Datasets:

ai-aerospace

/

ams_data_full_2000-2020

like 0

393 Figure 18. The overall architecture of the CCLL is shown above. The solid colored components show the launch lock in the unlocked position, with the ERM Cap released from the ERM body. The wireframe shows the position of the launch lock in the locked position, with the ERM Release Cap locked to the ERM body. The Rotor Arm rotates ~9.5 degrees between the locked and unlocked positions. Commercial grade tie rod end hardware was used in all of the linkages.

Document

AMS_2012.pdf

394 Pin Pusher Actuators Several concepts were traded, with a 3-point constraint architecture selected. This used 3 drive pins on the instrument going into three holes on the cryo-cooler keel structure posts. The dimensional clearance between the pins and the holes is only 0.002 inch (0.05 mm). Each point constrained 2 translations and 2 rotations, providing a 2-2-2 degree of freedom kinematic lock. To release the launch lock, the pins were pushed through the holes by spring pistons. To facilitate resetting, the pins were not completely removed from the holes, and the holes were given a toroidal profile. The drive pins were stepped down so that a thicker diameter matched the toroidal hole diameter, and a thinner diameter provided adequate free clearance. The pins were pushed the distance required to remove the larger diameter pin length from the toroidal bushing hole engagement diameter to a seating hole in the anchor fitting. Each restraint point had a pin pushing, compression-type spring, as shown in Figure 19. Each drive pin was tied to a central rotor disk via a drag link. As the pins are pushed out of the engagement bushings, the rotor disk rotated approximately 9.5 degrees, as shown in Figure 18. If the rotor disk is not allowed to rotate, the pins are held in place and the springs are compressed. If the disk is free to rotate the springs are free to expand and push out the drive pins. The disk also couples the motion of all of the pin pullers together; if any single spring fails, the rotor disk rotation, powered by the surviving springs, will pull the pin with the failed spring. Figure 19. The spring-powered, pin pusher actuato r shown locked (stowed) in the upper figure, and unlocked in the lower figure. The ERM restraint device prevents the compressed drive spring from pushing the the green Drive Pin out of the Toroidal Bushing. When the ERM releases the central rotor, the drive spring is free to expand and pushes the Drive Pin out of the Toroidal Bushing and into a guide hole in the anchor fitting. To reset the actuator, the the Drive Pin is manually pulled back through the Toroidal Bushing (v ia the Reset Linkage a nd the Rotor Arm) and connected to the reset ERM. The titanium Anchor FIttings and Toroidal Bushings had a Tiodize® type II (Teflon® impregnated) finish. The Drive Pins are made of aluminum bronze, CDA 63020 per AMS 4590B. A light film of Braycote

Document

AMS_2012.pdf

395 Micronic 602EF grease was applied to the Drive Pin, and to the sliding surfaces in the anchor housings and the toroidal bushings. Release/Restraint Device The ERM restraint device prevents the rotor disk from rotating. When fired, the ERM releases the disk, the pins retract, and the cryo-cooler unlocks. The ERM is not coupled directly to the rotor disk. An extension arm runs from the rotor disk through a clearance slot in the instrument -Y panel. A tie rod runs from the extension arm to the ERM mount. This allows the ERM to be mounted on the instrument exterior, facilitating reset access. A potentiometer is used to indicate rotor position, providing a positive indication of the locked or unlocked state. The ESDM potentiometer was used, as spares were immediately available. Telemetry lines to support this sensor were also available in the control electronics. Cryo-Cooler Launch Lock Mechanism Development Issues To reduce risk and expedite schedule, a functional breadboard unit was fabricated quickly to validate the overall architecture, and the flight unit and an engineering test unit were developed in parallel. The developmental intention was to qualify the design in parallel with the flight fabrication. There was no schedule available to environmentally qualify a unit and then fabricate the flight unit serially. It had to be done in parallel. The ETU unit was built onto a flight-like composite panel for environmental testing. The flight unit was built up directly onto the flight structure and was qualified at the instrument level. The ETU and flight units were identical. Figure 20. In the left figure, the flight Cryo-Cooler is shown in development at the vendor, mounted within the flight Keel structure. Vibration testing of the ETU Keel Assembly (with CryoCooler mass simulators) to verify post stiffness is shown in the center figure. The CCLL functional breadboard unit is shown in the right figure. It became evident during the initial development and analysis of the mechanism that 60-Hz stiffness was going to be a very diffcult requirement to meet, with initial concepts predicted to be ~42 Hz. Several design changes were implemented, including optimizing the Drive Pin/Toroidal Bushing clearance and the addition of the Triangular Frame, which rasied the system stiffness to meet and finally exceed the requirement. The CCLL first mode frequency was measured at 65 Hz. Upon actuation, the CCLL produces a peak torque of 99 N-m (877 lb-in), and an angular momentum of 12.4 Nms (110 in-lb-s), within the limits specified in Table 3. The CCLL ETU successfully passed qualification testing in September 2011 and the flight unit has been installed into the TIRS instrument. Many issues were encountered and ultimately solved in this unique mechanism.

Document

AMS_2012.pdf

396 Figure 21. The CCLL is shown during installation into the TIRS instrument, with the -Y panel removed. The left figure shows the rotor arm, anchor fittings, and prominent triangular frame. The Cryo-Cooler legs were integrated into the CCLL before the keel, and attached to it later. The drive springs are visible in the anchor fittings just beneath the triangular frame. The right figure shows the instrument with the -Y panel replaced, with the ERM mounted on the right and the rotor arm coming through the panel on left side, with the two joined by a linkage. The gold colored frames that partially obscure the ERM hardware are box enclosures that will keep thermal blanketing clear of the moving parts, while allowing access for manual ERM resetting. Use of commercial grade mechanical components for space flight Due to the compressed schedule, procuring flight quality tie rod ends and/or spherical bearings was impossible. The only option was to fly commercial industrial components. Traceability paperwork to verify material type and quality was generally not available, or was not sufficiently trustworthy for mission assurance. To reduce risk, many extra components were procured, and used not as spares, but for testing. Components were proof tested to measure yield and ultimate tensile strength, and chemically analyzed to verify material type, finish, and surface hardness. Parts that were selected from the lot for flight were static-load tested to qualification load levels before use. The commercial components used in the CCLL successfully passed mechanism component level and instrument level testing. Lesson Learned: Generally available, commercial industrial hardware may be qualified for spaceflight use if a comprehensive set of tests to verify base material type, coatings,finish, yield and ultimate strength are carried out. The hardware should be purchased in lots and many indentical units tested to determine unit-to-unit statistical differences and aid in flight part selection from the lot. Unpredicted behavior during vibration testing During vibration testing it was found that the frequency of the system would increase as full level input was approached. This occurred in the lateral directions only. The shifts in frequency for the Z lateral axes are shown in the Figure 22. The shift that was seen during testing in the Y-Axis was not as significant as the one seen for the Z-Axis. The Y-Axis shift was 10 Hz from -18 dB to Full Level. The shift for the Z-Axis was 31 Hz from -18 dB to Full Level. The predicted frequency was higher than the 65 Hz measured during testing, but that was for an idealized system, and did not take into account a number of factors that could be attributed to the difference.

Document

AMS_2012.pdf

397 Figure 22. The frequency of the stowed CCLL was shown to increase as the test level was increased to full level. Note how the first mode frequency peaks shift the right (increase) as the test level increases from -18 dB to full level. This behavior was not predicted, and occured in the 2 lateral (ZY mechanism plane) axis. It did not occur in the axial (X, thrust) axis. The shift was a result of the gap clearance around the Drive Pin to Toroidal Bushing interface. The gap was necessary to ensure that the mechanism would operate reliably at cold temperatures and was misalignment tolerant. As more energy was input into the system the gap would close, and the system would become stiffer. This characteristic made predicting the behavior of the system difficult, but established that smaller clearances made the sytem stiffer. Lesson Learned: Mechanisms which utilize sliding interfaces are difficult to model accurately, and will generally be less stiff than predicted . The measured system frequency will increase as test loads increase to full levels. This is due to sliding clearances being reduced in proportion to input level during vibration tests where the lines of action of the sliding device are close to the test axis. This was not seen when the test axis was perpendicular to the mechanism lines of action. Conclusions The SSM, ESDM, and CCLL span a broad range of mechanisms types; from microradian precision mirror positioning to coarse deployment of the large earth shield. Due to the compressed schedule, it was not possible to fully develop engineering test unit mechanisms before the flight mechanisms. A protoflight approach had to be utilized to save time, and even then some prudent development processes, such as the encoder electronics signal integrity analysis, and the SSM bake out, were sacrificed. Higher levels of risk had to be accepted to meet the delivery schedule. We were able to mitigate the consequences resulting from the analytical and contamination control omissions, and the heritage justification to forego critical analysis was proven false. The building of a simple breadboard SSM mechanism, which verified 1.00E-081.00E-071.00E-061.00E-051.00E-041.00E-031.00E-021.00E-011.00E+00 10.00 100.00 1000.00ASD (g2/Hz) Frequency (Hz)Z Axis Vibration Response Overlay for Cryocooler Mount 19Z from -18dB of Full 19Z from -12dB of Full 19Z from -6dB of Full 19Z from Full Level -18dB-12dB-6dB Full Level

Document

AMS_2012.pdf

398 the performance of the architecture to instrument management, had as much "political" value as technical value. Whereas heritage arguments hurt the SSM development, the ESDM benefitted strongly from heritage. It required little more than cosmetic changes from the heritage LRO design; its development proceeded comparatively painlessly compared to the SSM, yet some lessons were learned, and classic lessons reinforced. The relative simplicity of the mechanism allowed for adequate analysis to be undertaken, and many developmental performance tests to be done. The vendor-supplied components, such as bearings, potentiometers, and dampers, were known early and hence procured early, with adequate time for chracterization testing and bake out. A dedicated breadboard was (correctly) not needed, as the TIRS flight design was a derated version of the original LRO design. This was the classic, proper heritage application case; the new mechanism was as close to "build-to-print" identical to the heritage mechanism as was reasonably possible. The CCLL was just the opposite of the ESDM; There were no heritage mechanisms that would meet its requirements. The 6-month development schedule, and a fixed set of available interfaces, drove the team to a non-traditional approach that was difficult to analyze, but relatively easy to test. Here the early breadboard was absolutely essential to verify function, measure forces, and test critical sliding clearances. Methods were developed to lot-qualify commercial grade mechanical hardware (with little or dubious traceability paperwork) for flight use (albeit at higher risk), as flight-quality components were impossible to procure in the available time. The TIRS mechanisms development was undertaken knowing that a higher level of risk had to be assumed in order to minimize the development time. Many weeks were lost and the instrument schedule re-arranged several times to accomodate delayed mechanism deliveries due to the consequences of this decision. Ultimately, the difficulties were overcome, the mechanisms were integrated into the instrument, and have functioned flawlessly throughout instrument environmental testing. Acknowledgements The authors would like to sincerely thank the many individuals whom, through their Herculean efforts during this rapid development, were key to the success of the TIRS mechanisms. They are: Melissa Edgerton, Alissa Mitchell, and Monica Zuray, Mechanisms Engineers. Edwin Lee, Randy Frazier, and Scott Weedon, Mechanical Designers. David Maidt, Thermal Analyst. Scott Hoeksema and Milagros Silverio, Contract Oversight. Tommy Emmett, Co-op Student. Andy Wohl, Mechanisms Integration &Test, Gordon Bowers, Technician. We would also like to especially thank Dr. Michael Dube, and Anh Tran for their assistance in bearing lubrication and analysis, and John Sudey for his expertise in jitter control. References 1. Dennis Reuter, Cathy Richardson, James Irons, Rick Allen, Martha Anderson, Jason Budinoff, Gordon Casto, Craig Coltharp, Paul Finneran, Betsy Forsbacka, Taylor Hale, Tom Jennings, Murzy Jhabvala, Allen Lunsford, Greg Magnuson, Rick Mills, Tony Morse, Veronica Otero, Scott Rohrbach, Ramsey Smith, Terry Sullivan, Zelalem Tesfaye, Kurtis Thome, Glenn Unger, Paul Whitehouse "The Thermal Infrared Sensor on the Landsat Data Continuity Mission.” Geoscience and Remote Sensing Symposium (IGARSS) 2010 IEEE International, Honolulu, Hawaii 2. Budinoff, Jason G., R. Barclay, K. Bergandy, A. Matuszeski, J. Schepis, "Development of The Scene Select Mechanism for the Thermal Infrared Sensor Instrument.” Proc. 14th European Space Mechanisms & Tribology Symposium – ESMATS 2011, Constance, Germany (ESA SP-698, September 2011) 3. Capon, Thomas, “Design of a Digital Control System for Positioning a Scene Select Mechanism Professional Intern Program Level II Project” NASA GSFC Code 544, July 2011

Document

AMS_2012.pdf

399 Resolution for Fretting Wear Cont amination on Cryogenic Mechanism Charles S. Clark* Abstract The Near infrared camera (NIRCam) instrument for NASA is one of four science instruments to be installed into the Integrated Science Instrument Module of the James Webb Space Telescope (JWST) which is intended to conduct scientific observations over a five year mission. The NIRCam instrument incorporates multiple mechanisms that perform specific tasks as part of the observatory ground testing, instrument commissioning, and on-orbit science and diagnostics—all of which must operate between 293 and 37 K and be tested to typical launch and space environment standards. Two of these mechanisms, the pupil imaging lens assembly (PIL) and filter wheel assembly (FWA), use common bearing mounts designed for operation at ambient and cryo temperatures. Modifications to the existing bearing mounts were developed to address fretting damage and associated contamination between the bearing race inner diameter and fixed shaft interface. Comparative proto-flight level vibration testing of four (4) new design configurations was performed alongside a control configuration similar to the original design. To ensure the trial tests simulate worst case environments, the setup went through the equivalent of five 3-axis vibration tests, one single cryo cycle, and a post cryo-cycle vibe test. The final vibe test was run in a 5% relative humidity environment as requested by the customer to reflect latest thinking of the actual JWST launch environment. This paper presents details of the investigation, re-design trades, and trial testing that demonstrated that a titanium shaft with a diamond-like-coating along with a Nitronic-60 sleeve was the preferred configuration of the bearing mount design for both the FWA and PIL units. Introduction The near infrared camera (NIRCam) is one of five instruments aboard the James Webb Space Telescope (JWST) observatory which will be conducting deep space scientific observations from the second Lagrangian point of the Sun-Earth orbit at a passively cooled temperature of 37 K. The NIRCam instrument will process the light from the JWST 6.5-meter primary mirror at wavelengths of 0.6 to 5.0 microns. The NIRCam instrument is designed with two optical benches constructed of beryllium that are mirror copies of each other as shown in Figure 1. Each of the two NIRCam optical benches is configured with optical elements that divide the science beam into a shortwave and longwave path. The filter wheel assembly (FWA) and the pupil imaging lens assembly (PIL) are two different types of mechanisms being developed for the NIRCam instrument to support on-orbit calibration of the JWST observatory as well as conduct mission scientific observations. There are four FWA assemblies and two PIL assemblies installed on the NIRCam instrument, with two FWAs and one PIL located on each optical bench. As shown in Figure 1, one FWA is installed on the longwave path, and one FWA and one PIL are on the shortwave path. * Lockheed Martin Space Systems Company - Advanced Technology Center, Palo Alto, CA Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

400 Figure 1. The locations of two filter wheel assemblies and the one pupil imaging lens assembly are shown on the module A (top) beryllium optical bench. Stray light shrouds appear transparent in the figure to show the PIL located within. The FWA is designed to insert specific optical elements into the NIRCam optical beam. Each FWA unit consists of two independently driven wheels that each hold 12 optical elements and is required to insert each optic element into a target position with a repeatability of ±75 microns. Each FWA unit is configured with primary mirror wave-front sensing elements, optical calibration sources, and numerous optical filters dependent upon which longwave or shortwave path the FWA resides. An illustration of the FWA is shown in Figure 2. The primary function of the PIL assembly is to deploy a set of optics into the beam path. Once inserted, the NIRCam instrument focus is changed to image the 18-segment, 6.5-meter diameter, primary mirror instead of deep space. To achieve this function, the PIL must deploy its optics into the specified point in the beam path with a repeatability of 0.016 degree. The PIL assembly is shown in ). Figure 3 . Figure 2. The filter wheel assembly (FWA). Fi gure 3. The pupil imaging lens assembly (PIL). PIL located within optic shrouds on shortwave optic Shortwave filter wheel assembly Longwave filter wheel assembly

Document

AMS_2012.pdf

401 The FWA and PIL, along with all NIRCam components, are designed to support many demanding requirements. The units must operate after being subjected to the stresses of launch loads and in the vacuum of space. These units are also required to have a minimum first mode frequency of 100 Hz, operate with a maximum power input of 0.6 mW, and all actuators must have appropriate torque margins. While these requirements are not unusual for space flight mechanisms, both the FWA and PIL mechanisms are required to operate at 37 K to allow imaging at the near infrared spectrum. Furthermore, the FWA and PIL are not allowed to generate debris larger than 300 microns over their lifespan to ensure minimal image degradation through contamination. The PIL was given a unique requirement that specified the mechanism shall include a fail-safe system that, given a set of reasonable failure conditions while in the deployed state, the PIL optic would be able remove itself from blocking the science beam. Previous papers [1], [2], and [3], are available for more information about the FWA and PIL mechanism mission requirements and original design. The FWA and PIL are both rotating mechanisms that share common motor and bearing mount designs. The bearing and related bearing mount design ensures low friction rotation at ambient and cryogenic temperatures. The original designs for both mechanisms integrated a fixed titanium shaft, 440C cryo and space-rated bearings, and 455 CRES rotors. Due to different materials and the large operational temperature excursion, a flex bearing mount was chosen as illustrated in Figure 4 . For operation at cryogenic temperatures, a Teflon film bearing lubrication is employed through a transfer process from a Teflon and fiberglass composite ball-bearing cage to the balls and races of the bearing. With the back-to-back bearing configuration, the bearing preload is set by clamping the inner race of the bearing. This bearing preload is critical to ensuring smooth running, low-friction rotation. Selecting a spring with a spring-force just higher than the bearing pre-load, and applying the spring force through a close tolerance slip-fit bearing sleeve, one ensures the bearings have constant pre-load but allows for slight changes from differential coefficients of thermal expansion and possible Teflon lubrication buildup. While this design works well in quiescent environments, the forces induced during launch would exceed the capability of the spring system alone. Two features were added to the flex mount design to achieve the needed stiffness through launch. First, a hard-stop was added to the sleeve that only allows 75 microns of slip. The second feature is a cryo-release tube that effectively locks out the spring during the ambient launch temperatures but shrinks away at cryo temperatures, allowing the spring to pre-load the bearing. Through prototype and qualification testing, this design was proven to meet two-time-life testing with the required low friction rotation. Figure 4. The spring clamped bearing mount design for the pupil imaging lens assembly (PIL). Fixed titanium shaft Clam p nut and hardsto p Cryo release tube Bearing pre-load spring Duplex bearin g pair Bearing sleeve

Document

AMS_2012.pdf

402 Finding the Problem Testing of the FWA prototype, and the FWA and PIL qualification units successfully demonstrated all performance requirements, but a planned disassembly of the PIL revealed an unforeseen problem in the bearing mount. The PIL qualification unit was fully assembled and performance tested at ambient temperatures and then subsequently subjected to proto-flight random and sinusoidal vibration testing. Because of the unique optic configuration of the PIL, the first vibration test of the system was performed with a surrogate PIL optic installed. After successfully completing this initial vibration test, the PIL was partially disassembled to swap out the surrogate optic with the actual optic. At this point, contamination was found at the base of the PIL shaft as shown in Figure 5. Figure 5. Contamination found in the PIL qualifica tion unit after vibration testing, and a diagram of the PIL bearing mount illustrating the suspected contamination path. Investigation into the powdery contamination found in the PIL qualification unit initially focused on the bearing sleeve which showed evidence that the some of the anodic coating was worn away, but the investigation was quickly expanded to look at the entire bearing assembly. Magnified optical inspection and scanning electron microscope (SEM) analysis of the sleeve showed evidence of wear with debris in the form of conglomerate particles. This type of conglomerate particle debris which is made up of many sub-micron sized particles is characteristic of fretting wear 1. The wear debris found on the sleeve contained mostly titanium from the shaft and sleeve base material as well as silicon and oxygen from the anodic coating. However, it seemed unlikely that all the debris could come from the wear areas between the sleeve and shaft. Analysis of debris from other areas at the base of the PIL shaft showed the same conglomerate particles, but these particles consisted of titanium and iron, but did not contain silicon and oxygen. The iron was most likely from wear of the bearing. Titanium is a fretting-wear sensitive material, and it became clear that the microscopic motion of the flexure bearing mount during vibration testing resulted in the wear-generated particulate contamination seen on the PIL qualification unit. Since a similar bearing mount was used in the FWA design, the FWA qualification unit was disassembled to inspect for possible fretting wear. After disassembling the FWA qualification unit, it became evident that the assembly suffered from the same fretting wear issue. Much of the bearing-to-shaft interface surfaces were covered with similar wear 1 Fretting wear is a type of adhesive wear that will occur wh en contacting surfaces are undergoing small, oscillatory, tangential displacements. The relative sliding motion causes localized adhesion and disruption of the surface generating fine particulates which then oxide and become imbedded back into the surface causing further abrasion damage. Contamination particles Contamination path

Document

AMS_2012.pdf

403 particulates as seen on the PIL design. However, nodular buildup of fretting wear particles had collected in relief areas of the design as shown in Figure 6 . These 0.127-mm (0.005-inch) diameter nodules had the consistency of dry powder but would easily smear when touched or dissolve into a paste with isopropyl alcohol. The material composition, as determined from a scanning electron microscope analysis, was the same as what was seen in the PIL. Due to the stringent cleanliness requirement for these mechanisms and the optics that surrounded them on the NIRCam bench, this contamination from fretting wear of the flexure bearing mount was unacceptable. The next step was to resolve the issue. Figure 6. Similar fretting w ear found in FWA qualification uni t bearing shaft. The 0.127-mm (0.005-inch) nodules collected from the bearing relief c onsisted of nanometer-sized particles of titanium from the shaft and iron from the bearing and had the consistency of dry powder that smeared when touched. Fretting Wear Resolution Trade Studies The fretting wear contamination was clearly unacceptable, but what course of action could be taken that would ensure no contamination of the nearby optics, no performance degradation of the different functionalities of the FWA and PIL mechanisms, and not result in significant delays to the program? The team considered a number of possible solutions including using a conventional hard-clamp bearing mount, trying to contain the contamination, fabricating the shaft from a less fret-wear sensitive material, adding different wear resistant coatings to the shaft, as well as combinations of these ideas. The team was uncomfortable with a hard-clamped bearing design for two main reasons. First, the team was already considering changing the material of the bearing flexure clamp. Secondly, the test history of the current design had met all other performance requirements, and a hard-clamped bearing mount design would likely require a new set of qualification testing. Therefore, the design team focused on changes to materials and coatings and possible containment. There is a wealth of knowledge related to wear-resistance coatings and materials for many industrial applications as well as many aerospace applications. However, the knowledge base dwindles significantly for wear-resistant materials and coatings for applicatio ns at cryogenic temperatures that must operate in a vacuum and also survive launch vibrations in a dry air environment. By investigating known heritage materials and coatings used for near-ambient temperature space applications and more recent coating technology innovations that show promise, a selection of four candidate solutions where chosen. It was clear that test data at cryogenic temperatures for the selection was all but non-existent, and even the ambient test data did not provide wear performance related to the FWA and PIL fretting wear issue. The team quickly determined that a test was the best method to discern the best solution for the problem.

Document

AMS_2012.pdf

404 Down Selection Testing for Best Fretting Wear Performance The team devised a test for the worst case design configuration; the design that induced the most stress on the bearings. The FWA bearing design supported the higher mass spread out over large wheels and required the most stiffness to keep the dynamic deflections to a minimum. It was impractical, however, to build four filter wheel assemblies. Therefore, the team developed a surrogate design that would replicate the FWA bearing mount design to the exact dimensions and tolerances. The FWA design consisted of two independently driven optic wheel assemblies mounted on a common fixed shaft. Each optic wheel assembly was configured with the same flexible bearing mount and incorporated the same duplex bearing pair as used on the PIL, but with a thicker set of bearing spacers as seen in Figure 7 . Figure 7. FWA dual wheel bearing and motor mount configuration on solid titanium shaft. The test configuration replicated only half of the FWA design and included a surrogate wheel of the same mass that induced the same bearing forces and moment of inertia. The test configuration also replicated many of the other parts to exact dimensions and tolerances. The difference of each of the four test configurations was the unique change that presented possible solution to the fretting wear issue. Because the hardened 440C steel of the bearing was already very resistant to fretting, no changes to the bearings were incorporated. The test configurations focused only on changes to the titanium shaft and anodic coated titanium bearing sleeve. Three coatings and two material changes were chosen as candidate solutions for the test. The coatings were chosen for their wear resistance and adherence properties on titanium down to cryogenic temperatures. The coatings selected were titanium-nitride (TiN), ion-plated gold (Au), and an amorphous carbon-based tungsten-carbon/carbide (WCC). The Au and TiN coatings have been used successfully as wear-resistance coat ings at near ambient aerospace applications and have demonstrated good adherence to titanium. Despite having no direct aerospace heritage, the excellent wear and adherence data demonstrated by the WCC coating warranted a spot on the candidate list. Bearin g OD Clam p Ring Shaft end ca p (Ti) Solid Shaft (Ti) Bearing & bearing spacer (440C) Cryo Release Sleeve (Ti and Pol ymer) ½-28 Shaft Nut (Ti) Clam p Disk (Ti) FWA H ousing Motor Rotor (455 CRES )

Document

AMS_2012.pdf

405 The design team was very hesitant to change materials due to the differential CTE issues, but the wear resistant properties of Nitronic-602 and MP35N3 were hard to ignore. A previous aerospace application demonstrated the wear characteristics of an MP35N shaft and a Nitronic-60 sleeve, and the FWA solid shaft design could benefit from the higher strength of the MP35N. So for the fret wear test configuration, all sleeves were to be made of Nitronic-60, and one shaft was to be made from MP35N. However, because of the differential CTE issue, and the lack of direct CTE measurements of this material down to 35 K, samples of each of these materials were tested with the results shown in Figure 8. Figure 8. Measurements of thermal expansion down to 35 K for both Nitronic-60 and MP35N. The dL/L of Nitronic-60 at 35 K is 0.28% and 0.21% for MP35N. Although each unique candidate solution was representative of the original design, the team judged that there were enough differences in the test that a control configuration was needed. Therefore, a fifth configuration was added to include the original, un-coated titanium shaft and anodic coated sleeve. After completing the assembly of each of the configurations, each was mounted onto a common vibration block that was to be used to subject all units to the required three-axis vibration test. A thorough analysis of the five-wheel test block, as shown in Figure 9, was performed to confirm block input values would induce representative loads into the system. Figure 9. Diagram of test configuration and picture of the five test configurations mounted to a vibration block and sealed for contamination prevention. 2 Nitronic-60 is an austenitic stainless steel alloy known for its wear and galling resistance and is used in many high temperature and corrosion resistant applications. 3 MP35N is a nickel, cobalt, chromium, and molybdenum, allow which has a inherent high strength, outstanding corrosion resistance, and excellent cryogenic properties.

Document

AMS_2012.pdf

406 In preparation for testing the candidate solutions, the team recognized that the fret wear contamination was observed after only one 3-axis vibration test, but the FWA and PIL flight units will undergo four vibration tests during ground preparations and then launch. Additionally, the flight units will be subjected to a number of cryogenic test cycles in between the various vibration tests. Furthermore, the JWST review team volunteered that the launch capsule may be purged with dry nitrogen through ascent, and it is documented that WCC coating friction levels increase and TiN coatings friction levels decreases in low humidity environments. With all these factors in mind, the team developed a test plan that would subject the candidate solutions to all representative atmospheric and vibration conditions and be subsequently dis-assembled to determine which configuration produced minimum fretting wear contamination. Test Results The test configuration was subjected to the equivalent of five 3-axis vibration tests, a cryo test, and one final vibration test with all units purged to less than 5% relative humidity. The data review from the set of five 3-axis vibration tests yielded no anomalies with first mode frequencies at 200 Hz, maximums of a little over 30 Gs rms, and only small variations from predictions. Additionally, signature run comparisons throughout the test showed modest variations in amplitude and no changes in frequency. The tests were deemed successful, and before progressing to the cryo testing, the assemblies were disassembled for inspection of the bearing mounts. The control configuration with a titanium shaft and anodic coated titanium sleeve showed similar fretting wear debris as before, shown in Figure 10. The ion-gold coated titanium shaft with Nitronic-60 sleeve also fared poorly due to apparent coating failure and fretting wear debris as shown in Figure 11. The TiN coated titanium shaft showed improved fret wear performance, but a nominal amount of fret wear debris was still found on this configuration as shown in Figure 12. The WCC-coated titanium shaft, however, showed no indication of coating wear, and very little fret wear debris as shown in Figure 13. Finally, the MP35N shaft showed very small amounts of wear and little wear debris. After initial inspection, the WCC-coated titanium shaft showed the best performance. Figure 10. Scanning electron microscope (SEM) anal ysis of the titanium shaft and anodic-coated titanium sleeve. Titanium and iron debris (1 & 2), and conglomeration of fine particles are characteristic of the fret wear seen before on original design.

Document

AMS_2012.pdf

407 Figure 11. SEM analysis of ion-plated gold ti tanium shaft and Nitronic-60 sleeve. The photo shows the titanium shaft with gold coating failure. The SEM micrographs showing bare Ti surface (1), iron debris (2), and ion gold plating and Ti debris (3). Figure 12. Photo and SEM analysis of TiN-coated titanium shaft and Nitronic-60 sleeve. The TiNcoated titanium shaft showed only small amounts of wear. The SEM analysis indicates iron, Ti, and TiN debris (1 & 2), and TiN plating on the shaft bearing seat (3). The small particulates observed on the shaft consisted of conglomerates of sub-micron sized iron and TiN particles typical of fret wear. The last micrograph shows an example of the numerous voids and protrusions prevalent in TiN coatings.

Document

AMS_2012.pdf

408 Figure 13. SEM analysis of the WCC-coated shaft and Nitronic-60 sleeve. The WCC-coated titanium shaft showed no wear marks on the shaft surface. Only small amounts of tungsten and iron debris were found on the shaft bearing seat. Figure 14. SEM analysis of the MP35N shaft and Nitronic-60 sleeve. The MP35N shaft surface showed wear only at the microscopic level, and very little wear debris. The debris consisted predominately of iron with some cobalt alloy constituents. The next test was to subject the different shafts to cryogenic temperatures. Of the four initial candidate shafts, only three were placed in a cryogenic chamber and brought down to 35 K at a maximum of 40 K per hour as shown in Figure 15. The ion-plated gold shaft was not advanced due to poor performance. After returning them to ambient conditions, the shafts were reviewed under visual magnification as well as SEM analysis. The results of the inspection after cryo yielded no adhesion anomalies, and the shafts were re-installed into their respective surrogate FWA configuration for the last vibration test.

Document

AMS_2012.pdf

409 Figure 15. Cryogenic test fixture with candidate shafts mounted to cold head (shrouds removed). The gold shaft was replaced with a spare titanium shaft for temperature instrumentation. The last test for the candidate configurations was a shortened version of the previous three-axis vibration test but in dry air. During this vibration test, the assemblies were only subjected to one random and one sine vibration test per axis. However, these vibration tests required that each configuration be plumbed to dry air to ensure the tests were conducted in less than 5% RH dry air. This was stipulated by the customer to address the possibility that the JWST observatory would be launched in dry air conditions combined with the fact that the friction coefficient of WCC coatings increases in dry environments. As shown in Figure 16, each configuration was plumbed with independently controlled dry air input and monitored with independent humidity meters. As a side note, the MP35N configuration was eliminated from the test due to the higher density (i.e., weight) of this material. After this vibration test was completed, the units were disassembled and inspected. Figure 16. Second vibration test of the TiN-coated, WCC-coated, and bare (control) titanium shaft configurations. Note the dry air input as well as the humidity meters for each configuration. Conclusion After finding contamination in the bearing mounts of the FWA and PIL assemblies through testing, the team was able to identify the root cause and modify the design to resolve the issue. Fretting wear between the titanium shaft and bearings and sleeve was found to be the source of the contamination. The team selected a set of candidate designs that could be implemented to minimize the fretting issue after conducting a study of possible remedies. Due to limited cryogenic data on the candidate solutions, a series of random vibration tests on a simulated wheel assembly were run to verify the improved fretting

Document

AMS_2012.pdf

410 wear performance at 35 K. The original configuration was included in the tests for a control comparison. A clear winner was found in the WCC-coated titanium shaft with Nitronic-60 sleeve and original 440C bearings, as shown in Figure 17. The cobalt alloy (MP35N) configuration also showed little wear and only small amounts of debris, but it did not perform as well as the WCC-coated shaft. It also had the disadvantage of a differential CTE and higher density. The titanium-nitride coated titanium shaft showed more wear than the WCC-coated shaft and more debris was observed with this configuration. The micro-pits and protrusions were also a detractor for this configuration. Lastly, the ion-plated gold configuration showed the poorest performance with significant adhesion problems and large amounts of wear debris. The hard WCC-coating proved to be the best solution to minimize the risk of wear contamination for this application. Figure 17. Four candidate configurations showing relative fret wear performance. The WCCcoated titanium shaft demonstrated the least susceptibility to fretting wear. References 1. Clark, Charles S., “Redesign and Test of Cryogenic Mechanism for Improved Stiffness”, Proc. SPIE Paper 8150-19 (August 2011). 2. Clark, Charles S., “NIRCam Pupil Imaging Lens Actuator Assembly”, Proc. SPIE Paper 7439C-46 (August 2009). 3. Sean McCully, "Experimental Development Tests of a Cryogenic Filter Wheel Assembly for the NIRCam Instrument and James Webb Space Telescope", Proc. Aerospace Mechanisms Symposium (May 2006).

Document

AMS_2012.pdf

411 Design and Manufacturing Considerations for Shockproof and Corrosion-Immune Superelastic Nickel-Titanium Bearings for a Space Station Application Christopher DellaCorte* and Walter A. Wozniak* Abstract An intermetallic nickel-titanium alloy, 60NiTi (60wt%Ni, 40wt%Ti), is a promising tribological material for space mechanisms. 60NiTi offers a broad combination of physical properties that make it unique among bearing materials. 60NiTi is hard, electrically conductive, highly corrosion resistant, readily machined prior to final heat treatment, easily lubricated and is non-magnetic. It also falls within the class of superelastic alloys and can elastically endure large strains (beyond 5%) making it highly resistant to excessive and unexpected (shock) loads. Key material properties and characteristics such as elastic modulus, tensile fracture sensitivity and residual stress behavior, however, differ from conventional alloys such as steel and this significantly affects bearing design and manufacturing. In this paper, the preliminary design and manufacture of ball bearings made from 60NiTi are considered for a highly corrosive, lightly loaded, low speed bearing application found inside the International Space Station’s water recyling system. The information presented is expected to help guide more widespread commercilization of this new technology into space mechanism and other applications. Introduction Recent research on binary nickel-titanium (Ni-Ti) alloys has identified them as promising candidates for bearings and mechanical components [1]. The nickel-rich alloy, 60NiTi in particular, exhibits a remarkable combination of properties and characteristics relevant to rolling element bearings for space mechanism applications. For instance, 60NiTi is hard, electrically conductive, highly corrosion resistant, readily machined prior to final heat treatment, easily lubricated and is non-magnetic [2,3]. 60NiTi is also in the family of superelastic alloys and can elastically endure large strains (beyond 5%) making it highly resistant to shock loads. Table I lists many of these properties as they are currently known alongside the conventional shape memory alloy 55NiTi and more traditional bearing materials. As an emerging material, some of these published properties are estimated or preliminary. Table I – Representative Thermophysical and Mechanical Properties of Bearing Materials Property 60NiTi 55NiTi 440C Si3N4 M-50 Density 6.7 g/cc 6.5 g/cc 7.7 g/cc 3.2 g/cc 8.0 g/cc Hardness 56-62 Rc 35-40 Rc 58-62 Rc 1300-1500Hv 60-65 Rc Thermal Cond. W/m-°K 18 9 24 33 ~36 Thermal Expansion ~10x10 -6/°C ~10x10-6/°C 10x10-6/°C 2.6x10-6/°C ~11x10-6/°C Magnetic Non Non Magnetic Non Magnetic Corrosion Resistance Excellent Excellent Marginal Excellent Poor Tensile/Flexural Strength ~1000 MPa ~900 MPa 1900 MPa 600-1200 MPa (Bend Strength) 2500 MPa Young’s Modulus ~114 GPa ~100 GPa 200 GPa 310 GPa 210 GPa * NASA Glenn Research Center, Cleveland, OH Proceedings of the 41st Aerospace Mechanisms Symposium, NASA Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

412 Poisson’s Ratio ~.34 ~.34 .30 .27 .30 Fracture Toughness TBD TBD 22 MPa/ √m 5-7 MPa/ √m 20-23 MPa/√m Max. Use Temp ~400°C ~400°C ~400°C ~1100°C ~400°C Elect. Resistivity ~80x10-6 Ωcm ~80x10-6 Ωcm ~36x10-6 Ωcm Insulator ~60x10-6-6 Ω-cm TBD means “to be determined” Shock loading is a significant design challenge for space mechanisms. On orbit bearing loads, in the absence of gravity tend to be very low but the severe launch vibration environment can lead to bearing damage through the Brinell effect in which hard rolling elements dent more vulnerable races [4]. For applications that demand long life and ultra smooth operation, such Brinell damage can be catastophic. In such cases, great care is taken using vibration isolation features and tie-down systems to avoid the problem. These add weight and complexity. At times, load capacity design margins are increased for bearings and this leads to increased mass and power consumption. Clearly, the development of more resilient bearings is an advantage for aerospace bearings and mechanisms. Recent investigations into the hardness and Brinell damage sensitivity of 60NiTi show a potential pathway to engineer ball bearings that are highly resistant to indentation damage [1]. In preliminary indentation tests either Si 3N4 or 60NiTi balls (12.5mm diameter) were pressed into flat plates made from 60NiTi and the bearing alloys 440C, M50 and Stellite 6B. Initial experiments were carried out at light loads that were increased on subsequent trials. In this manner, both the classic Brinell Hardness Number (BHN) and the threshold load to achieve the first observable dent were obtained. The data, tabulated in Table II, shows that 60NiTi provides significantly more static load capacity (higher threshold load) than the other materials tested. Table II. Hertz contact stresses and contact diameter at the threshold load for various plate and indenter material combinations. Plate Indenter Threshold load, kgf (lbs) Peak Stress, GPa (ksi) Contact Dia., mm (in) Avg. Stress, GPa (ksi) Stellite 6B Si 3N4 10 (22) 2.06 (299) 0.30 (0.012) 1.37 (199) 440 C Si 3N4 51 (112) 3.48 (504) 0.52 (0.021) 2.32 (336) M50 Si 3N4 150 (331) 5.09 (738) 0.74 (0.029) 3.39 (491) 60NiTi Si 3N4 552 (1214)* 5.56 (806) 1.36 (0.054) 3.71 (537) Stellite 6B 60NiTi 15 (33) 1.56 (226) 0.42 (0.017) 1.04 (151) 440 C 60NiTi 150 (331) 3.33 (483) 0.92 (0.036) 2.22 (322) M50 60NiTi 501 (1102) 5.02 (728) 1.37 (0.054) 3.35 (486) 60NiTi 60NiTi 1512 (3327) 5.90 (856) 2.19 (0.086) 3.94 (571) *Using 1.2 m (50in) dent depth fatigue criterion. Threshold load is ~360kgf (~800 lb) using more stringent 0.5 m (20in) dent depth criterion for quiet running bearing. The reasons for this behavior are complex and discussed in detail in Reference 1 but briefly, the large elastic deformation range of 60NiTi combined with its high hardness and relatively low elastic modulus result in an increased contact area, reduced peak and average stresses and enhanced static load capacity. Given these somewhat unexpected results and capabilities, the next logical steps are to design and fabricate bearings from 60NiTi for evaluation. Bearing design and manufacturing is a field that is mature and well developed for steels and, in the case of rolling elements (balls and rollers), silicon nitride ceramics. 60NiTi is a superelastic material in the class of intermetallics (neither a metal nor ceramic with respect to atomic bonding) and has key physical properties and attributes that differ from the traditional bearing materials. Among the relevant differences are a relatively low elastic modulus, limited tensile ductility (a tendency for brittle failure in tension), and a

Document

AMS_2012.pdf

413 requirement for a rapid quench during heat treatment which leads to significant residual stresses. These three key differences along with other second order attributes affect bearing design and manufacturing processes. The following sections describe the general design and manufacturing process for 60NiTi bearings by considering a specific space mechanism bearing application; the Distillation Assembly (DA) centrifuge that is part of the ISS Environmental Control System (ECLSS) on the International Space Station (ISS). The ball bearings used in this application operate in a highly corrosive environment at low speed under very low average loads thus fatigue is not a great concern. However, the bearings must endure high launch loads without damage. Based upon preliminary data, 60NiTi appears to be a viable candidate bearing material. This paper lays out the design process and the results of pathfinder manufacturing investigations to develop a more generalized methodology for incorporating superelastic materials into space mechanisms. Materials and Procedures Bearing grade 60NiTi is manufactured via a proprietary high-temperature powder metallurgy (PM) process roughly similar to that described in the literature [5]. Pre-alloyed 60NiTi powder is hot iso-static pressed (HIPed) into various shapes and sizes depending upon the desired end product. To make 60NiTi balls, the powder is HIPed into rough, spherical ball blanks that were then ground, polished and lapped. Because the PM process yields ball blanks that have isotropic mechanical properties high quality (Grade 5) bearing balls can be readily produced. The finished 60NiTi ball specimens, shown in the photograph in Figure 1, are bright and shiny in appearance and resemble conventional polished steel balls. Figure 1. 60NiTi Grade 5 test balls. To make other shapes such as bearing races and mechanical and thermo-physical property measurement specimens, 60NiTi rods and ingots were first made using the same PM process. Figure 2 shows such specimens produced by the PM process.

Document

AMS_2012.pdf

414 Figure 2. 60NiTi ingots and blanks produced by PM process. Once the ingots and other shapes are produced, a series of steps that include wire electrode discharge machining (EDM), conventional machining using carbide tools and grinding are employed. A multi-step thermal process (heat treatment) generally occurs after rough machining to near final dimensions but before final grinding and polishing in the case of balls and bearing raceways. For parts that are not dimensionally critical, simple wear plates for example, final grinding after heat treatment may be unnecessary. A typical heat treatment includes solution treating at 1000 C in vacuum or inert gas atmosphere followed by a rapid quench in water. The solution treating dissolves precipitate phases like Ni 4Ti3 and Ni 3Ti forming the preferable NiTi phase. Rapid quenching locks in the dominant NiTi phase and discourages the formation of the other phases that can lead to brittleness and low hardness. Details regarding the processing and resulting properties are the subject of ongoing research and are partially described elsewhere [6]. Using the generalized processes described above, specimens are being fabricated to enable rolling contact fatigue stress limits, bend strength, Charpy impact toughness, compressive strength and thermal properties such as thermal expansion, thermal conductivity and thermal diffusivity. This data will be available in the future but initial bearing design can proceed using information available and estimated. Design and Manufacturing Considerations A baseline bearing can be designed using the preliminary and estimated 60NiTi materials properties. This design can then be reviewed and analyzed for its general appropriateness the DA bearing application. Using the preliminary design, a manufacturing method must be then developed and tested. Since key structural properties of 60NiTi differ from bearing steel (notably the elastic modulus is low and high machining forces are needed for material removal), the manufacturing process and tooling may differ from the norm. If the manufacturing investigation identifies design deficiencies it can be revised in an iterative manner. Further, as additional material properties are obtained, such as rolling contact fatigue stress limits, the design can be revisited. The following sections expand upon the preliminary design analysis and a pathfinder manufacturing trial. Results and Discussion The DA bearing application currently under consideration is characterized as highly corrosive and mechanically benign. The bearings are nominally 50-mm bore deep groove ball bearings operating at low speed (a few hundred rpm) under very modest axial preload. In space, there is virtually no imposed radial load. Table III gives the representative baseline bearing conditions.

Document

AMS_2012.pdf

415 Table III - Representative Application Bearing Parameters And Operating Conditions Parameter Value or Condition Outer Diameter, OD ~80 mm Inner Diameter, ID ~50 mm Width, W ~16 mm Ball Size, D ~9 mm Ball Material Si3N4 Race Material Cobalt Alloy Cage Snap fit, polymer Lubricant Lithium based grease Ball-Race Stress Limit ~2 MPa Ball-Race Mean Stress ~1 MPa Axial Preload ~200 N Radial Load (terrestrial) ~100 N/bearing Speed 100-300 rpm Environment Warm, highly acidic aqueous solution Ambient Pressure Slight vacuum When reviewing the data and parameters in Table III it becomes clear that the environmental stresses on the bearings, namely the warm and highly acidic environment, far outpace the modest mechanical loads and stresses. In fact, during early system development, the originally specified martensitic stainless steel bearings (440C) experienced unacceptable surface corrosion. Following bench top corrosion studies in a simulated environment, the steel bearings were replaced with the cobalt race-Si 3N4 ball hybrid bearing design as a baseline. Figure 3a shows a photograph of the DA assembly during ground tests. This device is essentially a rotating drum evaporator that utilizes a belt drive to provide a low pressure, warm internal chamber that boils wastewater on the ISS. The steam coming off of the drum section is collected and condensed and is then further treated before re-use. Further details of this system can be found in the literature [7-9]. In cross section (Figure 3b shows a representative design), the subject rotor system is a simple configuration of two shielded ball bearings that are lightly spring preloaded and operate in the warm, moist and acidic environment.

Document

AMS_2012.pdf

416 Figure 3a. ISS Distillation Assembly in ground tests. Figure 3b. Representative ISS Distillation Assembly cross-section taken from reference 9.

Document

AMS_2012.pdf

417 Figure 4 shows a photograph of the centrifuge bearings after removal from service. The bearings are a straight ball bearing design and utilize a cobalt alloy for the races and silicon nitride for the balls to achieve maximum corrosion resistance. Figure 4. ISS Distillation Assembly centrifuge bearings. The performance of the current DA bearings has not been completely satisfactory. The very first system launched suffered from raceway damage that occurred during bearing assembly and installation. Though this damage did not result in system failure, long life could not be assured. The second system launched was returned to earth after a short period of use for repair of an unrelated system component. During a routine bearing inspection significant race wear was observed. Figure 5 shows a photograph of the surface of a worn inner race in which the ball has created a pronounced wear track. Figure 5. Photograph of an ISS Distillation Assembly centrifuge bearing showing wear track following operation on orbit. The wear was attributed to the lack of hardness of the cobalt-based races compared to the ceramic balls. The bearings still functioned, but again, long-life was not assured. It is these shortcomings of the current bearing race materials that provide the impetus to consider 60NiTi. Among the first material selection criteria to be assessed is corrosion resistance to the process fluid.

Document

AMS_2012.pdf

418 A variety of candidate bearing alloys including steel, 60NiTi and the baseline cobalt alloy were evaluated for corrosion resistance. In this test, ball sized specimens were immersed in a warm and highly acidic aqueous solution that simulates the application. The specimens were weighed before and after exposure for times up to one week and the immersion fluid was analyzed to determine its metal ion content as a means to assess corrosion behavior of the alloys. Table IV shows a summary of the results. Table IV - Corrosion Mass Loss for Balls (12.7-mm dia.) Soaked in simulated DA solution (115 hr, 65 C, pH 1.0 ) M50 Tool Steel 27,600g Cobalt Alloy 389g 60NiTi 224g Not surprisingly, the corrosion resistance of 60NiTi is excellent. Unlike stainless steels, nickel and cobalt alloys do not rely upon the formation of chrome rich passivation layers for their corrosion resistance. Rather, such alloys employ intrinsically corrosion resistant constituents in their chemistry. For Ni-Ti alloys, such as 60NiTi, both nickel and titanium are regarded as highly immune to aqueous acidic corrosion thus the corrosion resistance for the alloy is to be expected. For corroboration, the current cobalt alloy bearing races have not shown any evidence of corrosion problems even after months of service. Since the 60NiTi fares better, in terms of corrosion, than the current cobalt alloy is it justifiable to expect the superelastic bearings will not suffer corrosion. In terms of mechanical properties, however, deeper consideration is warranted. The bulk hardness of 60NiTi (table I) ranges from Rc 58 to Rc 63 depending upon the heat treatment employed and the dimensions and geometry of the specimen. This is comparable to traditional bearing steels and much harder than the baseline cobalt alloy race material. Rolling and sliding wear tests of 60NiTi under dry and grease lubricated conditions yield tribological behavior comparable to that for hardened bearing steels like 440C [2, 3]. For these reasons, the wear observed in the current bearings is not expected for bearings made from the much harder 60NiTi. 60NiTi, however, has a much lower elastic modulus than both traditional steels and the cobalt alloy. This could affect bearing operation. The elastic modulus for 60NiTi closely resembles titanium and is less than half that of steel and superalloys (Table I). Thus under load, one expects to encounter deformations and deflections that are approximately twice the level of comparable steel or superalloy components. For mechanical systems in which rigidity and position control is paramount, such a change in elastic behavior can be a design challenge. In a bearing, the radial and axial stiffness are direct functions of the material elastic modulus and thus shaft deflections, for a given load, will increase when going to a superelastic bearing design. At the ball-race contact within a bearing, the lower modulus results in larger hertz contact areas and commensurately lower stresses. Recently reported analyses of 60NiTi roller bearings suggest that comparing all steel to all 60NiTi cases can be complex [10]. The reduced modulus of 60NiTi can be viewed as a higher compliance. Under load in a full bearing system, every rolling element deforms to a higher degree than a steel roller and thus shares the bearing load more readily with its neighboring rolling elements. This effectively reduces the load on each rolling element and may actually increase the ultimate load capacity beyond that estimated from the static load capacity data contained in Table II. From a design perspective, larger contact areas, lower stresses and higher deflections arising from the use of low modulus materials implies that the a careful and thorough detailed design review must be undertaken for all highly loaded bearing applications especially those in which precise orbit control and positioning is vital. For the DA bearing application, the loads are very low and the positioning (bearing stiffness) requirements are minimal. The only continuous loads on the bearings arise from spring preload washers and modest fluid motion-dynamic unbalance forces during normal rotation (relatively low speed of 200 rpm). For completeness, a cursory design stress and stiffness review was done and it concluded that from a mechanical perspective, for this application, the elastic modulus of the bearing race material is

Document

AMS_2012.pdf

419 not a critical parameter as long as the static load capacity exceeds that of the baseline superalloy and it does (Table II). Based upon the results from the mechanical loads and stresses review and the preliminary corrosion studies, it appears that a 60NiTi bearing utilizing the very same geometry and dimensions as the cobalt-hybrid baseline design is a good initial baseline design to consider for manufacturability. Before a manufacturing pathfinder trial can begin, a computer-generated model of the target bearing was developed. This model helps the manufacturing engineer visualize the bearing geometry and devise a manufacturing plan. The bearing model is shown in Figure 6. Figure 6. Computer model of DA bearing. The manufacture of 60NiTi balls is a fully commercialized process and many standard ball sizes are available. The steps needed to make precise raceways from 60NiTi to a desired geometry are not yet fully developed. Decades of experience with various steels has resulted in numerous rules-of-thumb (ROT) for their manufacture. Preferred annealing, rough machining, hard turning, heat treatment, grinding, polishing, acidic surface passivation, normalizing and other processes for steels are well understood and accepted by the manufacturing community. For 60NiTi these ROT’s remain undefined. The following paragraphs attempt to layout an acceptable, by not necessarily an optimal, processing path driven by 60NiTi’s several unique properties and characteristics, namely its low elastic modulus, resistance to metal cutting and deformation (i.e., low ductility), and its intrinsically brittle tensile behavior. The basic manufacturing steps employed to go from ingots to finished bearings are as follows: rough machining to near final shape and size using a combination of wire electro discharge machining (EDM) and carbide tool based machining, heat treating to develop high hardness, finish grinding and polishing. Because 60NiTi exhibits limited ductility and is susceptible to brittle tensile behavior, heavy machining operations such as drilling, high removal turning and milling should be avoided. Instead, we have successfully employed a combination of plunge and wire EDM to generate simple blanks from larger ingots and these are then rough machined to near final shape using more conventional machining methods. Figure 7 shows a slice of 60NiTi made by wire EDM and subsequently cut into cylindrical (donut

Document

AMS_2012.pdf

420 shaped) race blanks. The ingot, made by powder metallurgy has a rind and center core of steel with 60NiTi filling the annular space. This is part of the PM process and enables more efficient use of the relatively high cost 60NiTi materials. Figure 7. 60NiTi race blanks cu t from ingot using EDM processes. The resulting ring blanks are then shaped into race rings using properly shaped carbide tooling through a plunge turning operation. With this operation, the ring inside diameter is first turned or ground to ensure it is uniform and close to finish dimension (within ~0.01 mm). The ring is then mounted on an expanding type mandrel and care is taken to avoid excessive clamping forces. Because the elastic modulus is half that of steel, excessive clamping force on the inside diameter can result in considerable stretching of the ID leading to unpredictable dimensions after turned race is removed from the mandrel. Once mounted, the OD is plunge turned using a carbide tool previously shaped to match the desired race ring profile. In our tests, we used wire EDM to manufacture the plunge tool from a standard carbide lathe bit. Figure 8 shows the process for turning an inner race ring prior to the heat treatment step.

Document

AMS_2012.pdf

421 Figure 8. Plunge turning of 60NiTi race rings using a carbide tool profiled to match the desired race geometry using wire EDM. The final steps include heat treatment and finish grinding and lapping. The heat treatment has been described in detail in reference 6 and briefly includes a solution treatment at 1000 C followed by a rapid water quenching. A vacuum or inert atmosphere is used to prevent excessive surface oxidation. Figure 9 shows the microstructure of 60NiTi prior to and after heat treatment. The heat treatment dissolves undesirable higher order phases resulting in a largely homogenous NiTi phase structure with hardness in the range of 58 to 63 on the Rockwell C scale.

Document

AMS_2012.pdf

422 Figure 9. Microstructure of 60NiTi before (left) and after heat treatment (right). After heat treatment, standard industry grinding and polishing are employed to yield a finished bearing. Care must be taken to avoid excessive tensile stresses during bearing assembly. Unlike steel that can endure high tensile deformation without fracture, 60NiTi is brittle in tension and outer races, in particular, can break during bearing assembly when hoop stresses are high. To overcome this problem, either separable bearing designs or the use of differential inner ring cooling and outer ring heating are recommended. Beyond that, no other special considerations are needed. The bearings behave and perform as one expects based upon the geometry and lubricant selected. Future efforts to fully characterize the rolling contact fatigue stress-life relationships and gain a better understanding of the tensile strength and toughness behavior of 60 NiTi are planned to guide more mechanically demanding bearing applications. Summary Remarks 60NiTi offers a viable path towards rolling element bearings that have exemplary corrosion and wear resistance. Though sensitive to brittle fracture in tension, research suggests that extreme static load capacity and resilience imparted by its superelastic behavior is possible. The reduced elastic modulus, compared to steel, may impact highly loaded bearing internal geometry but for the DA bearing application being considered, a direct 60NiTi replacement is reasonable. Lastly, manufacturing research underway suggests that with proper care, modern manufacturing methods are capable of producing high precision ball bearings from 60NiTi. Acknowledgements The authors wish to acknowledge the support of the NASA Engineering and Safety Center (NESC) and the NASA Fundamental Aeronautics Program (Subsonic Rotary Wing Project). Their continued funding and technical and programmatic guidance has contributed greatly to bringing this new technology from the lab and closer to commercialization.

Document

AMS_2012.pdf

423 References 1. C. DellaCorte, R.D. Noebe, M.K. Stanford, and S.A. Padula: “Resilient and Corrosion-Proof Rolling Element Bearings Made From Superelastic Ni-Ti Alloys for Aerospace Mechanism Applications,” NASA TM-2011-217105, September 2012. 2. C. DellaCorte, S. V. Pepper, R. D. Noebe, D.R. Hull, and G. Glennon: “Intermetallic Nickel-Titanium Alloys for Oil-Lubrication Bearing Applications,” NASA TM-2009-215646, March 2009. 3. S. V. Pepper, C. DellaCorte, R. D. Noebe, D.R. Hull, and G. Glennon: “Nitinol 60 as a Material for Spacecraft Triboelements,” ESMATS 13 Conference, Vienna, AUstria, September 2009. 4. A. Palmgen: ”Ball and Roller Bearing Engineering, Chapter 8, ”Bearing Failures,” pages 217-225, 3rd Edition, SKF Industries, Philadelphia, PA, 1959. 5. M.D. McNeese, D.C. Lagoudas, and T.C. Pollock: ”Processing of TiNi from Elemental Powders by Hot Isostatic Pressing,” Materials Science and Engineering, A280 (2000), pages 334-348. 6. M.K. Stanford, F. Thomas, and C. DellaCorte: ”Processing issues for Preliminary Melts of the Intermetallic Compound 60-NiTiNOL, NASA TM- 2011-2 xxxx, A ugust 2011. 7. D.L. Carter: ”Status of the Regenrative ECLSS Water Recovery System, SAE 2009-01-2352, 39th International Conference on Environmental Systems, July 12-16, Savannah, GA, 2009. 8. C.D. Wingard: ”Compatibility Testing of Polymeric Materials for the Urine Processor Assembly (UPA) of International Space Station (ISS), NASA report 20030106448-2003127090, NASA Marshall Space Flight Center, Huntsville, AL, 2003. 9. D.W. Holder and C.F. Hutchens: ”Development Status of the International Space Station Urine Processor Assembly,” SAE 2003-01-2690, 33rd International Conference on Environmental Systems, July 7-10, Vancouver British Columbia, 2003. 10. T.L. Krantz: ”On Calculation Methods and Results for Straight Cylindrical Roller Bearing Deflection, Stiffness and Stress,” Proceedings of the ASME 2011 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference IDETC/CIE 2011, August 29-31, 2011, Washington, D.C., paper number DET2011-47930.

Document

AMS_2012.pdf

425 Wear of Steel and Ti6Al4V Rollers in Vacuum Timothy Krantz* and Iqbal Shareef** Abstract This investigation was prompted by results of a qualification test of a mechanism to be used for the James Webb Space Telescope. Post-test inspections of the qualification test article revealed some loose wear debris and wear of the steel rollers and the mating Ti6Al4V surfaces. An engineering assessment of the design and observations from the tested qualification unit suggested that roller misalignment was a controlling factor. The wear phenomena were investigated using dedicated laboratory experiments. Tests were done using a vacuum roller rig for a range of roller misalignment angles. The wear in these tests was mainly adhesive wear. The measured wear rates were highly correlated to the misalignment angle. For all tests with some roller misalignment, the steel rollers lost mass while the titanium rollers gained mass indicating strong adhesion of the steel with the titanium alloy. Inspection of the rollers revealed that the adhesive wear was a two-way process as titanium alloy was found on the steel rollers and vice versa. The qualification test unit made use of 440F steel rollers in the annealed condition. Both annealed 440F steel rollers and hardened 440C rollers were tested in the vacuum roller rig to investigate possibility to reduce wear rates and the risk of loose debris formation. The 440F and 440C rollers had differing wear behaviors with significantly lesser wear rates for the 440C. For the test condition of zero roller misalignment, the adhesive wear rates were very low, but still some loose debris was formed. Introduction This investigation was prompted by results of a qualification test of a mechanism to be used for the James Webb Space Telescope. The mechanism is used to move a magnet for certain operations of the telescope’s near infrared spectrometer (NIRSpec) instrument. The motion of the magnet is guided by a set of preloaded steel rollers in contact with anodized Ti6Al4V. The qualification testing was accomplished in a cold vacuum chamber to match as closely as possible the extreme deep space environment. Post-test inspections of the qualification test article revealed some wear of the steel rollers and mating Ti6Al4V surfaces, and some loose debris was found. The NASA Engineering Safety Center (NESC) is investigating the potential risk of the wear and debris toward hindering the full capability of the NIRSpec instrument. This article will discuss the NESC investigation of the wear phenomena including dedicated laboratory experiments. Qualification Testing Article and Testing Results. The mechanism of interest is a translator assembly that features a set of 11 rollers to guide the motion of a magnet (Fig. 1). The translation of the magnet is used to affect the positions of micro-shutters. A motor and linkages provide the motive force. The translator assembly undergoes oscillatory motion in a straight line. The translating distance is approximately 200 mm on each stroke (total of 400 mm of travel during back and forth motion) and translation time in one direction during qualification testing was 10.5 seconds. The qualification test included 96,000 of such forward and return excursions. The translator assembly motion is guided by a set of eleven rollers. The in-plane guide is provided by a set of four rollers (two of these rollers are visible to the far right side of Fig. 1). These four rollers contact both sides of a guide rail defining the direction of motion. The out-of-plane position and motion is guided by a set of seven rollers. Three rollers are in contact with a base plate defining one plane and four rollers are in contact with a cover defining a second parallel plane. The roller preloads are set by a shimming * Glenn Research Center, Cleveland, OH ** Bradley University, Peoria, IL Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

426 procedure to a nominal normal load of about 18 N. The roller is a cylinder with diameter of 21.2 mm. The roller profile is a crown radius with a flat feature in the center of the crown. Assuming line contact across the flat feature, the Hertz contact maximum pressure is about 240 MPa. The rollers are made from 440F steel (a free machining variant of 440C). The roller material was in the annealed condition and passivated. A pair of deep groove ball bearings supports each roller. The fastener securing the roller bearings to the axle is a locking style bolt assembled in a manner allowing for a small amount of axial play. The rollers are in contact with Ti6Al4V that had been stress relieved and anodized. After completion of the qualification test, wear was visible on the rollers of the translator assembly and on the mating surfaces. Fig. 2 shows the condition of a qualification test roller. Some loose particulate debris was also found, which was collected and analyzed [1-2]. The worn surfaces and wear debris provide evidence that the wear process was primarily adhesive wear. In total it appears that the steel has transferred to the mating Ti6Al4V surfaces. The loose wear debris included both the steel and titanium alloy materials. The severity of wear was not the same on all rollers. From observations of the wear patterns and wear debris of the qualification unit, study of the literature, and engineering assessment of the kinematic configuration, manufacturing tolerances, and roller mounting details, the roller alignment seemed likely to be a key variable influencing the wear rate. To better understand the wear phenomena and to explore possible mitigation strategies, a set of laboratory roller tests were conducted in vacuum. The next sections will discuss the testing apparatus, specimens, procedures, and the corresponding results. Figure 1 – Translator assembly [1]. Figure 2. Condition of roller after completion of qualification test [2]. Laboratory Test Apparatus, Specimens, and Procedure Test Apparatus for Roller Pairs Testing was done using the NASA Glenn Research Center Vacuum Roller Rig (Figure 3). The rig allows for application and measurement of a load pressing the rollers together while having a purposely misaligned and adjustable shaft angle. The rig is depicted in schematic form in Figure 4. A drive motor provides motion to the driving roller. A magnetic-particle brake attached to the output shaft imposes torque on the driven roller. The rig can be operated with the brake not energized. For such a condition the torque transmitted through the roller pair is the drag torque of the output shaft (drag of the seals and support bearings). The normal load pressing the rollers together is provided by an air cylinder. The cylinder acts through a gimbal point to rotate the plate that mounts the driving shaft and drive motor. The rotation of the drive motor plate displaces the driving roller toward the driven roller shaft in an arc motion. The pressure to the cylinder, and thereby the load between the contacting rollers, is adjusted by a hand-operated valve (open-loop control). A turbomolecular pump assisted by a scroll pump provides vacuum in the test chamber. The typical condition in the test chamber is a pressure of about 3x10 -7 Torr. The most

Document

AMS_2012.pdf

427 prevalent remaining constituent in the chamber during testing is water vapor as determined by residual gas analyzer [3]. Figure 5 provides a simplified schematic of the test rollers labeled with some of the nomenclature used herein. Figure 3 – Vacuum roller rig. A set of sensors on the test apparatus monitors the test conditions. The outputs of the analog sensors were digitized and stored via a data acquisition system at a rate of up to 0.66 hertz. Each of the sensors will be described in turn. The misalignment of the driving roller shaft and driven roller shaft is depicted in an exaggerated manner in Figure 4(b). The misalignment is measured via a linear variable differential transformer (LVDT). The transducer housing is attached to the bedplate, and the translating, spring-loaded transducer tip contacts against a mechanical stop on the turntable. To establish the aligned condition, special tooling blocks were machined to locate the roller-mounting surfaces of the two shafts as parallel. With the shafts aligned by the tooling blocks, the transducer circuit balance was adjusted to provide an output of zero. The precision of this method for aligning the shafts was limited by the dimensions of the roller mounting surfaces used as the reference planes. From analysis of the test rig drawing tolerances and geometry, the alignment procedure using the tooling blocks to define the zero-degree position has a precision within 0.08 degrees. Rotation of the turntable from the aligned position moves the LVDT sensor. The angular displacement of the turntable was determined by mounting a laser light source on the moving shaft at the roller mounting location and directing the light onto a paper placed at a known radial distance from the center of the turntable. The movement of the laser light was marked on the paper and distance between the points measured and used to relate the sensor output to the angular motion of the turntable. The torque on the output shaft is monitored by a strain-gage type torquemeter of 22 N-m (200 in-lb) torque capacity. Calibration was done in place using deadweights acting on a torque arm of known length attached at the test roller position and reacting the output shaft to ground. The load pressing the rollers together is termed herein the “normal load” (Figure 5). The normal load is applied via an air-pressure actuated piston. The air piston acts through a load cell against the drive motor plate that is gimbal-mounted relative to the test chamber (Figure 4(a)). In this way the air piston moves the roller on the input shaft in an arc motion toward the test roller. The arc motion of the input shaft table is measured by an LVDT. Once the rollers are in contact, additional force commanded to the air piston increases the normal load between the test rollers. Careful calibration processes allow calculating the normal load on the test roller based on the sensor outputs from the load cell and the LVDT that measures the input shaft position.

Document

AMS_2012.pdf

428 When rollers operate in a misaligned condition a force will develop in the direction of the shaft axis [4-7]. In such a condition points on the two rollers in intimate contact and within a “stick” zone of the contact patch are constrained to move in unison. If the points were not in contact the kinematic constraints would provide a slightly different path of motion. The difference in the actual path of motion and that defined by the motion if the points were not in contact gives rise to surface strains and a resultant axial force. A sensor to measure this force is labeled as the “axial force” sensor in Figure 4. The axial force sensor is co-located on the output shaft with the torquemeter sensor. The configuration of the rig did not allow for direct deadweight calibration in place. To calibrate the sensor in place, the following procedure was used. First, a load cell was calibrated via deadweights and then was placed on the free end of the output shaft to act as a reference load cell. A threaded jackscrew acted against the reference load cell and a hard stop in the vacuum chamber. Adjusti ng the ja ckscrew length allowed for changing th e force imparted on both the reference load cell and the rig’s axial load cell and to the machine frame. In this manner the same force was applied to both load cells, and the reference cell output was used to calibrate the axial load cell sensor in place. The preceding two paragraphs describe the sensors (and sensor calibrations) to determine two mutually perpendicular forces acting on the driven test roller. A force also acts along a third axis. This is the force directed tangential to the roller diameter and is termed here as the “tangential” force. The tangential force on the input shaft roller acts through a gimbal point (Figure 4(b)). The rotational motion about the gimbal point is restrained by a mechanical link to the turntable structure. A load cell is used to sense the force in said mechanical link to the turntable structure. This sensor was calibrated in place by using a pulley-cable system and dead weights to relate the tangential force applied at the test roller position to the sensor output. During testing, this sensor is also affected by spin moments [4-5] that can develop in roller contacts. The data from the tangential force sensor was recorded for possible future use, but such data were not of immediate interest and are not reported herein. Shaft speeds and total number of shaft revolutions were measured using encoders on each shaft. The encoder pulses were counted and recorded via a digital pulse counter. The encoder pulses were also monitored by a frequency converter to provide a convenient shaft speed display to the test operator. The encoders provide 6,000 pulses for each shaft revolution. Test roller conditions were photographed at regular intervals through a viewport. The images were captured digitally using a single-lens reflex camera with a 150 mm micro lens and a 12 million effective pixel image sensor. A debris pan was used to capture debris created from the roller pairs. A video camera recorded the condition of a portion of the debris pan. In spite of several attempts to adjust video camera setting and lighting, the video images failed to capture all that could be observed by eye through the viewport. Test Specimens The test specimens used for this research had a nominal geometry of 35.6 mm outer diameter and a 12.7 mm width. The apparatus requires that at least one roller has a crowned profile to avoid edge loading. The roller on the drive motor (input) shaft was provided a crown radius of 200 mm and was made from steel. The roller on the lower (output) shaft had a flat profile and was made from Ti6Al4V alloy.

Document

AMS_2012.pdf

429 (a) (b) Figure 4 – Schematic views of the vacuum roller rig. (a) Schematic, side view. (b) Schematic, overhead view with shaft misalignment depicted and exaggerated. Figure 5 – Simplified schematic view including some of the important sensed data. (a) Schematic, front view. (b) Schematic, side view. ferrofluid seals and bearingstorque/force sensor brakepivot point for drive motor plate relative to bed plate drive motor bedplate hydraulic cylinder“normal load” force sensorvac chamber turntable bedplatebellows ferrofluid seal and bearingtorque/force sensor brakedrive motor bedplateturntable on bearingload cell for “tangential” force, mounted to turntable normal load braking torquetangential force normal loadnormal load braking torquetangential force normal load normal load axial force normal loadinput shaft output shaft normal load axial force normal loadinput shaftinput shaft output shaft (a) (b)normal load braking torquetangential force normal loadnormal load braking torquetangential force normal load normal load axial force normal loadinput shaft output shaft normal load axial force normal loadinput shaftinput shaft output shaft (a) (b)

Document

AMS_2012.pdf

430 The material condition, manufacturing details, and surface roughness can all influence the wear behavior. The details of the roller conditions are given in Table I and additional details follow. The qualification test unit has rollers made from 440F steel in the annealed condition and the working surface specified to have a roughness of 0.4 µm maximum. The qualification test rollers were passivated per ASTM A-380-06. The 440F alloy is similar to 440C but includes the introduction of either selenium and/or additional sulfur to alter the machining characteristics relative to 440C. From x-ray photoelectron spectroscopy both the qualification test unit rollers and the 440F laboratory test rollers were found to have selenium as a constituent. The surface hardness of a 440F laboratory test roller was measured as 24 RC. Laboratory test rollers were made intending to enhance resistance to adhesive wear and to reduce debris generation. These rollers were made from 440C, hardened to RC 58, and provided a fine ground surface. Roughness of the test rollers was documented by stylus profilometer inspections. Typical roughness profiles in the rolling direction are provided in Fig. 6. The roughness data of Fig. 6 were prepared filtering the data using 0.8 millimeter cutoff and 300:1 bandwidth ISO standard filter. The roughness of the 440F rollers (0.15 micrometer Ra) was significantly larger than the roughness of the 440C rollers (0.07 micrometer Ra). The test rollers that mated with the 440 series rollers were made of Ti6Al4V. The Ti6Al4V rolling surfaces of the qualification unit were observed to have distinctive machining marks, and so the laboratory Ti6Al4V rollers were made using a turning operation as the resulting surface texture of the test rollers were judged similar to the surfaces of the qualification unit. The rollers were stress relieved per AMS 2801 after machining. The roller hardness was RC 35. Some of the Ti6Al4V rollers were anodized per Tifin 200 matching the qualification unit treatment. A typical roughness profile for the Ti6Al4V rollers after anodizing is provided in Fig. 6. The typical roughness average value was 0.44 micrometer Ra. Some testing was done with Ti6Al4V rollers that were not anodized to make project progress previous to the completion of the anodize process. Procedure to Install Test Rollers Test specimens were cleaned and installed using careful procedures to provide clean test surfaces. The test rollers were cleaned just prior to installation into the rig using de-ionized water and 0.05 micron alumina powder. After appropriate hand scrubbing, the cleaning powder was rinsed with deionizer water making sure that the entire roller surface wetted uniformly to confirm complete cleaning of surface oils. The water was removed from the roller using dried pressurized nitrogen. Test rollers and mounting hardware were handled only with gloved hands and clean tools to complete installation into the test apparatus. Procedure for Testing Rollers The first step for testing after installation of the test rollers was to immediately isolate the testing chamber and provide a vacuum, using the scroll roughing pump, to approximately 50x10-3 Torr chamber pressure. This isolation step was done even if test scheduling required some delay between the time of installation of rollers and the time for testing to minimize exposure of the cleaned surfaces to any contaminants that might be present in the atmosphere. Prior to testing the turbomolecular pump was used to bring the testing chamber to approximately 3x10 -7 Torr.

Document

AMS_2012.pdf

431 Figure 6 – Typical surface roughness of test rollers, profile traces obtained using a 2-micrometer radius tip conisphere stylus traci ng in the direction of rolling. (a) Ti6Al4V roller. (b) 440F roller. (c) 440C rollers. Contact analyses were completed to select roller normal loads, speeds, and alignment angle and to relate those choices to the operating conditions of the qualification test unit rollers. The contact conditions for the qualification test were studied assuming that the contact pressure distribution was affected by the moment produced by the axial load caused by misalignment. To estimate this effect, it was assumed that the ratio of axial load to normal load was 0.6 and such load produced an overturning moment by acting through the ball bearing center. It was also assumed that the supporting structure provided 90 percent of the reaction to the overturning moment while the pressure distribution provided 10 percent of the reaction. The pressure distribution for such a loading condition while accounting for the roller profile having a crown but modified with a flat section was calculated using the method of Vijayakar [9-10]. The contact condition for a perfectly aligned roller having zero axial force was also solved. The predicted contact pressure distributions are provided in Fig. 7. The misalignment of the roller axis will cause a shift of the pressure from the central flat section onto the crowned region, and a maximum contact pressure of about 450 MPa occurs near the transition from flat to crown geometry. Because of test rig limitations, the laboratory test conditions could not match these conditions exactly. The laboratory testing was done at maximum contact pressure of about 770 MPa. The laboratory tests were used to study trends and fundamental qualitative wear behavior. The test rig speed was selected using the idea that the “contact time” of the test unit and laboratory rollers should be matched. Here the “contact time” is the time required for a point on the roller to pass through the Hertz contact region. The contact time on the qualification test unit for condition of misaligned rollers was 0.015 second. To match this condition the test rig was operated at 1.6 rad/s (15 rpm) for the majority of laboratory testing. -3-2-10123 (a)height (micrometer) -3-2-10123 (b) profile position (mm)02468 1 0 1 2-3-2-10123 (c)

Document

AMS_2012.pdf

432 Figure 7 – Contact pressures on qualification test unit roller for cases of misaligned and aligned rollers. The pressures shown are at the midpoint of the Hertz contact in the direction of rolling. The lab apparatus can be operated with the misalignment angle at a maximum value of about 1.5 degrees. The influence of the misalignment angle on the contact conditions depends on overall system stiffness and the contact dimensions. The sliding distance in such a condition can be approximately quantified as proportional to the product of the misalignment angle and the length of the contact patch in the rolling direction. Using this concept the misalignment angle of the lab test apparatus times a factor of 1.7 provides the approximate simulated misalignment angle of the qualification test unit. The lab test was conducted for misalignment angles up to 1.5 degrees that would simulate a misalignment angle of about 2.5 degrees for the qualification unit. A summary of the test conditions is provided in Table 1. The tests are listed sequentially in the order of testing. In general one would prefer to have a randomly chosen testing order, but in this work roller availability dictated the testing sequence. In some cases the same roller pair was tested at more than one misalignment angle. In those cases the angles were tested sequencing from smallest to largest. Post-test documentation of the rollers included recording of the mass of each roller, profilometry, and photographs. Some rollers were inspected via scanning electron microscope. Debris was collected. In some cases the debris was swept from the debris tray, collected to a glass vial, and total mass of debris determined. In some cases debris was collected using a square piece of tacky material that could be placed onto the tray and lifted to collect the debris. The collected debris could then be subjected to automated analysis of debris particle counts and sizes. At completion of one test there was no debris readily visible on the brass-hued debris tray, but a swiping of the tray with a cotton-gloved finger revealed debris. The glove fingertip was saved to retain the debris. For later tests a dark grey anodized plate was used for the debris tray. The darker and matte-finished plate allowed to more easily see the small particles. Test Results Roller wear. Photographs of the test rollers were recorded through a view port during test operations at regular intervals. Figure 8 provides a set of photographs documenting the progression of wear during tests with test ID 3, 4 and 8 (test ID per Table 1). These photos show the difference in wear resulting from material condition. The first two rows of Figure 8 can be used to compare and contrast the wear when using annealed 440F rollers (first row) to the wear when using hardened 440C (second row). In these photos the upper roller is the one made from steel alloy and the lower roller is the titanium alloy. The steel rollers took on a relatively uniform appearance of wear while the titanium rollers had patches of irregular appearance. The wear rate can be qualitatively judged by the width of the wear track of the upper roller location along roller (mm)-1.0 -0.5 0.0 0.5 1.0 pressure (MPa) 0100200300400500misaligned aligned

Document

AMS_2012.pdf

433 that, when new, had a circular crown. As this roller loses material the wear track widens. The worn 440C roller appears slightly smoother than the 440F roller suggesting that steel wear debris would in general be smaller for the case of 440C compared to 440F. Recall that the 440F is alloyed to be “free machining” and to thereby produce chips more easily and of greater length during machining. The last row of Figure 8 shows dramatically reduced wear when the rollers are exactly aligned (final row of Figure 8). Even after significantly longer running time the aligned rollers appear almost like the running-in condition (first few hundred cycles) of the misaligned rollers. Still, for the aligned condition one can notice in the very center of the wear track a narrow band that exhibits adhesive wear. Table 1. Summary of Test Conditions To study the wear, rollers were inspected using a scanning electron microscope (SEM). By using x-ray photoelectron spectroscopy the constituents of particular regions of the rollers could be investigated. In all cases, two-way transfer of material was observed. That is, titanium alloy substrate was found on the 440 steel rollers and steel alloy was found adhered to the titanium alloy rollers. An example from inspection of the titanium alloy roller from test ID 3 of Table 1 is provided in Figure 9. An overall view of a region near the center of the wear track is given in Fig 9(a), and also noted are regions where spectroscopy was assessed. The bright regions noted as #1 and #2 in Fig. 9(a) produced the respective spectra of Fig. 9(b-c). Here in the spectra of Fig. 11(b-c) we note the clear presence of iron, chromium, and alloying elements of the 440 steel. Region #2 had a brighter appearance suggesting a more complete coverage of the substrate, and the spectrum for region #2, Fig. 9(c), has a very strong peak correlating to iron. The darker region noted as numeral 3 in Fig. 9(a) produced the spectrum of Fig. 9(d). Here the titanium is still exposed, we observe peaks correlating to titanium and aluminum, and peaks correlating to iron and chromium are not present. The spectrum for the region marked with the numeral 4 in Fig. 9(a) is not shown here but was consistent with that of Fig. 9(d). These trends were true for all of many samples inspected. In all cases the worn regions included regions of the base material still exposed and also contained adhered material from the mating roller. As will be noted, in net, mass was lost for the 440 steel rollers. However, in all cases Ti6Al4V material was found on the tested steel rollers via SEM spectroscopy. test # roller pair test ID material outer surface conditionmaterial outer surface conditionspeed (rpm)misalignment angle (deg) 1 1 1 440C bare Ti6Al4V bare 15 -1.4 2 2 2 440F bare Ti6Al4V bare 15 -1.43 3 3 440F bare Ti6Al4V anodized 15 -1.44 4 4 440F bare Ti6Al4V anodized 15 0.05 5 5 440F passivated Ti6Al4V anodized 9 -1.46 6 6 440F passivated Ti6Al4V anodized 15 -0.4 7 7 7 440F passivated Ti6Al4V anodized 15 -0.9 8 8 8(a) 440F passivated Ti6Al4V anodized 15 0.09 8 8(b) 440F passivated Ti6Al4V anodized 15 -0.1 10 8 8(c) 440F passivated Ti6Al4V anodized 15 -0.211 8 8(d) 440F passivated Ti6Al4V anodized 15 -0.312 8 8(e) 440F passivated Ti6Al4V anodized 15 -0.7 13 9 9 440C bare Ti6Al4V anodized 15 -1.5 14 10 10 440C bare Ti6Al4V anodized 15 -1.415 11 11 440C passivated Ti6Al4V anodized 15 -0.9upper roller lower roller test conditions

Document

AMS_2012.pdf

434 (a) (b) (c) (d) 588 cycles 5,965 cycles 25,158 cycles 77,357 cycles (e) (f) (g) (h) 402cycles 6,140 cycles 24,468 cycles 96,193 cycles (i) (j) (k) (l) 500 cycles 93,854 cycles 250,196 cycles 347,925 cycles Figure 8 - Progression of wear during 3 tests. The cycle count is denoted below each photo. (a-d) 440F at 1.4° misalignment. (e-h) 440C at 1.5° misalignment. (i-l) 440F at 0.0° misalignment.

Document

AMS_2012.pdf

435 (a) (b) (c) (d) Figure 9 – Scanning electron inspections of the wear track on the titanium test roller for test #3. (a) SEM image. (b) Spectrum for region noted as 1 in 9(a). (c) Spectrum for region noted as 2 in Fig. 9(a). (d) Spectrum for region noted as 3 in Fig. 9(a). The roller masses were determined using a scale with digital readout to 0.0001 gram. Masses of the rollers were measured after cleaning and just before installation in the test rig. The roller masses were also measured just after removal from the test chamber. The change in mass at the end of the test quantifies the net transfer of material by adhesive wear. The sum of the mass change for the two rollers provides a calculated value of debris lost from the roller pair. Because of practical test considerations, the test durations were not the same for each test. To provide a method for direct quantitative comparison, the change in mass was divided by the total number of revolutions of the input shaft (that is the total number of contact stress passes) to quantify the wear rate. The mass change data and calculated wear rates are summarized in Table 2. Often wear is modeled as proportional to the sliding distance. In these roller tests, to a first order effect the sliding distance is proportional to the misalignment angle. The rate of mass change was plotted as a function of misalignment angle (Fig. 10). Note that in the net, the steel rollers lost mass while the titanium alloy rollers gained mass. We also note that the 440F material had a higher rate of wear compared to the 440C in all cases. Rabinowicz [8] has noted that the experimental evidence for wear rate being proportional to sliding distance (as often assumed true) is mixed. He states that usually the relationship is not perfectly obeyed but the proportional relationship represents experimental data “reasonably well”. This observation matches the behavior from this investigation in that the misalignment angle (and thereby the sliding distance) correlates to the wear rate, but the linear relationship is not exact. Not only the wear rate but also the wear behavior differed for tests with 440F steel vs. 440C steel. Adhesive wear tends to progress toward an “equilibrium surface” [8] meaning the surface roughness changes with running, and

Document

AMS_2012.pdf

436 the wear process can either roughen or smoothen the surfaces depending on the materials and starting conditions. The equilibrium surfaces for the tests with 440F vs. 440C differed as depicted in Fig. 11. Tests with the 440C material resulted in smoother running surfaces (Fig 11(b)), and the photos suggest that the particle size of wear debris is likely smaller, in general, for the case of 440C. The Ti6Al4V rollers when mated with the 440F steel tended to gain material by adhesive wear in a more uniform fashion, that is, the coverage of the surface was more significant and uniform in appearance. It is possible that the hardened 440C behavior differed from that of the annealed 440F not only because of the change of the elastic limit but also because of an alteration of the metallurgical structure and alteration of the adhesion compatibility with the Ti6Al4V [8]. Table 2 – Summary of Roller Wear Test Results Figure 10 – Change of mass per cycle (shaft revo lution) as a function of misalignment angle. (a) Mass loss of 440 steel rollers. (b) Mass increase of Ti6Al4V rollers, symbols denoting the mating material. test durationmass change; Ti6Al4V mass change; 440 steel calculated mass liberated calculated rate of mass liberated test # test IDrevolutions of input shaftgrams grams milligramsmicrograms per cycle 1 1 26,080 0.0049 -0.0054 0.500 0.019 2 2 60,228 0.0981 -0.1258 27.700 0.460 3 3 77,326 0.1350 -0.2072 72.200 0.934 4 4 347,925 -0.0006 -0.0052 5.800 0.017 5 5 55,411 0.0817 -0.1052 23.500 0.424 6 6 60,850 0.0151 -0.0210 5.900 0.0977 7 124,900 0.1314 -0.1664 35.000 0.280 8 8(a) 93,027 -0.0005 -0.0038 4.300 0.046 9 8(b) 24,001 0.0012 -0.0012 0.000 0.000 10 8(c) 24,000 0.0017 -0.0018 0.100 0.00411 8(d) 24,095 0.0025 -0.0027 0.200 0.00812 8(e) 93,117 0.0464 -0.0501 3.700 0.040 13 9 96,193 0.0682 -0.0698 1.600 0.017 14 10 41,055 0.0200 -0.0210 1.000 0.024 15 11 90,004 0.0672 -0.0688 1.600 0.018 misalignment angle (degree)-1.6 -1.2 -0.8 -0.4 0.0mass change (microgram / cycle) -3.0-2.5-2.0-1.5-1.0-0.50.0 440F 440C (a) misalignment angle (degree)-1.6 -1.2 -0.8 -0.4 0.0mass change (microgram / cycle) 0.00.51.01.52.02.53.0440F 440C (b)

Document

AMS_2012.pdf

437 (a) (b) (c) (d) Figure 11 – Condition of rollers at end of test showing differing behavior of 440F vs. 440C. (a) 440F, test ID 5. (b) 440C, test ID 9. (c) Ti6Al4V test ID 5. (d) Ti6Al4V, test ID 9. The test ID numbers refer to Table 1. Wear debris. During testing of the qualification test unit, some loose debris was formed. The scope of the NESC assessment included consideration of mission risk from the loose debris. The laboratory testing revealed aspects of the loose debris as follows. From Table 2, note that loose debris was generated at rates on the order of 0.5 micrograms per cycle for large roller misalignment. The wear rates were greatly reduced for the case of zero roller misalignment. However, the adhesive wear was not completely eliminated with aligned rollers, and some debris was formed. Figure 12 (a-b) provides images of the largest sized particles collected from the two tests operated with zero roller misalignment. Figure 12 (c) is a typical example of the large number and widely distributed debris that occurred for extensive run time and the largest misalignment angles tested. A portion of the debris pan was not within direct line of sight of the rollers, but that region still contained significant numbers of particles. Summary Laboratory testing of 440 steel rollers in contact with Ti6Al4V in vacuum was completed to study the wear behavior and to quantify wear rates. The tests also assessed the influence of material condition and roller misalignment. The wear rate was found to be roughly proportional to the roller misalignment angle. The rate of loose particle mass created was on the order of 0.5 micrograms per cycle. The adhesive wear was

Document

AMS_2012.pdf

438 a two-way phenomenon with titanium alloy material found adhered to the steel and vice versa. Loose particles with linear dimensions on the order of 200 micrometer were created even for the condition of zero roller misalignment. For large misalignment angles, large numbers of loose particles could be created, and debris was found in locations that were not within a direct line of sight. The wear behavior differed for the case of 440F and 440C steel rollers. With the 440C rollers, the adhesive wear rate was reduced, and the final surface textures of the 440C rollers were smoother than the 440F rollers. Figure 12 - Examples of loose particle debris. (a) Test ID 4 with zero roller misalignment. (b) Test ID 8(a) with zero roller misalignment. (c) Test ID 5, rollers misalignment -1.4 degrees. The test ID numbers refer to Table 1. Acknowledgements This research was supported by the NASA Engineering Safety Center. The experimental work was done with assistance from Mr. Richard Manco, Sierra Lobo Inc. Dr. Iqbal Shareef’s contributions were supported by the NASA GRC Summer Faculty Fellowship Program. References 1. McClendon, M., “NIRSpec MSS Magnet Actuator Life Test Unit Wear Particle Evaluation”, obtained by Krantz, T., May 16, 2011 2. Authors unstated, “Micro Shutter Subsystem (MSS) Qualification Unit Test Report”, JWST-RPT013819, Rev. A, June 2010. 3. Pepper, S., “Research Note-Characterization of the Test Environment of JWST Roller Wear Evaluation at NASA-GRC”, Aug. 1, 2011. 4. Johnson, K.L., Contact Mechanics , Cambridge University Press, 1985. 5. Kalker, J.J., “Rolling contact phenomena: linear elasticity”, Rolling Contact Phenomena CISM Courses and Lectures, Issue 411, Spinger-Verlag, 2000. 6. McGinness, H., “Lateral forces induced by a misaligned roller”, DSN Progress Report 42-45, March and April 1978, Jet Propulsion Laboratory, Pasadena, CA, 1978. 7. Krantz, T., DellaCorte, C., Dube, M., “Experimental Investigation of Forces Produced by Misaligned Steel Rollers”, proceedings of the 40 th Aerospace Mechanisms Symposium, NASA/CP-2010-216272, also NASA/TM-2010-216741, 2010. 8. Rabinowicz, E., Friction and Wear of Materials , 2nd edition, Wiley-Interscience, 1995. 9. Vijayakar, S., “A combined surface integral and finite element solution for a three-dimensional contact problem”, International J. of Numerical Methods for Engineering, vol. 31, 1991. 10. Vijayakar, S. “Multi-body Dynamic Contact Analysis Tool for Transmission Design – SBIR Phase II Final Report”, Army Research Laboratory ARL-CR-487, 2003. (a) (b) (c)

Document

AMS_2012.pdf

439 Angular Runout Test Setup for High-Precision Ball Bearings Scott W. Miller*, Jonathan P. Wood* and Stuart Loewenthal* Abstract Ball bearing runout error is generally a limiting factor for the precision of pointing systems and astronomical instruments. A test setup was developed to optically measure the angular runout of a series of off-the-shelf duplex angular contact bearing pairs each having the same cross-section, but different ABEC ratings and manufacturers. A total of 6 bearing sets ranging from ABEC 3 to ABEC 7 were tested. The repeatable, synchronous runout generally improved with higher ABEC quality bearings as expected, although occasionally significant variation were observed with bearings of the same ABEC ratings. Less significant differences were observed for non-repeatable runout as a function of ABEC quality rating. A major finding is that strictly relying strictly on ABEC rating may be insufficient for applications where bearing runout plays an important role. Introduction When developing precision pointing systems and astronomical instruments supported by ball bearings, angular runout is often a key factor affecting pointing accuracy, stability, and overall performance. Line-of-sight pointing accuracy on precision pointing platforms are adversely affected by bearing runout since irregular encoder code disk motion can seriously degrade readout accuracy. Pointing mirrors are also vulnerable to bearing wobble particularly non-repeatable angular runout which cannot be easily calibrated out. Estimating angular runout based on bearing catalog individual race radial worst case tolerance will generally lead to a significant over prediction of the angular runout of the bearing assembly. In the case of a duplex bearing pairs, matching up the bearing high spots as marked on precision bearing races will essentially cancel most of the angular runout of the bearing. This will happen at the expense of enhancing radial runout motion. However, radial centerline motion is less of an issue for most optical systems that are more affected by angular pointing errors. The American Bearing Manufacturer’s Association (ABMA) standards do not provide any requirements for the runout of an assembled bearing pair, but instead specify maximum raceway to mounting feature runouts for each bearing race. This will generally not predict the runout of the assembled bearing pair, thus data showing assembled bearing runouts for different Annular Bearing Engineering Committee (ABEC) ratings is not readily available to the designer. Machine tool spindle bearing runout and disk drive spindle bearing errors are well known limiting factors for their capabilities. Standards for determining the quality of precision spindles have been available for some time. [1]. Eric Marsh and his colleagues at Pennsylvania State University have been prolific contributors to this field of precision metrology [2] [5]. Disk drive spindle runout is traditionally characterized by using two or more non-contacting proximity probes referencing the hub of the spindle, e.g. see [3]. However the runout includes spindle hub machining eccentricities and surface finish errors that must be subtracted out in order to arrive at the bearing contribution. Precision pins or spheres can be fixed at the end of the spindle to avoid minimizing machining geometric errors, but this is not always possible with many spindle designs and the eccentricity of the target to the bearing spin axis is still present. In the approach reported herein the bearing spin axis angular errors are measured optically and the geometric accuracy of the hub is not a concern. The test setup shown in Figure 1 optically measures the * Lockheed Martin Space Systems Company, Palo Alto, CA Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

440 angular axis of rotation errors of an assembled, preloaded, duplex bearing pair. The test setup was originally developed to determine whether ABEC 3 bearings would have sufficiently low angular runout to be used in a prototype precision pointing gimbal. This paper will present measured angular runout data for preloaded ball bearing pairs of a unique diameter and cross section, but three separate ABEC ratings. The goal of this testing is to provide a correlation between assembled bearing runout and ABEC rating. This data will be presented in several forms, including repeatable or synchronous runout as well as non-repeatable or asynchronous runout. [1-3] Of particular interest to the authors are the repeatability of the angular runout errors and frequency content in terms of mechanical harmonics. The data presented will be useful for estimating angular runout during the design process and the test setup presented allows for bearing-level screening and selection based on minimizing angular runout in bearing applications. Figure 1. Test Setup Overview Test Fixture Overview Each bearing tested was of the same diameter and cross section, so that a single fixture could be used for all testing. The test fixture consists of an aluminum housing and shaft, an integral, direct drive brushless DC motor, and an encoder for motor control purposes. The outer housing is mounted to the bench with the rotation axis parallel to gravity. Figure 2 shows both a photograph and schematic of the test fixture. The features of the housing and shaft that contact the bearing races were diamond turned in order to obtain appropriately precision bearing fits. The diameter tolerancing method is ±5 µm (0.0002 in) from the basic diameter, which permits a line-to-line contact under maximum material conditions. A benign thermal soak allows installation or removal of the bearings without applying excessive force. A mirror with a micrometer adjustable tip-tilt stage is mounted to the shaft. Bearing angular runout is measured using an electronic autocollimator.

Document

AMS_2012.pdf

441 Figure 2. Test Setup Description The test setup procedure begins with aligning the mirror roughly perpendicular to the bearing axis (see Figure 3). The autocollimator launches a reference beam, which reflects off the mirror and returns to a detector. When perfectly aligned, the reflected beam follows the same path as the reference beam upon return and has no angular difference between beams. However, introducing an angular displacement between the bearing axis and mirror normal or shaft axis will cause the reflected beam to sweep a conical path during shaft rotation, which will project a circular profile on the autocollimator detector. The nominally aligned case and an angular perturbation case are depicted in Figure 3. Recognize that if the bearing set were geometrically perfect then the trace recorded by the autocollimator would remain circular. The deviation from this perfect circle is therefore a direct measure of the bearing angular error. Figure 3. Measurement Method Overview

Document

AMS_2012.pdf

442 The degree of mirror tilt is somewhat arbitrary and does not affect the accuracy of the readings. This degree of mirror tilt is measured by the autocollimator and provides the scale factor for the bearing measurements. Also it was determined that if mirror tilt was too small, bearing wobble would cause crossover patterns of the reflected beam, such as a “figure eight” pattern. Therefore, the mirror to bearing axis alignment was intentionally perturbed for all testing cases to trace a nominal 500-microradian diameter path, as a seen by the autocollimator detector. Note that the effect of this approach was to introduce a calibrated, constant tilt of the mirror normal axis relative to the bearing axis. Each test consisted of ten clockwise revolutions, followed by ten counter-clockwise revolutions. All tests were performed using a constant, low-speed rotation rate of 4.28 RPM (14 seconds per revolution), taking approximately five minutes to complete the full 20 revolutions. Note that the particular motor control speed was selected in order to determine slow-speed, or kinematical, bearing behavior, neglecting the dynamic effects.[4] The autocollimator azimuth and elevation signals were sampled by the data acquisition system at 250 Hz. Bearing Summary A total of six unique bearings were tested from three separate vendors. Although the same diameter, cross section and duplex arrangement were used throughout, each configuration had a unique retainer. ABEC ratings include 3, 5 & 7. Also, note that SN6 had a single outer race for the duplex set. The complete bearing summary table is shown in Table 1. Table 1. Test Bearing Summary Serial Number ABEC OD Cross Section Vendor Ball Dia Ball Retainer Type (in) (mm) (in) (mm) (in) (mm) Qty 1 3 3.0 76.2 0.25 6.35 A 0.125 3.175 52 Nylon DB 2 3 3.0 76.2 0.25 6.35 A 0.125 3.175 52 Nylon DB 3 3 3.0 76.2 0.25 6.35 A 0.125 3.175 52 Nylon DB 4 7 3.0 76.2 0.25 6.35 B 0.125 3.175 52 Brass DB 5 7 3.0 76.2 0.25 6.35 B 0.125 3.175 52 Brass DB 6 5 3.0 76.2 0.25 6.35 C 0.125 3.175 60 Teflon Toroid DB (1-Piece Outer Race) Test Results The type of results presented can be described as kinematic or quasi-static since they are done at a relatively low rotation rate. This is to avoid structural resonances in the system so that the measurements can be directly related to bearing behavior. Measured bearing wobble was processed using several methods. The primary method for evaluating synchronous and asynchronous runout was obtained from the elevation versus azimuth plots. A cursory, qualitative analysis compares the acquired test data trace to the perfect circle at 500-microradian diameter, which we will refer to as the “reference diameter”. Several performance aspects should be noted. First, the deviation from the average test data and the reference diameter is defined as “Synchronous Error”. It is synchronous in that it repeats itself every revolution. Second is the portion of the error that does not repeat itself, forming a radial hash band about the average test data. This is defined as the “Asynchronous Error”. These two definitions are illustrated in Figure 4. Finally, hysteresis is observed during cycle reversal, which is most evident in SN5 and SN6 traces. All six traces are shown in Figures 5 to 10, labeled sequentially from SN1 through SN6.

Document

AMS_2012.pdf

443 Figure 4. Error Nomenclature Figure 5. Elevation vs. Azimuth Test Data Figure 6. Elevation vs. Azimuth Test Data SN 1 (ABEC 3) SN 2 (ABEC 3) Figure 7. Elevation vs. Azimuth Test Data Figure 8. Elevation vs. Azimuth Test Data

Document

AMS_2012.pdf

444 SN 3 (ABEC 3) SN 4 (ABEC 7) Figure 9. Elevation vs. Azimuth Test Data Figure 10. Elevation vs. Azimuth Test Data SN 5 (ABEC 7) SN 6 (ABEC 5, Super-Duplex) Data Processing It was conservatively assumed that the radial runouts of the two bearings in the duplex pair act in opposite directions, using worst-case runout data from the ABMA standard. These predicted angular runouts are shown in Table 2. A summary of the actual runouts measured from each bearing pair is shown in Table 3. Each bearing pair significantly out-performed the expected angular runout based on this calculation method, most likely due to the fact that the bearings were installed in the test fixture with the high spot markings on each race aligned. The data presented shows that even the ABEC 3 bearings tested offer a very low level of asynchronous angular runout. It is important to emphasize that the angular runout estimation method presented in table is by no means an established approach. In fact, the data from this testing show that this estimation approach provides excessive conservatism and that the best way to determine angular runout is to perform testing on the bearings to be used. The angular runout performance of a given bearing set will depend on workmanship factors, including machining tolerances and ball size matching, but the results presented are also indicative of the order of magnitude one can expect with different ABEC rated bearings of the size range presented. Table 2. Predicted Angular Runouts Using Different Methods Angular Runout Summation Method Angular Runout RSS Method Pitch Diameter Bearing Width Radial Runout Inner Radial Runout Outer (in) (mm) (in) (mm) (in) (µm) (in) (µm) (microradian) (microradian) ABEC 3 3.25 82.6 0.25 6.35 0.00040 10 0.0006 15 4000 2884 ABEC 5 3.25 82.6 0.25 6.35 0.00020 5.1 0.0004 10 2400 1789 ABEC 7 3.25 82.6 0.25 6.35 0.00015 3.8 0.0002 5.1 1400 1000

Document

AMS_2012.pdf

445 Table 3. Measured Angular Runouts Synchronous Error AmplitudeAsynchronous Error AmplitudeSynchronous Error RMSAsynchronous Error RMS (microradian) (microradian) (microradian) (microradian) SN 1 3 116 13.0 52 3.0 SN 2 3 74 11.4 38 2.6 SN 3 3 90 8.4 38 2.0 SN 4 7 30 11.0 22 2.8 SN 5 7 88 12.6 36 3.0 SN 6 5 28 12.0 18 2.6ABEC As shown in Table 3 & Figure 11, one of the ABEC 7 bearings exhibited unusually high angular runout. This ABEC 7 bearing actually had higher angular runout than some of the ABEC 3 bearings. Of course the angular runout measured for this bearing does not necessarily mean that the bearing would be rejectable per the ABMA raceway groove runout requirements. It does however, demonstrate the variation that can be seen from bearing to bearing, even at higher ABEC ratings. Interestingly, the one ABEC 5 sample outperformed both ABEC 7 samples in terms of angular runout. However, the SN 6 ABEC 5 bearing had a one piece outer race which assured that the radial runout high point of the bearing set was perfectly matched eliminating any deleterious scissoring action. The inconsistency of angular runout performance serves to illustrate the importance of “cherry picking” of bearings for high-precision applications. If “cherry picking” is performed on a sample of bearings, it is important to mark the relative clocking between both adjacent inner and adjacent outer races respectively within a given duplex pair, as the race-to-race clocking can influence the angular runout performance. The cherry-picking approach may also be used to screen for linear runouts and drag torque properties. In this case, only one relative clocking between corresponding races was tested. However, it is conceivable that relative clocking could play an influential role in the angular runout performance, such that re-clocking of the inner or outer races relative to each other could increase or decrease the measured angular runout, depending on the relative orientation of the high spots between the two races. Figure 11. Maximum Synchronous Error Data 020406080100120140 SN 1 (ABEC 3)SN 2 (ABEC 3)SN 3 (ABEC 3)SN 4 (ABEC 7)SN 5 (ABEC 7)SN 6 (ABEC 5)Error Amplitude (microradian)

Document

AMS_2012.pdf

446 As shown in Figure 12, the asynchronous error was very low, and was nearly the same magnitude for each bearing pair, regardless of ABEC rating. The asynchronous error is most likely a manifestation of the ball accuracy (sphericity) and ball-to-ball diameter matching. The later in conjunction with raceway geometry errors dictates preload variations as the balls orbit the bearing. This is one example where the performance of the bearing pair is workmanship dependant and could vary greatly from one manufacturer to another, or even lot to lot. The repeatability of the test setup is another factor. Figure 19 shows the Asynchronous error versus angle plot for SN3. Figure 12. Maximum Asynchronous Error Data Figure 13. Synchronous Error Versus Angle Figure 14. Synchronous Error Versus Angle SN 1 (ABEC 3) SN 2 (ABEC 3) 0.02.04.06.08.010.012.014.0 SN 1 (ABEC 3)SN 2 (ABEC 3)SN 3 (ABEC 3)SN 4 (ABEC 7)SN 5 (ABEC 7)SN 6 (ABEC 5)Error Amplitude (microradian)

Document

AMS_2012.pdf

447 Figure 15. Synchronous Error Versus Angle Figure 16. Synchronous Error Versus Angle SN 3 (ABEC 3) SN 4 (ABEC 7) Figure 17. Synchronous Error Versus Angle Figure 18. Synchronous Error Versus Angle SN 5 (ABEC 7) SN 6 (ABEC 5) Figure 19. Asynchronous Error Versus Angle

Document

AMS_2012.pdf

448 SN 1 (ABEC 3) A Fast Fourier Transform (FFT) analysis of the data reveals dominate spikes at 1, 2, 3, and 4 cycles per revolution. The largest and most consistent spikes occur at 3 cycles per revolution. The ball pass frequency and ball spin frequency for the bearings tested are all above 10 cycles per revolution. The large spikes at 2, 3, and 4 cycles per revolution and absence of any significant spikes above 10 cycles per revolution indicate that the angular runout of the bearings is dominated by race machining tolerances rather than ball out-of-roundness. Typically, when bearing raceways are machined, a tri-lobing of the race is the largest source of machining error. This is consistent with the dominance of the 3 cycles per revolution FFT spikes. Figure 20. Magnitude vs. Cycles Per Rev Figure 21. Magnitude vs. Cycles Per Rev SN 1 (ABEC 3) SN 2 (ABEC 3) Figure 22. Magnitude vs. Cycles Per Rev Figure 23. Magnitude vs. Cycles Per Rev SN 3 (ABEC 3) SN 4 (ABEC 7)

Document

AMS_2012.pdf

449 Figure 24. Magnitude vs. Cycles Per Rev Figure 25. Magnitude vs. Cycles Per Rev SN 5 (ABEC 7) SN 6 (ABEC 5) Lessons Learned An economical bearing fixture was initially fabricated using standard lathe tolerances, in order to determine if exotic feature grinding was necessary. After processing the initial test data, it was determined that the fixture diametral clearance to bearing race was excessive, which was made evident by a cycle-to-cycle trace precession and large asynchronous error. The error trace precession reversed directions, corresponding to the bearing rotation direction. It was concluded that the economical fixture did not provide adequate relative concentricity control between the duplex pair races. The second bearing fixture, which is the one presented in this paper, had decreased the asynchronous error substantially over the economical fixture and proved that bearing mount tolerances have a large influence over the shaft runout performance to levels similar to the bearings themselves. It can be concluded that it is imperative that bearing mount features be precision machined and ground even when using ABEC 3 bearings and especially for slim section angular contact pairs. Test results showing relatively low asynchronous error of all three ABEC 3 bearings came as a pleasant surprise. Testing additional quantities beyond those presented is, of course, desirable for statistical data processing. From a design standpoint, the data gathered from this testing may allow much more economical ABEC 3 bearings to be used, while still meeting the same design requirements that based on the predicted angular runout, would have required ABEC 7 bearings. This serves to illustrate the development testing benefits, which lead to the relaxation of requirements and ultimately cost savings. Although designing for the worst-case predicted angular runout would have resulted in a conservative design, it would have unnecessarily driven the design to excessively tight tolerances and excessively expensive bearings. Conclusions Three ABEC qualities were measured for synchronous and asynchronous runouts. Although bearings with a higher ABEC rating did correlate with lower measured synchronous runout, it was also observed that asynchronous runout was relatively consistent, independent of the ABEC rating. Also bearings had significantly lower angular runout than estimated using ABMA individual race tolerances. This confirms that such analytical estimates are not a viable substitute for actual runout measurements of the assembled bearing set. Finally, in the case of ABEC 7 bearings, the behavior of two bearings acquired at the same time, from the same vendor and to the same part number did not show consistent performance. For this reason, it is recommended that for critical applications, all bearings be tested prior to assembly and that appropriate spares are procured, if the application demands superior performance consistency.

Document

AMS_2012.pdf

450 Acknowledgements The authors would like to thank their Lockheed Martin colleagues, including Kyle Brookes for his role in the design of the bearing test fixture and for the technical review provided by Dr. Jean-Noel Aubrun and Dr. Ken Lorell. References 1. ISO-230-7, “Geometric accuracy of axes of rotation” and ASME B89.3.4-1985, “Axes of Rotation, Methods for Specifying and Testing.” 2. Marsh Eric and Grejda, Robert “Experiences with the Master Axis Method for Measuring Spindle Error Motions, Precision Engineering 24 (2000) 50–57 3. "Bearing Runout Measurements" Agilent Technologies, Applications Note 243-7. 4. Harris, Tedric A. and Kotzalas, Michael N. "Advanced Concepts of Bearing Technology" CRC Press, 5th edition (2007), p182 . 5. Weiss, Jeffrey R. "Rolling Element Bearing Metrology" Penn State University Thesis, May 2005 .

Document

AMS_2012.pdf

451 LightSail-1 Solar Sail Design and Qualification Chris Biddy* and Tomas Svitek* Abstract LightSail-1, a project of The Planetary Society, is a Solar Sail Demonstration mission built on the CubeSat platform. Stellar Exploration Inc. designed, built and fully qualified the Solar Sail module for LightSail-1 which includes a boom deployer mechanism that stows a total boom length of 16 meters (4 x 4 meter booms) which are used to deploy a 32-m 2 sail membrane in a 2U package. This design utilizes the rigid TRAC boom developed by AFRL (Air Force Research Laboratory) for deploying and tensioning the membrane that makes up the sail. In order to maximize the size of the solar sail, the boom deployer took on a unique shape to maximize packaging efficiency and achieve an 80:1 deployed to pre-deployed ratio. This paper will discuss the design challenges, unique design features as well system verification for LightSail-1. Introduction LightSail-1 is a project of The Planetary Society and is privately funded by members of the organization (Figure 1). The Planetary Society has been a proponent of solar sails for many years and the LightSail program is dedicated to advancing solar sail technology. The main objective of LightSail-1 is to demonstrate the viability of solar sails by demonstrating a positive change in orbit energy, the ability to manage the orbit energy, and to control the spacecraft under solar sail power. These objectives will be achieved by developing and demonstrating key technologies such as sail deployment and sail material management during flight as well as the control of the spacecraft's attitude. Figure 1. Artist rendering of LightSail-1 (The Planetary Society). LightSail-1 follows work done for The Planetary Society on another program called COSMOS-1 which was also a solar sail demonstration mission. Unfortunately COSMOS-1 failed to reach orbit due to launch vehicle failure. COSMOS-1 was ~100 kg and had a total sail area of ~600 m 2. This gives COSMOS-1 a solar sail characteristic acceleration (a common metric for evaluating solar sail performance) of 0.047 mm/s2. * Stellar Exploration, San Luis Obispo, CA Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

452 LightSail-1 Program Requirements The main requirements and corresponding selected implementation for LightSail-1 are summarized in Table 1. Many of these requirements were derived from COSMOS-1. Table 1. LightSail-1 requirements. Requirement Selected Implementation Low Cost CubeSat Platform Greater Performance than COSMOS-1 3U CubeSat (maximum mass = 5 kg) requires 32 m2 sail Demonstrate Orbit Energy Change 2 On-board 3 axis accelerometers and optical tracking Demonstrate Thrust Control Requires ADCS with 90 ˚ Slew Maneuver Image the Sail during/after deployment 2 On-board Aerospace Corporation Cameras Based on the customer requirements and associated trades it was decided to build LightSail-1 on a 3U CubeSat platform with a 32-m2 square solar sail achieving a solar sail characteristic acceleration of 0.050 mm/s2. LightSail-1 Configuration The overall spacecraft was divided into three sections that include the avionics section, sail module, and payload section. The avionics section is ~1U and houses the avionics board, the sensor interface board, a transceiver, eight Lithium Polymer batteries and battery control boards, three torque rods, a momentum bias wheel, three single axis MEMS gyros, and a three-axis MEMS accelerometer. The sail module is ~1.5U and includes the sail storage cavity, and boom deployer. The payload section is ~0.5U and contains the deployer spindle drive motor and gear train, a three-axis MEMS accelerometer, a deployable monopole antenna, a deployable panel burn wire mechanism, and a storage compartment for two Aerospace Corporation built cameras mounted to the ends of the deployable panels. Design Challenges The following is a list of design challenges to the development of the sail module for LightSail-1.  How to package the sail and booms in the allowable volume  How to manage boom strain energy while stowed and during deployment  How to control the sail deployment  How to constrain sail material and booms prior to deployment (including during launch)  How to manage sail material during deployment It was determined early in the design phase that controlled deployment of the sail would be achieved by driving the deployer spindle that the booms are attached to with a brushless DC motor. Because the booms have a large stored strain energy when fully stowed, a method of constraining the booms while the motor was not energized was required to avoid auto-deployment of the booms. It was decided that this could be achieved by implementing a worm drive that could not be back driven in the gear-train. The worm drive provided the required gear reduction from the motor to the spindle as well as eliminated the need for an additional mechanism to constrain the booms during launch. In order to keep the sail material contained in the sail storage structure the deployable solar cell panels closed over the sail storage module constraining the sail material in place. Detailed Design The boom chosen for LightSail-1 is the TRAC boom developed by AFRL because of its packaging efficiency and specific stiffness. These booms were used in NanoSail-D and successfully flown on NanoSail-D2 last year. The TRAC boom consists of two Elgiloy metal strips each formed into a c-shaped curve and laser welded together back-to-back forming an inverted v-shape. The TRAC boom is shown in Figure 2.

Document

AMS_2012.pdf

453 Figure 2. TRAC boom. The TRAC boom collapses and is then rolled around a drum so it can be stored in an annular space. The TRAC boom has a self-deploying feature due to the stored strain energy in the boom while collapsed and rolled around the drum. This collapsible and rollable feature of the TRAC boom allows it to be stored in a very compact fashion. Due to the constraint imposed on the individual metal strips by the weld bead, the minimum bend radius is driven by the maximum strain of the material. The sail storage section consists of a section with a wedge shaped cavity cut on each of the four faces. Also included is a hole passed through the center of the section for the routing of the wire harness to connect the avionics board in the top of the spacecraft to the sensors and deployer motor housed in the payload section. This configuration was driven somewhat by the requirement to have P-POD interface rails at least 75% of the overall length of the spacecraft (Figure 3). Figure 3. LightSail-1 layout. The sail folding concept consisted of z-folding the sail in two directions while varying the fold widths from the center to the outside tips to take a wedge shaped cross-section matching the sail storage cavity (Figure 4).

Document

AMS_2012.pdf

454 Figure 4. Sail Folding Procedure. Prototype sails were built to empirically determine the magnitude of the fold height and width tolerances that were required, (and reasonable from a manufacturing standpoint) for the proper fitment into the sail storage cavity. Figure 5. Sail test-fit in engineering structure. The design of the sail was such that it had a slight interference fit in the sail storage cavity for sail management purposes during deployment. The idea being that each fold would be pulled out of the sail storage cavity one by one as the booms were deployed while the rest of the folded sail material was held in place. This would insure that the sail material would not billow from the cavity and potentially tangle around a deploying boom or get caught in the deployer and fail to deploy properly (Figure 6). Figure 6. Sail behavior during deployment.

Document

AMS_2012.pdf

455 The sail-to-boom connection is made using metal grommets in the sails and booms with split rings and extension springs in series as shown in Figure 7. The extension springs are required to maintain an appropriate tension in the sail during thermal cycling. Figure 7. Sail to Boom connection feature. Deployer Design The deployer has four main functions: (1) Provide an attachment point to the boom to react all deployment loads (2) Protect the booms from yield due to strain (3) Store the 4m TRAC booms within the CubeSat allowable cross-section, (4) Provide smooth unrestrained deployment of the TRAC booms and Sail material. The deployer is made up of a simply supported spindle mounted between two plates that the booms are attached to. The booms are clamped to the spindle and held with two stainless steel #4-40 bolts. This feature provides the attachment point for the booms and takes all the boom reaction loads. The spindle and clamp are machined with a flare at the bottom so as to not completely pinch the booms while mounted to the spindle. This increases the area moment of inertia of the boom over the pinched configuration providing a better boom root condition. The spindle flanges constrain and protect the booms from contacting any non-moving surfaces during deployment. The spindle with flanges and clamps are shown in Figure 8. Figure 8. Spindle with boom clamps and flanges.

Document

AMS_2012.pdf

456 The rest of the deployer assembly contains four Tensioner Assemblies that are arranged between the two plates 90˚ from each other as well as multiple Delrin rollers mounted near the four corners, and four flexure clamp assemblies mounted adjacent to the Delrin Boom Exit Guides. These components are shown in Figure 9. Figure 9. The deployer assembly with booms not shown. As mentioned above, the booms require the application of a normal force to first collapse the boom into its pinched configuration and then to force the boom to roll around the drum. This is accomplished with the tensioner which uses the reaction force of a deflected cantilever flat spring against the flexure contact pin. The force on the flexure contact pin produces a moment on the tensioner body which pivots around two shoulder screws with Delrin bushings. This moment is reacted at the Boom Roller causing a normal force against the boom wrap. The flexure spring passes though the space between the shoulder screws and contacts the contact pin mounted in the tensioner. Standoff rollers are used to isolate the boom wrap from the metal tensioner body and shoulder screw heads. The tensioner assembly is shown in Figure 10. Figure 10. Tensioner Assembly. Figure 11 shows a top view of the deployer in the fully deployed configuration with the top plate and spindle flange transparent, and the booms not shown for clarity. This fully deployed configuration refers to the state when the booms are just beginning to be rolled around the spindle to be stowed, or at the end of deployment.

Document

AMS_2012.pdf

457 Figure 11. Tensioner orientation at full boom deployment with the flexure springs in the relaxed state (red arrows indicate the path of the booms). As the booms are wound around the spindle, the increasing outside diameter of the boom wrap pushes the contact roller on the tensioner outward increasing the deflection of the flexure spring. As the boom wrap thickness reaches its maximum outside diameter and the flexure spring reaches its maximum deflection, the tensioner assembly is stowed within the CubeSat allowable cross-section as shown in Figure 12. Figure 12. Tensioner orientation for stowed booms with flexure spring at maximum deflection (red arrows indicate the path of the booms). The “rocker-arm” configuration of the tensioner allows for very compact packaging by taking up area around the boom wrap and not protruding past the allowable CubeSat cross-sectional area while in the stowed configuration. The deployer and elegant design of the TRAC boom resulted in a deployed to pre-deployed ratio of 80:1 since the packaged cross-section is 0.1 m x 0.1 m and the deployed boom length, tip to tip, is 8 m.

Document

AMS_2012.pdf

458 Figure 13. Deployer with top plate and flanges transparent to show booms in fully deployed configuration. Development testing of the boom deployer assembly consisted of measuring the drive motor current required to deploy the booms while varying parameters within the deployer including the flexure spring rate, and coefficient of friction between adjacent boom wraps. This testing revealed a strong correlation between the required motor drive current required to deploy the booms and (1) the coefficient of friction between the adjacent boom wraps, and (2) the flexure spring thickness. It was discovered that a large coefficient of friction was desired between adjacent boom wraps because a higher axial force could be transferred from the drive motor through the boom wrap to the deployed length of the boom. Since the sails had a slight interference fit with the sail storage cavity a small axial force was imparted on the booms in order to pull the sail material out of the cavity and unfold it. Initially this axial force would cause ballooning of the boom wrap inside of the deployer. This ballooning was controlled with a higher spring rate on the flexure spring which exerted a larger normal force on the boom wrap keeping it contained. This was critical for achieving reliable deployment of the sail. Figure 14. Flight deployer and payload as sembly in fully stowed configuration.

Document

AMS_2012.pdf

459 Figure 14 shows the deployer and payload section in the flight configuration with the booms fully stowed. Figure 15 shows the spacecraft fully assembled (with engineering deployable solar cell panels). Figure 15. LightSail-1 fully assembled. System Validation System validation consisted of full-scale sail deployment tests before and after random vibration testing, and TVAC testing while monitoring the drive motor current as the performance metric. Cold full-scale sail deployment tests while monitoring motor drive current were conducted as well to verify cold deployment performance. In order to conduct sail deployment tests a method of off-loading was required to simulate a zero-g environment. This was accomplished by building a deployment table to support the weight of the booms and sail during testing. The table is shown in Figure 16 and features removable panels in order to access the spacecraft while the sail and/or booms are deployed. Figure 16. Sail deployment table.

Document

AMS_2012.pdf

460 Random Vibration Testing Random Vibration tests were conducted to validate the structural design of the spacecraft and also to validate the deployer mechanism design. The random vibration test would also be used to validate the worm drive locking feature. Figure 17 shows LightSail-1 on the vibration table at Cal Poly. Figure 17. LightSail-1 during random vibration testing at Cal Poly. The structure and worm drive locking feature performed well. The worm drive locking feature was verified by observing that the spindle orientation had not changed after random vibration testing by lining up reference marks on the spindle and deployer top plate before random vibration testing. If the reference marks lined up after random vibration testing it confirmed that the spindle was held fixed by the worm drive, which is what occurred. Figure 18. Spindle reference marks before (left) and after (right) random vibration testing. While the structure performed well and the worm drive locking feature was validated, it was observed that the booms were pulled inside the deployer ~10 mm during the random vibration test. It was determined that the boom wrap was allowed to balloon slightly due to some compliance in the flexure spring/tensioner assembly. This was solved by packaging the booms in such a way that the tensioner was completely bottomed out against a hard stop while the boom wrap was pulled tight. Additionally, a pin was added to the tip of each boom to act as a hard stop to not allow the boom to retract into the deployer.

Document

AMS_2012.pdf

461 Figure 19. Boom hard stop. Following the implementation of the above mentioned solutions the random vibration test was repeated to validate the design with full scale sail deployment tests conducted before and after random vibration testing. The deployment tests were successful with no significant difference in the motor drive current between the two deployment tests. Figure 20. Drive motor current plot for Pre and Post Random Vibration deployment tests.

Document

AMS_2012.pdf

462 Figure 21. Sail deployment test. Cold Deployment Testing Cold deployment testing consisted of cooling only the deployer and booms to a prescribed temperature and then deploying only the booms while monitoring the drive motor current and comparing that with the current required to deploy the booms at room temperature. It was decided to focus on the deployer mechanism and booms, without the sail, because the sail behavior was not expected to change significantly over the temperature range being tested and because of the difficulty controlling moisture in a laboratory environment (without a vacuum chamber). The moisture in the system needed to be carefully controlled since the cold testing was conducted below freezing and the introduction of ice in the system could limit the performance of the deployer. The cold testing was performed by flowing cold nitrogen vapor into a container housing the deployer with small cutouts for the booms to pass through until the deployer assembly reached the appropriate temperature after which the booms were deployed. The drive motor current was monitored during cold deployment testing for comparison with room temperature deployment data. The deployer was qualified down to -6 ˚C where it was well within the allowable motor current margin. Many full-scale deployment tests have been conducted on the engineering model hardware for development. Full-scale "acceptance" deployment tests will be conducted on the flight hardware including before and after random vibe and TVAC. As of this writing the flight unit TVAC test has not been completed. These flight hardware tests are held to a minimum because of degradation of the sail material during re-folding of the sail after deployment testing. However up to 6 deployment tests on a set of sails is reasonable without significant degradation.

Document

AMS_2012.pdf

463 Conclusion The result of work done on the LightSail-1 program is a fully qualified 32-m2 solar sail packaged in a ~1.5U volume with a mass <3 kg (4.6 kg total spacecraft mass). The boom deployer assembly has a deployed to post deployed ratio of 80:1. It is planned to launch LightSail-1 by the end of 2012. Figure 22. LightSail-1 fully deployed in the lab at Stellar Exploration. Acknowledgments The authors would like to thank The Planetary Society and its members for sponsoring LightSail-1, Louis Friedman and Jim Cantrell for hiring Stellar Exploration as the sail module developer and spacecraft integrator, and Bill Nye (The Planetary Society's Executive Director) for his continued support of the LightSail program. The authors would also like to thank Jeremy Banik of AFRL for his guidance and support during the development of LightSail-1. References: Anderson, J.L. (11/29/11). NASA. NASA's NanoSail-D 'Sails' Home -- Mission Complete. Retrieved January 3, 2012, from http://www.nasa.gov/mission_pages/smallsats/11-148.html McInnes, C. R. (1999). Solar Sailing Technology Dynamics and Mission Applications. Chichester, U.K: Praxis Publishing. Montgomery, Edward E., Adams, Charles L. (2008, April). NanoSail-D. CubeSat Developers Workshop, San Luis Obispo, CA. The Planetary Society. Projects: LightSail-Solar Sailing. Retrieved January 3, 2012, from http://www.planetary.org/programs/proj ects/solar_sailing/lightsail1.html

Document

AMS_2012.pdf

465 A Novel Release Mechanism Employing the Principle of Differential Coefficients of Thermal Expansion Clint Apland*, David Persons*, David Weir*, and Michael Marley* Abstract APL has developed a novel miniaturized release mechanism that can be used in CubeSats, NanoSats and miniature space-borne science instruments. This miniature device is inexpensive, reusable, power-efficient, and doesn’t consume any flight parts. The principle of operation for the release mechanism is to use two parts that have complementary coefficients of thermal expansion (CTE). Requirements, key challenges, performance results and lessons learned are presented. Results from a total of 37 test actuations of the design under various environmental conditions demonstrated a robust device that performed reliably in a flight-like environment. Introduction While in the process of developing the first 3U CubeSats designed and built at APL, the authors became convinced they needed a unique release mechanism for use in restraining and releasing the CubeSat’s four solar array panels. Driven by power and volume limitations, the actuator developed in this original work is inexpensive, has a single moving part, generates no shock, uses little power, is re-settable, and does not consume any flight parts in its operation. In the process of qualifying the release mechanism, we revised the design twice to improve the performance, ease of operation and installation of the device, and to reduce the cost of producing the devices. The CTE Release Actuator (CTERA) has successfully completed functional testing in vacuum, self-actuation testing and static load testing. It was tested under three power profiles representing those expected during flight. It has been tested in an actuator only configuration, as well as with a deployable structure representing a possible mission use. The principle of operation for the release mechanism uses two parts that have complementary coefficients of thermal expansion (CTE). The material with a low CTE is inserted into a hole in the material with the high CTE after the high CTE material is heated. Once the high CTE part cools, the low CTE part is trapped due to an interference fit between the two parts. The interference is sized for the retaining force desired from the actuator. The low CTE part, called the plug, is fastened to the part or assembly to be separated from the space vehicle. When the high CTE part, called the cup, is re-heated, the low CTE part is freed, just as a bolt would be freed from a separation nut. Key challenges for the design included analysis to set the machining tolerances required, choice of the mechanical surface properties of the two interfacing parts, finding low cost heaters with high watt-density that operate at low voltage, and testing various power profiles to minimize the power drawn by the device. A MathCAD design tool was developed for initial sizing of the interference fit, and later, using a test-verified release coefficient, to predict the release temperature under the influence of the kick-off spring. * Johns Hopkins University Applied Physics Laboratory, Laurel, MD Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

466 Requirements As the packaging layout of the reference mission progressed, these key requirements for the design of this tiny actuator became evident:  Tiny envelope of less than 9.83 cm3 (0.6 in3) – the device needed to fit into an approximately 2.54 x 2.54 x 1.524 cm (1” x 1” x 0.6”) envelope.  Low power availability - Since power for the CubeSat reference mission was at a premium, we needed to minimize the amount of power drawn by the actuator. The conductive heat loss between the cup and the plug, as well as the cup and the spacecraft, was a priority.  Low upper temperature limit – The avionics on the reference mission didn’t have the capability to sense the temperature of the actuator; therefore we needed to use a set time for resistor power cutoff. This limitation forced the design to release at a much lower operating temperature in order to reduce the risk of damaging the resistor heating source.  A relatively large holding capacity for the actuator’s size – the load requirement with test factor of 1.5 was 143 N (32.25 lb).  A tension-only boundary condition – due to the short distance from the hinge, a spherical ball boundary condition was needed, with no room for a spherical bearing.  Low shock – The CubeSat specification prohibits pyrotechnic devices. APL also chose to minimize shock generated by the chosen release mechanism to protect CubeSat avionics and mechanisms. The CTERA produces negligible shock.  Easily resettable – The device used in the APL CubeSat needed to be easy to reset. Removing the device from the SV for re-set would cause unacceptable delays in the I&T process.  Use of non-magnetic materials throughout the mechanism. Trade Study Results We conducted a quick electrical characteristic comparison, shown in Table 1, between the CTE actuator and commercial devices, with a particular focus on the energy required to actuate them. The table shows five release actuators and their electrical characteristics. We did not attempt to differentiate mechanical characteristics, since only one other device in the table fits the size constraints to which the CTE actuator was designed. Table 1 - Device Trade Study Device Current (A) Voltage (V) Power (W) Resistance (Ω), Total Δtime (s) Energy (W-s) % Battery Energy Device #1 (SMA) 1.00 9.00 9.0 9 35 315 0.20% Device #2 (SMA) 5.36 7.50 40.2 1.4 0.032 1 0.0008% Device # 3 (Paraffin Initiated Actuator) 0.36 28.08 10.1 78 150 1516 0.94% Device #4 (SMA) 3.50 14.70 51.5 4.2 0.035 2 0.0011% APL CTE Actuator 2.60 7.28 18.9 2.8 17.8 337 0.21% Device # 1 can be used in applications similar to those for which the CTE Actuator can be employed. This device has lots of heritage, produces almost no shock, and is a highly reliable device in APLs experience. Its nominal power performance is slightly better than that of the actuator. This device must be removed from the space vehicle (SV) or instrument to be re-set and it consumes notched bolts with each actuation. This particular model would need to have a custom heater resistance to be used on a CubeSat, since the 9-ohm resistive heater wouldn’t dissipate enough power to actuate the device at minimum voltage (6V). This device, unless non-recurring engineering was expended to re-size it, simply would not fit in the volume allotted in the APL CubeSat.

Document

AMS_2012.pdf

467 Device #2 is another shape memory alloy device. This device uses a shape memory alloy to actuate a ball detent mechanism which releases a deployable item. Its load capacity is far beyond what is needed for many CubeSats or small instruments. It uses very little power and has an almost instantaneous actuation. It’s also much larger and more massive than could be used in this APL CubeSat. It draws far more current than a CubeSat could supply without dedicated capacitor banks. While it produces much less shock than a pyrotechnic device, it still produces shock, which must be evaluated in the design process. Device #3 uses a resistive heater that heats paraffin wax. When the wax is heated, it expands. The actuators are designed to produce linear motion with a defined output force. They are compact, highly reliable, and insensitive to contamination. They require a companion mechanism to affect the release of a deployable device. They consume much more energy than any other device compared. Device #4 uses shape memory alloy wires to rotate a cam which allows a four segment nut to separate and release a bolt. They have a much higher load capacity than is needed for the reference application. They use very little energy and are near-instantaneous actuators. They are much larger and massive than could be used in this APL CubeSat. They draw much more current than is available in a CubeSat. Prior to conducting the trade study, we eliminated pyrotechnic actuators from consideration, because the CubeSat specification forbids their use. We also considered two other custom designs, one that operated by cutting ‘fishing line’ and the other that operated by vaporizing a Ni-Chrome wire. Enough work was completed on each design to determine that for the particular combination of available volume, power and holding force of this application, the designs were overly complex and difficult to manufacture and assemble. We conducted creep testing on available polymer lines compatible with a low power line cutter, and concluded that the lines’ creep would result in loss of preload in the restraint, causing gapping and unacceptably high loads during flight. We also had contamination concerns arising related to melting plastic and its possible re-solidification on solar array cells or optical surfaces. While promising, given a different situation, these ideas were eventually shelved for this application. Design Function & Development The first generation device (Figure 1) used parts without any surface coatings and used a single 30-W thick film power resistor as the heater element. It served as a proof-of-concept, but was not suitable for flight use. The outer cup was fabricated from 6061-T6 aluminum, the inner plug from 6AL4V titanium. The machining tolerances for the cup (high CTE) and plug (low CTE) parts were critical, since the interference fit between the two was just 15 µm (0.0006”), nominal OD to ID. Because of this, we machined the parts to a diametrical tolerance of 3.81 µm (± 0.00015”). A simple tension-only test stand was used for testing the single first generation prototype. The first generation actuator consisted of an un-plated aluminum body fabricated on a manually operated lathe, and single 30-W 3- Ω thick film resistor, an interface plate with fiberglass thermal insulators, and an un-plated titanium plug. Figure 1 – First generation device with a bare aluminum body and a single 30-W resistor The second generation device featured redundant heaters, a kick-off spring and a vent hole, and is depicted in Figure 2. The heaters have a temperature limit of roughly 135°C at our highest operating current, and the operating range of the actuator needed to be well below this temperature during operation. Our reference mission didn’t have the capability to sense the temperature of the actuator (this Cup Resistor

Document

AMS_2012.pdf

468 was a self-imposed limitation implemented early in the program and will be reversed in future CubeSats); therefore we needed to design for a much lower operating temperature to reduce the risk of damaging the resistors. We therefore chose to limit the operating temperature to 115°C. In practice, actuation temperatures rarely reach 115°C, but setting the actuators for flight gets much easier as the temperature increases. The aluminum cup is thermally isolated from the host structure with G-10 isolators to minimize heat loss and the overall power needed to actuate the device. We are exploring the idea of coating the flight actuators with low emissivity coatings to reduce power required for actuation by limiting heat loss due to thermal radiation. The aluminum parts are plated with a hard coat anodize treatment, but the titanium parts were left bare. The device acted as its own tell-tale separation indicator. When the plug is fully seated in the cup, it contacts the kick-off spring, which is electrically connected to the cup. A circuit is completed from the CubeSat input-output card through the CTERA and its titanium plug, through the solar array substrate, back over the solar array hinge line, into the CubeSat harness into the IO card. When the plug is released, continuity is broken, indicating separation. We verified the concept through several cycles of operation on the tension-only test-stand. The design required great care to re-set. The single-piece plug needs to be precisely aligned with the cup, inserted in as short a time as possible to reduce plug heating. This second-generation design also required fighting against a 44.5-N (10-lb) kickoff spring force during insertion. We ended up galling the surface of the plugs in our efforts to accomplish this. The second generation device functioned unreliably in flight-like use with a solar array wing. The flight-like wing, with its hinge line parallel to the gravity vector, imposed side loads upon the release mechanism that the test stand did not. Figure 2 – Underside (left) and top side (right of the second generation device) The third generation device featured a lower mass high CTE part, a reduced length interface between the two materials, and less heat transfer between the high CTE and low CTE parts, and surface treatments on both of the mating parts. Elimination of the kickoff spring preload during the precision mate of the high and low CTE parts made the mechanism much easier to re-set. Redesign of the solar array fitting accommodated the range of motion constraints imposed by the flight solar array wing by using a custom ‘near-spherical bearing’ featuring a pin and dual conical hole. The new interface to the reference mission solar array wing allows more alignment adjustability between the wing and the actuator. This tolerance eliminates racking and binding during wing deployment. Figure 3 depicts the third generation actuator body and plug and the assembly of pin, lower interface bracket and plug cap. The space-saving pin and dual cone features provide ‘near spherical bearing’ behavior in much less volume. To minimize galling and cold welding the aluminum and titanium parts under a sustained 6.9 mPa (1000 psi) interface pressure, surface finishes were critical. Two versions were fabricated, featuring different combinations of surface treatments on the high and low CTE materials. Some limited functional testing was used to determine which surface treatment combination would be used for the final design. Metrics were part wear, the change in temperature required to achieve separation, and the electrical energy expended to achieve separation. We selected hard coat anodize for the aluminum parts, and Tiodize, Type 2 for the titanium parts. Reduction of the contact area between plug and cup significantly reduced the amount of energy required for the actuator to function. Additionally, the annular interface area between the bottom of the plug and the cup was reduced by the addition of 8 radial ridges, which dramatically reduce the amount of heat transfer between the cup and the plug. Reducing energy absorbed by the plug reduces the amount of delta-T required to separate the plug from the cup. Because of the solar array deployed Plug Cup Resistors G-10 Isolator

Document

AMS_2012.pdf

469 geometry, the hinge causes each of the panels to deploy 90 deg and to twist 45 deg in an axis normal to the initial deployment axis. Figure 3 – Third generation actuator showing interconnect PC board, tell-tale harness, kickoff spring and retainer, and upper interface assembly with “near spherical bearing” Because of this and because of the small radius the plug travels during deployment, the secondgeneration actuator had the tendency to bind during deployment. The reduction of the cylindrical interface area between the cup and plug reduced the likelihood of binding due to any lateral load or misalignment during actuator release. Binding was also reduced by changes to the lead in angles on the outer diameters of the cup and plug. The taper on the inner diameter of the plug reduces the probability of the cup hanging on the kick-off spring during deployment. The change to a two-piece plug, consisting of an annular plug with a cap allows insertion of the plug without compressing the kick-off spring. This eliminates the possibility of binding and galling while setting the actuator for flight, greatly facilitating reassembly. Figure 4 shows some of the features of the third generation actuator. Figure 4 – Four views of the third generation device. Electrical Interface Board Kick-off Spring Dual cone & pin 2-piece cup Kick-off spring and retainer Tell-tale jumper Ridges reduce thermal contact between cup and plug Kickoff spring & retainer Tell-tale jumper

Document

AMS_2012.pdf

470 Testing Results & Data Analysis for Each Generation First Generation The primary intention of the first round of testing was to verify our assumptions about the electrical and thermal aspects of the concept. The device would eventually reach the temperature we calculated was required for separation (approximately a 70°C delta), but we learned that a single 30-W device was insufficient in both power handling capacity and output power. The power output was considered insufficient because it took several minutes for the device to change temperature. We theorized that more power input would reduce the delta-T of the low CTE plug part, as increasing the temperature of the high CTE part faster would provide less time for heat transfer to the low CTE part. This would, in turn, require a lower delta-T in the high CTE part to achieve separation. This was proven later in modifications implemented in the third generation device. We consulted with electrical component engineers who helped us pick parts using appropriate levels of part de-rating. Before this, we failed a few resistors by running them above their current-temperature curves. Using appropriate de-rating guidelines, we increased the resistance to 5.6 Ω, doubled the number of resistors, and switched from 30-W parts to 50-W parts. After the intial thermal and electrical tests were complete, we verified that concept would work, as we were able to expand the aluminum part to the point that we were able to assemble the unit. When we performed the first simulated actuation by removing the plug from the actuator at ambient pressure, we realized we had forgotten a vent hole in the actuator body. We also learned that we couldn’t make the parts on a manually operated lathe. We needed a CNC lathe to achieve the required tolerances for consistent device operation. The first generation test rig is depicted in Figure 5. Figure 5 - First generation device shown in a bell jar type vacuum chamber. Striped wires are power, white wires are temperature sensors. Second Generation The second generation of the actuator was subjected to the tests listed below. Static load testing was conducted prior to functional testing on the actuator, and a load of 334 N (75 lb) was maintained for 15 minutes. Following static load testing, we performed several separations at ambient conditions with 2 A of current. As expected, it took a long time to separate (236 sec). The delta-T to separate was low, 66°C, which we think can be attributed to the virgin state of the actuator (blemishes accrue with uncoated parts). Initial functional testing at vacuum was then successfully completed (separations 4 and 5). Lastly, 12 separations were conducted in vacuum under flight conditions, with the results summarized in Table 2. Run 15 was a scrub. Power was cut off before the full actuation occurred. Note 4 indicates a smooth plug insertion. Note 5 indicates a rough plug insertion. Nine of the twelve separations can be considered to be in family with the majority of the sample population. There were three separations that were out of family. We think these three separations took longer because the insertion of the plug into the cup for these runs was rougher. The rough insertions likely were due to a misalignment of the parts. As the number of cycles increased, we noted some wear marks forming on the uncoated titanium part, indicating that some galling took place, primarily during insertions, but during ejection, as well.

Document

AMS_2012.pdf

471 Table 2- Test data from second generation actuator. Test # Vacuum (Torr) Current (A) Voltage (V) Power (W) Tamb (°C) Trel (°C) ΔT (°C) Δtime (s) ΔT/Δt Energy (W-s) % Batt Energy Note 4 2.50E-01 2.00 5.60 11.2 19.1 94.0 74.9 147 0.51 1646 1.02% 5 11 2.50E-02 2.16 6.05 13.1 23.6 94.0 70.4 100 0.70 1306 0.81% 4 17 2.50E-02 2.16 6.05 13.1 23.6 93.6 72.0 100 0.72 1320 0.82% 4 12 2.50E-02 2.16 6.05 13.1 21.1 104.0 82.9 123 0.67 1607 1.00% 5 18 2.50E-02 2.16 6.05 13.1 19.3 91.0 71.7 115 0.62 1502 0.93% 4 5 2.50E-01 2.58 7.22 18.6 24.0 87.8 63.8 70 0.91 1305 0.81% 4 6 2.50E-02 2.60 7.28 18.9 18.5 92.0 73.5 68 1.08 1287 0.80% 5 7 2.50E-02 2.60 7.28 18.9 18.9 89.2 70.3 66 1.07 1249 0.77% 4 8 2.10E-02 2.60 7.28 18.9 19.5 100.3 80.8 69 1.17 1306 0.81% 5 9 2.10E-02 2.60 7.28 18.9 20.0 90.2 70.2 62 1.13 1174 0.73% 4 10 2.50E-02 2.60 7.28 18.9 23.5 91.0 67.5 61 1.11 1155 0.72% 4 13 2.50E-02 3.02 8.46 25.5 18.9 88.7 69.8 46 1.52 1175 0.73% 5 14 2.30E-02 3.02 8.46 25.5 19.9 80.2 60.3 39 1.55 996 0.62% 4 16 2.50E-02 3.02 8.46 25.5 23.8 80.6 56.8 38 1.49 970 0.60% 4 Not a “standard” power case. Low power case Nominal Power Case High Power Case Outlier We performed a preliminary analysis of only the separations performed in vacuum. A B-basis time to separate was calculated for the whole data set, and each of the three battery state of charge cases. The B-basis time to separate was calculated as follows: T2σ = Mean + (2 * σ ) where T 2σ is the 2 sigma time to separate, Mean is the arithmetic mean of the times to separate, and σ is the standard deviation between the times to separate. Table 3 lists the data for the four cases analyzed (all vacuum separations, nominal power, high power, low power). Table 3 - Second generation statistical results. Data Set Mean Time to Separate Standard Deviation 2-sigma time to separate All vacuum 78.86 33.04 144.94 Nominal (2.6A) 65.2 3.56 72.33 Low (2.16A) 109.5 11.45 132.39 High (3.06A) 41 4.36 49.717 The lessons learned from the 2nd generation testing were:  This version of the actuator was too difficult to set for flight. The operator was required to precisely align the plug in the cup and then to insert it into the cup with enough force to overcome the preload spring. Having to push the plug into the cup with force makes the actuator too susceptible to galling.  Performance degrades over time because of galling produced during resetting operations. Both high pressure surfaces needed to be hard coated.  The actuator was susceptible to racking when coupled with the reference mission solar array engineering model.

Document

AMS_2012.pdf

472  The amount of energy required for release needed to be further minimized for the 3rd generation design.  The tell-tale function of the actuator was determined to be reliable through all testing.  Installation of the plug is critical to the consistent operation of the actuator. We determined to make insertion easier and more repeatable by a redesign that eliminated the preload spring force during insertion.  Use the highest possible insertion temperature during assembly. Third Generation The first 4 units fabricated were fabricated to select which actuator coatings should be employed for follow-on use of the actuators fabricated for flight. Actuator SN002 featured Chemical Conversion Coating (MIL-C-5541, Class 3) coating on the aluminum cup part and Tiodize, Type II (AMS-2488, Type 2) on the titanium plug part. For SN002 (Engineering Model), we conducted 7 separations, and then we retired the unit. Actuator SN004 featured Hard Coat Anodize (MIL-A-8625, Type 3) on the aluminum cup part and Tiodize, Type II (AMS-2488, Type 2) on the titanium plug part. We conducted 6 separations, the self-actuation test, the static load test, and 2 more actuations. The third generation tests used an engineering model CubeSat solar array to represent the correct boundary conditions for the device. Figure 6 depicts the solar array wing and test fixture. The SN004 actuator had much better performance, and its tolerances and coating combination were selected for the final design. We characterized performance of the SN004 actuator in the following sequence:  A single functional test at ambient conditions to verify workmanship (test 4.1). Time and delta-T were recorded for all functional tests.  Two separations in vacuum at nominal power (tests 4.2 and 4.3).  A separation in vacuum at high power (4.5).  A separation in vacuum at low power (4.6).  A self-actuation test to determine the maximum temperature the actuator can be exposed to without spontaneously actuating (4.7). Temperature at separation was recorded.  A static load test to verify that the actuator can withstand launch loads. Time and load were recorded (4.8).  Two separations in vacuum at nominal power to verify that the device was not affected by the static loads test (4.9 and 4.10). Figure 6 – The solar array wing and test fixtur e duplicated the boundary conditions for flight actuator testing. Substrate Stiffener Actuator PlugHinge Assembly Antenna Blade, Captured by Solar Array Actuator Cup Assembly

Document

AMS_2012.pdf

473 The test at ambient conditions verified that workmanship was acceptable. The two tests at ambient temperature in vacuum showed the expected reduction in delta-T and time to separate. These results also showed (despite the small number of tests) a tight data grouping, showing that the device is repeatable. The tests at 3 A and 2.16 A showed that, as expected, the time to separate is inversely proportional to input current, when the tests are run at the same initial temperature. Somewhat surprisingly, the delta-T for the 2.16 A case was lower than those of the 2.6-A cases. Figure 7 shows the test stand inside the vacuum chamber. Figure 7 – Test stand, actuator and CTERA shown in vacuum chamber. Following the tests at 2.16 A and 3 A, we conducted the self-actuation test. The purpose of the self-actuation test was to determine the temperature at which the actuator would release a deployable structure while unpowered. The actuator was required to reach a temperature higher than 50°C before self-actuating to demonstrate a 10°C temperature margin. The ambient pressure temperature cycling chamber was set to ramp temperature at a rate of 1°C per minute, starting at room temperature, which was approximately 19°C. We monitored the tell-tale circuit until the device actuated at 65.9°C. The highest predicted temperature for the CubeSat prior to solar array deployments is 40°C, which gives a margin of 25.9°C on self-actuation. Following the self-actuation test, we conducted a static loads test. The purpose of the static load testing was to determine if the selected coatings and tolerances of the serial number 004 actuator can withstand launch loads after 7 separations. Note that the interference fit for the actuator tested, SN 004, is significantly smaller than those of the flight devices (SN 011-018). Since holding capability is a function of the interference fit, the flight actuators are able to react more load than the SN 004 device. The fixture used for static load testing is depicted in Figure 8. Tension on the actuator was increased in ¼ load increments and held for at least 30 seconds until the final load of 160 N (36 lb) was reached. Load was maintained at or above 156 N (35 lb) for 5 minutes. Following the static loads test, the CTERA was cycled two more times at ambient temperature and nominal power in vacuum (tests 4.9 and 4.10). Results from these two tests indicated that the electrical and thermal performance of the device wasn’t affected by the static loads test. Results from tests 4.2, 4.3, 4.9 and 4.10 are tightly grouped. The raw data from the tests are presented in Table 4

Document

AMS_2012.pdf

474 Figure 8 – The CTERA withstood 1.5 times the maximum expected flight load for 5 minutes. Table 4 - Non-flight third generation actuator data, SN 004. Test # Vacuum (Torr) Current (A) Voltage (V) Power (W) Tamb (°C) Trel (°C) ΔT (°C) Δtime (s) ΔT/Δt Energy (Ws) % Batt Energy Interference Fit Note 4.1 ATM 2.60 7.28 18.9 19.5 71 51.5 33 1.56 625 0.39% 0.0003 4.2 2.50E-02 2.60 7.28 18.9 18.7 45.0 26.3 18.5 1.42 350 0.22% 0.0003 4.3 2.50E-02 2.60 7.28 18.9 18.5 44.0 25.5 18 1.42 341 0.21% 0.0003 4.9 2.50E-02 2.60 7.28 18.9 18.4 47.8 29.4 17 1.73 322 0.20% 0.0003 1 4.10 2.50E-02 2.60 7.28 18.9 18.5 46.0 27.5 17.5 1.57 331 0.21% 0.0003 1 4.5 2.50E-02 3.02 8.46 25.5 18.5 39.8 21.3 12 1.78 306 0.19% 0.0003 4.6 2.50E-02 2.16 6.05 13.1 18.8 42.2 23.4 26 0.90 340 0.21% 0.0003 4.4 7.55E+02 2.60 7.28 18.9 UN K UN K UN K UN K UN K UN K UNK 0.0003 2 4.7 7.55E+02 0.00 0.00 0.0 18.9 65.9 47.0 NA NA NA NA 0.0003 3 4.8 7.55E+02 0.00 0.00 0.0 19.0 NA NA NA NA NA NA 0.0003 4 1 After static load test. 2 Actuated while adjusting current delivered to the resistor. At ambient. No data recorded. 3 Will not self actuate until ~66°C. SV will not go above 40°C. Adequate margin. 4 Static Load Test. Requirement (with test factor 1.5) was 143.5N (32.25#). Held above 157N (35.3#) for 5 minutes. When the results from SN 004, with AMS-2488, Type 2 coating for titanium and hard coat anodize coating for the aluminum were compared to those for SN 002, with the same titanium finish and chemical conversion coating (MIL-C-5541, Class 3) of the aluminum, the SN 004 device was the clear winner. SN 002 was tested first, and we concluded testing after the first 7 cycles because the data were poorly grouped, the performance of the device was perceptibly deteriorating, and we observed increasing damage to the interfacing surfaces of the aluminum and titanium parts. SN 002 raw data are not presented due to space restrictions. Once the final design was selected, we fabricated eight flight units and conducted the following tests in the following order to validate the design. The flight mechanisms (SN 011 through 018) underwent the following testing to further validate the design and workmanship, and the raw data from these tests are presented in Table 5:

Document

AMS_2012.pdf

475  At least one separation at nominal current at ambient temperature in vacuum (for SN 011-018, green shaded portion of the table)  A single separation at minimum current at -20°C in vacuum (for SN 011-018, light blue shaded area of the table)  Random vibration test on the Serial Number 001 3U CubeSat in the NASA NLAS CubeSat dispenser (SN 011-014)A single separation in vacuum after soak at 11°C and transient to -1°C on the Serial Number 2 3U CubeSat (mechanism SN 015-018) Table 5- Data taken from third generation flight devices (SN011-018) Test # Vacuu m (Torr) Curren t (A) Voltag e (V) Power (W) Tamb (°C) Trel (°C) ΔT (°C) Δtime (s) ΔT/Δt Energy (W-s) % Batt Energy Interfer ence Fit Note 11.2 2.50E-02 2.60 7.28 18.9 21.8 81.3 59.5 46 1.29 870 0.54% 0.00055 11.3 2.50E-02 2.60 7.28 18.9 21.0 77.7 56.7 48 1.18 909 0.56% 0.00055 11.5 2.50E-02 2.60 7.28 18.9 19.6 86.9 67.3 45 1.50 852 0.53% 0.00055 12.1 2.50E-02 2.60 7.28 18.9 21.0 85.0 64.0 53 1.21 1003 0.62% 0.0007 13.1 2.50E-02 2.60 7.28 18.9 21.2 95.7 74.5 70 1.06 1325 0.82% 0.0007 14.1 2.50E-02 2.60 7.28 18.9 21.0 79.0 58.0 50 1.16 946 0.59% 0.00045 15.1 2.50E-02 2.60 7.28 18.9 21.4 79.6 58.2 55 1.06 1041 0.65% 0.00065 15.2 1.40E-03 2.60 7.28 18.9 18.8 87.4 68.6 52 1.32 984 0.61% 0.00065 16.1 2.50E-02 2.60 7.28 18.9 20.0 94.2 74.2 59 1.26 1117 0.69% 0.0007 17.1 2.50E-02 2.60 7.28 18.9 20.5 91.8 71.3 58 1.23 1098 0.68% 0.0006 18.1 2.50E-02 2.60 7.28 18.9 20.8 87.6 66.8 58 1.15 1098 0.68% 0.00065 11.4 1.50E-05 2.16 6.05 13.1 -20 73.0 93.0 120 0.78 1568 0.97% 0.00055 12.2 1.30E-06 2.16 6.05 13.1 -19.9 87.9 107.8 127 0.85 1659 1.03% 0.0007 13.2 6.00E-06 2.16 6.05 13.1 -20 98.8 118.8 140 0.85 1829 1.13% 0.0007 14.2 5.50E-06 2.16 6.05 13.1 -20.1 58.1 78.2 129 0.61 1685 1.04% 0.00045 15.2 1.70E-06 2.16 6.05 13.1 -20.1 78.5 98.6 130 0.76 1698 1.05% 0.00065 17.2 9.80E-06 2.16 6.05 13.1 -20 94.0 114.0 140 0.81 1829 1.13% 0.0006 18.2 3.40E-06 2.16 6.05 13.1 -20 133 1737 1.08% 0.00065 16.2 8.50E-07 2.16 6.05 13.1 -20.3 122 1594 0.99% 0.0007 15.3 6.00E-06 2.19 8.06 17.6 -1.0 82.0 81.0 71 1.14 1252 0.78% 0.00065 16.3 6.00E-06 2.16 8.00 17.3 -1.0 80.0 87.0 72 1.21 1243 0.77% 0.0007 17.3 6.00E-06 2.17 7.92 17.2 -1.0 96.0 89.0 78 1.14 1342 0.83% 0.0006 18.3 6.00E-06 2.14 7.90 16.9 -1.0 92.0 91.0 79 1.15 1335 0.83% 0.00065 11.6 ATM 2.26 6.33 14.3 50 57 814 0.50% 0.00055 12.3 ATM 2.26 6.33 14.3 50 63 900 0.56% 0.0007 13.3 ATM 2.29 6.40 14.6 50 84 1228 0.76% 0.0007 5 14.3 ATM 2.23 6.25 14.0 50 59 824 0.51% 0.00045 5 Automated shutdown failed because of a harness fabrication error. Manual shutdown was delayed. We performed a basic analysis of the raw data from the flight device (SN 011-018) separations at ambient, cold vacuum (-20°C), thermal balance and thermal cycling cases. From these data, we determined the mean, standard deviation and two-sigma data, for time to separate, delta-T, and power required to separate. The two-sigma time to separate is calculated as follows: T 2σ = Mean + (2 * σ ) where T2σ is the 2 sigma time to separate, Mean is the arithmetic mean of the times to separate, and σ is the standard deviation between the times to separate.

Document

AMS_2012.pdf

476 We recommended to our team that 2 sigma durations be programmed to assure that timed separations are successful. Table 6 lists the data for the four cases analyzed (2.6 A at ambient temperature in vacuum, 2.16 A at -20°C in vacuum, ~2.15 A during thermal balance testing at 0°C, ~2.25 A during thermal cycling in ambient pressure at 50°C). Table 6 – Rudimentary statistical analysis of CTERA temperature, time and energy data Vacuum, 2.6A, 20°C Vacuum 2.16 A, -20°C mean std dev X + 2 σ mean std dev X + 2 σ Time 54.0 7.2 68.4 Time 130.1 7.4 144.9 dT 65.4 6.6 78.5 dT 101.7 15.0 131.6 Energy 1022.1 136.0 1294.1 Energy 1699.9 96.6 1893.1 Vacuum, Thermal Balance, 2.15A, 0°C Ambient, Thermal Cyclling, 2.25A, 50°C mean std dev X + 2 σ mean std dev X + 2 σ Time 75.0 4.1 83.2 Time 65.8 12.4 90.6 dT 87.0 4.32 95.6 dT Energy 1292.8 52.8 1398.4 Energy 941.6 194.8 1331.2 We had already confirmed that actuation in vacuum reduces the energy required to actuate (see the difference between test 4.1 and tests 4.2-4.10). Looking further at the data, we noticed a correlation between input power and time to separate, and a correlation between initial temperature and time to separate. The following several figures pictorially represent relationships between initial temperature, delta-T, time, power level, and the different interference fits of the various actuators. The left chart of Figure 9 shows that release time is directly related to the energy expended, as one would expect. It also shows a correlation between the power applied to the device, illustrated here by various current levels, and time to release. The faster energy is applied to the device (more power), the less time and energy it takes to operate. The right chart of Figure 9 shows the relationship between release delta-T and power. As power increases, release delta-T decreases. It appears as the power applied to the high CTE part is increased, its rate of expansion is higher and less heat is applied to the low CTE part. Since less heat is applied to the low CTE part, it expands less and more slowly, resulting in a lower delta-T before the parts separate. Figure 9 – Strong relationships between energy, delta-T, and time to separate

Document

AMS_2012.pdf

477 The left chart of Figure 10 shows the effect the interference fit has upon the delta-T required for the parts to separate. As the analytical model suggests, the looser fit requires less delta-T for the parts to separate. The right chart of Figure 10 shows the effect the interference fit has upon the energy required to separate. The high CTE part doesn’t need to expand as much to free the low CTE part, which takes less energy. Additionally, when the interface pressure between the two parts is lower, less conductive heat transfer results between the two parts, so less energy is wasted heating the low CTE part. We initially expected less scatter in these charts, but realized static friction effects cause scatter (torque vs. preload tests exhibit this scatter, for example), and that minute surface imperfections have an appreciable affect on results, since the interface pressures are quite high in this design. These two charts also show two other relationships. The first, which we already discussed, is between input power and the time, energy, and delta-T required for separation. The second, discussed in the following paragraph, is the relationship between the initial temperature and the time, energy and delta-T required for separation. Figure 10 – Relationships between interference fit, delta-T and energy required to separate The left chart of Figure 11 shows the relationship between the initial temperature and the energy required to separate. The right chart of Figure 11 shows the relationship between actuation delta-T and the initial temperature. Since the parts were manufactured at 20°C, they will have a tighter interference fit with a lower initial temperature and will have a looser interference fit as they have a higher initial temperature. Figure 11 – Relationships between initial temperature, and energy to actuate Testing of the flight parts has thus far confirmed the following:  A single CTERA has been shown to not degrade in performance over 10 cycles. We intend to perform further life testing when given the opportunity.

Document

AMS_2012.pdf

478  Despite variability in interference fits from device to device, actuation data are tightly grouped and can be used, when factored, for timed separation commands when the bus is at room temperature in vacuum.  The tell-tale function of the actuators was determined to be reliable though the vibration testing and subsequent thermal cycling testing they underwent.  Tell-tale reliability is not affected by vibration or thermal cycling.  Operating condition effects: o The effect of supplied current on total energy and “on-time” is significant. o The effect of initial temperature on actuation times is also significant. o Since the “on-time” can vary wildly with initial temperature and supplied current, these devices should not be used open loop to avoid damaging the heaters . We recommend either: implement tell-tale reading circuits, as APL’s CubeSat did, or implement temperature sensors with each actuator and shut the devices off when the devices reach the maximum allowable temperature, or code in a two dimensional look-up table in the flight software, which sets “on-time” as a function of battery state of charge (resulting in current delivered to the device) and initial temperature. Lessons Learned  Don’t skimp on Ground Support Equipment (GSE) and test equipment. Test GSE are just as important as the flight hardware: o It would have been immediately obvious that the first (and second) generation devices were incompatible with the CubeSat solar array deployment motion if we had tested with a flight-like wing mechanism from the beginning. Representing boundary conditions correctly would have saved one design iteration. o Pay attention to data acquisition. We lost some data due to shortcuts in setting up the data acquisition system. We also would have been able to capture more data (such as temperature vs time curves) with a more advanced data acquisition system. o We could have saved money and time by testing more than one device at a time with more elaborate GSE o Enlist subject matter experts initially. We could have eliminated one design revision if we had known of electrical parts de-rating guidelines and selected the right resistors in the beginning.  Incremental development is cost effective. By making really cheap parts, testing a little and learning a lot, fabricating a slightly more elaborate second generation device, testing a little and learning a lot more, we were still able to develop and test the flight device in a short, inexpensive effort.  Manufacturing methods are critical to the success of a mechanism design; it wouldn’t have cost much more to fabricate the first generation parts on the newer generation CNC lathe, rather than the manually operated lathe. Conclusion A compact device for releasing deployable structures has been developed exploiting differential coefficients of thermal expansion of two dissimilar materials. The resultant device is economical, miniature, simple in concept and execution, uses little power, produces negligible shock, is easily re-settable, and is able to restrain a relatively large load, considering its size. Through a total of 37 test actuations, the third generation design demonstrated a robust device that performed reliably in a flight-like environment. This device is applicable to small deployable devices on instruments, CubeSats and Nanosatellites, and has the capability to be used as an initiation device for mechanisms used on conventional spacecraft and larger instruments. Future work will focus on alternate materials, such as invar, which has a very low CTE, and manufacturing processes, as well as integration into larger load capacity mechanisms.

Document

AMS_2012.pdf

479 A Nichrome Burn Wire Release Mechanism for CubeSats Adam Thurn*, Steve Huynh*, Steve Koss*, Paul Oppenheimer*, Sam Butcher*, Jordan Schlater** and Peter Hagan+ Abstract The nichrome burn wire release mechanism uses a nichrome burn wire which when activated heats up and cuts through a Vectran tie down cable allowing the deployable on the satellite to actuate. The release mechanism was designed from scratch with the goals to make it small, inexpensive, simple, reliable and easy to use by anyone including student-run University CubeSat projects. The release mechanism, shown in Figure 1, utilizes a two saddle design with compression springs to apply a spring stroke and force to the nichrome wire to thermally cut through the Vectran cable when heated. Through a test program and using a design of experiments (DOE) approach it was determined that the applied current to the nichrome wire and the diameter of the nichrome wire were the key parameters governing successful performance. To activate the nichrome wire a constant current of 1.60 ± 0.05 amps is applied to ensure a successful and reliable cut. The tight tolerance constant current source is necessary in order to reliably: 1.) thermally cut the cable and 2.) prevent overheating failure of the nichrome wire to allow the mechanism to be reusable for many actuations without replacing the nichrome filament. The tight tolerances on the current prevent failure of the nichrome wire from overheating under too much current in a vacuum while also providing adequate thermal margin to cut through the Vectran tie down cable in air which requires more current than in vacuum. The burn wire release mechanism has been tested in air at room temperature and in vacuum at temperatures as low as -50°C and as high as 70°C on two different Vectran cable thicknesses. The release mechanism has shown to have cut times ranging from 2.4 to 7.2 seconds under these operating conditions. The burn wire release mechanism has 400 firings in component and system level testing without a single failure. The mechanism has been qualified for flight on the TEPCE (Tether Electrodynamic Propulsion CubeSat Experiment) program to release two carpenter tape deployments and a stacer and tether deployment system. Figure 1. Assembled burn wire release mechanism * Naval Research Laboratory, Washington DC ** NRL Co-op Student from the University of Cincinnati, Cincinnati, OH + NRL Co-op Student from Northeastern University, Boston, MA Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

480 Introduction CubeSats have grown into a class of satellites which are rapidly introducing new, inexpensive technologies in space. This ever increasing complexity and functionality in a small package brings rise to numerous mechanisms and deployables necessary in order to meet mission requirements. Release mechanisms often add significant development cost to a satellite program and in the world of small satellites, such as CubeSats, the need for simple, reliable, and inexpensive release devices can be critical to the success of the program. The ability for a single mechanism to accomplish many of these releases is a desirable trait among CubeSat users. Whether the mechanism be used to release antennas, solar arrays, deployable doors, etc.; an inexpensive, simple and reliable release mechanism would help to further promote the class of small satellites throughout the community. Since there are currently no standard commercialized CubeSat class release devices on the market, one was developed at the Naval Research Laboratory (NRL) via internal fellowship funding, since these programs are never funded sufficiently to develop one with program funding. A spring-loaded nichrome burn wire solution was selected because it: 1.) could be actuated with standard CubeSat bus power and fairly simple, low cost electronics, 2.) was simple enough that University students could build and use it, and 3.) was inexpensive enough that CubeSat funding levels could afford it. With few moving parts, simply machined components and the majority of the hardware able to be purchased through low cost commercial suppliers such as McMaster-Carr, the nichrome burn wire release mechanism would allow CubeSat programs, including Universities, to keep down cost and complexity and ensure a high level of reliability for the release mechanisms needed for successful satellite operation. Design The release mechanism, utilizes a compression spring system in order to apply a force and a stroke to the nichrome burn wire. When a constant current is applied to the nichrome wire, it will thermally cut through a Vectran tie down cable allowing it to release the deployable it had secured. The nichrome wire used on the mechanism is 30 AWG type Chromel C with an allowable free length ranging from 10.0 (0.4”) to 32 millimeters (1.25”). The free length of the nichrome wire is configured into a V shape with the apex in the V being the primary area for cutting through the tie down cable (See Figure 1). The nichrome wire free length range is determined by the minimum length which will avoid problems with heat sinking the nichrome wire to the rest of the mechanism. At a free length of at least 10 mm the apex in the V of the nichrome wire will be sufficiently far enough away from the mechanism heat sinks to avoid heat loss and ensure a successful cut. The maximum length of the nichrome wire is limited by the free length which causes the wire to lose structural stability when heated. At a free length greater than 32 mm the apex of the nichrome wire when heated becomes very hot and the resulting loss in tensile strength can cause necking of the wire which would impact further use of the nichrome wire. Using this free length range the resistance of the nichrome wire as measured from the screw head to screw head of the release mechanism is in the range of 0.4 to 0.9  . When selecting the free length of the nichrome wire it is critical that the release mechanism has ample spring stroke to cut through the Vectran cable with the allowable deflection of the Vectran cable. Particularly in vacuum where convective heating cannot be taken advantage of, the nichrome wire must completely stroke through the Vectran cable in order to ensure a successful release. If the spring stroke on the wire is lost before the entire cable is cut then it is possible for the nichrome wire to remain stuck in the Vectran cable without severing it enough to have a successful cut. Therefore the tension on the Vectran cable must be enough that it does not allow for large, sagging bends in the cable which would allow for the springs on the mechanism to lose their preload and cutting stroke. The compression springs are held between two saddles which are positioned on a pair of stainless steel dowel pins using retaining rings. The saddles are machined from 6061 aluminum which has been hard anodized to prevent electrical shorting (See Figure 1). The dowel pins have a tapped hole on their upper end where a #0 ‐80 button head screw threads into. The head of the #0 ‐80 screw attaches the nichrome

Document

AMS_2012.pdf

481 wire to a ring terminal which is connected to two flying leads and provides the electrical connection to the release mechanism. The nichrome wire is secured between the ring terminal and the underside of the head of the #0-80 screw (See Figure 2). The ring terminals which can be purchased cheaply with tin plating will need to be stripped by a plating manufacturer for a small cost, approximately $1 per ring terminal. The #0-80 screw is secured into the tapped hole of the dowel pin using 3M Scotch-Weld™ 2216 B/A on the threads of the screw to prevent the screw from backing out and losing the secure connection of the nichrome wire. Figure 3 shows CAD models of the nichrome burn wire release mechanism in the deployed and stowed states. Figure 2. Nichrome wire secured between the button head screw and ring terminal Figure 3. Nichrome burn wire release mechanisms in the deployed and stowed configurations. The mechanism was designed to thermally cut through two different deniers of Vectran cable. Both Vectran cables were 12 strand tubular braids with either 200 or 400 denier strands and were

Document

AMS_2012.pdf

482 manufactured without an oil finish for low outgassing. The 200 denier Vectran cable has a tensile breaking strength of 300 N and the 400 denier Vectran has a tensile breaking strength of 600 N. Vectran cable was selected as the primary tie down cable due to its low creep over time, its relative strength to other materials, and its resistance against self-abrasion. The entire burn wire release mechanism once assembled has dimensions of approximately 32-mm (1.25”) long by 16.5-mm (0.65”) wide by 11.5-mm (0.45”) tall. Figure 4 shows two, redundant burn wire release mechanisms stowed on the TEPCE tether deployment system. Due to the small size, assembly and workmanship, inspection is best done with the aid of magnification such as a stereo microscope. Figure 4. Release mechanisms on Vectran tie down cable for TEPCE tether deployment system Testing The goals established for a generic CubeSat class release mechanism were: 1.) to be simple enough to replicate the build and assembly by all CubeSat users (including Universities), 2.) be inexpensive, 3.) reliably release in a few seconds but rugged enough to survive for a 30 second timer for each actuation, 4.) have a design life at least 50 actuations, 5.) work at the same power draw in both air and vacuum and 6.) fit into as small a volume as possible. A burn wire based release device was selected based on a trade study between numerous miniature actuators including bolt releases, pin pullers, and linear and rotary actuators. Then the parameters associated with the burn wire, the compression springs and the tie down cable were investigated. Since a large number of parameters needed to be tested in order to find the optimal design of a nichrome wire system, a design of experiments (DOE) approach was taken in order to test multivariable changes in the minimum number of experiments. The parameters chosen to test were the nichrome spring stroke and force applied, the nichrome wire diameter, the free length of the nichrome wire and its corresponding resistance, the tie down cable, the tension on the tie down cable, the current supplied to the nichrome wire, the number of actuations (life) for each nichrome wire, the performance in air and vacuum and the effect of the environmental temperature. Additionally, minimizing the cut time of the Vectran tie down cable was a goal. Using the JMP Statistical Discovery software a DOE test matrix was designed and tests were conducted and important factors in the design were determined.

Document

AMS_2012.pdf

483 Through the DOE analysis in JMP, it was found that the most important factor for successful cut times in both air and vacuum was the supplied current to the nichrome wire. As long as the compression springs supplied spring stroke all the way through the Vectran tie down cable and the nichrome wire had enough current to thermally cut through the tie down cable then the release mechanism would successfully release. The diameter of the nichrome wire was then based on that which provided a large enough electrical current margin to reliably release in air balanced against not failing in vacuum due to overheating the nichrome wire . An electrical circuit was designed to supply a constant current to the nichrome wire independent of the resistance of the wire and the voltage available from the spacecraft. As long as the minimum required power (0.9 W) was available from the spacecraft then the electrical circuit would supply a constant current to the nichrome wire. It was found that the 30 AWG nichrome wire provided an acceptable margin relative to overheating in vacuum versus failing to thermally cut in air. Since the design envelope for the two failure modes was not large, the constant current source was kept to tight tolerances. The minimum amount of current needed to reliably cut through the tie down cable in air (worst case) was 1.40 amps. The supplied current to the nichrome wire that would cause a failure due to overheating in vacuum (worst case) was 1.90 Amps. Therefore a constant current requirement of 1.60 ± 0.05 amps was selected to provide margin on either side of the failure modes and was used to design the electrical circuit. This allowed the mechanism to successfully operate in both air and in vacuum without having to change any circuitry or software. It is critical to note that the critical operating window – in vacuum/space where it must work – is larger than the worst case window that includes operation in air for ground testability. Additionally, in all tests to date where the nichrome wire has been overheated to failure, it has first cut through the Vectran cable for a successful release prior to the nichrome wire failing such that this has not actually resulted in a failure to deploy but rather ended the reusable life of the nichrome wire filament. This shows that the nichrome burn wire release mechanism has a much larger margin to work successfully in vacuum where it counts. The schematic for the constant current circuit design used on the nichrome burn wire release mechanism is given in Figure 5. Figure 5. Schematic for the constant current circuit design.

Document

AMS_2012.pdf

484 Results The design and optimization of the burn wire release mechanism revolved primarily around understanding the performance of the nichrome wire. The failure current of the nichrome wire under only the tension of its own weight was first investigated and the results for the 30 AWG nichrome wire is given in Table 1. This data shows that the upper bound of the nichrome wire failure current is established at 1.90 amps. Table 1. Failure currents of the 30 AWG nichrome wire under only the tension of its own weight. Nichrome Wire Resistive Length (mm/in) Resistance ( ) – as measured from screw to screw Failure Current (amps) Failure Time (sec) 13 / 0.50 0.8 2.15 7 19 / 0.75 1.1 2.05 7 25 / 1.0 1.5 1.95 7 32 / 1.25 1.3 1.95 9 38 / 1.50 1.5 1.9 10 45 / 1.75 1.6 1.95 14 51 / 2.0 1.8 1.9 20 57 / 2.25 2 1.9 17 64 / 2.5 2.1 1.975 30 70 / 2.75 2.2 1.975 19 76 / 3 2.2 1.975 13 83 / 3.25 2.4 1.9 15 89 / 3.5 2.6 1.9 19 Avg. Failure Current 1.960 Using the tight tolerance constant current circuit, 419 successful tests of the release mechanisms have been conducted without a failure. Of the 419 tests, 242 were conducted in air at room temperature and 177 were conducted in vacuum at various temperatures. Tests were also conducted to intentionally overheat the nichrome wire and cause it to fail but even in these cases; the nichrome wire always first cut through the tie down cable ensuring a successful release and then the nichrome wire would fail after the successful cut from overheating. Figure 6 shows test data taken for the air and vacuum operation of the burn wire release mechanisms cutting through 200 and 400 denier Vectran cables.

Document

AMS_2012.pdf

485 Figure 6. Release mechanism deployment times for varying currents of both Vectran cables in air and vacuum environments. Using the constant current circuit which supplied the nichrome wire with 1.60 amps with a 30 second powered timer, the cut times for the burn wire release device in air and vacuum and using both Vectran cables ranged from 2.4 to 7.2 seconds. For the critical operating window in vacuum this gives approximately a 10X margin on the cut time versus the 30 second powered timer limit. Table 2 gives a summary of the average cut times for the operating conditions. Table 2. Average cut times of the burn release mechanisms. Average Cut Time (sec) 200 Denier Vectran in Air 5.3 200 Denier Vectran in Vacuum 2.6 400 Denier Vectran in Air 6.2 400 Denier Vectran in Vacuum 3 The burn wire release device was also tested down to -50°C and up to 70°C without any noticeable effect on the performance. This is likely due to the nichrome wire heating up to over 750°C and having negligible thermal mass, such that the difference in initial temperature has a negligible effect. It was also

Document

AMS_2012.pdf

486 demonstrated that when using the tight tolerance constant current circuit the burn wire release mechanism can be actuated at least 50 times without having to change out the nichrome wire. Many of the burn wire release mechanisms were used past the 50 actuations, up to 80 actuations, and still performed nominally Figure 7 shows the release mechanism immediately following a successful cut through the Vectran cable in vacuum with the nichrome wire still glowing hot. Figure 7. Burn wire release mechanism immediately following a successful cut through Vectran cable. To be useful to all CubeSat users, including Universities the burn wire release mechanism needed to be simple to manufacture and assemble and to be cost effective. For the NRL TEPCE CubeSat program headed by NRL, the cost for all hardware associated with building 10 of the burn wire release mechanisms was $1600 making the cost per mechanism $160. This cost included having a local machine shop manufacture both the upper and lower saddles and the dowel pins. However, a student with reasonably proficient machining skills could make the mechanism components themselves and eliminate all but the anodizing costs which could drop the price for 10 mechanisms to approximately $200. Table 3 gives a price breakdown of the hardware associated with machining and building up the burn wire release mechanisms for use on a small satellite. The nichrome wire release device has been through component level testing in different configurations ranging from carpenter tape deployments to a stacer and tether deployment to the deployment of 3U solar array panels on a CubeSat. The release device has also undergone system level testing on the TEPCE CubeSat program.

Document

AMS_2012.pdf

487 Table 3. Cost breakdown for burn wire release mechanisms. Part Description Material Secondary Processing Quantity Total Cost Upper Saddle Aluminum 6061 Hard Anodize 10 $849.00 Lower Saddle Aluminum 6061 Hard Anodize 10 $200.00 Dowel Pin 18-8 Stainless None 20 $430.00 Compression Springs (CI013DE03M from Lee Spring) 302 Stainless Steel None 100 $101.00 1/8" External Snap Rings (McMaster-Carr) Beryllium Copper None 100 $12.50 #0 Stud Ring Terminals (McMaster-Carr and Stripping Manufacturer) Copper Tin Stripping 100 $45.30 #0 Fasteners (McMaster-Carr) 18-8 Stainless Steel None 100 $5.68 0.0100" Diameter Nichrome Wire Nickel Chromium None 1/8 lb. $18.65 Total Cost for 10 Mechanisms $1,662.13 Cost per Mechanism $166.21 Cautions and Areas for Potential Improvement A drawback to the burn wire release mechanism is that the successful operation of it relies largely on the workmanship and assembly of the mechanism in its intended application. A loss of spring preload is possible if the Vectran tie down cable is not tensioned properly. Additionally, the available play and stiffness of the electrical wiring for the flying leads or and improperly securing the nichrome wire under the screw head can all lead to potential failures of the burn wire release mechanism to successfully cut through the Vectran tie down cable. Therefore it is critical that throughout the assembly and installation of the burn wire release mechanism that careful attention is given to these areas of concern and risk of failure is mitigated. Properly tensioning the Vectran tie down cable, ensuring an appropriate free length of nichrome wire and proper placement of the burn wire release mechanism will help to avoid the threat of the springs losing preload. Conducting pull tests of the nichrome wire and verifying the connection under a stereo microscope after securing it between the ring terminals and the screw head will verify that the nichrome wire connection is reliable and the possibility of the nichrome wire coming unattached can be avoided. When installing the release mechanism it is vital to verify that ample play is given for the electrical wiring of the flying leads to the mechanism and that snag hazards of the wiring are avoided. If the electrical wiring does become snagged or runs out of available room to move, the small force associated with the burn wire release mechanism will likely not be sufficient to compensate for a snag and the release mechanism would fail to cut through the tie down cable. Improvements could be made to the burn wire release mechanism to lower the reliance on workmanship and add confidence and reliability to the mechanism. To minimize the risk of the flying leads being a snag hazard, one could investigate adding a jumper wire to the mechanism with a connection pad that would have all of the moving wires on contained to the mechanism itself and then the electrical leads could be mounted to the stationary portion of the mechanism. This would mandate a design change to the saddles and add slight complexity to the release mechanism but would help to alleviate the risk of failed release due to the electrical wiring to the mechanism. Another improvement could be a new way to mechanically connect the nichrome wire to the release mechanism instead of using the preload and clamping force of

Document

AMS_2012.pdf

488 the screw to hold the nichrome wire. While properly securing the nichrome wire and verification of this connection has proven successful in testing to date, the concern still exists that the nichrome wire could slip out from under the screw head and increase the nichrome wire free length which would in turn allow the compression springs to relax and spring preload could be lost. Conclusions The burn wire release mechanism was designed in order to be used by a broad range of small satellite users in a manner that was simple, reliable, and inexpensive. Using a nichrome burn wire design the release device is capable of cutting through various tie down cable materials and providing successful satellite mechanism deployments. The burn wire release mechanism specifications were chosen in order to be user friendly in both air and vacuum environments and to give the user the ability to implement a release mechanism which has been previously tested and will work in a number of applications. Drawings and assembly procedures have been produced in order to allow any small satellite user to conveniently reproduce the nichrome burn wire release mechanism for their specific application. The nichrome burn wire release mechanism will hopefully allow for a more universal release mechanism which can be reliably used in a wide array of applications for CubeSat and other small satellite users. By giving the engineer a mechanism which has proven reliability, the hope is that the cost and complexity of smaller satellite programs can be kept at a minimum. References 1. Vectran Technical Data Brochure. Kuraray America, Inc. Vectran Division. 2008 2. General Considerations for the Processing of Vectran Yarns. Kuraray America, Inc. Vectran Division. 2007 3. Fette, R.B., M.F Sovinski. Vectran Fiber Time-Dependent Behavior and Additional Static Loading Properties. NASA/TM-2001-212773. December 2004 4. ASTM B344-01. Standard Specification for Drawn or Rolled Nickel-Chromium and NickelChromium-Iron Alloys for Electrical Heating Elements. American Society for Testing and Materials. August 2010. 5. JMP 9 Statistical Discovery software. SAS. 2010

Document

AMS_2012.pdf

489 Antenna Deployment Mechanism for the Cubesat Xatcobeo. Lessons, Evolution and Final Design José Antonio Vilán Vilán*, Miguel López Estévez* and Fernando Aguado Agelet* Abstract This paper aims to explain the whole development process that has been undertaken over the past few years including the problems encountered and the solutions adopted in what was to become the final design of an antenna deployment system for the Xatcobeo picosatellite. This mechanism was first presented at the 40 th Aerospace Mechanism Symposia. At that time, we presented an evolved version of our antenna deployment mechanism, with many conclusions and lesson learned. However, only now can we present the final version of this mechanism with all the development problems, solutions and conclusions reached as well as the lessons learned over 3 years of work. Introduction In the paper published for the 40 th AMS we explained the mechanism's general operation. This included how the antenna retention system, the burning system, and the opening detection all worked. Although, in general terms, none of these features has been changed to any great extent, they are all different in some way from what we described then. This is due to the many problems we came across and the way their solutions have led to changes in all the device's parts. Modifications This article will analyze in detail all the changes the design has undergone throughout the development period. Namely:  The dimensions of some of the parts on the main piece (sub-chassis)  The sub-chassis design  New attachments  The materials used  The retention wire burning circuit. Two resistance types  Retention wire tying  Power Voltage  Thermo-vacuum test Also, totally new features will be explained that have been developed since the first article, such as:  Electrical interface  Mechanical interface  Design of the RBF. PEM nut  New part for resistance support  Thermal interface between the sub-chassis, the new support part, the resistance and the shielding panel  Calculations for avoiding the appearance of ESD following ECSS standards  Compliance with the standard for outgassing values  Final integration process and envelopes. Problems with dimensions and assembly.  Modification of the margin with the enveloping film  Qualification Tests * University of Vigo, Vigo, Spain

Document

AMS_2012.pdf

490 Mechanism Background This deployment mechanism is based on one main part, called the sub-chassis, which is designed to be built from a polymer using fast prototyping. All the elements needed for the mechanism's operation are installed onto this piece. This includes the antenna enveloping film, the burning resistance, and the opening detection switch. The sub-chassis can itself be installed in the shielding panel on one side of the Cubesat, which means that everything is easily inserted in a modular way. Operation is very simple. The retention film is attached to the sub-chassis and allows the antennas to be folded and retained against the edge of the sub-chassis. Once the assembly is closed, a nylon wire passes through both the film and the antennas, and it is this that keeps everything folded away until the time for burning. At that moment, the burning resistance will cut the retention wire and the antennas will be deployed, leaving the retention film joined to the sub-chassis unable to detach itself. Figure 1. Final Deployment Mechanism General view of the Mechanism General operation is, as mentioned above, basically the same as before. The main difference lies in the fact that the enveloping film, instead of being fixed to the sub-chassis together with one of the antennas with which it was in direct contact, is now attached independently onto the sub-chassis. This way avoids problems related to the noise that could be produced on the antenna, but which requires a new nut and bolt to be included. Furthermore, the enveloping film attachment moves from point S on the sub-chassis to point E. This is due to the fact that if it were attached at point S, the end of the antenna attached at point E would end beyond the retention point. The pressure from the retention wire would thus lead to this unpressed end opening towards the outside in such a way that it would exceed the limit given by the available envelope. In compensation, the film is roughly 80 mm longer. It needs to be perforated twice for the retention wire to pass through, which means its capacity for retention is much greater precisely because it passes twice over the retention point and none of the antenna's parts are left with no pressure against the sub-chassis. Sub-chassis The sub-chassis, as stated above is the cornerstone of this mechanism. Its special features allow it to be extremely light and act as the piece that enables the installation of all the other mechanism elements. This means it is of vital importance for endowing the assembly with modularity and simplicity. EW N S

Document

AMS_2012.pdf

491 Material and Manufacture One of the main changes in this piece involves the material chosen to make it. In the previous article we had chosen polyamide 6 as the material and Selective Laser Sintering as the manufacturing method. During the development of this piece research into the properties of various polyamides led us to opt for polyamide 12 with fiberglass reinforcement. There were several reasons for this:  The first reason is that PA12 is lighter than PA6 by about 10%, and absorbs less water. This was a step in the right direction which was to reduce the weight of the whole assembly as much as possible.  The second is that the prototypes manufactured using SLS on PA12 with Fiberglass reinforcement were more precise than those made from PA6 with Fiberglass. This favored the correlation between the CAD model and the actual model. Bearing in mind that the attachment of the various parts onto the sub-chassis was based on a tight fit, it was very important to suitably meet this correlation. Moreover, this was also a way of improving repeated similarity between the different prototypes that were commissioned. Figure 2. Sub-chassis detail. E point Figure 3. Burning receptacle. W point Design and Dimensions The changes made in the sub-chassis were aimed at solving problems arising during the development stage. Below we outline both the problems and the solutions in the design:  Burning receptacle (resistance, switch, wire, tying system, increase of material in the area in order to insulate the resistance as much as possible and make the parts that are heat sensitive as massive as possible).  New attachment for the film (equilibrium of the part, improvement in retention)  Modification of the tabs (longer at the corners and narrower)  The piece was reduced in width from 6.5 mm to 6.3 mm Retention and Burning System The previously published version of the system for burning and cutting the retention wire explained that a 0.125-W 9- Ω resistance had been chosen. The parameters used for dimensioning this resistance were early ones and the voltage for the power was later modified to 3.3 V. Once this value was accepted as the one which would produce the burn, a re-dimensioning of the resistance was begun to maintain the dimensions and nominal value of the power, i.e., the sought after resistance would be 0.125 W. Test In order to select the correct resistance from those available commercially, thermo-vacuum and room-temperature tests were carried out. The thermo-vacuum tests were redesigned until pressures even lower than those used earlier were achieved; a new vacuum pump was installed; the vacuum chamber and the

Document

AMS_2012.pdf

492 electrical connections were modified. This was all with the intention of achieving lower pressures, gaining greater control of temperature and eliminating losses in the vacuum chamber electrical connectors. To reach the desired low temperatures, liquid nitrogen was used. To apply the liquid nitrogen indirectly, a system was designed comprising a bath of the liquid in a container within the vacuum chamber into which an oval-shaped piece of copper was dipped. While the lower part of the copper was submerged in the nitrogen, the upper part had the deployment mechanism bolted to it. The copper acted as a thermal conductor and the device's temperature was quickly reduced. Finally, a probe was installed in contact with the sub-chassis, as this is the part with the lowest heat conductivity and therefore the one that would be the slowest at changing temperature. The probe would thus provide the most reliable temperature reading for the parts that have a thermal balance inside the assembly. This is the atmosphere in which the device was enclosed and connected up in order to be fed power for the deployment to take place. Figure 4. Thermo-vacuum chamber Figure 5. Thermal image of the resistance (ºF) Besides this device, various tests were carried out on the wire burning at room temperature using a thermal camera for observation. With the combined results from thermal-vacuum and room-temperature tests, in which the same potential was burned and the same resistance used, the temperature value was found that the resistance must reach at room temperature in order to burn the retention wire in the thermal-vacuum atmosphere. This temperature is directly linked to an electrical power value needed, and it is thus simpler to select a resistance from commercially available values. Thanks to these tests we detected that the resistances, on being subjected to much greater powers than the nominal ones, underwent deterioration and variation in their properties when the current was passed through them. This was helpful when determining the point at which the resistance was capable of burning the wire in the coldest environment and not destroy itself at room temperature. For this, the deterioration was related to the variation in the current flow by maintaining the power voltage constant over time. This fact can only be due to the variation in the resistance value over time due to the deterioration through overload. The graph in Figure 6 shows the results from observation of the resistance in a thermal-vacuum test at 3.3 V over more than a minute of burning, with two seconds of stop before the opening of the antennas

Document

AMS_2012.pdf

493 Figure 6. Monitored intensity during burning thermo-vacuum test Change of direction. New Burning Voltage Once the resistance was dimensioned to burn at 3.3 V, at a specific value of 5.6 Ω, changes were made to the project from outside our design process. The voltage supply was changed from being a set figure to one somewhere in the interval between 4 and 5 V. This not only caused a delay and meant the work done so far could not be applied in the Xatcobeo project, but it also posed a new problem for the task of guaranteeing correct resistance operation, which did not now have a fixed voltage value but instead had a rather wide range of values. The challenge set by the new resistance values lay in guaranteeing that the selected resistance would burn the wire at the coldest temperature and with the lowest voltage (now 4 V), while at the same time ensuring that the resistance would not be destroyed at the highest voltage (5 V) and at the highest temperature, i.e., room temperature. Furthermore, the limitations still existed of using commercial values and consuming the least power possible. One possible consideration was to select a resistance of 0.25 W, as it would then be possible to dissipate more power and, in theory, have more room for manoeuvre. Thus, two sub-chassis were designed independently, each one adapted to each resistance type. However, when the burning tests were carried out, it was observed that, the 0.25-W resistance was much more easily destroyed than the 0.125 W one contrary to what we had first thought. This meant this options had to be totally discarded and the tests became more centered on finding a suitable resistance at 0.125 W. Figure 7. Burning thermo-vacuum test Figure 8. Destroying a 0.125W resistance Thermo-vacuum tests were carried out at -100ºC using several resistance values to determine the wire burning. The burning tests consisted of a three-minute continual burn in order to check both that no damage was caused to the resistance and also the burning time at 4 V and 5 V. Observation was also made of the current variation produced by the deterioration of the resistance. In this way all possible effects were checked and operation was guaranteed in the face of much higher time values than the real

Document

AMS_2012.pdf

494 ones for the burning cycle, which were set at 20 seconds by these tests and which double the burning time in the worst possible conditions. Furthermore, numerous tests were carried out of the burning in a vacuum, at room temperature, and at the highest voltage in the range, in order to guarantee that the worst conditions for thermal dissipation, the highest temperature and the maximum power value did not destroy or serious damage the resistance. Burning was also for a continuous three-minute period, and some tests were even carried out at up to 15 minutes non-stop without destroying the resistance, although there was some relatively major deterioration after this time. All the testing and dimensioning work led to a resistance of 6.8 Ω and 0.125 W being selected, thanks to which burning was produced in 2-3 seconds at room temperature and 6-9 seconds at -70ºC. Retention system. Remodelling In the 2010 article we described the retention system as a film joined to the sub-chassis and sharing an attachment point with one of the antennas, namely antenna S. The material chosen was a polyvinyl acetate, which had the optimum mechanical properties but not the best electrical ones as its surface resistivity was too high and so posed an ESD risk. This was one of the main problems, along with the one mentioned above concerning the attachment of the film to antenna S, which had one serious inconvenience in that the antenna attached at point E is not completely pressed against the sub-chassis as it finishes its run along the side several centimetres beyond the burning receptacle, where the retention knot is tied. The local pressure applied by the knot deformed the antenna and made its end cross over the limit of the envelope allowed on the deployment device and go beyond the Cubesat rail. There was a third problem linked directly to the film attachment. This time with the antenna's shared attachment. The radio-communication and IT engineering group discovered that the film, on being in direct contact with the antenna and unearthed, would introduce a great deal of noise into the signal, which meant measures had to be taken and changes made. To solve the first of these problems we decided a change of material was needed. So we began to look for a polymer that would keep the lightness but also guarantee suitable antenna insulation, be dissipative, and, furthermore, have mechanical and thermal properties that were similar to the excellent ones of the polyvinyl acetate. We chose a commercially available polymeric film and carried out thermo-vacuum deployment tests with it. The results were excellent, for, despite being thinner than the acetate, it complied with the mechanical needs for correct antenna retention operation. In order to ensure correct surface resistivity, and even though we knew the material was catalogued as being anti-ESD, we decided to make the calculation needed to demonstrate that, in effect, the chosen film met the electrical requirements. We decided to use document ECSS as a reference, which states that in the worst case it should bear an electric potential of 100V between extreme and basis. Also the current density through the film should not exceed 10 nA/cm 2 in order to avoid ESD risk. Independent attachment of the enveloping film The independent attachment of the retention film is one of the greatest changes the sub-chassis underwent due to the above. For this there was a possibility of creating a new fixing point in any free area of the sub-chassis. Many options were considered, such as locating the attachment point in the free area just below the burning receptacle on the sub-chassis corners or on a prolongation of one of the antenna attachments.

Document

AMS_2012.pdf