenic operation. The motor shows a positive margin under all conditions. Friction Brake A small friction brake was added to the actuator to prevent the coarse drive screw from back-driving operation. The axial force on the ball nut applies a Geajmotor w all Screw Shaft Figure 10. Section View torsional component to the coarse drive shaft. The coarse drive gear was chosen for the brake location because it is the largest torsional element on the coarse drive shaft thereby requiring the least force to constrain. The brake uses a double cantilevered beam to support and apply force to two Vespel buttons. The buttons slide along a raised surface on the coarse drive gear. Torsional Stabilizer In order for the actuator to apply pure axial motion, the nut on the ball screw must be constrained to the main housing in such a way as to resist torsional loads but to allow axial movement. The torsional stabilizer provides this constraint by incorporating a thin flexible shear panel that resists the torque applied to the shaft while allowing the ball nut to translate. As the nut moves up and down the shaft, the flexure forms an "S" shape. A pivot flexure that is opposite the panel flexure minimizes the radial loading on the ball nut from the stabilizer. 246

Document

AMS_2006.pdf

Gear tooth loadinq The gear teeth were analyzed using a short cantilever beam formula from Roark. The three gear passes were analyzed using the 5 g load case. The stress levels for the three passes are as follows: First Pass = 17.0 MPa (2.47 ksi) Second Pass = 17.4 MPa (2.52 ksi) Third Pass = 11.4 MPa (1.66 ksi) All gears are made of titanium, which has a yield strength of 869 MPa (126 ksi). The minimum margin of safety is therefore 49. Axial load reuuirements Large axial loads are imparted into the actuator from two sources: The defined acoustic load for one mirror is 6480 N The angle of the actuator is 15 degrees The load is shared between all 6 actuators 1. Launch Load: 6480 N /6/cos15 = 1118 N (251 Ib) x 1.4 = 1565 N (352 Ib). 2. Ground Test Load: The maximum axial force on a single actuator during ground test is 378 N (85 Ib). A factor of 5 is put on this value to protect against damage during ground test operations. 378 X 5 = 1890 N (425 Ib). Since the ground test load is greater than the launch load, it was used for design calculations. Lenath The length of the actuator was calculated based on a nominal deployed actuator length of 138.8 mm. The bipod assembly is required to retract 12.5 mm to the stowed position and extend 5.0 mm to the maximum deployed position. The actuator is also required to translate the bipod at the same piston, which requires additional length. Launch Restraint The original concept was to pull the mirrors down upon some mechanical hard stop until the motor stalls. There are two reasons why this concept was discarded: 1. The large force exerted by the actuator would require a huge increase in mass in order to withstand the stall forces. 2. When a stepper motor stalls, step count is lost. Another concept was to incorporate some extra feature in order to stop the motor when a certain preload was reached. Several techniques were investigated and discarded including: load cell (adds extra wires, mass, electronics); slip or magnetic clutch (lose step count, adds mass); visual indicator (impractical for all but the outermost actuators). In all cases, the extra feature adds mass and complexity. Also, the hard stop must be designed to be loaded to some factor above launch loads multiplied by some safety factor. The current launch restraint design is such that motion is restrained in the plane of the mirror and allowed in the direction perpendicular to the mirror. The actuators, therefore, constrain the motion of the mirror in the perpendicular direction. The actuators must withstand axial loads in both tension and compression. Power-train Summary The stepper motor is integrated with a 60:l planetary gearhead to form what is called the gearmotor. The output of the gearmotor drives a 3:l spur gear in the first gear pass. The drive torque then branches to the fine stage and/or the coarse stage. The fine stage has a 1:l bevel gear for the second gear pass to change the axis of rotation by 90 degrees to drive the fine stage flexure. The coarse stage has a tumbler type coupling that has 324 degrees of backlash before driving an 8:l spur gear for the third gear pass to the coarse drive shaft. The coarse drive shaft is an 8-mm-diameter ball screw with a 2-mm pitch. 247

Document

AMS_2006.pdf

Power Train Capabilities Because of the numerous gear passes, it was convenient to summarize the load at each step of the power train. Starting with the gearhead output torque, Table 2 lists the load at each step. The input torque is converted to a tangential force by dividing by the radius of the pitch diameter of the gear. The Pitch Angle is also factored in as well as the resulting load on the bearings. The tangential force is then multiplied by the radius of the driven gear to obtain the torque on the driven shaft. The torque on the shaft is then used as the input torque of the next pass. The calculations shown are for the 1 g operational load. Table 2. JWST Power Train Calculations First Pass Calculations Gearhead Input Torque Shaft Radius Tangential Tooth Load Camshaft Gear Radius Efficiency Torque to Cam Shaft Second Pass Calculations Cam Shaft Input Torque Bevel Gear Radius Tangential Tooth Load Bevel Gear Radius Efficiency Torque to Coupling Third Pass Calculations Coupling Input Torque Pinion Radius Tangential Tooth Load Coarse Gear Radius Efficiency Torque to Lead Screw Lead Screw Calculations Lead Screw Input Torque Brake Drag Lead Screw Radius Ball Screw Ramp Axial Force 0.200 in-lb 0.02 N-m 0.166 in 4.22 mm 1.205 Ib 5.36 N 0.500 in 12.70 mm 0.950 0.950 0.572 in-lb 0.065 N-m 0.572 in-lb 0.065 N-m 0.166 in 4.22 mm 3.448 Ib 15.33 N 0.166 in 4.22 mm 0.950 0.95 0.544 in-lb 0.061 N-m 0.544 in-lb 0.061 N-m 0.156 in 3.97 mm 3.481 Ib 15.48 N 1.250 in 31.75 mm 0.950 0.95 4.133 in-lb 0.467 N-m 4.133 in-lb 0.467 N-m 3.000 in-lb 0.34 N-m 0.164 in 4.17 mm 0.077 in 1.95 mm 90.09 Ib 400.7 N 248

Document

AMS_2006.pdf

Actuator Testing Assembly stage Motor only Motor & gearhead ambient Motor & gearhead cold Actuator assembly, ambient Actuator assembly, cold Bipod assembly, ambient Bipod assembly, cold Testina Options Early in the program it became apparent that the testing of certain properties of the actuator could occur at several different stages of assembly as shown in Table 3. Because some of the components are life limited and because of the relatively large quantity of units involved, it was impractical to test every property at every level. Also, it was desirable to limit cryogenic testing because it is expensive and time consuming. Property to be tested Motor torque Motor torque, gearmotor torque Motor torque Motor torque, fine range of motion, coarse range of motion, coarse accuracy, Axial stiff ness, Axial backdriving capability Motor torque, fine range of motion, fine step size, fine accuracy, coarse range of motion, coarse step size, coarse accuracy, LVDT output accuracy Motor torque, coarse range of motion None Table 3. Actuator Testing Options 1 Assembly Stage: Motor only, Motor & gearhead ambient, Motor & gearhead cold, Actuator assembly, ambient, Actuator assembly, cold, Bipod assembly, ambient, Bipoci assembly, cold, Mirror hexapod assembly, ambient, Mirror hexapod assembly, cold. Mirror hexapod assembly, ambient Mirror hexapod assembly, cold I I Motor torque, TBD Property to be tested: Motor torque, Fine range of motion, Fine step size, Fine accuracy, Coarse range of motion, Coarse step size, Coarse accuracy, Axial stiffness, Axial backdriving capability, LVDT output accuracy Test Plan After much discussion and trades, a testing plan was established to identify those properties that would be tested, and at what stage. It was desirable to test certain properties early in the process to determine any potential problems. The final test plan is shown in Table 4. 249

Document

AMS_2006.pdf

i. I 11111111111111111- c Error in microns a Actuator Test Results To date, there have been four major test activities: ambient motor tests, cryogenic motor tests, ambient actuator tests and cryogenic actuator tests. Ambient Testinq Figure 11 shows the combination test stand at the actuator fabricator. The fixture is used to measure axial stiffness, fine range of motion and coarse ; accuracy under ambient conditions. To measure the coarse accuracy, the length of the actuator was measured from stowed to deployed at every cam shaft revolution. The theoretical position was then subtracted from the measured position and the results are presented in Fig. 12. Note each point represents 0.25-mm axial movement. The accuracy requirement is 2% for this level of move, which equals 0.005 mm (5 microns). The test was performed in both the deploy (CW) and retract direction (CCW). . Figure 11. Test Stand 111111111111111. - 5 micron requirement Crvoaenic testing After the actuator was delivered to Ball, more accurate tests were performed to verify cryogenic operation and fine motion characteristics. Nanometer-level measurement testing requires interferometric instrumentation, vibration compensation and special software. Up to seven actuators can be tested at once using the special setup shown in Figure 13. One of the fine motion tests is single step repeatability. The results of this test are shown in Figure 14. The motor was rotated one step clockwise then one step counter-clockwise several times. The results show an average step size of about 7 nanometers and a repeatability of 1 nanometer. Figure 13. Cryogenic Testing at Ball 250

Document

AMS_2006.pdf

Lessons Learned Mass Droduction One of the main lessons learned was that for mass production, it is worth some up-front time and expense to develop tooling and fixtures in order to save time later on. In order to more efficiently fabricate the 140 actuators, several mass production techniques were incorporated. For example, the application of dry film lubrication usually requires masking to prevent the dry lube from being applied to sensitive areas. This masking has traditionally been applied one unit at a time using tape or a painted on compound. For this program, however, special masking tools and holding fixtures were designed and built ahead of time to streamline this operation. Another aspect of mass production that we learned was that is well worthwhile to spend more at the machining stage to save time during the assembly stage. For example, match drilling for alignment pins is always problematic in titanium in a clean room. For this program the alignment holes were put in ahead of time while the parts were being machined. At assembly the parts went together very easily with no match drilling required as shown in Figure 15. Recast from EDM Electro Discharge Machining or EDM was used to machine many of the critical areas on the actuator because it results in very good dimensional accuracy and repeatability. However, the EDM process produces a recast layer on the surface of the part. For a flexure, this recast layer can reduce the fatigue life of the part by up to 10 times. Multiple passes in the EDM process at Ball have resulted in a thin recast layer of about 0.0033 mm (0.00013 in), which is easily removed with light chemical etching. Figure 15. Production Tooling For one particular flexure, however, standard etching did not remove the recast layer. Additional etching to remove the recast layer made the flexure area too thin. A review of the drawing and discussions with the vendor revealed a misunderstanding of the process. A specific EDM process sequence has since been defined, which results in a thin recast layer and acceptable flexure dimensions after recast removal. Inspection techniques have also been developed to simplify the verification of recast removal. Thermal straming Titanium a poor thermal conductor and the gearmotor generates heat during operation, which affects the length of the actuator. In order to test the actuators in a timely manner, heat straps were attached to several points on the actuator. Again with mass production as a goal, special clamps were made so that the thermal straps could be easily attached and removed. Pathfinder for the EDU It is usually good practice to have an Engineering Development Unit (EDU) as a precursor to building a flight unit. JWST took advantage of this "lesson learned" that has been cited by many authors. The EDU for JWST consisted of 18 units to be used as follows: 7 actuators for the Primary Mirror Assembly, 6 actuators for the Secondary Mirror assembly, 2 actuators for a life test and 3 actuators for spares. With so many units to build, it was decided to turn one of the spare actuators into a "pathfinder" unit, which is essentially a first article development unit for the EDU. One actuator was taken from the initial build and designated as the pathfinder. This unit was the first to be assembled and the first to be tested. Any new testing or fixture was proven out on the pathfinder. To maintain schedule and to have processes verified in advance, certain shortcuts were allowed on this unit. For example, although the components were cleaned to a flight level, the pathfinder parts were not subject to an external particle count. 25 1

Document

AMS_2006.pdf

Conclusion The design of the Cryogenic Nano-Actuator for JWST is now complete. The EDU units have been assembled and tested. The actuators will next be assembled into bipods, which will then be integrated into the primary or secondary mirror hexapod configuration. Long lead component procurement for the flight actuators is underway. Flight actuator fabrication is scheduled to begin in the last quarter of 2006. The design of the actuator was extremely challenging but test results show that optical level positioning can be reliably achieved using simple mechanical components. The unique combination of the patented fine stage flexure with the coarse coupling proved to be quite effective in achieving fine accuracy over a long range of travel. Mass production techniques greatly simplified the design and assembly of the actuator. Reliable operation was achieved by the use of robust components and supporting analysis. Acknowledgements This paper presents the design of the Cry0 Actuator for JWST. The design of the actuator is the result of important contributions from numerous people and organizations. I would like to acknowledge the contributions from those at Ball Aerospace including Robert Slusher, Scott Streetman, Lana Klingemann (nee Kingsbury), William Schade, Mike Matthes, Bruce Hardy and all the technicians, machinists, engineers and administrators. I would also like to acknowledge the outstanding work of outside vendors & suppliers including CDA Intercorp, All American Gear, New Hampshire Ball Bearing, Barden Bearing, ATK-Able, Next Intent, Schaevitz and Beaver Aerospace. The photographs in Figures 8, 11 & 15 are courtesy of Able. Finally, I would like to thank our customer, Northrop Grumman Space Technology (formerly TRW) and NASA Goddard. References 1. R. Slusher, “Motion Reducing Flexure Structure”, US. Patent Number 5,969,892, Oct. 19, 1999. 2. S. Streetman, L. Kingsbury, “Cry0 Micropositioner”, U.S. Patent Number 6,478,434, Nov. 12, 2002. 3. Scott Streetman, Lana Kingsbury, “Cryogenic Nano-positioner Development and Test for Space Applications”, in IR Space Telescopes and Instruments, Proceedings of SPlE Vol. 4850, Sep. 2002. 4. Lana Kingsbury, Paul Atcheson, “JWST Testbed Telescope”, in Optical, Infrared and Millimeter Space Telescopes, Proceedings of SPlE Vol. 5487, June, 2004. 252

Document

AMS_2006.pdf

Space Shuttle Body Flap Actuator Bearing Testing for NASA Return to Flight Timothy R Jett', Roamer E. Predmore'*, Michael J. Dube" and William R. Jones, Jr+) Abstract The Space Shuttle body flap (BF) is located beneath the main engine nozzles and is required for proper aerodynamic control during orbital descent. The body flap is controlled by four actuators connected by a common shaft and driven by the hydraulic power drive unit. Inspection of the actuators during refurbishment revealed three shaft bearings with unexpected damage. One was coated with black oxide on the balls and race wear surfaces, a second contained a relatively deep wear scar, and the third with scratches and an aluminum particle in the wear track. A shaft bearing life test program was initiated to measure the wear life and explain the 5.08-micrometer wear scar. A tribological analysis was conducted to demonstrate that the black oxide coated wear surfaces did not damage the bearing, interfere with the lubrication, or cause severe bearing wear. Pre-damaged (equivalent of 30 missions), commercial equivalent bearings and previously flown shaft bearings were tested at axial loads, speeds, and temperatures seen during flight operations. These bearing were successfully life tested at 60°C for 24 hours or 90 flights. With a safety factor of 4X, the bearings were qualified for 22 flights when only a maximum of 12 flights are expected. Additional testing at 23°C was performed to determine the lubricant life and to further understand the mechanism that caused the blackened balls. Test results indicating bearing life was shortened at a lower temperature surprised the investigators. Start\Stop bearing testing that closely simulates mission profile was conducted at 23°C. Results of this testing showed lubricant life of 12 flights including a safety factor of four. Additional testing with bearings that have the equivalent of 30 missions of damage is being tested at 23°C. These tests are being performed over the Shuttle load profile to demonstrate the residual bearing life in the actuators exceeds 12 missions. Testing showed that the end of the shaft bearing life was characterized by bearing temperature rise, preload drop, and the onset of a severe wear bearing failure mechanism. The severe wear failure mechanism is characterized by rough wear scars, extensive bearing wear and steel transfer between the balls and the races. Introduction The BF is part of the Space Shuttle Orbiter control system that operates primarily during the critical descent maneuvers. The Shuttle body flap is supported and controlled by four body flap actuators (BFA) sharing a common, segmented drive shaft. The common shaft is driven by the hydraulic power distribution unit. During inspection of these actuators, one of the input shaft bearings was discovered to have blackened balls and a blackened wear track in the race after the actuator had completed approximately 20 missions [I, 21. It was deemed unacceptable for service, replaced and the used grease was discarded. Visual examinations of most of the BF actuators in the fleet revealed no evidence of blacken balls or race wear tracks in the shaft bearings. The used shaft bearings were cleaned, inspected, re-lubricated and reinstalled in the actuators or scrapped. Subsequent microscopic inspection and metrology of these shaft bearings revealed significant wear and possible plastic deformation in a second shaft bearing [3]. A third shaft bearing was found to have scratches and an aluminum particle in the wear track. To determine if actuators with used bearings were acceptable to re-fly for twelve missions, a bearing test program was initiated at the Marshall Space Flight Center. Lf NASA Marshall Space Flight Center, Huntsville, AL + NASA Goddard Space Flight Center, Greenbelt, MD ++ SWALES\Glenn Research Center, Cleveland, OH SWALES/Goddard Space Flight Center, Greenbelt, MD Proceedings of the 38 Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 253

Document

AMS_2006.pdf

Body Flap Actuator Background The Body Flap actuator is shown in Figure 1. Several failure modes can end the life of the BF actuator bearings. Fatigue spalling, fretting fatigue, lubricant failure, high temperature oxidation, corrosion pitting, severe wear or combinations of these mechanisms can cause bearing failure. Corrosion pits or wear damage accelerate fatigue crack growth that can cause fatigue failure before the bearing reaches the B1 or B10 fatigue life. If the bearing wear life or the Braycote 601 grease life is exceeded, the bearings can fail. However, because the actuators possess very high driving torques, the worst bearing failure mode was expected to be a broken ball or bearing race that could jam the actuator as well as the common drive shaft and cause an actuator performance failure. The design criteria for the space shuttle (OV) BFA are 100 missions and 10 years. During the 1990s, excessive BF flutter during launch and excessive BF housing corrosion led to disassembly and inspection of BF actuators. In October 2002, bearings from the SIN 402 and SIN 405 actuators were cleaned and microscopically inspected to find out if ball bearings in the BFA assemblies needed to be replaced. As part of that investigation, shaft bearing PIN 5902050, SIN V6L009, from body flap actuator SIN 402, was found to possess unusually dark wear surfaces when compared to the rest of the bearings under examination (Figure 2) [I, 21. NASA's Engineering and Safety Center (NESC) became aware of this issue and initiate an Independent Technical Assessment to assess this bearing and all other bearings in the Shuttle's Rudder Brake and Body Flap Actuators [4]. qw aonn D nc~rumn snw~ INTEGRAL BAL BEARINGS OUTCLW PIANEf OUR C mnb OUTPUT HOUSING (STATOR) NPUT HOUSING CmTER HOUSINO (flAl-0 R) mom R) Figure 1. Space Shuttle Body Flap Actuator - Figure 2. Body Flap Shaft Bearing S\N V6L009

Document

AMS_2006.pdf

Analysis of BFA Bearing with Black Balls Visual inspection of P/N 5902050, SIN V6L009, revealed a black, tribological coating on the balls and race wear tracks (Figure 3). It should be noted that no other bearing in the Body Flap or Rudder/Speed Brake actuators was found to contain the black, tribological coating on the wear surfaces. Minor corrosion was noted on the inner and outer raceways. The grease present in the bearing was brown in appearance due to the tribological decomposition of the Braycote 601 grease and iron oxide wear debris. The recommendation based on microscopic examination analysis was to replace the bearing since the wear surfaces were black and it suffered Brinell damage during disassembly. I Figure 3. Black bearing after sectioning and cleaning Note black balls and the dark bands present on the raceways. Further work performed in 2004 by Hamilton Sundstrand, Inc. reported that the inner ring contained a black tribological coated band. The black band or wear track possessed a smooth matte finish with a grain boundary like structure. Outside the black wear track, the original surface was still evident along with the original finish marks. The outer ring contained a similar black coated wear band (Figure 4). The wear surface of the outer race was similar in appearance to the wear surfaces present on the balls. Figure 4. Magnified views of the black wear tracks in the inner race (left) and outer race (right) are shown. Note the false brinelling marks. The dark brown color is an optical aberration of the black color. Scannina Electron MicroscoDv GEM) Scanning electron microscopy was performed on segments of the black ball bearing by various groups [5]. The segments examined included the balls, inner raceway and outer raceway. Analysis of the balls by energy dispersive spectroscopy (EDS) revealed that the black tribological coating was composed of carbon and iron oxide. The coating was porous and appeared to be uniform over the entire ball surface. A few skid marks were noted (Figure 5). In addition, each ball possessed a moon-shaped spot that appeared to have penetrated the burnished coating. The spot was probable caused by a drop of rain 255

Document

AMS_2006.pdf

water absorbing corrosive chemical from the degraded grease and dissolving the black coating during Shuttle storage. Figure 5. Scanning electron micrograph of black ball showing spot (left) and skid marks (right) Figure 6. Scanning electron micrographs of the black ball (left) and magnification of particle present on the ball surface (right) Higher magnification of the ball revealed some particles present on the ball (Figure 6). The presence of aluminum and silicon, in addition to other elements detected by EDS on these particles, were likely attributed to the bentonite clay additive present in the Braycote 601 grease and not contamination as was originally proposed. Analysis of the inner and outer raceways showed surfaces similar to the black ball. EDS performed on those surfaces detected the carbon iron oxide coating found on the black balls. Examination of the inner raceway away from the burnished band showed the original finishing marks which gradually translated into a smoother region as the burnished area was approached (Figure 7). Figure 8 shows the surface for the coating present on the black ball obtained by surface profilometry. Figure 7. Scanning electron micrograph of the outer raceway in the region above the burnished band (left), and within the burnished band (right) I 256

Document

AMS_2006.pdf

1 --- 3 3311 , , , frnl7l 000 mm 0 361 Figure 8. Surface profilometry trace showing the coating present on the black ball including a segment of the grain boundary-like structure in the center Raman SpectroscoDy Raman spectroscopy of the black balls showed carbonaceous peaks from the carbon coating on the surface of the complex iron oxide, black, tribological coating [6]. Specifically, the diamond, D, and graphitic, G, peaks were observed confirming the presence of amorphous carbon (Figure 9). Figure 9. Raman spectrum for black ball showing the graphitic (G) and diamond (D) peaks associated with amorphous carbon. X-rav Photoelectron Spectroscopv [XPS) XPS was performed on the black ball at three different locations and on the inner race. Following the initial scans, depth profiling was performed by application of fifteen sputtering cycles, each lasting ten seconds, and a total sputter time of 450 seconds per location. The results show a carbon-rich coating on top of an iron oxide-rich surface. The carbon-rich top surface forms during the tribological decomposition of Braycote 601 grease and contains fluorine and sodium from the grease [7, 81. The Raman spectrum of the carbon-rich coating is reported in the Raman spectroscopy section. Below the carbon-rich coating is the black tribological iron oxide-rich coating that also contains fluorine from the grease. At the end of the sputtering cycles, sodium was no longer present and chromium was observed in both locations in addition to carbon oxygen, fluorine, and iron (Figure 10). Metalloaraphv A black ball and a segment of the race from the blackened bearing, S/N V6L009, were metallographic cross sectioned and the tempered martensitic 521 00 steel and black tribological coating examined (Figure 10 and 11) [5]. The microstructure of the black ball was consistent with properly processed 52100 temper martensite with finely dispersed carbides with a hardness of R, 64-65 from the surface into the ball. Tapered sections showed no evidence of annealing below the black tribological coating therefore the 257

Document

AMS_2006.pdf

temperature of the ball did not go above 204°C during the formation of the black tribological coating. The black coating appears to be a micron or two thick. Pits present on the cross-section were very shallow. 11K 10% 1M 9% LI.( 0.x BK 7.5K 7K 6.W 5w U 6 5.W BW 5 4% 4n 3s jn 23 w I.€# 1H 5a: 11 6 1-6 -6 EYE6 7W6 8p8.6 5-6 A986 3986 -6 lW6 W6 mw harw lev1 Figure 10. An XPS spectrum for black ball I Figure 11 a. Metallographic section of black bearing 258

Document

AMS_2006.pdf

1 Bearing Serial Number of Wear Depth Range Number Flights (micro meter) V6M002 20 5.08 V6L009 20 2.23 V89E006 19 2.23 V6L007 30 2.03 V6L025 32 1.27 V6M014 30 2.03 V6L013 32 Wear debris build- UP V6L001 19 .38 -. ::. . . . . . .- , . . ..> Microscopic Observations Mild discoloration Black complex oxide on balls and wear track Light gold color discoloration Superficial rust stains Slightly darken balls Moderate surface distress corrosion Moderate surface distress, discoloration likely to lubricant degradation Ball banding and mild discoloration and balls and races Figure 11 b. Metallographic section of black bearing Metrology of Previously Flown Flight Bearings After the discovery of the bearing with the blackened balls, seven additional BF shaft bearings were microscopically inspected and the wear depth profiles were measured. Wear was measured using a form- Talysurf measurement machine. Microscopic examination and wear depth results for these bearings are reported in Table 1. The bearings had mild surface distress but no spalling. The bearings appeared to be highly loaded with ball tracks running near the raceway shoulder. The balls showed discoloration, superficial rust stains but only S/N V6L009 was contained the black tribological coating on the ball and race wear surfaces. Metrology results show that Shuttle flights produced wear depths ranging up to 5 micrometers. The most damaged bearing was S/N V6M002. This bearing had 5 micrometers maximum wear on the inner race with evidence of plastic deformation in raceways. Table 1. Results of Metrology of Flight Body Flap Actuator Shaft Bearings 106 259

Document

AMS_2006.pdf

Bearing Testing of Body Flap Actuator Bearings A pre-existing test rig (Figures 12 and 13) was adapted to conduct life tests on the body flap output Shaft bearings (Size 106). The bearing tester supported an angular-contact bearing on each end of the drive shaft. Axial load was applied to the bearing pair and shaft by a locking mechanical screw. A drive pulley was attached to the middle of the drive shaft and was driven by a variable-speed AC motor. The bearing housing temperature was controlled by circulating ethylene glycol through coolant passages in the bearing test housings, which raised the test bearing temperature up to 70°C. During testing, the bearing axial load, shaft rotational speed, motor amps, and temperature of bearing outer races were monitored and recorded. I Figure 12. MSFC Bearing Test Rig - Figure 13. Close-up view of MSFC Bearing Testing Rig 48 Mission Life Test Results at 60°C The objective of the testing was to demonstrate that previous flown and re-lubricated was acceptable for 12 additional shuttle flights with a safety factor of 4. The temperature of 60°C was selected because it 260

Document

AMS_2006.pdf

was thought that Braycote 601EF grease was shorter at higher temperature and 60°C was the highest and only temperature recorded in BF actuator system, namely in the power distribution unit. Test Serial Number Max. Wear Depth during Flight Simulation (micro meter) BF-001 .66 BF-003 4.45 The life test conditions for a single mission were derived from the flight load spectrum for the body flap actuator as understood early in the test program. The body flap qualification test load spectrum was made worst case to establish the single-mission life test load profile for the body flap input shaft bearings shown in Table 1. MSFC Life Max. Wear after 48 mission (micro meter) Test Number life test LT002 2.54 LT002 5.72 Table 2. Single Mission Load Profile With a limited supply of post-flight bearings available, initial test were performed on commercial bearings with similar dimensions to the flight bearings. Balls and cages were removed from the commercial bearings and 18 matched balls were added to make full complement test bearings, Le., no ball separators. As shown in Table 1, S/N V6M002 had a maximum wear of 5.08 micro meters, which was higher than the other flight bearing. Two commercial bearings were loaded to 26700 N and operated for 10 minutes at 72 rpm to produce wear similar to wear measured in Shuttle flight bearing after 20 to 30 missions, After this testing, the maximum wear on a test bearing (BF-003, Table 3) was found to be 4.45 micro meter. These damaged bearings were cleaned, re-lubricated with Braycote 601 EF and successfully tested for 24 hours or 48 mission life at 60°C. The bearings successfully completed this testing and about 1.27 micrometer of additional wear was observed BF-003. During testing no evidence of degradation of bearing performance was observed. Post-test inspection and metrology of the bearings showed moderate additional wear in the bearings as shown in Table 3. A photograph of a ball from BF-003 is shown in Figure 14. The photograph showed some banding and blue discoloration due lubricant degradation. The test lubricant was found to be in good condition. It was creamy white in color with some browr rnrl arm=,= ,-=~lfpd +ha p=rlll stage of lubricant degradation. I Figure 14. Ball from BF-003 after 48 Mission life test 26 1

Document

AMS_2006.pdf

Two additional 48 mission life tests were performed on previously flown flight bearings. The results of this testing are shown in Table 4. Both tests were successfully completed. No indication of lubricant failure or severe wear was observed. During the life testing, all bearings performed without anomalies and no evidence of torque increase was observed with increasing test time. In addition, no rise in bearing operating temperature was observed, which also indicated no gross degradation in bearing performance. Post-test inspection of these bearings showed moderate additional wear (Table 4). No evidence of lubricant failure, severe wear and or fatigue spalling was observed. Test Serial Number V6M014 V6L013 V6L001 Table 4. Wear for flight bearings after Shuttle Flights and after 48 missions of Additional Life testing Max wear before test Max wear 48 mission life test MSFC Life Test (micro meter) (micro meter) Number 2.03 2.03 LT003 Wear debris build up .51 LT004 .38 Wear debris build UD LT004 After testing, the lubricant appeared to be in good condition. The grease in the bearings was still creamy white with only a few areas of reddish brown discoloration (Figure 15). Balls from flight bearing showed significant banding due to normal wear and possible lubricant degradation (Figure 16). - after 4imission cfe test Figure 16. A ball from flight bearing V6L001 after 48 Mission of life tests Results of this 48 mission life test showed that 3 previously flown flight bearings and 2 pre-damaged commercial bearings successfully passed life testing for 24 hours or 48 missions at 60°C. Based on these results, it was recommended that freshly re-lubricated, pre-flown body flap shaft bearings were acceptable for 12 more flights on the Space Shuttle with safety factor of 4. When the average operation time per mission was updated to 16 minutes per mission, the bearings were found to be acceptable for 22 missions at 60°C. This recommendation was based on the assumption that 60°C was the worst-case condition for bearing lubricant life. Subsequent testing proved this assumption to be incorrect. Life Testina of the Bodv FlaD Shaft Bearinas The 48 mission life testing did not reproduce the black tribological coating on shaft test bearings, so testing with partially lubricated bearings, different temperatures, speeds, and stop-start speed profiles was continued in an attempt to produce the black tribological coating on the bearing wear surfaces, i.e. SIN V6L009. The investigators were surprised to find the life of the bearings to be 10 to 20 hours rather than the 66 hours originally required for the BF actuators. Shaft bearing life testing was redirected to measure the life of the bearing and demonstrate that 12 more Shuttle missions could be successfully accomplished. In general, it was assumed the blackened balls were generated due to high local temperature, possibly due to lubricant degradation, and\or starvation or from an external heat source during re-entry. The objective of this testing was to generate blackened ball similar to those observed on BFA shaft bearing 262

Document

AMS_2006.pdf

SIN V6L009. A second objective of the testing was to determine the lubricant life of Braycote 601 grease used in BFA shaft bearings. Life Test Conditions: 1. Size 106 commercial 52-1 00 steel thrust bearings were life tested. 2. A shaft with bearings mounted on the ends was thrust loaded with a locking mechanical screw. 3. The drive shaft was rotated by an electrical motor driving a pulley mounted on the shaft. 4. The bearings were tested in air at room temperature. 5. The criteria for end of bearing life were the onset of a dropping preload, bearing temperature rise, laboring of the drive motor and the onset of severe wear in the bearing. The initial testing was performed at room temperature with continuous shaft rotation at 50 rpm. Tests were run at several thrust load levels. The tests were allowed to run until lubricant failure was detected. The results of testing are shown in Table 5. Table 5. Body Shaft Bearing Life Test Results Using Commercially Equivalent Bearings Bearing failures (lubricant failure) were observed much earlier than expected. These failures occurred much sooner than in previous 60°C, 48 mission life tests where no failures were observed. The mechanism continued to operate, but significant drops in axial load and bearing temperature increases were observed as well as copious amounts of wear debris. Post-test microscopic examination of the bearings showed two distinct wear modes. 1. Normal Wear Mode Characteristics a. Race Appearance - Microscopic examination of the wear tracks in the races revealed a smooth matte finish. In some cases, the wear track was bronze colored. The bearing fabrication grinding marks were worn away. b. Ball Appearance - The balls possess a shiny finish like a new ball. The shiny balls surface are often coated with a thin smeared black film covering the balls or concentrated in a wear ring pattern. The wear rings are produced when the rolls around a single axis. 2. Severe Wear Failure Mode (typical of failed bearings) a. Race Appearance - Microscopic examination of the race. wear tracks showed very rough surface. Wear scars form across the direction of ball motion. At higher loads and longer bearing lives, radial wear scars develop between the wear tracks and the shoulder and metal wear debris deposits on the bearing shoulder. Steel wear debris transfers from ball- to-race and/or race-to-ball. As the bearing test time in the severe wear mode increased to about 60 minutes, the transferred wear deposits turns black. The wear depth increased from 3 micrometers for normal wear to about 80 micrometers for severe wear. b. Ball Appearance - undulating rather than smooth ball surfaces were observed. As the test time in the severe wear mode increased to about 60 minutes, the balls turned black. 263

Document

AMS_2006.pdf

Apparently, the black surface was formed when the worn balls heated up and oxidized during the last few minutes of the test. In general, failed bearings exhibited severe wear. A failed bearing is shown in Figure 17 and 18. The onset of wear started when the lubricant between the ball and race was completely consumed or degraded. Shear plastic deformation occurred between the race and ball steel to cause adhesive wear or severe wear. Due to high friction and heat generation, the local ball temperatures rapidly increased and caused oxidation and discoloration of bearing balls. In some case, the balls were highly discolored and almost black. The test lubricant in the failed bearings was very dark in color with significant degradation and metallic wear debris. I Figure 17. Failed Test Bearing BF012 Figure 18. Failed Test Bearing BF012 (disassembled) Darkened balls that were produced in testing were not identical to the black ball from V6L009. Scanning Electron Microscope examination of a failed test ball showed a texture surface. The surface of the test bearings did not have a polished finished, but appears to have been smeared with transferred steel. Some microspalling and surface cracking was observed. The flight bearing with the blackened balls (S/N V6L009) was in fairly good shape. This flight bearing did not exhibit severe wear. Balls had smooth surfaces and the raceway wear track was covered with smooth black oxide coating. Life Testina at Elevated TemDerature Due to these unexpected results of lower life at room temperature, life tests at elevated temperatures were performed to confirm this trend. Tests were run at 60°C with 15569-N axial load at 50 rpm continuous shaft rotation. One test was run at 70°C under the same conditions. The same bearing test rig as previously described was used. The results are shown in Table 6 and Figure 20. Table 6. Results of Continuous rotation testing at elevated temperature 264

Document

AMS_2006.pdf

NESC Bearing Testing Effect of Temperature on Life 250000 50W 0 Room Temp, 50 rpm, Room Temp, 50 Room Temp, 50 140 F, 50 rpm Continous 165 F,80 rpm Continous 15569N rpm,20017N rpm,22241 N rotation, 15569N rotation, 15569N Figure 20. The effect of temperature on lubricant life The results of this testing indicated that temperature had a significant effect on bearing life. Increased temperature yielded a significant increase in bearing life. The cause of this effect has not been determined. One theory is the higher temperature afforded better lubrication either by a decrease in apparent grease viscosity or by an increase in oil separation rate. This resulted in more oil available in wear contact area thus extending bearing life. STOP\START Testinq at Room TemDerature The testing to determine the lubricant life was performed with continuous rotation. This differs from the 48 mission life tests that had been previously performed in several ways. In the 48 mission life testing, the bearing ran at several different speeds and loads. Also, in actual BFA operation the actuator is operated for short periods of time (several minutes) then sits stationary for extended periods (months). To evaluate the effect of startbtop operation on the extension of bearing life, a series of stop\testing was performed using the same bearing test rig as previously described. For this testing the bearing rig was run for 30 minutes at 80 rpm shaft rotation and then held idle for 30 minutes. This cycle was continuously repeated until bearing failure occurred. For a baseline comparison three additional tests were conducted with continuous rotation at 80 rpm. The results of this testing are shown in Table 7 and Figure 21. These results showed start\stop operation has a moderate effect on lubricant life. The average bearing life for stopktart tests was 92500 revolutions which was about 30% greater than the average bearing life for the continuous rotation testing, most probably because surface tension pulls lubricant into the wear track while the bearings were stopped. These results also show that shaft speed had moderate beneficial effect on bearing life most probable because more lubricant was squeezed out of the ball race contact at lower speeds. 265

Document

AMS_2006.pdf

Table 7. Results of Startistop lubricant life testing NESC BFA Bearing Testing I 140 F, Mission Profile, 48 140 F, 50 rpm Continous Room Temp, 50 rpm, Room Temp, 80 rpm, Room Temp. 80 rpm, Missions (no failures) rotation Continuous rotation Continous rotation stop\stati Figure 21. Results of startbtop lubricant life testing The start\stop testing at 80 rpm most closely simulated the actual BFA operation. The majority (92%) cycles on the BFA are applied during ground operation under stop\start conditions. All of these cycles occurred at 80 rpm with 15569-N axial load at room temperature. The remaining cycles occurred during flight. The BFA was never operated under maximum load (22241-N load). This load occurred during ascent when the actuator is in a lock position. The actuator was operated for several seconds in space to test the actuator, but the remainder of cycles occurred during re-entry where temperature probably 266

Document

AMS_2006.pdf

reaches around 60°C. Recent reexamination of the BF actuator operation time line by the NESC showed that the actual average mission actuator operating time was 16 minutes (1280 cycles) rather than the 24 minutes. Based on these facts, the shortest lived startktop test of 12.8 hours (61.4 cycles) was used to life the BFA shaft bearings. This calculation yielded a bearing life of 12 missions including a safety factor of 4. Conclusions 1. 2. 3. 4. 5. 6. 1. 2. Initial test results of full scale bearing testing at 60°C indicated that previously flown and re-lubricated Body Flap Actuator shaft bearings were acceptable for 22 Shuttle Missions with a safety factor of 4. Subsequent bearing testing at 23°C show the lubricant life is reduced at lower operating temperatures. Results of full scale bearing stop\start testing at 23°C showed that new bearings were acceptable for 12 Shuttle Flights with a factor safety of 4. Efforts to duplicate the blackened balls observed on one of the Shuttle Body Actuator shaft bearings were unsuccessful. The blackened balls were produced in testing, but only after lubricant failure and severe wear are initiated. SEM analysis showed that the test blackened balls are not very similar to the flight ball. The test blackened balls were in much worse condition and had a more textured surface with smalls cracks observed. The flight bearing with blackened balls did not suffer a lubricant failure. The flight blackened bearing was in relatively good condition with smooth wear surface. The Raman Spectroscopy, XPS and EDS analysis of the flight black ball detected the presence of carbon-rich layer on an iron oxide-rich layer and both with fluorine. The detection of amorphous carbon, fluorine and FeF3 in both the carbon-rich surface layer and the iron oxide under layer were significant since local degradation of the PFPE lubricant appeared to be in part responsible for the black coating observed on the balls and raceways of P/N 5902050, S/N V6L009, bearing removed from S/N 402 BF actuator. The origin of the thicker, black iron oxide-rich layer under the carbon-rich layer is not well understood. Because it occurred only on the wear surfaces, the tribological chemical reactions in the ballhace wear interface did cause the black coating. Most investigators felt that high temperatures below 200°C contributed to the black coating formation. Local lubricant starvation within the ball path of the bearing may have generated local heating with temperatures high enough to cause the remaining lubricant film to degrade and react with the 52100 steel surface, effecting formation of a complex oxide layer that resulted in the black coating on the wear surfaces. While the system may have recovered locally after the blackened coating event, the oxide layer formed was apparently effective in separating the working surfaces of the bearing, preventing its failure, and serving as a surrogate solid film lubricant. The strength, geometry, metallurgical structure and wear of the black coated bearing were found to be essentially identical to other BF shaft bearings after Shuttle flights. No evidence of severe wear was observed. Therefore, shaft bearings with black tribological coatings like S/N V6L009 are expected to perform successfully for at least 12 more Shuttle missions. The severe wear mode caused failure of the shaft bearing and determined the bearing life. The onset of severe wear is thought to be caused by ball/race metal adhesion after all the Braycote 601EF grease is depleted from the interface. Lessons Learned When space mechanisms are qualified by accelerated life testing, bearings should be qualified by real-time life testing. Bearings enduring the highest stresses, longest operating lives or highest temperatures in the mechanism application should be life tested as part of the qualification process. Bearing life and lubricant life was found to be shorter at lower temperature even though the lubricant tribological decomposes or degrades more rapidly at higher temperature. 267

Document

AMS_2006.pdf

Acknowledgments The authors thank F. Hernandez, J. McManamen and J. Figert of Johnson Space Center; B. Heitzman and L. Salvador of Boeing; P. Munafo, P. Hall. L. Moore, H. Gibson, M. Sharp, and S. Cat0 of Marshall Space Flight Center; E. Devine of Swales Aerospace; E. Zaretsky and K. Street of Glenn Research Center; J. York, D. Hill and S. Tollefson of Hamilton Sunstrand; and M. Sovinski of Goddard Space Flight Center for their assistance. Without this team’s support, this task could not have been accomplished. References 1. 2. 3. 4. 5. 6. 7. 8. L. Hughes, Hamilton Sundstrand; Destructive Analysis of Space Shuttle Body Flap Actuator Blackened Ball Bearing, P/N 5902050, S.N V6L009, Memorandum Report 66852, February 17,2004 J. York, J. Sikes, L. Hughes, Hamilton Sundstrand; Overhaul Inspection Findings for Components from Space Shuttle Body Flap Right-Hand Inboard Actuator Unit S/N 402, P/N 5003661A, Report 66442, Rev. A, January 24,2003 J. York, J. Sikes, L. Hughes, Hamilton Sundstrand; Overhaul Inspection Findings for Components from Space Shuttle Body Flap Left-Hand Inboard Actuator Unit S/N 401, P/N 5001 41 B Body Flap and Rudder Speed Brake Actuator Bearing Independent Technical Assessment Final Report, NASA Engineering and Safety Center, NESC ITA 04-076-1 S. N. Cato, Marshall Space Flight Center; Orbiter Body Flap Actuator Bearings K. Street and M. Sovinski; Analysis of Black Bearing Balls from a Space Shuttle Body Flap Actuator, Proceedings of World Tribology Congress Ill, Washington, D. C. September 12-16,2005 P. Herrera-Fierro, M. Masuko, W. Jones, S. Pepper, Glenn Research Lab.; XPS Analysis of 440C Steel Surfaces Lubricated with Perfluoropolyethers Under Sliding Conditions in High Vacuum, NASA Tech. Memo. 106548, April 1994 W. Jones, Glenn Research Lab and M. Jansen, Univ. of Toledo; Lubrication for Space Applications, NASNCR 2005-21 3424, January 2005 268

Document

AMS_2006.pdf

Bearing Development for a Rocket Engine Gimbal Christian Neugebauet, Manfred Falkner*, Ludwig Supper* and Gerhard Traxler* Abstract The bearings for a gimbal of a cryogenic upper-stage rocket engine are highly loaded, they are exposed to corrosive environment, and they have to cope with a wide temperature range down to cryogenic temperatures. A tailored single-row full-complement needle roller bearing design integrated by press fit mount technology was selected for this specific application. This paper describes the design considerations, the analysis approach, and the development test results of these bearings. Introduction Austrian Aerospace has the privilege to develop the gimbal for the next generation upper-stage engine of the ARIANE 5 Launcher family, the VlNCl engine. One of the key components of the gimbal are the bearings which - - - - are highly loaded by the engine operating loads and by inertial loads, are limited in space and mass due to stringent mass and envelope requirements, must tolerate considerable misalignments of the axis due to elastic deformation of the gimbal under mechanical load, and have to cope with harsh environmental loads and cryogenic temperatures. The chosen bearing is a tailored needle roller bearing with specially crowned needles. To transmit the high loads, the bearing pitch circle diameter should be as large as possible. This leads to a design with a bearing integrated by press fit mount technology to allow for an undivided bearing support to find a space and mass optimized assembly design. The design and dimensioning of the press fits was driven by the following constraints: - - - - - Maintenance of proper fit under any thermal load-case to prevent loosening of the bearing rings in the support when experiencing thermal gradients, The reduction of the bearing’s radial clearance during integration caused by the press fits, Elastic deformation of the race along the bearing axis due to the unequal stiffness distribution of the bearing housing. Such deformations may not disturb the radial clearance of the bearing. Roundness deviations of the race due to radial elastic deformation caused by the unequal stiffness distribution of the bearing housing. The unequal stiffness distribution of the bearing housing and the thin sectioned design also causes roundness deviations caused by the manufacturing process of the housing that sums up with radial elastic deformations due to the press fit. The design must allow the introduction of considerable axial integration forces during bearing insertion. - The development test results confirmed the chosen design. * Austrian Aerospace GmbH, Vienna, Austria Proceedings of the 3dh Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2005 269

Document

AMS_2006.pdf

Description of the Mechanism Bearing type Material of rings and rollers Lubricant Supplier of the bearing The main functions of the gimbal are to fix the engine in its defined position, to transfer the thrust and all inertial forces from the engine to the stage and to allow the gimbal operation. The gimbal consists mainly of an engine-side bracket, the gimbal cross, and the stage-side bracket, connected by two pairs of bearings. The distance between the gimbal interface plates amounts to 150 mm. Four single-row needle roller bearings allow gimbal motion about two axes. The development of these bearings is described in this paper. Single row needle roller bearing Cronidur 30 Sputtered Lead by ESTL, UK FAG Kugelfischer AG, Germany A i Figure 1. CAD model and development model of the Gimbal, with uncovered bearings Bearing Type, Layout and Design Table 1 presents the main parameters of the bearings described in this paper. Table 1. Bearing Parameters Bearina TvDe At an early stage of the development, the decision for rolling bearings was made due to the low friction torque and the low wear of rolling bearings. The decision for a needle roller bearing was driven by the load capability of the bearing at a given envelope restriction. This lead to a one-row full-complement needle roller bearing design. The race of the outer ring is equipped with borders to position the needle rollers in the axial direction. As no lateral force transmission by the bearing is required, the races of the inner ring are not equipped with borders. 270

Document

AMS_2006.pdf

Diameter / Roller Lenath Ratio To achieve a certain static load rating, the bearing diameter increases if the needle length decreases. As considerable axial deflection between inner ring and outer ring was expected, a design with short needle lengths was chosen combined with a bearing diameter as large as possible to still have high load capability even if axial deflection occurs by spreading the load to a higher number of needles. Using shorter needles reduces also the impact of edge pressing. Needle Roller Diameter Investigations on the needle roller diameter were performed. It turned out that, computing the static load of needle bearings with different needle roller diameters while keeping the outer race diameter constant and also keeping the needle length constant, the load capability increased with decreasing needle roller diameter: However, the static load capability is not the only parameter to be taken into account. The friction moment of the bearings increases with the number of rolling elements. As this effect can not be numerically evaluated for needle bearings we relied on experience from previous needle bearings. The ratio between roller diameter and roller length increases with higher roller diameters leading to less friction caused by misaligned tilted rollers. Bearina Outer Rina Thickness To distribute tensioning loads, lateral loads, and torsional loads of the gimbal on a high number of bearing needle rollers, the stiffness of the load-transferring elements, which are the bearing housing and the rings, should be as high as possible. A comparison of different thickness ratios between bearing outer ring thickness and housing thickness within the envelope constraints has been performed. The inner diameter of the bearing outer ring and the outer diameter of the bearing housing (formed by the yoke) have been assumed as fixed. The comparison showed that the smaller the bearing ring thickness, the more stiffness is gained. The higher Young’s modulus of the bearing steel Cronidur 30 (208000 N/mm2) compared to the Young’s modulus of the housing of Ti-GA14V-ELI (1 10000 N/mm2) has less effect on ihe stiffness than a thicker rectangular section of the yoke of Ti-6AI-4V-ELI. No shear force transmission was assumed between the bearing and the housing in this comparison. Following this, the bearing outer ring thickness has to be as small as possible. The lower limit for the bearing ring thickness is a manufacturing issue. The ring thickness must stay above 2 mm according to the bearing supplier. Bearina Inner Rina Thickness The bearing inner ring js supported stiffly by the gimbal cross, which is made of lnconel 718 (Young’s modulus 205000 N/mm ). Therefore, load transfer is not a driver for the inner ring thickness. The design driver for the bearing inner ring thickness was integration issues. The gimbal cross insertion process requires a relatively small shaft diameter. Bearina Sizinq The bearing pre-dimensioning was done keeping in mind above considerations and using standard formulas fo;full-complement roller bearings. The final bearing calculation (including Hertzian pressures in the roller needles) was then done using the results of the FEM calculation. A coefficient of static load f, of 0.7 was assumed as the application exhibits small angular oscillations and shock loads. The largest bearing that fits in the room given and having the necessary static load capacity was selected. Bearina Rina Desian The bearing inner side is not covered by the gimbal, so borders were designed to prevent intrusion of contamination. Shoulders on the outer and on the inner rings limit the travel during the axial insertion process. Figure 2 shows the gimbal bearing design: 271

Document

AMS_2006.pdf

"I Outer Rina Press Fit The operating temperature range of the gimbal is between 120 K and ambient temperature. A compression fit between the outer bearing ring and the housing, and the inner bearing and the shaft is desired at any temperature. The material of the housing is Ti-6AI-4V-ELI. The bearing steel, Cronidur 30, has a higher coefficient of thermal expansion than Ti-6AI-4V-ELI. This leads to a decrease of compression at lower temperatures as the bearing ring shrinks more than the housing. The compensation of this effect is done by a proper dimensioned compression fit between the bearing housing and the bearing outer ring. The dimensioning of this compression fit must also take into account the worst-case manufacturing tolerances; sufficient compression must still be granted even with the smallest outer ring diameter and the largest housing bore diameter. Combining the greatest outer ring diameter with the smallest bore diameter, the integration force of the outer bearing ring will increase. To keep this increase as small as possible, tight tolerances are given: P5 (9 pm) on the bearing outer ring and IT5 (13 pm) (in accordance with DIN 7151) on the housing bore. The housing bore diameter was chosen to still have compression even with the smallest outer ring diameter and the largest housing bore diameter at the lowest expected temperature. The seat dimensioning provides a compression of 8...30 pm at ambient temperature and 2...24 pm at the operating temperature of 120K. Inner Rina Press Fit A compression fit between the bearing inner ring and the shaft is desired. The material of the shaft is lnconel 718. lnconel 718 has a higher coefficient of thermal expansion than Cronidur 30. So the same thermal expansion situation as occurs between the bearing outer rings and the yokes also occurs for the inner bearing ring seat on the shaft of the gimbal cross. Also, for the inner ring fit, a compensation using a compression seat with tight tolerances is used: P5 (8 pm) on the bearing inner ring and IT4 (7 pm) on the shaft. The inner ring bore diameter is chosen to still have compression even with the smallest shaft diameter and the largest bearing bore diameter at the lowest expected temperature. The seat dimensioning 272

Document

AMS_2006.pdf

provides a compression of 21 ... 36 pm at ambient temperature and 2...17 pm at operating temperature of 120K. Selection of Bearing Material The selection of the bearing material had to consider the demanding mechanical and environmental loads. Among the bearing materials, firstly the traditional bearing materials AIS1 521 00 and corrosion resistant AIS1 440C were studied. It turned out early that these materials do not have the required corrosion resistance for exposure to salt fog. A preliminary salt environment test showed severe signs of corrosion as expected for the AIS1 521 00, but also, although less severe, for the AIS1 440C. For this reason it was decided to use a better corrosion resistant bearing steel for the bearings of the gimbal: Cronidur 30. This material is a FAG-proprietary bearing steel that combines excellent corrosion resistance with high strength and durability [l]. A standard grade of Cronidur 30 was chosen that is also used on cryogenic applications of the U.S. space shuttle program. The corrosion resistance of this material was demonstrated in a salt spray test according to the Ariane standard salt spray test procedure that has a duration of 4 days. No signs of corrosions were visible. Selection of Bearing Lubricant Liquid lubrication was not possible as the gimbal operates at cryogenic temperatures. After a pre-selection among available dry lubricants the following candidates were identified: sputtered MoS2, sputtered Ag (Silver), sputtered Pb (Lead), blasted WS2 (trade name "Dicronite DL-5"), TIC coating, or an uncoated design. An early development test showed that sputterevhlead by ESTL, UK was the suitable solution. The results of the tribological tests were presented on the 37 Aerospace Mechanisms Symposium [3]. Design of the Bearing Support The design of the bearing support was driven by stringent envelope constraints and demanding stiffness specifications. To cope with the high radial loads on the bearings, it was desired to find a design that allows for a maximized pitch size diameter of the needles. A bearing support consuming little space was necessary and leads to an unconventional design solution for the bearing support: 0 0 an un-divided yoke design, that requires axial insertion of the press-fit mounted outer rings, and thin walled outer bearing rings and yokes. The bearing support consists of a circular bore in housing that has varying stiffness along its circumference: It is thin walled at the lateral faces and it is rigid at the top and at the bottom of the bearing. Compression fits are used for the bearings to allow a design consuming little space while being more rigid than a conventional, separated housing clamped by bolts. Firstly the unequal stiffness distribution causes circularity deviations caused by changing cutting forces during machining. The weaker sections of the bearing support exhibit a larger diameter than the rigid sections caused by elastic deformation during the turning and grinding operations. These circularity deviations were measured to be 4.. .8 pm. Secondly, as described above, the bearing outer rings are situated within the bearing support using a compression fit. This compression fit causes radial load on the bearing support. This radial load deforms the bearing support made of titanium alloy. Deformation is larger in areas of the weaker sections than in the areas of the rigid sections. This circularity deviation, caused by elastic deformation was calculated by FEM analysis with 8 pm. The compression fit causes the outer bearing ring to follow these circularity deviations, so above two reasons lead to a circularity deviation of the race of the integrated bearing outer ring that has to be considered for the dimensioning of the bearing clearance. 273

Document

AMS_2006.pdf

Figure 3. Bearing Support (located in the gimbal yokes) Optimization of the Needle Roller Design The gimbal transmits the thrust of the rocket engine to the stage. This force bends the gimbal cross. The bearing inner rings are attached to the deflecting gimbal cross whereas the bearing outer rings that are attached on the gimbal yokes do not perform a corresponding deflection. This leads to a considerable angular misalignment between the axis of the bearing inner ring and the axis of the bearing outer ring. The bearing is not pre-loaded but has radial clearance. For this reason, the load distribution along the loaded bearing needle rollers is not constant but approximately triangular shaped. It was decided to study the load distribution in detail and to optimize the shape of the needle rollers to minimize the Hertzian contact stress. The load distribution was analyzed in a detailed FEM analysis of the bearings, which included a model of each needle roller within the complete gimbal assembly so that also the impact of the stiffness of the adjacent structure was adequately modelled. In this model, each needle roller was divided into sub- segments. Each of these sub-segments has adequate stiff ness and is equipped with gap-elements to simulate the bearing clearance. Figures 4 and 5 present a typical FE model of the needle rollers attached to the bearing inner ring, and the resulting load distribution within the bearings under worst case loads. This model was used to optimize the load distribution and reduce the peak values of Hertzian stress on the needle roller by local reductions of the stiff ness of the bearing support structure, and to predict the load distribution of the most loaded needle roller of the bearing. The resulting stress distribution was used as input for the optimization of the needle roller shape. An adequate crowning of the needle was defined by the bearing supplier FAG. Thus the bearing allows angular misalignment between the inner ring and the outer ring while transmitting the operational loads with permissible Hertzian stresses. 274

Document

AMS_2006.pdf

Figures 4 and 5. Finite element model of the needle roller and analysis result for the needle loads Dimensioning of Clearance To reduce the bearing friction torque it was required that the bearings have positive radial clearance during the whole operation cycle. A mathematical model was established that considered: 0 0 0 0 0 0 The reduction'of the outer ring race diameter by compression of the bearing outer ring due to its compression fit in the housing. As the compression depends on the temperature, this value had to be evaluated for the whole operating temperature range of the gimbal. The increase of the inner ring race diameter by tension of the bearing inner ring due to its compression fit on the shaft. Also this value had to be evaluated as temperature dependent. The thermal elongation of the bearing steel, the Titanium alloy of the housing, and the lnconel 718 of the shaft. The manufacturing tolerances of the fit diameters of the outer ring and the housing as well as the inner ring and the shaft. The roundness tolerances of these diameters. The elastic deformation of the outer bearing ring in the housing due to the unequal stiffness distribution of the housing as described above. The clearance was evaluated for the whole mission profile, using the thermal model results to apply the appropriate thermal gradients. Figure 6 shows the bearing clearance results for the whole mission profile. Integration of the Bearing As highly pre-loaded compression fits are used for the bearings, integration by a press tool is mandatory. The gimbal yokes and the gimbal cross are fixed in a special tool. The complete bearing assembly, comprising of the inner ring, the outer ring, and 37 needle rollers is pressed axially into the gimbal assembly by a spindle press. To ease the press process and to prevent cold welding of the parts, all contact areas are greased with Fomblin Y 225 by Solvay Solexis, Italy. The residue is removed by cleaning after the press process. During integration, the integration force is measured by a load cell, and recorded. The press process stops when the measured integration force increases rapidly, which is an indication that the shoulders on the bearing rings are in contact with the according counter-faces of the housing and of the gimbal cross. Figure 7 shows a typical load trace as experienced during bearing integration: 275

Document

AMS_2006.pdf

Time Figure 6. Bearing clearance results for the whole mission profile Figure 7. Bearing integration force versus insertion depth trace The trace shows a continuous increase in insertion force as the bearing slides into its seat. It is overlaid by vibrations of the setup. At a sudden point, the integration force increases rapidly signalling the completeness of the integration process. Development Test Setup Testing of the bearing on bearing level was not considered necessary. The development testing of the bearing was part of the development test campaign of the gimbal assembly. For the development testing of the gimbal assembly, a dedicated test rig was developed and manufactured that allows applying the following loads simultaneously: 0 0 Dry Nitrogen atmosphere 0 Gimbal movement of ~k6O Mechanical load, consisting of forces in longitudinal (X), lateral (Y, Z) direction and moment load, Thermal load, applied as temperature gradient between the upper and the lower gimbal interface plane, 276

Document

AMS_2006.pdf

During application of these loads the following measurements are continuously recorded: 0 0 0 0 Forces and moment load 0 0 Friction torque of the bearings Deformation and clearance of the gimbal assembly Gimbal angle and gimbal count Temperature of the interface planes and of 8 measurement points on the gimbal Strain of the gimbal structure To allow measurement of the bearing friction torque under load and under low. temperature, a special design of this test rig was developed. The gimbal assembly forms one joint of a pendulum support. The second joint is a large spherical, case-hardened contact that has a very low friction torque compared to the bearing friction torque. The pendulum is equipped with a quartz measurement plate that measures the bending moment within the pendulum. Rotation of the gimbal assembly around a gimbal axis causes friction torque that induces bending moment in the pendulum. By measuring the bending moment, the friction torque can be determined. As the location of the quartz measurement plate is distanced from the gimbal assembly and it is thermally isolated, it is possible to operate the gimbal at low temperature while the quartz measurement platform is at ambient temperature, which is necessary for high accuracy friction torque measurement. Accuracy is 10 Nm, measurement range is 300 Nm. The mechanical loads, which are forces and a moment load, are applied by a computer-controlled hydraulic cylinder system consisting of three cylinders arranged on the test rig structure, which allows applying all the required load combinations. The loads are continuously measured by separate load cells. The gimbal motion is applied by a crank driven by an electrical motor that is supplied by a frequency converter to allow adjustment of the gimbal frequency. The individual temperature of the interface plates is controlled by two separate temperature controllers that control the flow of liquid nitrogen through pipes embedded in the interface plates. These interface plates are thermally isolated from the test rig structure. The gimbal assembly is placed in a housing that is purged with dry nitrogen during low temperature testing. Figure 8 shows a schematic view of the test rig. Figure 8. Gimbal Test Rig: Schematically View and Photograph 277

Document

AMS_2006.pdf

By testing the bearings at the gimbal assembly level, the bearings could be tested with representative integration such as the compression and the tension from the press fits, and with representative deflections of the rings and between the inner ring and the outer ring caused by the deflections of the gimbal structure under load. Development Test Resu Its The development tests comprised the environmental tests and the life test of the bearing. After these tests, the performance was verified as end-of life characteristic. An overload test concluded the bearing development test campaign. Environmental Tests Before the mechanical tests, environmental tests were performed to simulate on-ground and flight environmental loads such as dry heat, damp heat, thermal cycles in damp air, thermal shock and salt fog. The tests were performed in a climatic chamber. No visual degradation of the bearings was observed. The selected bearing steel Cronidur 30 did not show any signs of corrosion. The lubricant (sputtered lead) formed lead salt as expected. Particles of this lead salt escaped from the bearing as can be seen in Figure 9. A Figure 9. Formation of lead salt on the lubricated needle rollers, escaping lead salt The formation of the lead salt as a consequence of the environmental testing was observed before during component level testing of the lubricants. These results were presented at the 37th Aerospace Mechanisms Symposium [3]. It is considered acceptable as the lead salt acts as protective layer which encapsulates the underlying lead. The lead salt has lubricating properties and acts as a lubricant itself. Functional performance was verified during the subsequent performance tests. Load Test The bearings were exposed to the mechanical design loads at ambient and at low temperature. No structural damage occurred. Friction Toraue Test The friction torque of the gimbal assembly around one axis was measured. The result is the friction torque sum of the engaged pair of bearings. A lateral support transmits the lateral loads as the bearings support radial loads only. The friction torque of this radial support is also included in the results; however it is supposed to be negligible. The friction torque was measured at room temperature, and at low temperature, and result was well below the allowable friction torque. This means that the lubricant performed as expected even after the harsh environmental tests. Life Cycle Test The life cycle test was intended as a tribological test to verify the lifetime at worst-case loads and temperature. A thermal gradient simulating the worst-case operating interface temperatures was applied. 278

Document

AMS_2006.pdf

Figure 10 presents the friction traces during the test (friction torque peaks and cycle count vs. load cycles). Gimbal operations with the worst case non operating load applied and additional gimbal operations under the worst-case operating loads were performed. The bearing friction torque was recorded during this test. The friction torque keeps approximately constant during the load cycles. No significant increase in bearing torque could be observed. Figure 10. Bearing friction trace during the life cycle test under operational load Clearance Test The clearance was measured after the life cycle test as an end of life value. The measured clearance was within the clearance requirements and was according to the test prediction. Stiffness Test The stiffness of the gimbal assembly was measured. It is mainly determined by the bearings' stiffnesses. The stiffness was measured in all load directions: compression of the gimbal, tension of the gimbal, lateral force, and moment load on the gimbal. As the bearing stiffness is determined by the deformation of the contact between the races and the needle rollers, which is a Hertzian line contact, the spring rate is not constant. For this reason, the stiffness was predicted and evaluated for 50% and 100% of the maximum loads. Figure 11 shows a typical deformation versus load graph of the gimbal assembly. Flex Y-100% RT Figure 11. Deformation versus load graph The graph shows the deformation versus the load in both load directions, thus the hysteresis is also shown in the graph. While recording this graph, the load direction was swapped so that during at the zero crossing of the load the bearing clearance can be evaluated from the graph. As a result of the test, it turned out that the stiffness in longitudinal direction and in lateral direction met the predictions of the FEM analysis. However, the measured torsional stiffness was noticeably smaller than predicted. 279

Document

AMS_2006.pdf

The reason for this difference was investigated in detail. It turned out that the torsional stiffness strongly depends on the crowning of the needle rollers. The local rotation of the bearing housing is determined by the contact point of the needle rollers which is the point of torque introduction into the bearing housing. un-deformed shape Cardan cross connected with bracket Figure 12. Simulated deformation of the bearing housing and structural analysis model Figure 12 shows on the left hand side the locally rotated bearing housing, and its un-deformed shape. On the right hand side, a model detail of the structural analysis model is shown. Each needle roller is simulated by a row of 7 rod elements and 7 gap elements connected to each other. The rod element simulates the stiffness of the racehollerhace contact and the roller whereas the gap element simulates the radial clearance of the bearing. The load introduction point is as initially resulting of the structural analysis model of the gimbal assembly is marked with a full arrow. This analysis model assumed cylindrical rollers. The dashed arrow presents an expected point of load introduction into the bearing housing as can be assumed from the stiffness test result because of the crowning of the rollers. A refined analysis model was established to evaluate the impact of the roller crowning by variation of clearance along the needle similar to the real bearing needle shape instead using a cylindrical model for the needle rollers. It turned out that the calculated stiffness decreases due to this refined simulation of the needles. This could explain the test result. The needle shape is optimized for stress distribution and not for optimized stiff ness. Life Cvcle Marain Test To verify the friction and coating behavior under higher load, the life cycle test was continued. The load was increased up to 150% of the qualification load in steps of 10%. 500 gimbal operations were performed with each load step. Figure 13 shows the friction envelope traces in red and blue color, the counter in green line, and the compression forces (negative) in black color. The graph shows the envelope curves of the measurement platform torque which are not identical with the friction torque of the gimbal bearings themselves, however they are proportional. It showed that the friction torque increases nearly proportional with the increasingly applied test load. The friction torque kept far below the specified value even at 150% of the qualification load. The hardware under test showed no visible damage. 280

Document

AMS_2006.pdf

Thu. 25.AwUsl2W5 - H(lku).m - HO*; *own' - ZuDan -2- -uM - FWTX Figure 13. Torque traces of the gimbal assembly during the life cycle margin test Summary and Conclusion It was demonstrated that the gimbal bearings were adequately designed and dimensioned and that the tribological system is capable to sustain the environmental and mechanical loads and meets the related performance requirements. b b 1. 2. 3. The sputtered lead coating on the Cronidur 30 substrate provides excellent tribological behavior in terms of friction and life even after being exposed to harsh environmental loads. The expected but unintentionally produced lead salt acts also as a lubricant. The un-conventional bearing mount design using press fit mounts proved successful however, a careful design was necessary as dimensional changes of the bearing rings occur due to elastic deformation and due to summation of manufacturing tolerances. The stiffness of needle roller bearing strongly depends on their axial alignment between inner and outer ring. The crowning of the rollers must be included in the structural analysis model to allow accurate prediction of the stiffness behavior. References Trojahn, W. et al. "Progress in B:aring Performance of Advanced Nitrogen Alloyed Stainless Steel, Cronidur 30." Presented at the 5' International ASTM Symposium on Bearing Steels, New Orleans, Louisiana, USA - Nov. 19-21, 1996. N.N. "Space Tribology Handbook, 3rd Edition." AEA Technology plc - ESTL Cheshire, UK, 2002. Neugebauer, C. and Falkner M. ,,Lessons Lhearnt on Cryogenic Rocket Engine's Gimbal Bearing Lubrication Selection." Proceedings of the 3? Aerospace Mechanisms Symposium, (May 2004), pp. 137-1 41. Remark & Acknowledgement Ariane 5 is an ESA program, managed by CNES, prime contractor of VlNCl is SNECMA. The gimbal is part of the VlNCl Thrust Chamber with prime contractor EADS-ST. The VlNCl gimbal has been developed under contract to EADS-ST of Ottobrunn, Germany. Austrian Aerospace gratefully acknowledges the continuing support of all parties. 281

Document

AMS_2006.pdf

Effect of Test Environment on Lifetime of Two Vacuum Lubricants Determined by Spiral Orbit Tribometry Stephen V. Peppe; Abstract The destruction rates of a perfluoropolyether (PFPE) lubricant, Krytox 1 43ACTM, subjected to rolling contact with 440C steel in a spiral orbit tribometer at room temperature have been evaluated as a function of test environment. The rates in ultrahigh vacuum, 0.21 3 kPa (1.6 Torr) oxygen and one atmosphere of dry nitrogen were about the same. Water vapor in the test environment - a few ppm in one atmosphere of nitrogen - reduced the destruction rate by up to an order of magnitude. A similar effect of water vapor was found for the destruction rate of Pennzane@ 2001 A, an unformulated multiply alkylated cyclopentane (MAC) hydrocarbon oil. introduction The destruction of liquid lubricant molecules by tribochemical attack (tribochemistry) is well-recognized, especially for perfluoropolyethers (PFPEs). The destruction or degradation can be manifested by the finite lifetime of a mechanism with a limited supply of lubricant. It is also indicated by the observation of chemical reaction films on the bearing surface, friction polymer and molecular fragments emitted into the environment. Destruction of PFPEs, in particular, has been observed in ball bearings for vacuum service in spacecraft and in eccentric bearing tests that operate in the starved or boundary lubrication regime, depleting the lubricant supply and causing bearing failure [l , 21. Destruction of PFPEs is also a matter of concern in hard disk magnetic storage media [3] The destruction rate of the lubricant molecules is a function of both the lubricant’s chemical structure [4, 51 and bearing substrate chemistry [6, 71. It can also depend on the test environment [&lo]. Understanding environmental effects on lubricant’s destruction rate is vital to ensure that earth-based life tests of vacuum hardware provide meaningful results. Some tests are performed in ‘inert’ gas, such as dry nitrogen, because of significant cost and convenience advantages versus vacuum testing. However, the test chamber may still contain trace impurities that compromise the inertness of the test environment, even though the chamber may have been backfilled or purged with an inert gas such as nitrogen. Impurities such as oxygen or water vapor may be tribochemically active and change the lubricant’s lifetime compared to what would be exhibited in vacuum. In this paper, the destruction rates of Krytox 143ACTM, a popular PFPE vacuum lubricant, in rolling contact with 440C stainless steel in a spiral orbit tribometer are studied for test environments of ultrahigh vacuum (UHV), dry oxygen, dry nitrogen, nitrogen containing water vapor and pure water vapor. The destruction rate is expressed as the lifetime of a finite charge of lubricant in a test and the results give a ranking of the lifetimes as a function of test environments. Some results are also presented for the effect of water vapor on the destruction rate of Pennzane@ 2001A, an unformulated multiply alkylated cyclopentane (MAC) hydrocarbon oil. Experimental The test instrument is a spiral orbit tribometer (SOT) depicted in Fig. 1. This rolling contact tribometer is a retainerless thrust bearing with one ball and flat races whose elements and kinematics have been described [l 11. The SOT was used to observe the tribochemical destruction of lubricants in vacuum and to give a ranking of the lifetimes, or degradation rate, of two different PFPEs (Krytox 143ACTM and Fomblin Z-2EiTM) and the unformulated Pennzaneo 2001A on 52100 steel [4}. NASA Glenn Research Center, Cleveland, OH Proceedings of the 38’ Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 283

Document

AMS_2006.pdf

Guide Plate Load I- Force Transducer Figure 1. Schematic of the Spiral Orbit Tribometer The 12.7-mm (.5-in) diameter ball and the plate specimens were 440C steel. All tests were run at room temperature at a mean Hertz pressure of 1.5 GPa and a ball orbit rate of 30 rpm (rolling velocity 0.071 m/s). The stainless steel, metal-gasketed test chamber was evacuated to ~0.133~1 0-8 kPa (1 x108 Torr) with a turbomolecular vacuum pump. A gate valve between the test chamber and the turbomolecular pump could be closed to permit the evacuated chamber to be backfilled with a particular gas. The nitrogen and oxygen supply bottles were specified to have a water vapor content of c 2 ppm. The gasses were admitted into the test chamber through an evacuated and baked stainless steel transfer line and a variable leak valve. Pure water vapor was admitted into the test chamber through a variable leak valve from a water supply that had been thoroughly degassed by the freeze-pump-thaw method. Water vapor concentration in the nitrogen atmosphere test environment was determined with a thin film hygrometer (Kahn Cermet II) inserted directly into the test chamber. The hygrometer's readout in dewpoint temperature, Td, could be converted to either partial pressure of water vapor or Concentration of water vapor in ppm through the Magnus formula and then to relative huyidity. Test :hamber total pressure P was determined by a cold cathode ionization gauge for P-d.064~10- kPa (8x1 0- Torr), a Pirani gauge for 1.06~1 0-3 kPa (8x1 0-3T~rr)<P<0.266 kPa (2 Torr) and a diaphragm gauge for P>0.266 kPa (2 Torr). The lowest reading of the hygrometer is Td = -99.9"C which corresponds to a water concentration of .014 ppm. This reading is achieved in vacuum and initially upon backfilling the chamber to an atmosphere of nitrogen. Eventually, however, water vapor accumulates in the valved-off chamber as shown in Figure 2. The accumulation is detectable after about 2x1 O4 orbits (-1 1 hours) and continues to increase thereafter. This water vapor is due to desorption from the chamber's unbaked interior surfaces. The lowest rate of accumulation is observed after the chamber had been evacuated for many days and then exposed to room air for a minimal time - approximately 10 minutes - during insertion of test specimens. 284

Document

AMS_2006.pdf

0 40000 aoooo 120000 160000 200000 Orbit Number 0.25 - Figure 2. Accumulation of water vapor in an atmosphere of nitrogen 0 A lubricant charge of -25 pg for the PFPE and -20 pg for the MAC was deposited only on the ball from a gas-tight syringe containing a dilute lubricant solution in a volatile solvent. This lubricant amount was consumed in a reasonable time - from a few hours to a few days. The speed and temperature of the test are similar to values that would be seen in spacecraft bearings, with test acceleration achieved through limited lubricant. Friction is recorded as a function of ball orbit and results in a ‘friction trace’, shown in Figure 3.The trace exhibits constant coefficient of friction (0.134) until an abrupt increase at -1230 orbits. A friction coefficient exceeding 0.2 is defined as test failure, the point when the lubricant is totally consumed. Normalized lifetime is obtained by dividing the orbits to failure by the initial lubricant charge. In this test, failure occurred at 1287 orbits and the initial lubricant charge was 25 pg, resulting in a normalized lifetime of 52 orbitdpg. Normalized lifetime is inversely proportional to the lubricant destruction rate. I i 25 pg Krytox 143ACTM, run in 1.6 Torr O2 I 1287 orbits/ 25 pg = 52 orbits/p.g ‘Lo 0.15 i -- 1 0 I. 285

Document

AMS_2006.pdf

Results Krvtox 1 43ACTM The results of four tests of Krytox 143ACTM run in each of the five specified environments are presented in Figure 4. The shortest lifetimes are exhibited by tests run in ultrahigh vacuum (UHV). The longest lifetimes are observed by running in the presence of water vapor, with increasing lifetime corresponding to increasing water vapor concentration in the test environment. Only 16 ppm water vapor in an atmosphere of nitrogen, corresponding to a relative humidity of c0.07%, is sufficient to extend the lifetime by an order of magnitude relative to testing in vacuum. Pure water vapor itself, without its being present in nitrogen gas, also leads to lifetimes much longer than testing in vacuum. Testing in one atmosphere of ‘‘dw nitrogen, in which the hygrometer indicated no additional water vapor above its lowest reading of Td= -99.9% (0.014 ppm), gives lifetimes greater than the tests in vacuum, but only marginally so. Testing in 0.21 3 kPa (1.6 Torr) dry oxygen, corresponding to -0.2% trace contamination of oxygen in a nitrogen atmosphere, also has little effect on lifetime relative to that in vacuum. The results presented here thus indicate that water vapor in the test environment exercises a “protective” effect on the PFPE lubricant, permitting a longer life before failure. Finally, the coefficient of friction of 0.134 was independent of the test environment and thus of the lifetime. 700 600 500 a u) s e c .- 400 .- f E c J 300 .- - m z 200 100 0 (ppm H20 in N2) + (16) 0 (Total pressure H20, milliTorr) (4) 0 (14) 0 (5) (.W i (.3) 8 0 - - - 8 - UHV 1.6 Torr O2 1 atm ”dry” N2. 1 atm N2 wlH20 Pure H20 Figure 4. Normalized lifetimes of Krytox 143ACTM when testing in different environments Pennzane@ 2001 A Friction traces for Pennzane@ 2001A are plotted in Figure 5 for testing in three environments - ultrahigh vacuum, an atmosphere of nitrogen with a minimum amount of water vapor and an atmosphere of nitrogen with a greater concentration of water vapor. The normalized lifetimes for the tests are indicated above the individual traces. The minimum concentration of water vapor in an atmosphere of nitrogen was achieved simply by closing the gate valve between the chamber and the turbomolecular pump and allowing the water vapor to accumulate in a nitrogen atmosphere as shown in Figure 2. The test at 75 286

Document

AMS_2006.pdf

ppm water vapor was achieved by deliberately admitting water vapor to the isolated test chamber from the degassed water vapor source and then backfilling with dry nitrogen. The test in ultrahigh vacuum lasted much longer than that of the test of the PFPE in ultrahigh vacuum as demonstrated in [4]. The tribodegradation rate of the MAC is much lower than that of the PFPE. The lifetime for the test in the minimum concentration of water vapor in the nitrogen atmosphere - whose evolution of concentration is shown in Figure 2 - is greater than the lifetime for the test in ultrahigh vacuum. The lifetime for the test in 75 ppm water vapor is greater still. Thus water vapor in the test environment extends the lifetime of the MAC as it does for the PFPE. As with the PFPE, the initial coefficient of friction of 0.08 is the same for all test environments, regardless of lifetime. 0.3 0.2 LL 6 0.1 0.0 2925 Vacuum Normalized Lifetimes, orbitdpg 7902 18,180 1: . Run in 1 atmosphere nitrogen ;/- 0 100000 200000 300000 400000 Orbit Number Figure 5. Friction traces of Pennzand P2001 A in different environments Discussion Contact Conditions The low constant friction coefficient exhibited in the tests prior to failure is an indication that the contact is lubricated. However, the state of the lubricant in the contact is not known. Cann and coworkers [12] have studied similar systems under a state of reduced availability of liquid lubricant using optical EHL. Although the lubricant charge used in the SOT is three orders of magnitude smaller than used in their investigations, the SOT system still appears to be in a state of lubrication. Gann, et al, presented an expression for the film thickness in the contact in the fully starved regime. An evaluation of their expression for the present conditions yields a film thickness that is unphysically small (<<I nm), confirming that the contact regime here is fully starved or parched [13]. The specific concentration of lubricant molecules in the contact is not known, except for the fact that there are enough of them to 287

Document

AMS_2006.pdf

lubricate the contact with a friction coefficient close to that exhibited by a fully flooded contact [14]. However, there are probably few enough so that an appreciable fraction of them are in direct contact with the substrate, where they are eventually consumed by tribochemical attack, indicating a failure by elevated friction (Figure 3). In this study, the friction coefficient is independent of the degradation rate. Evidently, the process responsible for the establishment of the friction force (possibly of rheological origin) and the process responsible for the degradation (tribochemical origin) proceed independently of each other. The tribochemical events may be considered second order effects - events that occur relatively seldom compared with the rheological friction that is continuous in the contact. If the tribochemical rate of attack was higher, then the coefficient of friction might well be different for different test environments that determine different lifetimes. Krvtox 143ACTM The lifetime increase is evident for even very low concentrations of water vapor in the test environment. A concentration of 0.03 ppm in an atmosphere of nitrogen corresponds to a partial pressure of water vapor of 0.02 milliTorr and lower concentrations may be difficult to achieve. Indeed, the somewhat longer lives for the tests in atmospheres of “dry” nitrogen (atmospheres in which the hygrometer maintained its base reading of Td = -99.9’C) could be due to water vapor concentrations too low to be registered. The extrapolation of the data in Figure 2 to small orbit numbers indicates that there is some lifetime-extending water vapor present even for the relatively short lifetime exhibited by this PFPE. This would indicate that the lifetime tests conducted in this manner are more sensitive to water vapor than the hygrometer itself. The tests conducted with Krytox 143ACTM in pure water vapor support the assertion that water vapor really is the cause for extended lifetimes in nitrogen atmospheres and not some other tribochemically active molecule that was introduced into the chamber along with the nitrogen. These results also indicate that life testing in “soft” vacuums of about Torr or less (for which water vapor constitutes a majority of the residual gas) can be as valid as testing in a better vacuum. The null result of testing in 0.21 3 kPa (1.6 Torr) oxygen establishes that such trace amounts of oxygen in nitrogen cannot protect the PFPE lubricant against tribochemical attack. This result relates to a current idea that tribological stress exposes clean metal by removing the native oxide, allowing the clean metal to initiate tribochemical attack on the lubricant. However, such clean metal would be oxidized by exposure to 0.21 3 kPa (1.6 Torr) oxygen thus quenching any tribochemical attack by clean metal. If the oxidation is fast enough to oxidize clean metal in the contact, then these results imply that the “exposure” mechanism is not operative here and that the ball is probably rolling on the native oxide of the steel during the test. Pennzane 2001 A@ The friction traces in Figure 5 indicate that water vapor in the test environment also extends the lifetime of the MAC. It would bedesirable to determine if there was any effect on the lifetime by testing in an atmosphere of a truly inert gas or oxygen. However, the MAC’S long intrinsic lifetime allows the accumulation of water vapor in the test environment: - as presently constructed - that would mask the possible effects of these species. A study of the effects of oxygen or an inert gas must be conducted in the absence of water vapor and remains to be done. Mechanisms A consideration of the mechanism by which water vapor retards the tribochemical degradation of the lubricant molecules begins with the observation that the effect is present for both the PFPE and the MAC. The mechanism is not specific to the particular lubricant chemistry and attention is thus directed to the substrate. Substrate sites (probably on the steel’s native oxide) are evidently passivated by the water vapor. Water molecules might adsorb on the sites physically, that is without chemical reaction, and simply sterically block the interaction of the site with the lubricant molecule. Although this physical adsorption is well known, appreciable coverage at room temperature occurs at much higher partial pressures of water vapor [15], rendering this physisorption approach unlikely. An approach that considers the passivating chemical reaction of water with the active surface sites may be more fruitful. Lewis acid sites have received much attention in the context of PFPE degradation [16]. This approach may also work out for the sites that can attack both the PFPE and the MAC. The purely tribological results presented here offer limited chemical insight. 288

Document

AMS_2006.pdf

Other studies [8, 91 of lubricant life extension under a nitrogen cover gas have emphasized the role of an oxygen impurity component in the cover gas interacting with both an additive in the lubricant and with clean iron exposed by tribological action. The results presented here on oils without additives indicate that neither oxygen nor clean iron is a necessary aspect of life extension under an impure cover gas. Neither is the presence of an additive an essential aspect of the life extension process. Rather, the passivation of active sites on the surface of the bearing steel by water vapor is the governing mechanism that reduces the attack on the lubricant molecule itself. Amlication to Testinq The first implication of these results for testing of space mechanisms is that life-testing devices such as bearings in a nitrogen environment to simulate vacuum is permissible with sufficient -absence of water vapor. However, the severe restrictions required on the water vapor levels may eliminate any cost savings versus vacuum testing. The second implication is that the device housing may contribute water molecules to the device’s environment, since water always desorbs from an unbaked surface. This water flux captured in a housing with limited access to vacuum can provide a degree of protection to the lubricant that is not available in a more open geometry such as in the present test arrangement. In this sense the present test arrangement provides a very severe test. Conclusions Both the PFPE and the MAC lubricants are tribochemically degraded in vacuum as evidenced by the finite lifetimes they both exhibit in these tests. Tests in controlled environments indicate the lifetime of both lubricants is extended if water vapor is present in the environment. The effect of water vapor is evident at quite low levels (4 ppm) in the present tests. These results imply that sites must exist on the bearing steel’s surface that can attack the lubricant molecules. These sites are chemically passivated by water vapor, thus reducing the attack rate and allowing a longer lubricant life. An independent understanding and chemical characterization of these sites is not available at present. These results imply that testing of those mechanisms whose life is limited by lubricant consumption in an environment with water vapor, even at very low concentrations in a nitrogen atmosphere, can lead to mechanism lifetimes that are much longer than will be realized in vacuum service. Thus in terms of a “lessons learned, it is advisable to provide water and oxygen sensors for the mechanism test environment. Acknowledgments I thank Dr. Kenneth W. Street for requesting tests whose results motivated this study. References 1. 2. 3. 4. 5. 6. 7. Carre, D.J. “Perfluoropolyalkylether Oil Degradation: Inference of FeF3 Formation on Steel Surfaces under Boundary Conditions”, ASLE Transactions, 29 (1 986), 121 -125. Carre, D.J., “The Performance of Perfluoropolyether Oils under Boundary Lubrication Conditions”, Tribology Transactions 31 (1 988), 437-441. Novotny, V.J., Karis, T.E. and Johhnson, N.W., “Lubricant Removal, Degradation, and Recovery on Particulate Magnetic Recording Media”, Journal of Tribology, 1 14 (1 992), 61 -67. Pepper, S.V. and Kingsbury, E.P., “Spiral Orbit Tribometry - Part 11: Evaluation of Three Liquid Lubricants in Vacuum”, Tribology Transactions, 46 (2003) , 65-69. Bazinet, D.G., Espinosa, M.A., Loewenthal, S.H., Gschwender, L., Jones, W.R., Jr., Predmore R.E., “Life of Scanner Bearings with Four Space Liquid Lubricants”, Proceedings of the 3fh Aerospace Mechanisms Symposium, Johnson Space Center, May 19-21,2004. Carre, D.J., “The Use of Solid Ceramic and Ceramic Hard-coated Components to Prolong the Performance of Perfluoropolyalkylether Lubricants”, Surface and Coatings Techno/ogy, 43/44 Jones, W.R., Jr., Jansen, M.J., Chen, G-S., Lam, J., Baker, M., Lo, J., Anderson, M. and Schepis, J.P., “The Effect of 17-4PH Stainless Steel on the Lifetime of a Pennzane@ Lubricatyf Microwave Limb Sounder Antenna Actuator Assembly Ball Screw for the Aura Spacecraft”. 7 1 (1990), 609-617. 289

Document

AMS_2006.pdf

European Space Mechanisms and Tribology Symposium (ESMATS 2005), Lucerne, Switzerland, September 21 -23,2005. 8. Carre, D.J., “Lead Naphthenate Tribochemistry Under Vacuum and Gaseous Nitrogen Test Conditions”, Tribology Letters, 16 (2004), 207-21 4. 9. Carre, D.J., “Effect of Test Atmosphere on Moving Mechanical Assembly Test Performance”, Proceedings of the 3fh Aerospace Mechanisms Symposium, Johnson Space Center, May 19-21, 2004. 10. John, P.J., Cutler, J.N. and Sanders. J.H., “Tribological Behavior of a Multialkylated Cyclopentane Oil Under Ultrahigh Vacuum Conditions”, Tribology Letters, 9 (2000), 167-1 73. 11. Pepper, S.V. and Kingsbury, E.P., “Spiral Orbit Tribometry - Part I: Description of the T ri born e te r” , Tribology Transactions, 46 (2003), 57-64. 12. Cann, P.M.E., Damiens, B. and Lubrecht, A.A., “The Transition between Fully Flooded and Starved Regimes in EHL”, Tribology International, 37 (2004), 859-864. 13. Kingsbury, E. P., “Parched Elastohydrodynamic Lubrication”, Transactions ASME, J. Tribology, 14. Spikes, H.A., “Comparison of Krytox 1 43ABTM and 1 43ACTM in Prolonged Mixed Sliding/Rolling”, Report TS03 1/99 of Tribology Section, Department of Mechanical Engineering, lmperial College, London. 15. Israelachvili,, J.N., Intermolecular and Surface Forces, 2“d Ed., Academic Press, New York, 1992 and Gregg, S.J. and Sing, K.S.W., Adsorption, Surface Area and Porosity, 2“d Ed., Academic Press, New York, 1982, Chapter 5.5. 16. Kasai, P.H., “Perfluoropolyethers: Intramolecular Disproportionation”, Macromolecules, 25 (1 992), 107 (1 985), 229-233. 6791 -6799. 290

Document

AMS_2006.pdf

Influence of Oil Lubrication on Spacecraft Bearing Thermal Conductance Yoshimi R. Takeuch?, Matthew A. Eb;, Benjamin A. Blake*, Steven M. Demsk; and James T. Dickef Abstract Increasing demands on bearing performance and a lack of thermal conductance data for bearings in space applications motivated The Aerospace Corporation to study heat transfer across angular-contact ball bearings for space systems. Tests were conducted under controlled conditions including rotational speed, temperature, axial load, and vacuum environment. Bearings with Nye Pennzane SHF2001 synthetic oil were compared with dry (non-lubricated) bearings. These comparisons show that dry and oil lubricated bearings vary in thermal conductance by up to an order of magnitude. Experimental measurements also indicated that sensitivity to other variables, such as axial load and temperature, depends on whether the bearing is dry or oil lubricated, and whether it is in a static or dynamic (rotating) state. Mechanisms of heat transfer are discussed for each of these states. Introduction In contrast to typical terrestrial applications, the absence of convection shifts the focus for thermal analysis of rotational space hardware. In vacuum environments, conductance through the bearings often provides the primary heat transfer path between the shaft and the housing. As such, temperature predictions for rotating components, such as satellite instruments, or the bearing itself require knowledge of the bearing thermal conductance. However, published literature provides little help on the subject, and thus bearing thermal conductance is usually the significant unknown in the development of a thermal model of a rotational system in space. When available, engineers often use heritage information for comparable systems with similar bearings. Significant uncertainty arises with the advent of design changes, different bearing geometry, different lubricant type or quantity, or dissimilar operational conditions. An absence of thermal conductance data leads to challenges in using thermal models for guidance in the design process. Existing literature yields limited thermal conductance information for static and low-speed bearings [l-91 and none for high-speeds. Yovansvich [I ,2] developed a mathematical model for the thermal conductance of a non-lubricated, static (non-rotating) bearing. Experimental work followed, including studies on spacecraft bearings performed by Stevens and Todd [3], from the European Space Tribology Laboratories (ESTL). They measured thermal conductance across a bearing, up to a maximum speed of 2,500 RPM, using an experimental setup designed by Deli1 et a1 [4-51. ESTL continued to study this subject over the years, focusing on static or low-speed and large thin cross-section bearings [6-81. Demand on bearing performance has grown and thermal concerns have increased as systems have reached higher speeds. Current momentum wheels and control moment gyroscopes typically operate at 6,000-9,000 RPM [lo]. Future wheels, including energy storage flywheels, are envisioned to run at even higher speeds, above 15,000 RPM. In addition to the need for high-speed data, low-speed applications could benefit from a greater range of experimental data, and additional sources of counter verification from other researchers. To address these concerns, an experiment was designed to assess thermal conductance of bearings in vacuum, at speeds ranging from 0 to 6,000 RPM. Controlled studies were conducted by varying variables such as axial load, thermal boundary conditions, and rotational speed, individually for parametric studies. The tests identified variables of importance for different bearing conditions, including dry, lubricated, static, and dynamic states. Finally, qualitative theories for the mechanisms of heat transfer for each of these conditions emerged. * The Aerospace Corporation, Los Angeles, CA Proceedings of the 38th Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006. 291

Document

AMS_2006.pdf

Nomenclature G - conductance across the bearing N - number of balls in the bearing Ri -thermal resistance at the contact between the inner race and one ball R, -thermal resistance at the contact between the outer race and one ball Rb - thermal resistance across inner to outer race of one ball k, - thermal conductivity of the inner race material kb - thermal conductivity of the ball material k, - thermal conductivity of the outer race material a - major axis of a Hertzian contact area q - major axis of the Hertzian contact area between the inner race and one ball a, - major axis of the Hertzian contact area between the outer race and one ball b - minor axis of a Hertzian contact area b, - minor axis of the Hertzian contact area between the inner race and one ball bo - minor axis of the Hertzian contact area between the outer race and one ball Experiment Reference [l 11 provides a detailed description of the experimentaj setup and measurement tecQniques. All tests were conducted in vacuum environments of at least 1 xl0 Torr (approximately 1.3~10- Pa). As a summary, the experimental design matrix is outlined as follows: . 1. Vary rotational speeds between 0 and 6,000 RPM. The maximum speed of 6,000 RPM is typical of a spacecraft control moment gyroscope (CMG) or momentum wheel. 2. Apply a constant pure axial load ranging from 40 to 129 N. 3. Accommodate bearings of different sizes, namely the 101 and 204-size ball bearings. 4. Test dry or oil lubricated bearings. 5. Vary the average bearing temperature. Oil lubricated bearings were tested in either the virgin or fully run-in states. A fully run-in bearing was established by continually running the bearing at a constant speed until heat generation, torque, and thermal conductance remain unchanged (about a week of continual operation). Test Bearings A 521 00 steel bearing and a hybrid bearing, consisting of silicon nitride balls and 521 00 steel races, were tested. Table 1 provides the specifications of the two different angular contact ball bearing sizes used. Table 2 summarizes the relevant material properties. Table 1. Ball Bearing Specifications 292

Document

AMS_2006.pdf