Datasets:

ai-aerospace

/

ams_data_full_2000-2020

like 0

87 Development of a Family of Resettable Hold -Down and Release Actuators Based on SMA Technology and Qualification of Different Application Systems Marcelo Collado(1), Cayetano Rivera(1), Javier Inés(1), José M. San Juan(2), Charlie Yeates(3), Michael Anderson(3), Francisco Javier Rivas(4), Mónica Iriarte(5), Jens Steppan(6), Calem Whiting(7) and Karine Murray(7) Abstract The present work summarizes the development of a complete family of resettable Hold Down and Release Actuator s (HDRA), ca lled REACT, based on Shape Memory Alloys (SMA), from its conception to its qualification as component and subsequent integrat ion in several final users’ systems. The paper details different topics involved in this development, including technological devel opment s, design conceptions of the actuators, their unit qualification and their integration and qualification in different subsystems. Introduction The work presented in this paper was performed in the frame of an European project were a consortium of seven partners collaborated in the development of a family of HDRAs for general use. The partnership covered the definition of requirements for a variety of typical applications used by system integrators , the development and assessment of the required technologies, such as SMAs and the tribology of release mechanisms , the development up to qualification of the actuator and the qualification of several systems integrating the new device. The resulting REACT actuators have several advantages with respect to existing alternatives: • Easily resettable by manual operation through the separation plane without disassembly from the system, enabling the test of the final units before flight. An important reduction in Assembly Integration and Test (AIT) costs at the system level is enabled bre eliminating disassembly and refurbishment operations. • Low-shock compared to other technologies. • Non- explosive devices, therefore they do not require special safety and security measurements during installation and handling. • Wide range of operating temperatures. Two different temperature options are offered to customers: standard temperature and ex tended temperature. Development of REACT family The new design started from Arquimea’s experience with the design of a previous HDRA, which was completely redesigned to overcome the limitations found in the first version and optimize the final performanc as detailed in [1]. Three different models were developed: 5kN, 15kN and 35kN, in two different variants: standard operation temperature ( -90ºC to +65ºC), based on NiTi , and extended temperature (-90ºC to +120ºC), motorized by SMARQ®, which was evaluate d in previous ESA activities [2]. (1) Arquimea Ingeniería, Leganés, Spain; mcollado@arquimea.com (2) University of the Basque Country UPV/EHU, Leioa, Spain (3) ESTL | ESR Technology Ltd, , Warrington, United Kingdom (4) Airbus Defence and Space, Madrid, Spain (5) AVS - Added Value Solutions, Elgoibar, Spain (6) SpaceTech GmbH, Immenstaad, Germany (7) Surrey Satellite Technology Limited, Guildford, United Kingdom Proceedings of the 45th Aerospace Mechanisms Symposium, NASA Johnson Space Center, 2020

Document

AMS_2020.pdf

88 The REACT design employs a segmented nut configuration, where the load is connected with a standard threaded interface. The nut is split in 3 segments. The mechanism is based on an over -center configuration. The thread segments are connected by link bars to a third element, called the external ring, which has a single degree of freedom in the longitudinal axis. The external ring is blocked in the preload or reset configuration by a set of balls, maintaining the nut segment s closed and therefore supporting the external load. Once a trigger element, motorized by means of SMA initiators, unblocks the balls, the external ring is moved to a second position by internal springs. At this position, the nut segments are open and the load is released. The new mechanism allows its reset by simple manual operations from the interface plane, by returning the external ring to the reset position with a simple tool. This reduces the complexity of AIT operations, reducing costs and improving system operations. Figure 1. REACT QM Models. Different sizes and mechanical interfaces. Development of Shape Memory Alloy Technologies for High Temperature Environments REACT actuator uses a trigger system based on a SMA wire, which at the environment temperature is in the low temperature phase called martensite. When heat ed, this SMA fibre contracts by shape memory effect once its transformation to the high temperature phase (austenite) is reached. During this contraction, the SMA fibre pro duces a force high enough to trigger the mechanism. Two different SMA can be used in REACT. A commercial NiTi alloy is used for operation temperatures below 65ºC. Another High Temperature Cu- based SMA, SMARQ, is mounted in REACT allowing operating temper atures up to 125ºC because it has actuation temperatures over 140ºC. In Figure 2a, the thermal transformation of the SMARQ alloy is plotted, showing the small thermal hysteresis between cooling (direct transformation, in blue) and heating (reverse transfor mation, in red), obtained by integration of differential scanning calorimetry measurements. The last part of the direct thermal transformation during cooling extends along a broad temperature range, because the martensite plates have to nucleate against the internal stresses created during the progress of the transformation. However, it is enough to apply an external stress to promote the nucleation of the oriented martensites, getting a large transformation strain. In Figure 2b, a transformation strain close to 8% is measured for a stress of 23 MPa. This curve represents the thermal transformation under a constant load and is representative of the behavior of the material in a situation similar to the working condition of REACT. This kind of experiments wer e performed in a specific test setup working in an ultra-high vacuum (UHV) of 2∙ 10-8 mbar [1], and many measurements were performed at different temperatures in such an UHV environment, showing that the SMARQ material exhibits a good reliability in the space environment.

Document

AMS_2020.pdf

89 Figure 2. Behavior of SMARQ fibres used in the REACT extended temperature range. (a) Thermal transformation cycle without load. (b) Thermal transformation cycle under 23 MPa load. Two examples of SMA fibre activation under UHV are presented in Figure 3, for two different environment al conditions and activation parameters. In Figure 3a, the fibre was under UHV inside a cryostat at -110ºC, working against a load or 62 MPa, and in this case the electrical current on the heater surround ing the SMA fibre was a pulse of 3 A with an increasing ramp of 0.5 s, a constant current time of 6 s and a decreasing ramp of 0.5 s (curve red). The delay of the response of the triggering SMA fibre is rather long (about 5 s) because of the very low initi al temperature of the SMA fibre, requiring to be heated above 125ºC before actuation. However, once the actuation starts, the complete strain contraction of 3.5 % (limited by a sensor) was accomplished in 1 s (curve blue). In the second example of Figure 3b, the SMA fibre was in UHV at 125ºC working against the same load of 62 MPa. Nevertheless, in this case, although the current pulse sent to the heater has the same profile than before, a lower maximum current is used, 1.5 A. The delay to start the respons e was shorter, as expected from the higher environment temperature, but the actuation time was longer, as a consequence of the lower heating current. In both cases (a) and (b), four cycles are plotted superimposed to show the reproducibility of the actuati on process. The parameters of the activation current pulse can be easily tunned in order to get the response required by the customer. Figure 3. Plots of the activation current pulse (red), and strain- time response of the SMARQ fibres, measured in UHV of 2 ∙ 10-8 mbar. (a) Cryogenic test at - 110ºC. (b) High- temperature test at 125ºC.

Document

AMS_2020.pdf

90 Analysis of L ubrication A set of tribological test activities were performed at ESTL with the following objectives: • to carry out testing at a fundamental (pin on disc) level to evaluate and compare candidate lubricants. • to carry out testing on sub- assemblies equivalent to t he Qualification Models. Five lubricants were applied to EN31 steel disks and Ti -alloy disks and they were evaluated in the pin- ondisk tests. Testing was carried out in air and in vacuum: • Sputtered MoS 2 – applied by ESTL using PVD process. • Everlube 620C – bonded lubricant applied by Arquimea. • Molykote D321R - bonded lubricant applied by Arquimea. • Molykote D106 – water -based bonded lubricant applied by Arquimea (note that this is a different product from Molykote 106). • X54 – PTFE based lubrican t. The test program was completed and the results are summarised in Figure 4. Table 1. Pin on D isc Test Conditions Motion type Reciprocating Motion cycle static, +5mm/s, static, - 5mm/s (TBC) Stroke length 10 mm Number of cycles 250 in air, followed by 250 in vacuum Test Temperature Laboratory ambient Figure 4. Pin -on-Disk Results on EN310 S teel and Ti -Alloy Based upon the results from this tribometer test program together with existing flight heritage and test data for Molykote 106 (solvent -based, butyl alcohol), it was agreed that this lubricant was the preferred candidate for the REACT application and it was therefore selected. Three REACT units were tested, one of each size, defined as 5kN, 15kN and 35kN units and equivalent to the Qualification Models. The assemblies were tested under the following conditions: 1. Air at ambient temperature (ISO 7 Class Cleanroom: 55% ± 10% RH, 22°C ± 3°C). 2. Vacuum (target 1x10- 5 mbar or less) at ambient temperature (22°C ± 3°C). 3. Vacuum (target 1x10- 5 mbar or less) at 120°C. 4. Vacuum (target 1x10- 5 mbar or less) at - 90°C. Each test comprised a single actuation driven by a stepper motor and the actuation torque was measured. The results are provided in Figure 5.

Document

AMS_2020.pdf

91 The three units, 5kN, 15kN and 35kN, all actuated successfully at each of the test temperatures. The results indicated that the temperature and environment did not have a significant effect on the actuation torque. A post-test examination revealed that although the Molykote 106 lubricant was worn, it was still present and protecting the critical sliding surfaces (examples are illustrated in Figure 6). Figure 5. Actuation Torques Measured in the T hermal Vacuum Tests Figure 6. Comparison of S liding Interface with S teel Ball Before (left) and After Testing (right)

Document

AMS_2020.pdf

92 REACT Qualification C ampaign A complete qualification campaign has been defined for each model and temperature variant, for a total of 6 campaigns. Eac h campaign includes 3 different units covering different mechanical interfaces for the customers. Figure 7 summarizes the qualification campaign sequence for each variant. Figure 7. REACT Qualification Campaign The qualification campaigns of the REACT 5kN and 15kN Standard Temperature units have already been completed, while the rest of campaigns are on- going. The main results and findings are: • The units are being qualified to a lifetime over 50 cycles, which enables for at least 10 cycles on ground and one in orbit, according to ECSS. Extension to 70 cycles during qualification to enable up to 15 on ground uses is being implemented on new campaigns. REACT 15kN Standard temperature has already completed its qualification for 70 cycles. • All the units withst ood their proof load (6.5 kN, 19 kN and 43.5 kN, respectively) with high stiffness and were able to release even under such preloads. Axial stress tests have shown that the strength of each unit are well above the specification values . REACT 5kN has shown a strength over 18 kN, where the test resulted in external bolt fracture without any sign of damage to the device, which was still showing normal behavior actuating in nominal conditions. The REACT 15kN test r esulted in the breakage of the external bolt at 45 kN and the device show ed nominal behavior after the test. In the case of REACT 35kN, a maximum preload of 100 kN was reached, at which point the interface between the bolt and the tensile machine was broken. The unit was still fully operative after such load. These values show the robustness of the design and a high confidence in the device mechanical strength due to the application of design margins. • Vibration and shock environment s were succesfully applied to all the models, show ing that the device holds the rated load without issues. • The units were subjected to 6 thermal cycles over the operating range, followed by actuations at low and high temperature extremes . The devices showed good behavior during the cycling and actuated successfully. • Misalignment and creep tests were performed to characterize the devices . The results during misalignment tests show ed that the units were able to accomodate half-cone angles of more than 4.5º during integration and release. Creep tests were run for more than 6 months succesfully, showing very low initial preload loss followed by stabilization . Nominal actuation has been achieved after the test. Creep test s are being extended to 12 months. • Force margin tests have been completed by assembli ng SMA initiators with a fraction of the nominal cross- section, and therefore a fraction of the available output force. This way, the force margins of

Document

AMS_2020.pdf

93 the real devices can be obtained. These tests were completed in different environmental conditions for each model: ambient conditions, minimum temperature with air (worst case for the lubricant) and minimum and maximum temperature in vacuum. The models showed margins above 6X in all cases, as required by ECSS [4], with some models showing more than 11 X margin. This value shows the reliability of the device to operate under worst conditions. Figure 8. REACT Qualification – Left: Vibration test. Right: Missalignment test Launcher Separation System For the qualification of REACT as the release device for an Airbus DS separation system, a fully flight representative clamp- band type LPSS937* was chosen as the test item. The LPSS* Separation System is the Airbus DS solution for clamp- bands where low shock and high load carrying capability are essential. Airbus DS is a provider of Separation Systems in the European market for Ariane- 5, Soyuz and VEGA launchers. Also, in the American market , Airbus DS provides Separation S ystems for Space- X and Orbital - ATK. Finally, in the Japanese market , Airbus DS clamp- bands fly in HIIA and will soon fly in HIII launchers. Figure 9. Airbus DS LPSS* Separation Systems (left). Endurance results during LPSS937* qualification. To enable the use of REACT on LPSS*, some mounting/ adaptation hardware was designed and manufactured. Later, the LPSS937* clamp- band, employing a REACT 15kN Extended Temperature model, were subjected to the standard qualification test campaign applied to all Airbus DS flight separation systems. The test campaign began with a functional release test, to verify that the REACT enabled proper performance of the clamp -band. Although the functionality of the system for minimum, nominal and maximum band tensions was correct, an unexpected hooking of the “key” was detected. The “key” is a part that trasmits the tension load of the clamp band to the REACT through a shock reduction mechanism. The

Document

AMS_2020.pdf

94 main outcome of these tests was the need for a device to eject the key faster than the movement of the shock reduction mechanism. This device could be a simple pull -out spring. The next step was the life test campaign consisting of ten tensioning/release tests to demonstrate that the clamp- band will survive to launch after ten releases. The main outcome of this campaign was the confirmation of the hooking issue (Figure 10 left). The LPSS937* plus REACT assembly was subjected to a random vibration level of 20- G RMS showing a stable and adequate dynamic behavior without high amplification. After the successful environmental test, the clamp -band was released to demonstrate the survival of all its components including the REACT release device. Figure 10. LPSS93 7* qualification setups using REACT. Vibration test (left). Thermal Vacuum Test (center). REACT detail connected to the clamp- band (right). The thermal qualification of the LPSS937* plus REACT assembly was performed in vacuum conditions. The assembly was submitted to four cycles between LPSS* extreme cold and hot qualification temperatures of -55°C to 120°C. Functional releases were performed both at cold and hot conditions demonstrating the correct function of the REACT as the LPSS* release device. Finally, the clamp- band was tensioned to its nominal band tension and this parameter was measured and recorded for more than one month to determine its relaxation. The observed behavior was normal , with a relaxation of just 0.3 kN in the mentioned time period. Conclusion • Tensioning operation torque/preload relationship and the repeatability of this parameter in consecutive operations is similar using REACT to the release devices previously used on LPSS* . • It is necessary to implement an extractor for the key bolt (bolt retained by REACT) either installed in REACT body or as part of the LPSSS937* mechanism. The lack of this extractor creates an incompatibility between the velocity of the key bolt driven by the opening of the band mechanism and the movement of REACT fixed to the clamp- band main- beam. The result of this incompatibility is an unexpected hooking between two pieces that causes an off -nominal band release. • Main outcome of this work is that, although there are still a set of activities to be done prior to the conclusion of formal REACT qualification for Airbus DS clamp- bands, it is clear that the device is a n option, at least for a given type of missions wher e the activation time of several seconds is acceptable. Planetary Exploration Subsystem Added Value Solutions (AVS) participated in the development and qualification of a Hold Down and Release Mechanism for a planetary exploration subsystem. AVS designed a HDRM based on the REACT 5kN to hold and release the Sample Canister of an interplanetary sampling tool. The “Sampling Tool Mechanism for Low Gravity Bodies” developed by AVS collects a minimum of 100 g of regolith from the surface of a

Document

AMS_2020.pdf

95 low gravity celes tial body for later transfer to the next step in the sampling chain. Therefore, once the regolith has been sampled successfully, it is necessary to perform two disengagement sequences: • The first disengagement releases some parts of the mechanism leaving the canister exposed. For doing this three HDRM's located 120˚ apart are needed. • The second disengagement releases the sample canister and here is where the new REACT Hold Down and Release mechanism (HDRM) design is considered. Figure 11. Sampling Tool Mechanism (left), 1st Disengagement (cent er) and 2nd D isengagement (right) Two models were developed in order to qualify the 5kN REACT HDRM: an Engineering Model ( EM) to test the functionality of the design and a Qualification Model ( QM), which was subjected to a qualification campaign. The functional test campaign was successful as all the requirements were met during the execution of the complete series of measurements. The HDRM EM released every time after the REACT EM 5kN actuator fired. The onl y issue encountered was related to the resetting procedure of the actuator, but they were solved for the QM phase. The results of the qualification tests are explained hereafter. A total of 22 releases were carried out, two of them in vacuum (10-6 kPa) an d at a temperature of +120°C and -90°C. All of the deployments were successful. The HDRM survived to 25- G quasi -static acceleration along all axes, as well as to 31- G RMS during the random- vibration testing. Low -level sine-sweeps were performed before and after each random- vibe test. The pre- and post - sine sweeps were compared to verify that there were no eigenfrequency shifts greater than the predetermined tolerance of 5%.

Document

AMS_2020.pdf

96 Figure 12. Vibration Y Direction (lef t). Shock Test Setup (right) Figure 13. Sine Survey Comparison ( Before and After 2 S hock Tests) at SA1 Channel (left). Test Setup for the Thermal Vacuum Cycling (center) and Cycles C arried Out During the TVac (right) Conclusion In conclusion, it can be stated that the 5kN REACT successfully passed the functional and qualification campaign. The issues with the resetting procedure that occurred during EM testing were resolved prior to starting the qualification campaign. The QM testing vallidated thes e changes and this time the reset s were simple and repeatable. Large Structure Deployment System As part of the H2020 REACT project, SpaceTech GmbH (STI) verified the use of the new release mechanism for the separation of large solar generators, as typi cally used in space travel. A wide- ranging test campaign examined the possible uses for this actuator. The main objective was the qualification of the REACT actuator, preferably for multiple deployment jig designs for STI projects. The tests were performed using a 15-kN series actuator (QM). The basis for the physical design of the actuator mounting to large deployment structures were the interface points, for example as in the Sentinel S5P -project. Adaptat ors and test benches (GSE’s) for all mechanical tes ts were designed (CAD), analyzed (FEM) and manufactured by SpaceTech GmbH accordingly. After the test series was carried out, the following was determined for the respective tests: Electrical Test: The REACT actuator (QM) has two redundant electrical connections in case of failure of one circuit. The resistanc e of the nominal and redundant lines were measured. Both circuits indicated the required values. Additionally, isolation tests were performed to ensure that the conductive lines and the mechanical structural parts maintained electrical isolation. The insulation measurement revealed full isolation.

Document

AMS_2020.pdf

97 Functional Test: The REACT actuator (QM) was connected to a standard power supply (40V/2.5A). Aft er approximately 4 seconds, the HDRM was released successfully. The actuator reset carried out after the release performed as expected without issue. Worst Case In- Plane Release Test: The deployment jig was mounted with a gap on both main hinges to creat e an angle offset between the screw and inner threads of the actuator. This was done via a practical simulation of a solar panel in- plane shift by replacing the positions from the actuator to the HDRM. After assembly, the actuator release was successful with an actuation time of 4 seconds . Vibration Test: In order to show that the actuator can withstand the mechanical qualification loads without damage, frequency shift, damping change etc., the actuator was subjected to sine and random vibration loads. As a result, it can be stated that no damage or debris occurred (visual inspection), no amplitude shifts greater than 20% and no frequency shifts greater than 10% were found. T he resistance measurement (nominal/redundant) after vibration indicated no change. Release Shock Test (after vibration): To determine the release shock level at the interfaces (Spacecraft and Photovoltaic assembly simulated interfaces) and the substructures, release shock tests were carried out immediately after the vibration test (examined frequency range: 100 Hz to 10 kHz). During the first attempt to measure the actuator release shock level, the release was not completed. After investigation, it was confirmed that the screw thread engament exceeded the maximum allowed by the actuator. After an incomplete release of the actuator, the actuator was newly pre- tensioned. Special attention was paid to the thread depth control before the preload was initiated. The second attempt to release was successful. Further releases were carried out. There were no more malfunctions. The redundant cir cuit was used to trigger the release. It should be noted at this point that the Engineering Model (EM) at SpaceTech GmbH is in "continuous use". Of the approximately 150 releases, about 80 were triggered electrically and the other 70 were triggered mechani cally. In these tests, the shock accelerations were measured only sporadically. Life Cycle Release Test: To guarantee that the actuator can safely and reliably execute several releases in succession with no negative impact on the assembly performance, the actuator was pre- tensioned and released several times. Thermal Cycling Test: During the thermal vacuum test, the actuator showed no noticeable problems. After completion of the thermal cycle test, it should be noted that a n undersired (thermally -induced) self-release did not take place and the preload force was fully maintained. The visual inspection was carried out and showed no visible changes. The subsequently performed release test could be carried out without any problems. Hot / Cold Self-Release Test: The actuator was subjected to thermal cycling with no self -actuation, and released successfully after thermal cycling when the electrical pulse was applied. Long Storage T est: After a storage period of several weeks, no self -release and no signifi cant loss of preload was observed.

Document

AMS_2020.pdf

98 Figure 14. REACT Actuator (QM) T est Setup Vibration Test, Sensor Positions Conclusion The initial problems lay with the misunderstanding of the narrow tolerance for the screw -in depth during system assembly, which led to functional failures during the tests. This was identified and solved. Based on this, suggestions were made that should lead to an improvement in the monitoring of the actuator during the various test phases. The release shock test results indicate an exceedance of the required limits at the interface points (input), especially in the upper frequency range. Whereas on the dummy mass, the shock loads remained well below the requirement. These exceedances can be explained by the fact that the interface measurement points were located directly on the bracket for the tests performed in this project. At the output interface point s, the shock response spectrum levels are likely to be significantly reduced by the damping of the structure. The final actuator mass and dimensions are significantly higher than the actuator of other competitors, which makes it especially difficult to use the REACT as a standard deployment solution for SpaceTech at the moment. If the mass of the actuator could be reduced, it would be an extremely attractive solution compared to the products of the competition. Above all, the extremely fast -recoverable pr eload is an enormous advantage. As already mentioned above, the reliability has been tested at SpaceTech GmbH in a lifetime comprising either electrical or mechanical releases far above the qualification limits. Small Satellite Deployable Subsystem The project undertaken at SSTL tested the 35kN standard temperature variant of REACT from Arquimea. SSTL were involved from the start of the project, defining requirements necessary for use in small satellite deployable systems. Test campaign of an EM could not be completed due to the mechanism not holding the maximum preload after a certain number of tests. This was mainly due to the selection of an excessively fine thread pitch. Arquimea further refined the design after these issues and sent a Q M for SSTL to test. The qualification tests were completed and confidence was gained that the mechanism could perform consistently.

Document

AMS_2020.pdf

99 The qualification tests used REACT as the release mechanism within a solar array (HDRM). HDRMs are used by SSTL to secure deployable solar panels against the spacecraft during launch, and release once commanded in orbit. SSTL specifically adapted the HDRM design and qualification test set up to integrate the REACT actuator. Historically, many of the release actuators used by SSTL are not resettable by the end user and are must be returned to the manufacturer after one use, causing large impacts on schedules. REACT has the advantage of being easy to reset and non- explosive. The qualification test plan was driven by the requir ements set out at the start of the project. This included testing the 35- kN preload, making sure the release and reset functions were as expected, and verifying that there was no degradation in performance after experiencing simulated in- orbit environments through vibration and thermal vacuum testing. After first confirming that the actuator received met expected properties such as mass, dimensions and wire resistance, it was set up in a tensile test machine to verify it could hold the maximum preload (Figure 14 left). The actuator was fixed in the machine using bespoke MGSE which held the actuator down and pulled a load of 35 kN on a bolt fixed within the mechanism, and held for a period of time. This was repeated including application of a low no- fire cur rent of 0.8 A on primary and secondary circuits to make sure that a small accidental current would not inadvertently cause deployment. Tests showed that the actuator held the load perfectly with no reduction in load. The actuator was fitted into SSTL ’s solar array deployment test setup (Figure 15 center). The actuator was mounted within a hold down and release system incorporating a cup (spacecraft side) and cone (array side), kick- off springs to help separation, and a bolt catcher to trap the deployed bol t from the actuator. REACT was attached to the cup within the setup. The final deployment angle was set by SSTL tape spring hinges. Deployment was confirmed by an integrated micro- switch, which allowed for measurement of the mechanism ’s actuation time. The functional tests were carried out using preloads of 16, 22 and 35 kN on both primary and secondary circuits . During these tests, shock was measured with the use of two accelerometers attached to the system. Each planned functional test successfully deploy ed the panel, and shock values remained under the required limit. After the successful deployment tests, the system set up was moved to the vibration test machine and underwent low level sine tests, qualification sine tests and random vibration in all three axes, at a nominal preload of 22 kN. The system passed all vibration qualification tests. The preload on the REACT mechanism did not decrease during testing and the deployments performed post testing were successful. The final qualification tests occurred whi le attached to a thermal plate within a vacuum chamber over a period of 4 days. Initially, one full thermal cycle at vacuum was carried out, with a maximum temperature of 60°C and minimum of - 85°C, after which deployment tests occurred at a nominal preload of 22 kN at hot, cold and ambient. All thermal vacuum tests were successful with actuation time varying with temperature as expected. Ambient pressure and temperature deployment tests on primary and secondary circuits were carried out after the thermal vacuum testing to prove function had not been compromised. Both tests were eventually successful, although the primary test required an increase of voltage to the actuator to get a successful deployment. Conclusion The REACT mechanism successfully completed qualification testing whilst integrated within an SSTL HDRM for a small satellite deployable solar panel. The qualification testing demonstrated that the mechanism could perform as required during all stages of operation on ground and in simulated orbital environments. There were some occasions in which difficulty was experienced in actuating the device due to lack of margin between expected actuation time and maximum allowed pulse duration, however, with a more flight representative electrical setup, these issues would most likely be removed.

Document

AMS_2020.pdf

100 Figure 15. Proof Test (left). Function Deployment Test Setup (center). Use of Reset Tool (right). The resetting of the actuator (Figure 14 right) was easy and the feedback from the mechanism gave the operator confidence that the device had been reset properly. At no point was there any concern that the reset may not work or that the mechanism may not hold preload after a reset. Throughout the testing the REACT QM device was released 15 times (6 Pre- EVT, and 9 either during or Post-EVT), against its specified life of 10 releases on ground and 1 in orbit, which gives good confidence of the device’s life. Conclusions A whole family of HDRAs has been developed and qualified in this activity, including their quailification in systems for different applications. One of the key advantages of REACT is its resettability, which allows the users to perform integration and multiple tests in a short time, espec ially when compar ed to other existing solutions. This characteristic is especially interesting as a cost -efficient solution to improve the Manufacturing, Assembly, Integration and Test operations at system and satellite levels. Moreover, the resettability enables a test -what -you-fly and fly -what -you-test philosophy. In addition to that, the temperature range is competitive for most applications, giving different actuation options for different applications and uses (shorter actuation time and power for lower temperature environments, different load levels, mechanical interface options including compatibility with existing solutions,…). Another important advantage is its proven reliability, the high mechanical and force margins shown by the devices in the w orst environment conditions. The devices have shown successful operation under a wide range of preload levels, from 0N to levels over their nominal preload, allowing loads over proof levels to be released during the campaign. In addition to that, the devi ce showed capabilities to withstand loads much larger than nominal, maintaining functionality, which gives an additional robustness level , showing survivavility to any unexpected event. Finally, the device shows good stiffness and creep behaviour. During the work, several lessons learn ed have been identified: • A good analysis of the concept and its potential failure modes should be done as early as possible in the development. • Reducing the contact pressures in the critical friction points is very import ant. • The selection of lubricants together with a proper definition of contacts, materials and preparation of the surfaces is critical for a good and reliable space mechanism, as shown after the issues detected on the design selected at the beginning of t he activity.

Document

AMS_2020.pdf

101 • Thermoelastic effects have to be well considered from the beginning of the design, especially for devices with wide operation range. • Adaptation to typical electrical interfaces of customers is important to reach the market. • The development of many variants of a new product in parallel is really difficult to handle in terms of cost, time and risk. It would be recommended to develop a first variant and then develop the rest of them. • Testing as early as possible during the development is k ey to find potential issues and problems. • A good definition of tooling and its interface with the thermal chambers is critical to reach good thermal cycling and to save time and money during test cycles. • EGSE sometimes is as critical as the flight dev ice. The risk associated to poorly designed electrical equipment may compromise the integrity of the devices, with the consequent impact. • EGSE can be a sub- product of the activity. It is an important asset to improve the market access of REACT, reducing the costs and complexity of customers to test the new product. • It is important to tune the shock with a representative device prior to qualification, in order to avoid excessive levels during test. It is important to keep high frequency responses within limits. • It is needed to improve the clarity of the documentation shared with customers. The requirements for assembly of the unit into the system and its use (adjustment of electrical pulses for different conditions) is critical. • It is a good practice to offer guidance to customers for the adaptation of their designs to the characteristics of the product. Finally, there is a list of lessons learned coming from the users’ experiences, where collaboration between partners has been key for the identificat ion of requirements, issues and the evolution of the product: • Better definition of critical points in user manuals, to avoid any misunderstanding that may lead to handling or operation errors. These manuals were tuned after the first users experiences. • Implementation of a quick user guide to summarize the most critical points for the use of REACT. • Elaboration of white notes to guide the users in the particularities of the device and its integration in systems. • Evolution of the test setups to be offered to customers as EGSE. It will allow easier use of the devices in addition to reduce the adoption time for customers. • It is important to collaborate with customers to ensure a proper design and use of the equipment. • Improvement of key details description in ICDs. Currently, REACT qualification campaign is ongoing for some models and different milestiones for each model have been accomplished: • REACT 5kN Standard Temperature has reached flight heritage as part of eSAIL mission in 2020. • REACT 5 kN Extended Temperature has also reached flight heritage in two missions in 2019. • REACT 15kN Standard Temperature has succesfully completed its qualification. • REACT 15kN Extended Temperature model is in the middle of qualification campaign and has already been selected for several flight missions. • REACT 35kN Standard and Extended Temperature are running their qualification campaigns. Acknowledgement The project showed in this work has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 640241 (project REACT – Resettable Hold- Down and Release ACTuator).

Document

AMS_2020.pdf

102 References 1. Collado, M., Rivera, C., Inés, J., Sánchez, J. " Evolution of a resettable Hold- Down and Release Actuator Based on SMA Technology ." 18th European Space Mechanisms and Tribology Symposium, 2019 2. Collado, M., Cabás, R., López -Ferreño, I., San Juan, J. " Functional Characterization of a Novel Shape Memory Alloy. " Journal of Materials Engineering and Performance, Volume 23, Issue 7 (July 2014), 2321- 2326; DOI: 10.1007/s11665- 014-1104- 7. 3. I. López -Ferreño, U. Urrutia, P. Lorenzo, M. Collado, C. Rivera, N. Escudero, T. Breczewski, M.L. Nó, J. San Juan, “Ultra- High-Vacuum experimental equipment to characterize shap e memory alloys for space applications”. Materials Today: Proceedings 2S (2015) S953- S956. 4. ECSS -E-ST-33-01C - Mechanisms.

Document

AMS_2020.pdf

103 Development and Post -testing Anoma lies of the Parker Solar Probe Clamshells Mark Bryant* Abstract The Johns Hopkins University Applied Physics Laboratory built and flew a novel deployable antenna containment mechanism specifically for use on Parker Solar Probe. These mechanisms came to be known as “the clamshells”; a moniker earned for a deployment that was reminiscent to the movement of their ocean- bound namesake. This document describes the fundamental design and development efforts associated with th ese mechanisms . Furthermore, it will delve into to the subsequent investigation of the deployment anomaly that that occurred during protoflight testing and resulted from dimensionally noncompliant parts and effects of titanium -to-titanium galling. Introduction The NASA Parker Solar Probe (PSP) spacecraft, built by the Johns Hopkins University Applied Physics Laboratory (JHU/ APL) , was launched from Cape Canaveral, FL on August 12, 2018. The spacecraft, shown partially in Figure 1, will provide new data on solar activity and make critical contributions to our ability to forecast major space weather ev ents that impact life on Earth1. During the development of PSP , APL conceived, fabricated, tested and integrated a unique, partially deployable antenna containment system for launch and testing environments . These mechanisms were successfully deployed two days after the launch of Parker Solar Probe. Formally referred as the Antenna Retention S ystem (ARS) , this late-stage development effort served the purpose of attenuating the anticipated excursions of the ~2100 mm ( ~82 in) niobium V1, V2, V3 and V 4 FIELDS instrument antennas (provided by the Space Sciences Laboratory at the University of California, Berkeley ) during spacecraft (SC) testing and launch. A result of the flexible nature of the antennas, their stowed mounting location, and the fact they spanned multiple SC interfaces, containment was necessary to prevent damage to other SC components. The ARS system , as depicted in F igure 1, is comprised of a one- time deployable , SC structure- mounted clamshell mechanism, and a series of static fork-shaped snubbers mounted to the thermal radiators directly adjacent to the instrument hinge. The deployment of the clamshell mechanism is completely independent of the FIELDS deployment hinge; the former is deployed first, enabling the latter to individually deploy each antenna from the stowed launch position to the final, radially extended configuration beyond the umbra of the Thermal Protection System. Each of the four FIELDS antennas have a corresponding clamshell mechanism, and a set of carefully aligned forks adjusted to the natural shape of the stowed antenna. The clamshells at the FIELDS V1 and V2 locations shared similar mounting bracket geometry, V3 and V4 mounting brackets were completely unique to the others ( Figure 2) . Though the mounting brackets varied, the clamshell mechanism at each location were identical. * Johns Hopkins University Applied Physics Lab, Laurel, MD ; mark.bryant@jhuapl.edu Proceedings of the 45th Aerospace Mechanisms Symposium, NA SA Johnson Space Center, 2020

Document

AMS_2020.pdf

104 Figure 1: PSP Side view: The primary components of the Antenna Retent ion System (ARS) and FIELDS instrument on Parker Solar Probe Figure 2: The three variations of the clamshell mechanism mounts for 4 different mounting locations Clamshell Mechanism Design The PSP clamshel l mechanism is comprised of two, half -circular machined aluminum ha lves, one deployable or active hinge, two follower hinges, redundant deployment telltale switches , and plastic antenna guides that are mounted to the top end of the clamshell. The clamshell halves are 762- mm (30- in) long, have a 22- mm (.875- in) external diameter, and are fabricated from 6061 aluminum, with aggressive mass - reduction features on the inside. The clamshell mounts to the PSP SC bus via features on the top and bottom hinges, on brackets that position each clamshell on a natural tangent of each stowed antenna. These brackets were designed with multiple bolted interfaces to provide multiple degrees of in-plane angular and translational adjustability to ensure proper in-situ alignment with each stowed antenna. Clamshell Mechanism Containment Forks FIELDS Antenna FIELDS Antenna SC +Z Axis

Document

AMS_2020.pdf

105 Figure 3: Top view: The clamshell mechanism that corresponds to the V1 FIELDS antenna position. When deployed, the clamshells hinge open along the length of the tube created by the halves . They do not open much; only 15 degrees per half ( Figure 4) , but this is more than suffic ient to release the 3.2- mm (.125in) diameter antenna. They are not designed to physically clamp or restrain the antenna, but only to loosely contain it. The 6.4-mm (.25 -in) clamshell ID enables to the antenna to rattle slightly and freely translate axially , as not to impart any additional loads i nto the FIELDS instrument from the SC deflections that occur during SC vibration events . The closed side of o ne clamshell half features an “anti -trapping” tab along its length, to prevent the antenna from getting hung up or caught in the gap formed when the mechanism is deployed. Similarly, t he yellow Ultem® blocks on the end of each clamshell half ease the transition of the antenna into the end of the tube; a location d etermined to have the highest dynamic contact. Extensive material testing was performed to ensure Ultem® compatibility with niobium antennas . Figure 4: Top view: The clamshell mechanism depicted in the closed (left) and open (right) condition. The heart of the mechanism, the active hinge, utilizes a TiNi Aerospace P10 P inpuller to provide the motive force to deploy the greater mechanism. A TiNi P10 pinpuller is designed to retract 9.5 mm (.375 in) with 44 N ( 10 lbf) of pulling force2. Due to the shear load limitations of the pinpuller pin as specified by TiNi , it was necessary to isolate the vibration loads imparted by the mass of the clamshell from the pinpuller pin. This isolation is performed by a titanium slider cup, whic h is fastened to the end of the pinpuller via a loose spherical interface and engages two titanium legs built to each clamshell hinge half. When the pinpuller is actuated, the sliding cup translates along the axis of the pin, disengaging the clamshell legs, and enabling each spring- loaded clamshell half to pivot open (see Figure 5) . The spherical interface between the

Document

AMS_2020.pdf

106 Pinpuller pin and the cup prevents binding of the sliding cup in the housing by eliminating all potential over - constraining degrees of freedom. The intentional loose fit of the spherical interface ensures that no external moments or side loads can be imparted onto the pinpuller pin due to assembly tolerances or launch shifts. Only the retraction force can be imparted into the cup. Figure 5: The details of the active clamshell hinge, in stowed and deployed positions . Left: Cross -section, Right: Housing not shown The other two hinges are spring- loaded follower hinges that simply provide additional opening torque via torsion springs at each location. These hinges feature no release mechanism or lock, and the mass of the center hinge is supported entirely by the clamshell halves. The bottom hinge is designed to accommodate ±3.2 mm (.125 in) of change in the axial length of t he clamshells ( Figure 6), and t his eliminates the potential for the mechanism to bind from temperature- induced dimensional changes (CTE). Furthermore, the bottom shaft was cut with a profile of spherical undercuts, to further minimize the potenti al for binding, and provide additional angular adjustment during SC integration. Figure 6: A side view of the bottom hinge illustrates the bottom mounting bracket and the gaps designed to accommodate CTE effects of the aluminum clamshells “Slider Cup” Engagement Legs TiNi P10 Pinpuller Spherical Interface

Document

AMS_2020.pdf

107 Also a ttached to the active hinge bracket are four Honeywell 9HM1 micro switches to provid e redundant deployment telltale signals of each clamshell half. Resistors of different values were placed in- line in the harness, as to indicate “left -open”, “right -open” and “fully -open” states. A small aluminum lever arm attached directly to the clamshell engaged the switches directly when stowed (Figure 7) and disengaged only when the cl amshell reached the fully open, 15-degree state. These switches functioned perfectly upon the actual deployment in space but proved to be very difficult to set and tune on the ground. The #2 attachment screws needed to be torqued enough to prevent shifting of the switch, but not so much to crush the thin exterior case. The risk of a false telltale reading was also partial ly mitigated by analyzing the SC IMU data to show the shock of each at deployment. Figure 7: A bottom isometric view of the redundant telltale switches Materials and Coatings The mounting brackets and clamshell halves are fabricated from aluminum, and are iridite coated and electroless nickel plated, respectively. All of the mechanism parts are 6Al -4V titanium, including the active hinge housing, the sliding cup, the hinge halves and shafts. This fact becomes an important detail in the anomaly investigation to be detailed later. The decision to uniformly fabricate all mechanism parts from titanium was made to eliminate the effects of CTE mismatch withi n the tightly -fitting, high- precision mechanism components , and in some cases, for strength considerations . The two halves of the titanium spherical interface screwed onto the pinpuller are unfilled PEEK, the springs are E lgiloy ®, and all the fasteners are stainless steel. All titanium mechanism components are titanium anodized with a type 2 T iodize ® process, with local application of Dicronite ® (tungsten disulfide) dry film lubricant on all moving or sliding surfaces. No additional lubrication was used in the assembly, except on the fasteners.

Document

AMS_2020.pdf

108 Development Effort As a result of the late schedule addition of the Antenna Retention S ystem, the clamshell mechanism was developed on a very aggressive timeline. Two engineering model (EM) versions of the clamshell mechanism w ere built and tested before the flight units were fabricated. The first EM provide d a successful proof of concept of the mechanism, and the environmental testing scheme. T he second EM implemented most of the flight -like mass reduction efforts and added the telltale switch assembly . The differences between the second EM and the flight implementation were intentionally very minor, including maintaining EM2-to-flight continuity with the fabrication shops selected for each part . Each iteration of the design was fully vibration (Figure 8) and thermal vacuum ( TVAC ) tested, and dozens of ambient temperature, post -vibration and hot/cold deployments (Figure 9) were executed. Additionally, ARS system -level containment tests were conducted at three points during the PSP development effort: during the SC modal test, the full SC vibra tion test , and a specially built SC “stiffness simulator”, as shown in Figure 10. This simulator provided a test bed to perform ARS -level tests at higher vibration levels, while eliminating the risk to the actual SC bus. In each case, s urrogate FIELDS antennas were captured in EM clamshells and forks. Efficacy of the retention system was determined through high- speed video and measurements from the video of the antenna maximum excursions as observed during the modal test, and the SC stiffness sim ulator vibration tests . The flight clamshells were qualified separately and added to the SC after the complete SC vibration test, but before the complete SC TVAC test. Figure 8: V1 and Spare flight clamshells on a vibration table

Document

AMS_2020.pdf

109 Figure 9: Flight clamshell mechanisms staged for TVAC testing. With the exception of some erroneous telltale indi cations observed during the EM2 clamshell development, no anomalies occurred during mechanism- level testing. The telltale design was modified and follow up thermal and vibration testing was conducted to verify the correction. This was considered to be a reasonable step in the development process, and while additional attention was given to the telltale configuration as a result, it was not considered to be a serious anomaly. No deployment failures occurred through EM2 development, and the clamshells enjoyed a perfect record of functionality. Figure 10: Left: EM Clamshell with preliminary fork configuration. Right: EM Clamshells on structure “ stiffness simulator ”

Document

AMS_2020.pdf

110 Anoma ly Investigation Despite a successful testing campaign of the first and second EM iteration s of the clamshell mechanism, an anomaly of the V3- position flight clamshell occurred immediately following protoflight vibration testing. The unit failed to actuate; the proper v oltage was sent to the TiNi Pinpuller , a faint click was heard, but no movement was observed. The electrical ground support equipment ( EGSE) was checked, determined to be functional, and v oltage was applied to the redundant (backup) circuit of the pinpuller, yielding the same result, but with no additional audible click . The frozen clamshell was carefully photographed in this state, and a course of action was devised. It was unknown at that time if the failure was contained to only the pinpuller, or if was problem at a higher -level assembly and something in the rest of the mechanism was bound. As the mounting screws were being carefully (no shocks or bumps) removed from the top mount to the titanium hinge housing, the clamshell deployed. From this event, it was theorized that perhaps the distortions in the housing caused by the screws prevented the cup from sliding down a perfectly straight bore. This also immediately indicat ed that the problem was not isolated to the pinpuller, which was maintaining a constant 44-N (10-lbf) pull force, and that there were external effects acting on the greater mechanism. Despite the deployment, a fish scale was attached to the back of the pinpuller to measure the extraction for ce necessary to remove the sliding cup and pinpull er assembly (Figure 11) . 36 N ( 8 lbf) was measured; it should have been nearly zero. With the addition of the predicted 18-N (4 lbf) disengagement force necessary to actuate the hinge legs, the 53-N (12-lbf) combined force is outside the oper ational capabilities of the P10. Along with the extraction of the cup, dark, powder -like FOD was observed to come out as well. This FOD was collected for chemical analysis, and i t was determined t o be comprised primarily tungsten disulfide powder, with a combination of trace amounts of titanium. This was FOD produced from the mechanism itself , and not from an external source. This was not an anomaly that was caused by mishandling or a dirty environment. Figure 11: Removal of the sliding cup and pinpuller assembly, immediately after deployment anomaly

Document

AMS_2020.pdf

111 An extensive investigation effort was initiated to determine the root cause of the V3 clamshell anomaly, focusing on the following aspects: the design of the mechanism, the manufacturing process, the testing environment, and the selected material s. A fishbone diagram (Figure 12) was created, and each potential contributor was subsequently investigated. The theory that the housing distorted enough to bind the cup was quickly discounted through additional FEA analysis. Figure 12: Completed c lamshell Failure Review Board investigation fishbone diagram Upon removal from the housing, the pinpuller was evaluated electrically and mechanically for functionality. It was quickly determined that it was destroyed; it physically could be reset and maintained a constant spring force, but it could not be actuated via either circuit . Though initially suspect, t he damaged Pinpuller was determined not the root cause of the f ailure, but a side effect that resulted. TiNi Pinpuller s are designed to automatically disconnect and shut off the flow of power upon actuati on. If the Pinpuller were physically prevented from actuation, as was the case in this situation, the electrical contacts will remain closed , overheating the actuation element . The nature of the EGSE being used to conduct this deployment (power supply with push button) resulted in power being applied to the devices applied for much longer period of time than designed. Typical Pinpuller actuation times are 30 to 80 msec (dependent on device temperature)2; so an uncontrolled button press of approximately 1 second is an order of magnitude too long, resulting in irreversible damage to the nickel titanium shape memory actuation wire. One of the lessons learned, and immediate changes made as a result of this anomaly, was t o implement EGSE and on-board SC commands that provide a pre-determined pulse of current, instead of a simple button press . This has the effect of protecting the Pinpuller from electrical damage in the case of a frozen mechanism, both during testing, and in space. It was appar ent however , during c lose inspection of the sliding cup that was forcefully extracted from the housing that there were very small ( 0.1 mm, 0.0039 in) areas of damage at the bottom of the cup. These three areas were spaced about 120 degrees apart, radially, on the circumference of the sliding cup. There were corresponding areas of similar damage on the housing as well. Instead of material buildups like on the cup, the housing damage was manifested as shallow pits. It was obvious that there was a transfer of material, and that t hese build -up locations ( Figure 13) on the cup were substantial enough to close the gap between the sliding cup and the housing; effectively seizing the cup in place. These areas were photographed and documented with the aid of a microscope.

Document

AMS_2020.pdf

112 Figure 13: Locations and micrograph of damaged area on the circumference of the sliding cup Dimensional Analysis A detailed investigation of the dimensional clearance between the cup and the housing began, to determine what the realized clearances between the two parts were, and if they were designed to be too tight. All previous clamshell mechanisms (EM1, EM2, and F light) were disassembled and each housing and cup part measured. Up to this point, each iteration of both the housing ID and the sliding cup OD were only spot checked; only one measurement of each were taken of one part in the lot. In addition to the full postinspection, it was determined that the profile, or effective shape of the parts should be measured and ascertained as well . This was accomplished by taking multiple diameter measurements of each part, in the critical areas of part interaction. As depi cted in F igure 15 these diameter measurements were made in 1.27mm ( 0.050 in) step increments along the length of each part . Figure 14: Snippets of original flight drawings of critical diameter dimensions and tolerances

Document

AMS_2020.pdf

113 Figure 15: Detailed measurement s and shape of all sliding cups fabricated up to the time of the anomaly This inspection process was illuminating, as it became immediately clear that the specific sliding cup that seized was out of family from all of the other sliding cup parts that were fabricated during the entire mechanism development effort. In addition to the effects of a larger outside diameter sl iding cup, which effectively reduced the diametrical gap between the parts to half (.022 mm, .0009 in), (Figure 16) from the average nominal (.045 mm, .0018 in), but the back half of the sliding cup was tapered over .0254 mm, ( .001 in) diametrically over the distance of approximately 8 mm (.32 in). The spots where the damage occurred corresponded exactly where the diameter was the largest. The resulting tapered- shape adds credence to the theory that the affected sliding cup was contacting with the housing on a line profile only, as opposed to the distributed as -designed surface contact of a correctly fabricated cylinder. A line contact would effectively magnify the contact stresses to the reduced areas of contact during the vibration testing that was performed immediately prior. .

Document

AMS_2020.pdf

114 Figure 16: Relationship of all housing/slider fits EM1 through flight (dimensions in inches) Material Discussion With the dimensional issues with the anomalous part in revealed, the investigative team focused on the selected materials as a contributor to the root cause of the problem. A mentioned earlier, both the sliding cup and the housing were coated and dry film lubricated 6Al -4V titanium; a decision made to eliminate any potential CTE effects within the mechanism. The tendency for bare titanium to w ear and gall to itself is well documented, particularly for space applications, where solid- phase welding can occur at areas of high friction and elevated pressure despite temperatures well below the melting point2. The effects of this material phenomenon were considered to be mitigated in this mechanism through proper selection of surface coatings and lubrication; a position supported by t he lack of any other example of titanium- titanium galling or during the clamshell development . Despite the presence of those coatings , it is theorized that the increas ed contact pressure that resulted from the tapered- cup line contacts, was enough to transfer titanium material from the housing to the sliding cup and lightly cold- weld the two together. This deposition filled the reduced gap between the p arts and increased the sliding force beyond the capabilities of the 44-N (10-lbf) pinpuller. The combined effects of a singular out of tolerance part and localized titanium -titanium galling was determined to the root cause of this anomaly. These effects were further exacerbated by control issues observed during testing; some clamshell configurations experienced higher dynamic loads than others, but that is a topic worthy of a different paper. Corrective Action Several Failure Review Boards were held as a result of this incident , with reviewers invited from outside organizations. With the root cause identified, a mitigation plan was developed, and unanimously supported by the board. Ultimately, a complete set of new active hinge halves, housings and sliding cups and were re-fabricated as a result of this anomaly. In an additional risk mitigation step, an entire new set of TiNi P10 pinpullers were procured as well, because the previous set had uncontrolled- length current pulses applied to them from the EGSE . Each mechanism part was fully inspected to the same level of scrutiny as the parts measured during the investigation, to ensure a minimal amount of part taper and shape. The inspection data was then used to match the cups and the housings to each other to guarantee a minimum .055- mm (.0022- in) diametrical gap in each assembly. Strategic material changes were considered, like a leaded

Document

AMS_2020.pdf

115 bronze sleeve between the cup and the housing, but the review board deemed such a drastic change to be too risky. There was no available volume for the additional thickness of a sleeve. Despite being considered a risk, t he original coated and lubricated titanium parts remained in the design. The clamshells were already slated to be integrated late in the SC I&T schedule, so the delays associated with this investigation and corrective action necessitated a c onsiderable amount of re- planning and schedule modification. Though EM2 clamshells were present in two locations on the flight structure during SC vibration testing, the new flight units had to be qualified individually, separate from the SC . The final integration of the clamshells occurred in earl y 2018, weeks before the full SC TVAC test. This was a hard deadline; four deployed clamshells had to be present on the SC to gather relevant thermal balance data. Final alignments of the complete ARS system were performed months later, in May 2018. Conclusion and Lessons Learned There is an adage that states generally, “All tes ting failures are good failures”, and this anomaly is a perfect example of the truth in that statement . Though severity of this issue can be debated, t he marginal nature of the failure (the pinpuller almost had enough force to overcome the cold- welds) presents a potential scenario that had this failure not occurred when it did, it could have resulted in a much larger problem later in the mission. Had the V3 clamshell deployed as expected, the damage may not have been detec ted until it was too late. That damage could have further compounded during SC vibration tests and launch, resulting in a worst -case scenario wher e the mechanism didn’t deploy when it needed to in space. In addition to a partial loss of Level -1 mission requirements on a flagship NASA mission, the failure investigation would have been impossible to complete with certainty. A lack of comprehensive inspection data, an d the parts impossible to recover, only postul ations could be made about a true final root cause. Ultimately, this anomaly and subsequent investigation resulted in a more robust , and reliable mechanism that enabled mission requirements to be met. Primary Lessons Learned: • Avoid dynamic titanium on titanium interfaces, and do not rely on surface coatings and lubrication to fully mitigate galling and cold welding effects. • Fully inspect all size and form critical mechanism components . • Anomalies can beget additional anomalies, which can become red herrings and distractions in an investigation. This was the case with the EGSE- damaged pinpuller. • Take steps, however conservative, with GSE to protect the hardware. Both MGSE and EGSE used during this development exacerbated the anomalous conditions . Acknowledgements The fantastically supportive and professional Parker Solar Probe development team at the Johns Hopkins University Applied Physics Lab. References 1. Parker Solar Probe, parkersolarprobe.jhuapl.edu/index.php#the- mission. 2. “Pinpuller.” TiNi Aerospace, Inc., 27 Oct. 2017, tiniaerospace.com/products/space- pinpuller/. 3. P. D. M iller and J. W. H olladay . " Friction and Wear Properties of Titanium." Wear, vol. 2, 1958- 1959, pp. 133.

Document

AMS_2020.pdf

117 Mars 2020 Rover Adaptive Caching Assembly: Caching Martian Samples for Potential Earth Return Milo Silverman* and Justin Lin* Abstract The Adaptive Caching Assembly (ACA) is part of the Sampling and Caching System on the Mars 2020 Perseverance Rover and consists of multiple stations that process , hermetically seal, and store sample tubes containing collected Martian material, either rock cores or regolith samples, in preparati on for caching on the surface of Mars. The ACA stations consist of seven active degrees -of-freedom, as well as a large number of passive mechanisms that must operate in extreme Mars temperature and pressure conditions . A robotic arm within the Rover manipulates the sample tubes between ACA stations as part of an end- toend sampling sequence and utilizes a compliant end effector to accommodate misalignments during station interactions . Stringent hardware c leanliness requirements were dictated to ensure collected samples would not be compromised, which significantly impacted the design, assembly, and test operations of the ACA. Three ACAs were assembled to support ground testing and flight operations , which were exposed to environmental testing to validate functionality in Mars -like conditions . A number of challenges existed from design through test, including volume constraints, mechanism controllability and operation, the effects of tight tolerances, and c leanliness requirements. Introduction The Adaptive Caching Assembly is part of the Sampling and Caching System (SCS) on the Perseverance Rover (Figure 1), which successfully launched on July 30, 2020, and will land on Mars in February 2021 . The Rover’s primary mission has a duration of one- and-a-half Martian years, or approximately three Earth years . The Mars 2020 Perseverance Rover design is heavily based on the Mars Science Laboratory (MSL) Curiosity Rover, which has been operating on the surface of Mars since August 2012; however, Perseverance is outfitted with a new sampling system and scientific instrument suite to address the new mission goals . The Perseverance Rover has four main science objectives: looking for habitability, seeking biosignatures, caching samples, and preparing for humans [1] . Seven science instruments are located on the Rover to address many of these objectives via remote and in- situ observations and operations . Sampling and Caching System The Sampling and Caching System’s purpose is to create the cache of scientifically selected and documented Martian materials by collecting rock cores and regolith samples, packaging these materials into hermetically sealed sample tubes, and depositing them on the Martian surface for potential return to Earth . These packaged samples must meet stringent science- driven contamination control requirements which drive the physical architecture and hardware implementation: less than 10 ppb organic carbon, less than 1 terrestrial viable organism per sample, and limits on inorganic contamination on elements that impact returned sample science. SCS must also perform rock abrasion and dust removal operations to prepare surfaces for sci entific instrument assessment , and then position the instruments for the assessment. SCS consists of two robotic systems which work in conjunction with each other to perform the mission functions: one MSL- like robotic system on the outside of the Rover c onsisting of a Robotic Arm and Turret, and one new autonomous r obotic system on the inside of the Rover called the Adaptive Caching Assembly. * Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA Proceedings of the 45th Aerospace Mechanisms Symposium, NASA Johnson Space Center, 2020

Document

AMS_2020.pdf

118 The robotic system on the outside consists of a 5 degree- of-freedom (DOF) Robotic Arm (RA), similar to the roboti c arm implemented on the Curiosity rover, and a Turret assembly mounted to the end of the RA . The Turret assembly contains a new rotary -percussive drill ( also referred to as the Corer), components of which are described in [2], [3], and [4]; a ground contact sensor; a gas dust removal tool (gDRT) , using puffs of nitrogen to remove dust from abraded surfaces ; and two instruments, PIXL (Planetary Instrument for X -ray Lithochemistry) and SHERLOC (Scanning Habitable Environments with Raman and Luminescence for Organics and Chemicals) , for in -situ science. The robotic system (i.e., ACA) on the inside of the Rover consists of a 3- DOF Sample Handling Assembly (SHA) that manipulates sample tubes, gloves, and covers . Sample tubes with collected Martian material are processed within the ACA by moving the tube between stations using the SHA to support the following functions : assess the volume of sample material collected, image the sample tube (multiple instances) , dispense a hermetic seal into the sample tube, activate the hermetic seal to preserve the collected sample, store the sealed sample tube until ready to drop, and finally drop- off the sealed sample tube to the surface of Mars at prescribed locations. Figure 2 illustrates the SCS architecture and how the two robotic systems interact . Sample tubes stored inside the ACA are inserted through the lower Bit Carousel (BC) door into sampling bits stored in the BC . The BC rotor rotates the bit/tube to the upper door where the RA docks the Corer to the BC and the Corer acquires the bit/tube using its actuated degrees -of-freedom . The RA undocks the Corer and then places and preloads the Corer on the intended target . The Corer acquires a sample into the sample tube inside the bit . After sample acquisition, the RA docks the Corer with the BC and the Corer transfers the filled sample tube (still in the bit) to the BC . The RA undocks the Corer, and the BC rotates to orient the filled sample tube for removal from the bit . After BC rotor motion, the filled sample tube is removed from the bit by the SHA within the ACA and is ready to be processed using the ACA stations (Figure 3) . Once the acquired sample has been imaged in the sample tube, its volume has been measured, the sample tube has been hermetically sealed, and stored for future deposit on the Martian surface, the process repeats for the suite of sample tubes available within the ACA . The remainder of this paper will focus on the ACA portion of the Sampling and Caching Syst em. Figure 1. Perseverance Rover with SCS components highlighted, and Turret Assembly details .

Document

AMS_2020.pdf

119 Figure 2. Sampling System Architecture. Adaptive Caching Assembl y Description and Operations The ACA consists of stations/components as identified in Figure 3 (CAD images) , and Figures 4 and 5 (asbuilt images) : Figure 3. ACA CAD image s with stations identified , rotated to a bottom up view for visibility . Left image: SHA stowed for launch , and bit carousel doors closed. Right image: SHA extended for sample tube operations , and bit carousel doors open.

Document

AMS_2020.pdf

120 Figure 4. As-built ACA – Bottom Up View. Figure 5. As-built ACA – Front View (mounted on test fixture). The Caching Component Mounting Deck (CCMD) is a monolithic titanium structure to which all ACA stations interface, and interfaces to the Rover via three bipod assemblies . All ACA stations are made from titanium t o minimize thermally induced position errors for robotic operations using the SHA . While the A CA is designed to fit within the Rover for launch through landing, surface operations on Mars require the SHA to extend approximately 200 mm below the Rover’s bellypan. Therefore, an ejectable bellypan was implemented directly below the ACA volume, which is released after landing to provide the SHA with an unobstructed volume to extend into during operations . Surface features are assessed via Rover imaging prior to SHA motion to prevent contact with potential obstacles below the Rover .

Document

AMS_2020.pdf

121 The ACA consists of 7 active degrees of freedom via gear motor assemblies (planetary gearboxes mated to brushless motor s with magnetic detent brakes ) described in [5] . The gear motors support actuation of the Bit Carousel , SHA, End Effector ( EE), and Sealing Station. A significant number of passive ball lock mechanisms are implemented throughout the ACA to support locking and unlocking of hardware from various stations/c omponents via SHA interaction. Hardware using ball lock mechanisms/features include the bits, sample tubes, gloves, covers, EE tube gripper, and the seal dispenser . A set of ball lock mechanism design guidelines were developed across the ACA to ensure cons istency for each use case and included details such as recommended geometry and clearances for mechanism features . These guidelines also accounted for dust inclusion in the mechanisms, which is a real concern for proper functionality . ACA Operations A high- level end- to-end flow of the sample collection, processing, storage, and tube drop- off for a single collected sample, focused on ACA operations, is shown in Figure 6. Every step, with the exception of the three Sample Collection steps inv olving the Corer, requires a robotic interaction between the SHA and an ACA station utilizing appropriate force and position limits to safeguard the hardware. The entire operation from Sample Collection through Sample Storage executes autonomously in a few hours . If a fault were to occur during the sequence, the ACA is capable of recovering autonomously in some cases; otherwise, operations will stop and the Rover will “ phone home” for assistance before proceeding. All Sample Processing steps identified as i maging or assessment based are for documentation purposes only (to support sample return selection), and are not decisional for proceeding through the full sequence, meaning operations will not stop based on the results of those activities unless a non- recoverable fault were to occur. Figure 6. End-to-end sequence for caching each sample tube on Mars Instead of dropping off each sample tube once it’s processed, the current expected strategy is to place the processed sample tube into storage until enough samples are collected in a particular region of interest, at which point a group of samples will be dropped off at a designated location for caching. The location of drop- off will be meticulously documented to support potential future retrieval and return to Earth, since the sample tubes could become covered with dust and not easily visible. More than one drop- off location for groups of sample tubes may be designated during the mission based on the number of regions of interest visited for sampling activities.

Document

AMS_2020.pdf

122 Planetary Protection and Contamination Control Because integrity of the collected samples is essential for future scientific evaluation on Earth, stringent Planetary Protection (PP) and Contamination Control (CC) requirements were levied on t he ACA hardware to ensure samples would not be compromised by e arthborn contaminants . Contamination c oncerns significantly drove design and material choices as well as assembly methods and resulted in less than straightforward test and qualification flows . Special hardware handling techniques were developed for both assembly and testing of contamination sensitive hardware. ACA hardware is classified in terms of cleanliness needs based on proximity to collected sample material (i.e., Martian rock and dust) . Hardware that directly contact s sample material is defined as Sample Intimate Hardware (SIH) . Hardware that contact s SIH items, but not directly sample material is deemed Sample Handling Hardware (SHH), while all remaining ACA hardware is considered Other . SIH items , including the sample tubes , hermetic seals, bits, and volume station, have the most stringent handling constraints and required sterile assembly techniques in an ISO 5 cleanroom . SHH items include the hermetic seal dispensers, gloves and covers, and the vision station, which were assembled in similar conditions to the SIH hardware. Other hardware included the Sealing Station, SHA, E E, Drop- off Station, and Parking Lots , which did not require sterile assembly and handling constraints . While t he Sealing Station directly contacts a sample tube prior to seal activation , it is considered Other since a hermetic seal will be installed in the sample tube at this point and the potential for introducing contamination into the sample tube is not credible. The BC actually falls into both SHH and Other categories, with internally and externally located hardware requiring different levels of cleanliness . Since the BC houses the bits used to collect material, internal cleanliness was important to not introduce contaminants to sample tubes during interactions; however, aseptic techniques were not levied due to the materials required for mechanism functionality over environments, such as wet lubrication. External cleanliness was equivalent to the Other items since the BC enclosure provides protection to the internal hardware . All piece parts were precision cleaned prior to assembly using approved solvents for the implemented materials and coatings . SIH hardware was cleaned to very stringent cleanliness levels, which all owed approximately one particle of no more than 50 microns in size per 0.1 m2, defined as Particle Cleanliness Level (PCL) 50. Other hardware was cleaned to typical cleanliness levels, which allows for one particle of no more than 300 microns in size per 0.1 m2, defined as PCL 300 . As the particle size decreases , the number of allowed particles per 0.1 m2 increases, as defined in industry standard IEST -STD-CC1246, Product Cleanliness Levels – Applications, Requirements, and Determination. In addition to a s tringent PCL count, SIH hardware required Non- Volatile Residue (NVR) cleanliness verification to ensure contaminants such as oils and greases were sufficiently removed to a level of A/10, which means no more than 0.1 mg per 0.1 m2 remained after cleaning . For a number of SIH items, precision cleaning more than once was required to achieve the required levels . Hardware was not released for assembly until cleanliness verification was completed. To ensure SIH cleanliness could be maintained as the hardware was manipulated per the operational sequence , steps were taken to avoid contact between SIH and Other items . A glove, which is attached to each sample tube, was implemented to prevent the EE tube gripper from directly interfacing with a sample tube. The glove is “clean” on the sample tube interface side, and “dirty” on the EE interface side . Each sample tube with glove attached is stored within a sheath. Clearances between the sample tube and sheath ensure th e sample tube only contacts surfaces considered clean, while also accounting for position accuracy of the SHA during sample tube manipulations at a sheath. The glove serves as a Fluid Mechanical Particle Barrier (FMPB), which based on Computational Fluid D ynamics analysis constrains contamination from reaching clean hardware. The FMPB does this by restricting flow into the sheath via specific (tight) hardware clearances between the glove and sheath in critical locations , which results in a maximum expected penetration height for particles that may enter a sheath, deemed the “bug line” . During insertion into and removal from a s heath, a sample tube may contact sheath surfaces if they are located above the “bug line” , which are deemed clean. The hermetic seal dispensers and volume station employ covers with

Document

AMS_2020.pdf

123 a similar FMPB feature to maintain hardware cleanliness requirements for hermetic seals and the volume probe prior to and between usage on Mars . In the dispensers and volume station, the sample tube may only contact surfaces considered clean similar to within a sheath. Refer to Figures 19 , 20, and 21 for depictions of a glove and cover s in their respective installed cases. In addition to precision cleaning and FMPB hardware features to achieve and maintain SIH hardware cleanliness , a high temperature bake- out at 350°C in air for a minimum of 1 hour was prescribed. This bakeout initiate s combustion cleaning for the elimination of viable organisms and the reduction of organic carbon on hardware surfaces and occur s as the last process prior to hardware delivery to the Kennedy Space Center for installation into the Rover . As a result , the SIH and associated SHH items must participate in the bake- out together as an assembly, which includes the STSA and the DVT assembl ies. Material selection was critical for all STSA and DVT hardware to ensure compatibility with the high temperature bake- out. Spring materials such as E lgiloy were implemented versus typical stainless steel options . High temperature compatible materials were implemented such as Ti -6Al-4V and A286, as well as d ry film lubrications such as molybdenum disulfide (MoS 2) and tungsten disulfide (WS 2). Non- metallics were not permitted d ue to the temperature requirement. All other ACA hardware was exposed to both a station- level and a full ACA -level bake- out at 110°C for durations up to 141 hours . These bake- outs were a Dry -Heat Microbial Reduction for PP purposes, as well as to meet surface cleanliness and outgassing requirements for CC . Thermoelectrically Controlled Quartz Crystal Microbalance was implemented to verify outgassing rates were met during bake- outs in thermal vacuum environments . Overall, t he PP and CC constraints introduced a number of challenges that resulted in novel hardware solutions, as well as hardware processing changes along the way due to lessons learned discussed later in this paper . ACA Station Descriptions Bit Carousel The Bit Carousel (BC) is mounted to the CCMD on the ACA and extends through a cut -out in the Rover top deck and front panel to accommodate both tube and bit exchange operations . A non-permeable fabric close -out covers the gap between the BC and Rover structures to prevent contamination from entering the ACA volume within the Rover from above. The docking assembly is the critical interface between the ACA and Turret for bit exchange operations with the Corer and is mounted to the front of th e BC (Figure 7). During bit exchange, four docking posts on the Corer structure engage the docking cones on the docking assembly as the RA brings the Turret closer to the BC. Due to the kinematic s of the 5- DOF RA, continued advancement of the Corer results in a rotation of the Corer on the end of the RA, meaning the docking posts are now rotating with respect to the docking assembly . The docking assembly therefore rides on a passive bearing mechanism that compensates for this Corer rotation. The bearing mechanism consists of a dry film lubricated four -point contact (X -type) bearing from Kaydon , implemented due to packaging limitations . Docking is considered complete once contact switch mechanisms located at the bottom of each docking cone are fully activated and the desired preload force is achieved . A minimum of 3 out of 4 switch activations are required to proceed. Because of the induced rotation during docking, a single speed variable reluctance resolver from Ducommun, Inc. is implemented on the docking assembly to measure the resulting amount of rotation, which is used to adjust for Corer -to-bit orientation during bit exchange. When the Turret undocks from the BC , a return spring mechanism brings the docking assembly back to its original position to support the next docking operation. The return spring mechanism consists of a cam roller design, with a detent in the cam to center the mechanism, and the roller located at the end of a flexure to enforce re- centering.

Document

AMS_2020.pdf

124 Figure 7. BC Docking Assembly. Internal to the BC is a rotor mechanism assembly which actuates the carousel ’s rotor structure to a desired position for either tube or bit exchange operations (Figure 8). The rotor structure is the primary interface within the BC for the bit holders with bits installed, as well as a sample tube sheath with a sample tube and glove installed . The rotor mechanism assembly consists of a n electronically commutated gear motor, harmonic drive, and dual speed resolver from Ducommun Inc., with a back -to-back set of angular contact bearings spaced apart to carry both launch and operational loads . Wet lubrication is used within the rotor mechanism components , which requires heating for proper functionality at cold temperatures on Mars . A spring energized Teflon seal is used to prevent debris from entering the mechanism within the rotor assembly . Additionally, spring energized Teflon seals are used to prevent wet lubrication from migrating into the carousel , which would be a contamination concern for the bits . The rotor mechanism does not contain a hard stop and is free to rotate clockwise (CW) or counterclockwise (CCW) indefinitely as commanded. The gear motor contains no brake, and only relies on a magnetic detent for holding position. Inside the BC resides a suite of nine bits to support coring (6x), regolith collection (1x), and abrading (2x) operations . These bits are “locked” within bit holders when not in use to survive not only launch vibration and Rover traverse loading, but both bit exchange with the Corer and tube exchange with the SHA. The bit holders contain both axial and radial cam rollers to align a bit within a holder and minimize tipping which could be detrimental for tube- to-bit insertion . In addition, an axial s pring mechanism provides compliance to accommodate misalignments during bit insertion and removal activities from a bit holder via the drill. The BC assembly contains an upper and lower opening in its structure. The lower entry point allows for sample tubes to be inserted into bits for sample acquisition, as well as removal from a bit once a sample has been collected (Figure 8). The upper entry point supports bit exchange operations with the Corer assembly . Both of these openings utilize a one-time deploy ment door to maintain cleanliness of the hardware within the BC until release on the surface of Mars . Both doors consist of a passive spring- actuated hinge mechanism held closed by a latch that is deployed via a release mechanism. The release mechanism is a series of passive spring- actuated mechanisms initiated by a separation nut non- explosive actuator (NEA) device from Ensign- Bickford Aerospace & Defense Company .

Document

AMS_2020.pdf

125 Figure 8. Tube Exchange Configuration in the BC (coring bit in tube exchange position at lower door opening, locked in a bit holder). When the doors are closed, a HEPA filter allows for venting during launch depressurization, and to minimize contamination from entering the BC prior to door actuation on the surface of Mars. Once the doors are deployed on Mars, crushable honeycomb absorbs the shock and determines the final resting position of the doors during surface operations on Mars . This position has been factored into the overall assembly to ensure inadvertent contact does not occur with surrounding hardware , and sufficient clearance exists for interfacing hardware, such as the Turret /Corer during docking. Sample Handling Assembly The SHA is a 3- DOF robotic arm approximately 0.5- meter long with a compliant, single DOF E nd Effector (EE) mounted at the end of the arm (Figure 9) . The SHA is responsible for all sample tube, glove, and cover manipulations between ACA stations . In order to reach the required operational work space in the ACA, the SHA uses a prismatic joint , identified as the Z -stage, which is actuated by an electronically -commutated gear motor assembly ; and two revolute joints , identified as the shoulder and elbow joints , which are actuated by electronically -commutated gear motors with harmonic drives and dual speed absolute position resolvers . The Z -stage is responsible for the linear motion of the SHA , and is designed to allow the EE, or any component (i.e., sample tube, glove, cover) on the EE, to interact with the ACA stations utilizing a minimum preload capability of 350 N . This applies only during station interactions when the SHA is retracted and not fully extended. A minimum clearance of 4 mm between the top of a hermetical ly sealed sample tube attached to the EE and the lowest feature of each ACA station is accommodated when the SHA is fully extended such that the sample tube can be manipulated safely in free space within the ACA volume. The Z -stage consists of an outer housing that directly interfaces to the CCMD , and an internal translation tube called the slide adapter that mounts to the shoulder (Figure 10). A welded metal bellows between the outer housing and the shoulder closes out the Z-stage and protect s the internal components from Martian dust. The slide adapter translates against the outer housing using two sets of eight track roller bearings: one set is attached to the end of the outer housing and preloaded against the external surfaces of the slide adapter and the other set is attached to the slide adapter and preloaded against the internal surfaces of the outer housing . These track roller bearings are set sufficiently apart to carry the cross -axis moment and side loads, and wit h the slide adapter and the bore of the outer housing having square cross -sections, the Rotor Mechanism

Document

AMS_2020.pdf

126 preloaded track roller bearings support torsion loads and minimize rotations between the slide adapter and outer housing. A lead screw mechanism drives the slide adapt er with the lead screw and actuator assembly mounted to the outer housing and the nut mounted to the slide adapter . Both the leadscrew and nut are mounted with gimbals to isolate the lead screw from any bending and radial loads . To limit the travel of the Z -stage, hard stop tabs are mounted directly to the free end of the lead screw and directly contact tabs on the nut gimbal when fully extended and tabs in the shoulder when fully retracted. The shoulder and elbow joints provide precise lateral positioning and placement of the EE at all stations with sufficient torque capability during station interactions , as well as holding torque during Rover traverse stow. Both joints are identical in design and utilize the structural member of the arm as housings for the joints . The gearmotor directly attaches to the gearmotor housing with the output connected to the wave generator of the harmonic drive through an Oldham coupling. Downstream of the harmonic drive is the output shaft t hat is supported by a back -to-back duplex bearing pair separated by a spacer for increased stiffness and reduction in wobble. To further reduce misalignments, the output shaft is pinned and match drilled to the flexspline of the harmonic drive . Embedded in the output shaft is the dual speed resolver from Ducommun Inc. to provide precise position feedback and knowledge of the joint . When stowed for launch, the SHA is restrained by two NEA separation nut mechanisms , similar to the devices used for the BC door releases . One NEA is used to launch- restrain the Z -stage assembly, while the other is used to launch- restrain the SHA forearm, located at the end of the arm before the EE interface. Figure 9. SHA Overview (EE not installed) .

Document

AMS_2020.pdf

127 Figure 10. SHA Z-Stage Cross -Section (shown sideways). End Effector The EE is mounted at the end of the SHA and interfaces with gloves , with sample tubes via gloves , and covers to be manipulated and moved throughout the ACA . The EE is comprised of two primary components: A Tube Gripper Assembly (TGA) which secures covers or s ample tubes via gloves during manipulation via a ball lock interface; and the Remote Center of Compliance Mechanism (RCCM) which provides compliance for all SHA to station interactions (Figure 11). A third component, the Lockout System, secures the RCCM during launch and traverse (Figure 1 2). The TGA has a n electronically commutated gearmotor that is mounted to the angular deflection plate of the RCCM with the output of the gearmotor directly connected to a ball screw mechanism (Figure 13 ). A pair of angular contact bearings support the ball screw mechanism as well as the gearmotor . The nut of the ball screw is attached to a plunger, which when the ball screw rotates either extends or retracts the plunger . In the retracted state, the ball lock groove of the plunger lines up with the ball sockets of the upper housing, which puts the ba ll locks in the released state. When the plunger extends, the balls extend out of the upper housing to allow the EE to grip against a cover or a glove (with sample tube) . The TGA also contains a 6 degree- of-freedom force- torque sensor with the inner diamet er mounted to the TGA upper housing and the outer diameter mounted to the TGA lower housing. The force/torque sensor protects the SHA, EE, and stations during operations against inadvertent loading during SHA free space motion and limiting loads during station interaction. The RCCM compliant mechanism is designed to lower the necessary loads required to align the EE during station interactions (Figure 14) . The concept comes from robotic end effectors that use a remote center of compliance approach to insert a peg into a hole as described in [6] to minimize the chance of jamming in close toleranced parts. The RCCM compliance is achieved through two sets of ma chined flexures: Lateral Flexures, which provide the lateral /translational compliance to the mechanism, and Angular Flexures, which provide the angular compliance to the mechanism . Both sets of flexures can be twisted about the z -axis for rotational compli ance. To reduce the chances of a fatigue failure, the flexures of the RCCM are locked out during launch as well as Rover traverse. This is achieved through the Lockout System mechanism internal to the EE that interfaces with a cam mounted to the shoulder of the SHA (Figure 12) . The main structural component of the Lockout System is the RCCM housing, which has two spring loaded lock ro ds, one bolted and one floating, that slide into bushings in the RCCM housing. The cover plate interfaces with the opposite end of

Document

AMS_2020.pdf

128 the lock rod. Affixed to a clevis directly adjacent to the lock rod and cover plate interface is a roller supported by a shou lder bolt and locking nut. This roller rides on a clearance- fit bushing and is supported on both faces by thrust washers . When the roller engages the cam during stow operations, the cover plate compresses towards the RCCM housing and restrains the RCCM by a series of pins and holes present at both the lateral and angular stages . These interfaces restrain motion of the RCCM via a load path back to the primary structure, the RCCM housing. Sealing Station The Sealing Station supports two functions within the ACA : to activate a hermetic seal within a sample tube, and to drop -off a sample tube from the Rover for retrieval by a potential future sample return mission. To accomplish these functions, t he Sealing Station consists of two mechanis m assemblies , a ram mechanism to conduct seal activation, and a gripper mechani sm to support the sample tube during seal activation as well as perform sample tube drop- off, as described in [ 7]. The ram mechanism is capable of generat ing more than 20 kN with a stroke of approximately 16 mm (Figure 15). To achieve this, a gear motor drives a gear train that connects to a planetary roller screw from SKF, Inc. that provides the linear Tube Gripper Assembly EE Harness Tube Gripper Plunger Force -Torque Sensor Remote Center of Compliance Mechanism (RCCM) – within housing End Effector Structure and Interface to SHA RCCM Cover Plate & Lockout Rods (shown locked out) Compression Spring Bolted Lockout Rod Cover Plate Cam Roller Floating Lockout Rod Lower Housing Plunger Upper Housing Force -Torque Sensor Gearmotor Lead Nut Angular Contact Bearing Pair Angular Deflection Plate RCCM Housing Angular Flexure Lateral Flexure Lateral Deflection Plate Lower Cover Figure 11. End Effector Assembly . Figure 12. End Effector Lock -out System. Figure 13. Tube Gripper Assembly Cross -Section . Figure 14. RCCM Cross -Section.

Document

AMS_2020.pdf

129 motion necessary to mechanically activate a hermetic seal within a sample tube, within a small package. The gripper mechanism is actuated via a gear motor connected through a gear train to the gripper, which opens and closes to “grip” or release a sample tube within the station . A kicker is attached to the gripper mechanism, which rotates with the actuated gripper, to enforce a sample tube to release from the station during the drop- off sequence. Figure 16 shows a cross -section of the Sealing Station with the ram in contact with a hermetic seal at the start of the seal activa tion sequence within a sample tube, and the kicker in a retracted position. The ram will travel approximately 3.5 mm from this position until the ram hard stops on the gripper at which point the hermetic seal has been fully activated within the sample tube. Full activa tion occurs when the hermetic seal knife- edge feature has fully expanded into the sample tube wall, thereby creating a seal. Figure 15. Sealing Station Ram and Gripper Figure 16. Cross -Section of Sealing Station with Sample Tube in Position for Hermetic Seal Activation. Sample Tube Storage Assembly The STSA houses thirty -nine flight sample tube assemblies within storage sheaths (Figure 17 ). Within each sheath is a sampl e tube and a glove. Figure 1 9 shows a cross -section of an empty sample tube with glove attached in a sheath. A glove serves two purposes: 1) Prevents contamination from entering a sheath to maintain sample tube cleanliness, and 2) Provides an interface between the SHA and sample tube for manipulation within the ACA . Due to cleanliness concerns, a sample tube is not to directl y contact the EE of the SHA. The sample tube consists of a passive ball lock mechanism with two states, engaged or disengaged . The engaged state has the balls pushed outward into a groove to “lock” the sample tube into a sheath or bit . The disengaged stat e allows the balls to retract to support sample tube removal from a sheath or bit via the SHA. With the glove attached to a sample tube, a multi -stage ball -lock activation process is required for sample tube manipulation. To manipulate a sample tube, the E E tube gripper first extends its plunger into a glove which ball locks the two together . With the tube gripper and glove locked together, the tube gripper plunger continues to extend, which pushes on the glove plunger . Depending on the amount of tube gripper plunger travel, the glove can be engaged or disengaged from a sample tube. Continued extension of the tube gripper plunger eventually results in the glove plunger pushing on the sample tube plunger, thereby disengaging the sample tube ball lock all owing for sample tube removal from a sheath or bit . Retraction of the tube gripper plunger provides the opposite effect and allows for sample tube ball lock engagement into a sheath or bit. Ram Gripper (Closed) Hermetic Seal in Sample Tube Kicker

Document

AMS_2020.pdf

130 Dispense, Volume, Tube Assembly The DVT assembly consists of hermetic seal dispensers (D), a volume station (VS), sample tubes (ST) with gloves in a sheath, and a sample tube parking lot (STPL), all integrated to a single mounting plate (Figure 18). There are seven hermetic seal dispensers, with each dispenser contai ning seven seals in a stack protected by a cover when not in use (Figure 20) . A cover is only removed when a sample tube, with an acquired sample inside, is ready to retrieve a hermetic seal from a dispenser. Covers remain installed otherwise to minimize h ermetic seal exposure to the environment and maintain cleanliness . The seal dispenser uses a two-stage passive ball lock mechanism to allow for the advancement of a single hermetic seal for dispensing into a sample tube, while also preventing the remaining seals in a stack from falling out . The SHA brings a sample tube into a dispenser which activates the ball -lock mechanism, and results in release of a hermetic seal into a sample tube. Once a seal is dispensed into a sample tube, refer to [8] for seal activation details at the sealing station . Volume assessment is conducted via the SHA bringing a sample tube into contact with the volume probe to a prescribed load limit . From this operation, the amount of sample material collected can be determined, while minimizing potential damage to a sample. A cover protects the volume probe when not in use (Figure 21). Covers utilize a passive ball lock mechanism to lock and unlock from the dispensers, volume station, and the cover parking lots . Three sample tube storage locations exist on the DVT similar to those on the STSA . A Sample Tube Parking Lot (STPL) is a modified sample tube storage sheath that allows for a sample tube to be stored without a glove. This capability allows for standalone glove operations in the ACA. Figure 17. STSA Assembly (bottom view with glove bottom flanges visible). Figure 18. DVT Assembly

Document

AMS_2020.pdf

131 Figure 19. Cross -Section of Sample Tube and Glove in a Sheath. Left: Figure 20. Hermetic Seal Dispenser Cross -Section . Right: Figure 21. Volume Station Cross -Section . Assembly and Test Program Three ACA units were assembled to support ground testing and flight operations: an Engineering Model (EM), Flight Model 1 (FM1), and Flight Model 2 (FM2) . The EM unit was assembled first, consist ing of flight - like stations, and was used for early ground testing of the flight hardware design, including algorithm development of station interactions for flight software purposes . Successful first -time end- to-end operations were conducted on the EM ACA in an Earth ambient environment . The EM unit was moved into a thermal vacuum chamber where it is currently part of an SCS testbed for full end- to-end testing at Mars pressure and temperature using analog samples . This testbed is known as QMDT, or Qualification Model Dirty Testing, where dirty testing is conducted to allow for evaluation of hardware life and performance through dust interaction. Drilling into rocks generates a lot of dust, whi ch can be monitored to understand where it goes, and how well the mechanisms continue to operate over time in that environment . This testbed is as flight -like as possible, utilizing a QM (Qualification Model) Corer attached to an QM Turret on an EM RA, with the EM ACA . The FM1 ACA was assembled after the EM ACA and was the first ACA to be exposed to 3-axis random vibration and thermal vacuum functional testing. Since station interaction sequences were not developed in time for FM1 functional testing, only range of motion and aliveness checks were performed for active mechanisms , and no passive mechanisms were tested, which would require robotic manipulation via the SHA. The functional check -outs were conducted before and after vibration testing, as well as at cold temperatures ( -110°C and -70°C) and Mars pressure ( between 5 and 10 torr) in a thermal vacuum chamber to verify functionality . Heat-to-use operations were verified below -70°C for wet lubricated mechanisms . Upon the completion of ACA -level testing, the FM1 ACA was installed in the flight Rover to support Assembly, Test, and Launch Operations (ATLO) testing with the flight vehicle. This testing included random vibration and functional testing of the Rover in thermal vacuum at Mars press ure and temperature and is part of validation for flight readiness . During the Rover thermal vacuum test campaign, limited station interaction was performed on the FM1 ACA to validate force and position sensing of ACA hardware via flight avionics, which was the only opportunity with the flight vehicle at Mars conditions prior to launch . Since the entire Sampling and Caching System was available on the Rover during thermal vacuum testing, bit exchange operations were conducted, which allowed for the flight R A to dock with the FM1 BC and the Corer to remove and return bits to the carousel . The opportunity to test this functionality on the flight vehicle Sample Tube Sheath Sample Tube Ball - Lock in Sheath Glove Cover Hermetic Seal Stack (7 Advancing Mechanism for Seal Dispensing Volume Probe

Document

AMS_2020.pdf

132 was a significant risk reduction effort; therefore, the flight bits were installed in the FM1 BC just prior to delivery to ATLO . While the FM1 ACA was in ATLO testing, the FM2 ACA was being assembled . The FM1 and FM2 ACA units were intended to be identical, but this did not occur due to findings from the FM1 ACA thermal vacuum test. A modification to the linear stage mechanism on the FM2 SHA was implemented to improve robustness over temperature and life. Because of this improvement, the FM2 SHA was designated the flight unit, and ultimately the decision was made to upgrade the FM2 ACA to flight as well . This decision was not made lightly and would mean that the FM2 ACA would not participate with the flight Rover at environments until actual launch and surface operations at Mars . This was deemed acceptable for the following reasons: 1) ACA units are interc hangeable in the Rover at the three bipod interfaces, with no mass impact or structural feature differences between the FM1 and FM2 units, 2) the FM2 ACA was exposed to a complete environmental test program at the ACA -level, which included 3- axis random vi bration and thermal vacuum functional testing which incorporated full end -to-end sequences (with the exception of bit exchange and sample acquisition operations) , and 3) after the FM2 ACA is installed in the flight Rover in ATLO, a ground test was conducte d that include d an end- to-end functional test within the ACA using a non- flight sample tube, hermetic seal and seal dispenser hardware, which exercised/verified force and position sensing using flight avionics . Because the FM1 BC had been verified in the Rover thermal vacuum test with the flight bits installed, and is interchangeable between ACA units, the FM1 BC was designated the flight unit . Removal of the B C from the ACA is required as part of Rover integration/de- integration, so an interface break is necessary regardless . As a result, the eventual flight ACA will be a mix of FM1 and FM2 ACA hardware. Maintaining a consistent test flow and history was critic al for understanding mechanism life as well as for documentation/configuration management purposes. Due to PP and CC constraints on SIH and SHH items, additional STSA and DVT assemblies were required to support ground testing and not compromise cleanliness of the flight hardware. In addition to an EM and QM STSA and DVT assembly, an ATLO flight model was developed to support Rover level activities in ATLO, including environmental testing, while a sterile flight model was established for flight with extensive PP and CC oversight to ensure stringent cleanliness constraints are being adhered to. To verify functionality of the sterile flight model hardware prior to flight, a test program was conducted that exercised sample tube interac tions with sheaths and bits, as well as seal dispensing at both ambient and Mars temperature and pressure conditions . The sterile flight model assemblies were also exposed to a 3-axis random vibration test for launch dynamics verification. Because of the s tringent cleanliness requirements that had to be maintained for the hardware, clean tents and special handling requirements were implemented. At the completion of the test campaign, the hardware went through a final cleaning operation followed by bakeout. The sterile flight model hardware is integrated into the Rover as late as possible prior to launch, directly from the bake- out, and requires a clean air purge within the ACA volume up to the moment of launch to maintain cleanliness . Functional testing onc e integrated in the Rover is prohibited to avoid compromising cleanliness . As a result of this limitation, the QM STSA and DVT assembly will be exposed to the flight bake- out as well followed by testing at Mars pressure and temperature in QMDT to verify functionality and performance by similarity. Challenges The ACA has had to overcome a number of significant design and test challenges, including volume constraints, tight tolerances , hardware cleanliness and operational issues . As a result, quite a bit of planning and re- planning was necessary to ensure requirements could be met, and the right tests were conducted. Being flexible is key when working in this mode. Having two ACA flight models, FM1 and FM2, and the ability to swap hardware between t he two, allowed for some flexibility . As previously noted, the final flight ACA configuration is a combination of FM1 and FM2 ACA stations.

Document

AMS_2020.pdf

133 SHA reachability and Z -stage travel constraints were a significant driver of the ACA configuration. From initial flight concept to final configuration, the ACA went through well over 50 different configuration layouts . The ACA had to fit within the defined volume in the Rover, while preventing ACA stations from interfering with each other, particularly during robotic operations . SHA geometry was dictated by the ACA volume in the Rover, with link lengths maximized to allow for motion within the Rover with clearances to structure of approximately 10 mm. Placement of the SHA on the CCMD was the result of ensuring the SHA could successfully reach all ACA stations and operate with sufficient position accuracy and force margin . This placement required the SHA to flip (invert) in the Rover, meaning the SHA would fully extend and rotate back on itself in the opposite direction at the elbow joint , extending the reachable workspace in the available volume. While the SHA Z -stage extends below the Rover bellypan during operations, it must fit within the Rover volume for launch through landing which limits overall available travel . The resulting SHA Z -stage travel drives ACA station and component heights to ensure sufficient clearances exist during robotic operations . A hermetically sealed tube on a glove on the EE results in the largest stack -up on the SHA requiring the largest Z -travel, while maintaining approximately 10 mm of clearance between the top of the seal and bottom of the ACA stations . In a few locations, clearance is reduced to as low as 4 mm, which is the result of design maturity of ACA hardware within exist ing constraints . Analysis and test verified acceptability of these close clearances. Because of SHA -driven hardware height limitations, sample tube, glove, and cover ball lock mechanisms were constrained such that their required spring forces could not be achieved with a single spring solution in the available volume. As a result, nested compression springs were implemented that are counter wound to avoid unwanted interaction (i.e., intertwining) . In addition, the sample tube ball lock mechanism could not support a sufficient length- over-diameter (L/D) ratio for close clearance sliding components, which increases the chances of wedging/jamming. The glove and cover mechanism designs achieved adequate L/D ratios, minimizing their risk of issues . During hardware testing, sample tube ball lock mechanisms exhibited jamming in the ball lock mechanisms, which was mitigated by the addition of sputtered moly - disulfide dry film lubrication to hardware that was previously bare within the assembly to reduce friction tha t was attributed to a number of failures . Follow -on testing showed no issues across environments, providing confidence in the corrective action implemented. SHA operations were conducted from the start using flight software to ensure the hardware and software worked in sync , as well as ensured fault protection and positioning required to operate the SHA within the ACA volume limits were in place to protect the hardware. While this is beneficial for proving out hardware/softw are behavior, t his introduced a number of challenges since the ACA was operated using Actuator Electrical Ground Support Equipment ( AEGSE) versus the flight avionics Rover Motor Control Assembly ( RMCA ) during ACA -level testing. Behavior differences between controllers existed that impacted position error during operation. For example, if a motor gets bogged down and needs more current to reach a position, the RMCA will allow a temporary increase (i.e. , boost) in current that allows a motor to reach position within about ±1 hall count . While AEGSE attempt ed to match the RMCA boost current behavior , significant position errors occurred during hard stop calibration activities affecting functionality . Therefore, two operational modes were developed for the AEGSE to achieve similar position accuracy as the RMCA for all functional activities: boost current was disabled to accommodate hard stop calibration, while boost current was enabled for nominal functions. Force control response was impacted when using flight software for SHA operations . Response times for force control were unexpectedly high as a result of filters on the force torque sensor signal causing force overshoots . Additional force overshoots occurred as the motor controller accelerated t he motor through the speed range where the motor detent brake interacts with the motor to avoid erratic behavior . The combined overshoot s reduce force limits , which results in a reduction of the SHA -to-station interaction operational margin by approximatel y 10% , as well as more occurrences of force overshoot nuisance faults . Software updates were made prior to launch to correct these overshoots , regaining operational margin required for surface operations.

Document

AMS_2020.pdf

134 While ensuring the ACA functions properly has been a challenge on Earth, this is only exacerbated by having to operate in a Mars environment where extreme temperature differences as well as exposure to, and operations within, a dusty environment are commonplace. With an operating range of -110°C to +50°C, this can result in significant CTE effects and/or gradients , especially due to heat -to-use conditions at cold, that need to be accounted for to ensure proper fits and functionality . Testing at Mars temperatures was conducted to verify hardware functionality and performance, which identified a number of issues . One example is significant drag increases at cold temperatures due to Teflon spring- energized seals used in dynamic applications , which are implemented in a number of locations within the BC. Ultimately, the springs were removed from the seal, leaving the Teflon jacket , which still exhibited too much drag on its own. Therefore, the jacket was cut to allow for expansion and contraction significant ly reduc ing drag, while not detrimentally affecting the performance of the seal . These changes were successfully validated by test . Using spring- energized seals in a dynamic application over a large temperature range, and particularly cold, must be evaluated closely for potential signs of fit or drag issues due to CTE. Because of cold temperature exposure, all of the wet lubricated mechanisms require heating for operation, with a minimum allowable heat-to-use temperature of -70°C used for ground testing. Braycote 600 and 601EF are used in various ACA mechanisms, which were found to contribute to high drag conditions impacting performance at cold temperatures (starting below approximately -40°C) . The SHA and EE both contain high life linear motion mechanisms where cold temperature testing revealed operational issues resulting in warmer heat-to-use temperatures being prescribed. These changes are implemented to preserve mechanism life and/ or to support mechanism functionality with appropriate force/torque margins for operations on Mars . During ACA testing in thermal vacuum at Mars pressure and temperature, a number of sample tubes failed to insert properly into bits and sheaths at - 110°C . These failures did not occur during Earth ambient testing using the exact same hardware prior to thermal vacuum. An investigation identified two key findings. The first was related to the tight clearances between the tube and sheath/bit, which were required for proper ball lock mechanism functionality . However, system errors due to factors such as tolerances, deflections, controller errors, etc. exceeded the clearances of the mating hardware. In addition, the geometry of the hardware included non- axisymmet ric features which offered catch opportunities due to the tight clearances and resulting machining. As a result, the compliant mechanism on the EE struggled to compensate. Performing lateral offsets prior to sample tube insertions favorably aligned tube features with the bits and sheaths resulting in successful operations . While the operational fix was successful, being mindful of what a compliant mechanism is being asked to do within a very constrained design might lead one to different implementation strat egies. The second key finding was related to a critical hardware configuration that was incorrectly modified for testing . The EE contains a cable bundle that exits the force- torque sensor and was strain relieved with specific tie -downs such that the cabling would not affect EE operations , including at cold temperatures where stiffness could increase by about a factor of two. For the thermal vacuum test, thermocouple (TC) wires were routed along the EE cable bundle and tied down in a few locations . The TC wires were added just before the thermal vacuum test and therefore not part of any ambient testing prior to environments . As a res ult, the EE RCCM ended up being stiffer than the intended design point at cold temperature, which contributed to tube insertion failures . The test was actually stopped to allow for correction of this issue once identified . The cable bundle tie downs with t he TCs were removed and the test continued successfully . When instrumenting hardware for a test, be very careful that the introduction of test components (i.e., TCs and their respective cabling) will not impede expected functionality of the hardware under test . Friction values were found to be much higher than expected during testing with hardware that had been exposed to both stringent precision cleaning and the high temperature bake- out at 350°C in air for 1 hour (for combustion cleaning) . This was determined during FM2 ACA testing using non-flight sample tubes and gloves that had been processed flight -like (cleaning and bake- out) and failures in the ball lock mechanisms were observed. This was the first opportunity to test hardware processed in a flight -like manner . The final flight hardware cleaning and bake- out processes were developed concurrently with hardware development .

Document

AMS_2020.pdf

135 While development testing had been conducted early on to validate the cleaning and bake- out efficacy , the overall process was not 100% flight -like, and may account for the discrepancy in findings . A detailed investigation of the hardware from FM2 ACA testing yielded a number of findings . The stringent precision cleaning operation cleaned the surfaces so well that higher friction was exhibited between sliding surfaces even during assembly operations . Inspection of components from the suspect sample tubes identified that the bare 440C balls in the ball lock mechanisms had significantly oxidized , resulting in surface changes contributing to higher friction. In addition, moly -disulfide dry film lubricat ion on ball lock mechanism surfaces had degraded. Pin-on-disk tribometer testing was conducted using flight -like material combinations, which confirmed friction increases of 1.5 to 2 times (reaching up to approximately 0.4) as a result of the bake- out. Per analysis, this increase in friction could be detrimental to proper mechanism functionality, as observed in test . This finding was very late in the project; therefore, a quick solution had to be identified and implemented. Ultimately, two corrective actions were implemented. The 350°C bake- out in air for a minimum of 1 hour was reduced to a 150°C bake- out in vacuum f or 24 hours, which eliminate d the adverse material affects previously observed . Concurrently, methods to reduce friction in the ball lock mechanisms were assessed. As a result, the balls within the ball -lock mechanisms for the sample tubes, covers, and dis pensers were modified with the addition of a sputtered moly -disulfide dry film lubrication to help reduce friction. Pin-ondisk tribometer testing of these modifications showed a significant decrease in friction with results less than 0.1 (details of tribo logy findings in work for publication) . Implementing this change actually addressed a few concerns identified through design and test, such as the aforementioned L/D concern in the sample tube ball lock mechanisms, and a concern with worn dry film lubricated surfaces observed in the seal dispensers after test . Validation of these changes using both flight -like and the flight hardware was performed over environments without issues . With the change in the bake- out criteria, additional cleaning steps w ere implemented as required for PP and CC compliance just prior to the bake- out. Tight tolerances were used throughout the ACA, on the order of microns in a number of passive mechanism locations required to ensure position accuracy could be achieved with SHA interactions . Working in microns meant a lot of manufacturing challenges and discrepancies . Verification of tolerances was critical to ensure hardware was acceptable to use and required very precise CMM (coordinate measuring machine) inspection (on the order of less than 1 micron in some cases, and typically less than 3 microns) . When working with tolerances so tight, verifying assembly -level dimensions may be warranted vs. relying on piece part inspection data. This became apparent for hermetic seals w hich contain piece parts designed in 5 micron increments as required to meet sealing requirements . It was unknown that hermetic seal components were expanding during seal assembly . This issue was not identified until hermetic seals failed to dispense successfully into sample tubes during test . Root cause was identified as fit issues where the seal was too large to fit within a sample tube. Seal-to-tube pairing takes into account a number of factors, but ultimately the seal has to fit in the sample tube. Inspection of hermetic seal assemblies yielded larger diameters for the components that must fit within a tube which had been CMM inspected previously . Therefore, all hermetic seals were reinspected at the assembled state using optical measurement equipment in the cleanroom to not compromise hardware cleanliness . With this information, the flight hermetic seals were assessed against the flight sample tubes to ensure sufficient clearances exist for dispensing, with ideal pairing arrangements identified for s urface operations. Dust accumulation on hardware can become a significant problem during operations . Accommodating anticipated dust in the ACA during operations resulted in designs either attempting to provide a path for dust to exit if possible, or the i mplementation of seals to prevent dust intrusion in the first place to protect sensitive mechanisms (i.e., bearings , screw mechanisms, etc. ). Standalone dirty development testing was conducted on ACA hardware that will be exposed to significantly dusty env ironments during the mission. This testing include d the BC exterior mechanisms (i.e., docking assembly and upper door release), which will not only be exposed to the dusty environment for the duration of the mission, but also susceptible to debris strikes during the Rover landing event . These mechanisms exhibited robustness to excessively dusty conditions . Dirty testing in the QMDT venue revealed significant dust accumulation on critical sample tube and glove surfaces . While some dust accumulation on sample tube surfaces was expected and

Document

AMS_2020.pdf

136 accommodated for in statio n designs and/or operations, excessive dust accumulation on sample tube and glove surfaces after removal from a bit post -sample acquisition can detrimentally impact operations . Therefore, operational dust mitigation efforts are being developed for validat ion in QMDT to ensure robust solutions are available for the flight hardware should an issue exist on Mars . Summary The ACA is a complex assembly of mechanisms required to perform a number of functions to meet Mars 2020 science objectives . A number of challenges existed for the development, assembly, and test of the ACA. Detailed test campaigns were completed to validate the flight hardware and software prior to integration onto the Rover, which identified a number of issues that were address ed as quickly as possible to not impact the mission. After the launch of Mars 2020, the remaining ACA units will continue to provide critical ground- based learning opportunities to improve the robustness of operations on Mars. Acknowledgements This work was performed at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. The Sample Handling Assembly (SHA) was provided by Maxar , Pasadena, CA under contract with JPL. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not constitute or imply its endorsement by the United States Government or the Jet Propulsion Laboratory, California Insti tute of Technology . Copyright 2020 California Institute of Technology . U.S. Government sponsorship acknowledged. Designing and developing the ACA, as well as the subsequent ass embly and testing, relied on a significant number of people who dedicated many long hours to make the ACA a reality . It would be impossible to list everyone on the ACA team ; however , listed here are the hardware/station leads and critical ACA staff: Nick Haddad, Grayson Adams , Erich Brandeau, Rebecca Perkins , Kayla Andersen, Sarah Sherman, Pavlina Karafillis , Brad Kobeissi , Suzie Kellogg, Will Green, Jeff Seiden, Jesse Grimes -York, Sean O’Brien , Sivan Kenig , Eric Roberts , Alex Bielawiec , Ken Glazebrook , Ed Dorantes, and Mary Magilligan. In addition, Louise Jandura, K eith Rosette, Matt Robinson, Richard Rainen, Don Sevilla, Matt Orzewalla, Mark Balzer, and Justin Kaderka have made substantial contributions towards making the ACA a success. References 1. https://mars.nasa.gov/mars2020/mission/overview 2. Barletta, A . “Design and Development of a Robust Chuck Mechanism for the Mars 2020 Coring Drill.” Proceedings of the 45th Aerospace Mechanisms Symposium, (2020). 3. Chrystal, K . “Percussion Mechanism for the Mars 2020 Coring Drill.” Proceedings of the 45th Aerospace Mechanisms Symposium, (2020). 4. Szwarc, T., Parker, J., Kreuser, K . “STIG: A Two -Speed Transmission Aboard the Mars 2020 Coring Drill.” Proceedings of the 45th Aerospace Mechanisms Symposium , (2020). 5. LoSchiavo, M., Phillips, R., Mikhaylov, R., Braunschweig, L. “Mars 2020 Maxon Commercial Motor Development from Commercial -Off-the-Shelf to Flight -Qualified Motors, Gearboxes, and Detent Brakes: Overcoming Issues and Lessons Learned.” Proceedings of the 45th Aerospace Mechanisms Symposium, (2020). 6. Drake, S . “Using Compliance in Lieu of Sensory Feedback for Automatic Assembly.” Diss. Massachusetts Institute of Technology , (1977). 7. Grimes -York, J. “Sealing Station Mechanisms for the Mars 2020 Rover’s Sample Caching Subsystem. ” Proceedings of the 45th Aerospace Mechanisms Symposium, (2020) .

Document

AMS_2020.pdf

137 Sealing Station Mechanisms for the Mars 2020 Rover Sample Caching Subsystem Jesse Grimes -York* and Sean O’Brien ** Abstract The Sealing Station is composed of two mechanisms: the Gripper Mechanism, for orientation and support of the Sample Tube during the Hermetic Seal activation activity, as well as ejecting Sample Tubes out of the Rover, and the Ram Mechanism, which produces and applies a high force for seal activation. The Ram Mechanism is a compact high for ce linear actuator that uses a planetary roller s crew (PRS) as the output linear stage , tapered roller bearings (TRB) to support the PRS at the high loads , a set of spur gears supported by cylinder roller bearings , and a brushless motor with a planetary gearhead and magnetic detent brake. This paper describes the role of these mechanisms in the Mars 2020 Rover mission as a part of the Sample Caching Subsystem (SCS) and Adaptive Caching Assembly (ACA) and their interactions with the Hermetic Seal a nd Sample Tube hardware. We report on development testing performed with the PRS that show it has capability of supporting cross -moment loads, something manufacturer s of these components do not rate. Also discussed is the interaction of the passive holding torque from magnetic detent brakes and hardstop preload, setting of a very light preload in the TRB , and the complex linear dyno test equipment used, and how that testing could have been improved. Introduction The Mars 2020 Rover will gather rock core and regolith samples from the surface of Mars for a potential return to Earth via the Mars Sample Return mission. Samples are collected into tubes that are hermetically sealed in- situ to capture both solid material and any volatiles that would otherwise evaporate and exit the sample during the long journey to Earth. The Sealing Station, pictured in Figure 2, is a two-mechanism assembly with the duty of activating th ese hermetic seal s and dropping the filled and sealed Sample Tubes out of the Rover and onto the surface of Mars. Sample Tubes are moved into position within the station by the Sample Handling Assembly (SHA), having already placed the Hermetic Seal onto the tube. The Sealing Station is a subcomponent of the Adaptive Caching Assembly (ACA) described in [1] and pictured in Figure 1 which is in turn a part of the larger Sample Caching Subsystem (SCS) . Figure 1. ACA with SHA pictured holding Sample Tube near Sealing Station * Formerly of the Jet Propulsion Laboratory, currently at NASA Johnson Space Center, Houston, TX ** Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA Proceedings of the 45th Aerospace Mechanisms Symposium, NASA Johnson Space Center, 2020

Document

AMS_2020.pdf

138 The primary mechanism of the Sealing Station is the Ram Mechanism, which is responsible for pushing the Hermetic Seal’s ferrule into the ‘sealed’ position. The Ram Mechanism is composed of a Planetary Roller Screw ( PRS), as the output linear stage, supported by a Tapered Roller Bearing ( TRB) pair and driven by spur gears and a brushless motor. The second ary mechanism is the Gripper, which orients and supports the Samp le Tube during the high force seal activation, alleviating the Sample Handling Assembly (SHA) of these requirements. The Gripper additionally performs the Sample Tube drop -off function, to eject the Sample Tube from the Rover. The Gripper has a rotational joint that moves half of the Sample Tube support structure, allowing ingress/egress of the Sample Tube. Figure 2 shows steps in this sequence. The Ram Mechanism is a high force density device tested to a capability of ~21 kN for a stroke of 16 mm in a 3. 3-kg package (Ram Mechanism only; total station mass is 5.5 kg). In this paper we will show the method for setting preload in the tapered roller bearings, development testing of the PRS linear -to-rotary component, for compatibility with this application and for a loading scenario past the vendor’s recommended limit for these components . Additionally, dyno testing to these high forces across warm and cold temperatures is presented with recommendations for how this could be performed more simply for future applications. An Engineering Model, a Qualification Unit, and Flight Unit have all been designed, assembled, tested, and integrated into their respective higher -levels of assembly, the ACA. The Mars 2020 Rover is scheduled for launch in July of 2020 and landing on the surface of Mars in February of 2021. The hardware described in this paper will be used during the surface mission portion of the mission. Figure 2. (left) Sealing Station with Sample Tube during Seal Activation (upper right) prior to seal activation Ram motion the ferrule is shown in the un- activated position. (lower right) end of seal activation Ram motion the ferrule has been pushed into final position.

Document

AMS_2020.pdf

139 Figure 3. The Hermetic Seal assembly shown separate from the Sample Tube and prior to seal activation. The ferrule expands the knife edge outward as it is pushed through the smaller diameter of the ramp profile in the seal cup. Ferrule shown in staring position. Hermetic Seal & Sample Tube Design The Hermetic S eal is compos ed of two major components : the seal c up and the f errule, held together by the ferrule retainer, shown in Figure 3. A retaining spring holds the Hermetic Seal in place inside the Sample Tube. The spring washer permits the f errule to only move deeper into the Seal Cup, thus retaining the ferrule after assembly and throughout launch. The seal cup has a knife e dge feature that, upon activation, is pressed against the inner wall of the Sample Tube creating a hermetic seal ( <1×10−8 cc He/sec). The knife edge i s coated with gold, which acts to fill in scratches in the interior wall of the Sample Tube caused by the rock core as it is ingested duri ng coring. The interior of the seal c up has a ramp profile opposite the exterior knife edge, which is sized and shaped to expand the k nife edge out radially, into the t ube wall. The geometry of this ramp profile, ferrule to seal cup to tube interference, lubricant and materials selection was a considerable challenge but is not the focus of this paper. The notable aspects of this seal for the purposes of seal activation and the Ram Mechanism is that the f errule must be pushed 3.5 mm further into the seal cup r amp after insertion into the sample tube in order to create the hermetic seal , a task that requires up to 7300 N. Also of note, this load must be reacted within the Sealing Station as the SHA , the robotic arm moving the Sample Tube from station to station within the ACA, cannot support loads of this magnitude. Figure 4 shows a successful seal activation performed by the SCS during ground testing. Figure 4. CT scan image of Sample Tube with Rock Core Sample and activated Hermetic S eal. The Hermetic Seal was successfully activated by the EM Sealing Station integrated to the EM SCS .

Document

AMS_2020.pdf

140 Mechanism Designs Ram Mechanism The output function of this mechanism is to push the f errule of the Hermetic Seal into the s eal cup; a stroke distance of ~3.5 mm with a load of up to 7300 N. Additional stroke length, 16 mm total, is required to retract the Ram away from the Sample Tube and Hermetic Seal to allow for ingress/egress of the tube. Ferrule depth in the c up upon seal a ctivation is determined by the Ram contacting its extension hardstop. The Ram Tip geometry is designed to control this depth, targeting the middle of a ~2- mm positional band, where any position within this range results in a successful Hermetic Seal activation. This stroke range allows for variation in piece part dimensional variation between the 45 Sample Tubes flown on the mission, and dust accumulation. Figure 5. Cross Section view of Ram/Sealing Mechanism showing details of the gear transmission. Output force is produced by a PRS where the screw is the translating element and the PRS nut is driven in rotation. The screw is a 15- mm pitch diameter with a 5- mm pit ch and a 5- start thread, a catalog item that is only modified to replace steels with stainless steels . The screw rotation is restricted by two cam follower bearings running in tracks, see Figure 5. The PRS nut is driven and supported by a spur gear that is in turn supported by a pair of tapered roller bearings. The output gear is driven by an idler gear supported on a cylinder roller bearing, which is in t urn driven by a pinion gear mounted to a gear motor. The gear motor is described in [2]. The Ram mechanism has been shown, in test, to be capable of producing ~21 kN , for an input torque of about 7.8 Nm at the pinion gear, across its qualification t emperat ure range of - 70 to +70° C. Lubrication throughout the geartrain is accomplished with Braycote 600 EF grease applied by greaseplating, at a 10% grease- to-solvent level, with additional grease, 30% void- volume fill, applied directly to the rollers of the TR B and cylinder roller bearing. All geartrain elements are CRES except for t he roller cages in the TRB and cylinder roller bearing which are silver plated 4340 steel. Sensing of Ram position is accomplished by Hall sensors at the motor. The only other telemetry available for this mechanism, and the station as a whole, is motor current. An example of this telemetry can be seen in Figure 6 where the distinct Hermetic Seal activation force profile is plainly visible . The mechanism is designed to be robust to contacting hardstops in either direction at full force and uses the retract hardstop for position zeroing. Stowing of the Ram for launch and rover traversing is accomplished by pressing the Ram against the extension hardstop, up against the Gripper , where it has the added benefit of restraining

Document

AMS_2020.pdf

141 the flexurized portion of the gripper output , refer to Figure 7 for the gripper flexure. A holding force of 2900 N is maintained by a passive magnetic detent brake located at the motor . The R am Mechanism is not required to produce high output load while simultaneously overcoming startup drag, especially at low temperatures. The first motion of the mechanism is to un- stow and retract , unloaded, to allow the tube ingress prior to any high force actuation. Test data shows that b y this time any start -up drag from cold grease is overcome. If the mechanism is ever required to begin operation where high load is required immediately , it can move backwards away from the load, “warming up” the mechanism before proceeding forward under high load. Figure 6. Telemetry of Hermetic Seal Activation on the FM1 Sealing Station during ground testing at 20° C Gripper Mechanism The Gripper is a much lower force mechanism compared to the Ram but serves the important role of supporting and locating the Sample Tube during the high force seal activation activity. The SHA is incapable of supporting loads much higher than about 300 N while seal activation can require as much as 7300 N. The actuated motion of the Gripper allows the Sample Tube to be brought into, and out of the Sealing Station . The Gripper ’s second primary function is to eject the tube out of the r over to the Martian surface. The ACA can return a sealed sample tube to storage and eject it later as mission operators choose. The Sample Tube interface design is highly constrained as it must interact with a large variety of hardware throughout the SCS on the Mars 2020 miss ion in addition to interacting with the Mars Sample Return mission . The seal activation function differs from all others in the mission, giving rise to different design constraints for ingress/egress than for Sample Tube insertion into a Coring Bit , Sample Tube Storage Sheath, or Seal Dispenser . Figure 7. View of Gripper Mechanism showing details of the gear transmission and flexurized output .

Document

AMS_2020.pdf

142 During the Seal Activation sequence , the Sample Tube enters the Sealing Station laterally at an elevation higher than its seated position. The Gripper c loses around it, enclosing but not squeezing, and the t ube is pulled down by the SHA, until the t ube bearing race is seated against the top of the Gripper with a preload of about 50 N. Because of position accuracy challenges with the SHA in the Martian thermal environment there can be residual load between the Sample Tube and the Sealing Station due to deflection in the S HA’s end effector , which features a remote center of compliance mechanism. Because of this residual load the Gripper must exert a torque high enough to center the tube under the Ram, and hold it there passively, necessitating the need for a passive brake described later. The Ram moves to activate the seal, retracts, then the Gripper and t ube reverse the ingress motions to extract the now sealed Sample Tube. For tube drop-off a similar ingress procedure is used, after which the SHA detaches from the tube leaving it in the Sealing S tation. T he G ripper then opens , allowing the t ube to fall to the s urface of Mars. To prevent the tube from sticking in the non- moving half of the Gripper , a kicker mounted to the gripper rotational output motivates the tube out of the Gripper. To prevent t he tube from sticking to the moving half of the Gripper output, the Ram extension hardstop is used as a lateral guide to ensure the tube remains on center of the egress path . Rotary motion of the gr ipper output is accomplished through a spur gear set supported by angular contact bearings , which are in turn driven by a motor with a planetary gearhead and a magnetic detent brake. The motor is described in [ 2]. The gripper output, green in Figure 7, is positioned to rotate into the closed positio n just above the supportive shelf meant for reacting seal activation forces from the ram. This gap is intended to prevent jamming of the m echanism. The output is flexurized to allow the gap to close while still controlling rotation. The flexures permit ver tical motion of the gripper output, allowing the high seal activation loads to be reacted by the static gripper housing as well as preventing the output shaft , colored magenta in Figure 7, and bearings from being in the ram load path. The rotational axis of the gripper output shaft and the bore of the gripper, used for tube locating, and tube support features are offset from one another, making the mechan ism self -closing during seal activation and tube seating. Figure 8. Hardstop/detent interaction plots for theoretical cases. Blue line depicts active motor torque at mechanism output with variation due to the passive detent brake. Grey line depicts the passive holding torque of the mechanism, see also Figure 9. Orange line depicts reacted torque from hardstop contact and preload. The Gripper uses a passive magnetic detent brake to maintain a preload between the Gripper output and the Gripper housing during both launch, rover traverse accelerations , and loads during seal activation. The interaction of the detent step size and the hardstop stiffness is an important design element that was not well understood until flight har dware testing. For high hardstop stiffness relative to the detent step size it may be nearly impossible to guarantee preload against the hardstop. To ensure preload, it is desirable to have a relatively soft hardstop to ensure that the detents can “catch” the reacted preload. See Figure 8. The intersection of orange and blue is the maximum torque at which the motor, with detent brake, can push into the hardstop, denoted by a red mark. When the motor is powered off, system load follows the orange

Document

AMS_2020.pdf

143 line down to where it intersects the grey line. The hardstop acts as a spring, pushing the geartrain backwards giving it momentum which can cause it to move past the first orange and grey intersection. The system may come to rest at any of the green dots or it may completely unload. Figure 8 left shows a system with a soft hardstop relative to the detent step size, where ensuring preload when motor power is removed is easier than where. Conversely, Figure 8 right shows a system with a stiff hardstop. The Ram Mechanism more closely resembles the left plot, while the Gripper more closely resembles the right. During testing of the Gripper, it was observed to back drive for certain commanded preload positions, which are now avoided. Figure 9. Convention for stable and un- stable nodes for a rotational system with magnetic detent brake. Circle markers indicate un- stable nodes, where, while the system torque is zero, any small perturbation will result in significant motion to the stable node indicated by the square markers. The motor hall sensor locations are centered on the stable nodes. Component Development Testing Roller Screw Component Development Testing At the start of mechanism design, the PRS was selected because it offers a higher load capacity densi ty than ball screws or lead screws with a high efficiency . Packaging of the screw and nut and supporting elements for each proved difficult , as the manufacturer s of these components recommend that cross - moment loading and radial loads be prevented between them . In the pursuit of a smaller mechanism package a prototype mechanism was built and tested to determine if omitting screw support was viable for this application. A view of the PRS nut is visible in Figure 14. First, a characterization test was performed to measure the loads during a 2X worst case aligned seal activation using the test equipment shown in Figure 10. With Hermetic S eals being a limited resource the test setup was reconfigured to replicate these observed loads without them with the test equipment shown in Figure 12. More cycles were executed on this prototype than is expected to ever be used on the flight unit, to an axial load near to that of the flight units max capability, and to a cross -moment ranging from matching the flight expected load to 4X that , see Figure 11 and Figure 13. Both lubricated and nonlubricated ram t ips were used, and each was tested pas t 1X life. No degradation in the performance of the mechanism PRS was observed. The need for additional support of the s crew is not required for this application, and the PRS is shown to have a non- zero cross moment load capacity , contrary to the limit specified in the manufacturers catalogs [ 3][4][5]. These tests were not intended to extensively characterize the cross -moment capacity, only to evaluate compatibility for this specific application. Critical to these results and this application is the high axial load relative to the cross- moment levels tested, and the large diameter of the supporting bearings , since the axial force acts as a restorative force, balancing the cross moment. This application of the PRS has a very low total life cycle count of just a few hundred, at most, load cycles where ec centric loading can occur. A higher cycle count application may not be compatible with supporting the PRS screw in this fashion and would warrant further development testing.

Document

AMS_2020.pdf

144 Figure 10. Test setup for characterization of the hermetic seal activation at off -nominal angular alignment. Figure 11. Data for hermetic seal activation at off -nominal alignment. The tests at higher level of lateral load and cross -moment were performed at twice the expected misalignment during flight operations. (left top) measured force along PRS axis, (left middle) measured lateral force at ram tip, orthogonal to PRS axis, (left bottom) calculated net cross -moment reacted by the PRS, (right) histogram of the number of tests performed at various cross -moment measurements. Indicated in red is the expected max cross - moment for expected misalignment during flight and ground test operations.

Document

AMS_2020.pdf

145 Figure 12. Test setup for reproducing flight -like loading conditions on the PRS in the Ram Mechanism. Load levels, misalignment levels, and ram tip lubricant were varied across tests. Push plates were single use components due to the fact that each of the 45 hermetic seals contact the ram only once. Figure 13. Data for tests performed on the test setup shown in Figure 12. (left top) measured force along PRS axis, (left middle) measured lateral force at ram tip, orthogonal to PRS axis, (left bottom) measured net cross -moment reacted by the PRS, (right) histogram of the number of tests performed at various cross- moment measurements. Indicated in red is the expected max cross -moment for expected misalignment. 1.25X life was performed and no degradation in performance was observed, nor any detrimental wear on the components.

Document

AMS_2020.pdf

146 Figure 14. PRS Nut. Axial key for torque transmission used in place of commercial key. Design of Preload for Tapered Roller Bearings With the functional load case characterized through the testing described in the previous section , the bearing preload could be chosen and analysis could be completed. Results show that s ince t he design load cases are high enough in axial force to counter balance the cross -moment load, these bearings are not required to support the cross -moment through prel oad, allowing the preload to be low, between 44- 88 N (10-20 lb) . Preload is therefore set to maintain ram t ip alignment throughout the stroke of the mechanism, which is critical for insertion into the seal c up during seal activation motion, in addition to keeping drag torque low, below 0.5 Nm at - 70°C. COBRA AHS bearing analysis software, [ 6], was used to ana lyze the contact stress in the tapered roller bearings , and was paired with MATLAB to run multiple analysis cases that vary input parameters such as: load definition, temperature, roller crown profile, and ID & OD & axial interference/clearance*. This M ATLAB code al lowed for thousands of analysis cases to be computed overnight with the COBRA software , that otherwise strictly requires manual input of analysis parameters. Analysis results show that , at the limit load, the contact stress in the upper bearing can be as high as 3310 MPa ( 480 ksi), and a more typical high load of 2070- 2410 MPa ( 300-350 ksi). This contact stress is high but the overall intended life of the mechanism is very low . Most of the total revs of the mechanism are unloaded, retracting the ram from stow to allow the Sample Tube and Hermetic Seal into the Sealing Station. The high force motion is only 7%, 175 revs, of the total life of 2500 revs of the output bearings. Life testing of this mechanism to a level 200% greater than expected lif e was recently completed at the time of this publication, with no sign of degradation in performance in the mechanism telemetry. Lesson Learned on Load Dependent Drag A parasitic drag source, which was unintentionally omitted from the early detailed design phase efficiency estimate for the Ram Mechanism, was later found to be a significant source of drag. This drag source is the load dependent drag within the rolling element components, specifically the Tapered Roller Bearings and the PRS†. Analysis of dyno test data showed that the peak output force measured was significantly lower than predicted. At a force level of 16,400 N, a drag source of about 1 Nm was calculated to be the discrepant drag magnitude, as observed at the mechanism pinion gear. An estimate for the omitted drag source was computed by, 𝜏𝜏𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 =𝐹𝐹𝑑𝑑𝑎𝑎𝑎𝑎 𝑏𝑏𝑏𝑏𝑑𝑑𝑑𝑑𝑏𝑏𝑏𝑏𝑑𝑑 𝑑𝑑𝑟𝑟𝑎𝑎𝑎𝑎𝑏𝑏𝑑𝑑𝑟𝑟⋅𝑟𝑟𝑏𝑏𝑏𝑏𝑑𝑑𝑑𝑑𝑏𝑏𝑏𝑏𝑑𝑑 𝑝𝑝𝑏𝑏𝑝𝑝𝑝𝑝ℎ 𝑑𝑑𝑏𝑏𝑑𝑑𝑑𝑑𝑏𝑏𝑝𝑝𝑏𝑏𝑑𝑑⋅𝜇𝜇 * The author would like to note the helpfulness of the COBRA software creators with aiding our efforts to extend the functionality of their software in this way. † The cylinder roller bearing and the cam followers was found to contribute less than 1% of the total omitted drag and is thus excluded from the discussion.

Document

AMS_2020.pdf

147 Where the friction factor, 𝜇𝜇, was varied between a typical range of rolling element bearings o f 0.0025 to 0.005. For friction values of 0.0021 to 0.0035, the increased drag corrects the mechanism efficiency prediction to match test data for the 3 flight units tested. The TRB was the significant contributor to this drag source as the radius of the bearing is much larger than other components in the mechanism. The TRB has an OD of 68 mm and contributed ~87% of the load dependent drag, while the PRS has a pitch diameter of 15 mm and contributed ~13%. Assembly Setting Preload in Tapered Roller Bearings Preload is set in the Tapered Roller Bearing pair by a two-step procedure , where a drag torque to preload relationship is established, and then the preload is set and locked based solely by the drag torque measurement. A load cell is used during t he measurements in the first step , see Figure 15, that is not included in the final design and is thus removed for the second step. Due to the placement of the gear between the two opposing bearings, preload was not controlled by precise grinding of the ring widths. Preload is controlled by the installation of the two-piece retainer installed after the preload- to-drag torque relationship has been establish ed, and from this relationship, a target drag torque is established. The retainer is adjusted to achieve the desired drag , corresponding to the desired preload range, and locked into place. The installation torque of this threaded retainer is very low and prone to backing out if not retained in some additional way. To lock this retainer in place, the preload between the retainer and housing threads is increased by installing a second retainer close to, but not touching the first retainer ; a gap size of 0.25 to 0.5 mm (10 to 20 thousandths of an inch) is used. The two retainers are then pulled toward one another by installing small screws at a standard preload and with a standard secondary retention method (Arathane 5753 for this mission subsystem). Installat ion of the small screws does affect the final preload in the bearings and must be accounted for and corrected in the final installation step. The fully installed retainer is capable of resisting at least 56 Nm (500 in- lb) whereas it was installed by hand w ith less than 1 Nm. Between the characterization measurements and the retainer installation the drag torque measurements have a different characteristic shape. The characterization measurements are much smoother than with the flight retainer. This inconsistency is due to the change in bearing clamp. The permanent retainer is threaded and is thus oriented by those threads , which imposes an alignment on the bearing race such that both may not be perfectly square to the rotational axis. Additionally, it is much more rigid than the characterization test setup where a ball bearing is used to intentionally eliminate this kinematic cons traint, see Figure 15. Figure 15. (left) Test configuration for characterizing preload to drag relationship. (right) retainer and lock.

Document

AMS_2020.pdf

148 Testing Dyno Testing Equipment Dyno testing of the Ram Mechanism involved custom built equipment to apply up to 28 kN of resistive force across the full stroke of the mechanism, while also capable of being reduced to zero force, and physically mount ed to an access port on the side of a thermal chamber. The linear d yno equipment was placed outside of the thermal chamber both for access of the operator to vary the resistive load and to place the load and position sensors in a room temperat ure environment. See the hardware pictured in Figure 16. A compound lever system is used to amplify a force applied by the user via a threaded rod and nut. This force squeezes on a rectangular cross -section shaft applying a friction force, resisting motion of the coupled mechanism. The mechanism’s intended high force output applies a compressive load to the dyno equipment requiring good support of the shaft to ensure good motion and prevent ing binding or buckling of the shaft. Two bushings are used to support the l inear brake shaft as well as the load cell shaft while couplers eliminate over constraints on each shaft. Figure 16. Linear brake equipment used for thermal dyno testing of the Ram Mec hanism. Lesson Learned with Linear Dyno One unforeseen issue with this hardware was that the compliance in the brake pad material used at the interface of the rectangular cross -section shaft and the compound levers. This COTS high friction brake and clutc h lining material while appearing somewhat metallic and feeling metallic to the touch is in fact significantly less stiff compared to steel or aluminum. A compression of 0.76 mm ( 0.030 in) at the brake pad for a resistive sliding force of nearly 20 kN is amplified at the user input to about 7.6 cm ( 3 in). This meant that to sweep the resistive load from zero to maximum the nuts must be moved up to 10 cm ( 4 in) along the threaded rod on the l eft and right side. A test can take be as short as about 15 seconds of mechanism motion time, requiring the operator to turn the nuts very quickly, which proved very difficult and introduced high irregularity of test conditions. The hardware was re- worked to thin the brake pad to reduce total deflection, giving the effect of increas ing the operator ability to apply a changing load throughout the

Document

AMS_2020.pdf

149 brief tests. But even with this fix the rate of successful tests was less than 100% and required several retests to achieve a good sweep of resistive force from low to high or stall. What c ould be better than the Linear Dyno? Given the less than satisfying experience with this dyno equipment we suggest a simpler approach: a spring box. The spring box would consist of a spring mounted shaft supported by bushings that is pushed by the mechanism. Bushings prevent the shaft from binding under the high compressive load. The springs, a stack of Belleville washers, and should be sized to provide some stroke, ~4- 8 mm would be enough for this application and reach full displacement at a force above the predicted peak force of the mechanism. A load cell would be placed between the mechanism and the spring box shaft to measure for ces. The short stroke of the spring box , an obvious difference to the linear brake, can be mounted in various locations within the mechanism stroke to characterize points of interest. For the Ram Mechanism the stroke of inter est is the end of travel where ferrule -to-ram interaction happens. The rest of travel is unloaded, thus, not necessitating high force dyno testing in these locations. The spring box would be a much simpler and quicker to design and fabrication alternative to the linear dyno as well as s impler to operate. Such a spring box was used in early development testing successfully; however, it was not selected because it lacks bi -directionality and the ability to apply load across the whole stroke of the mechanism. Requirements that lead to the much more complex linear dyno. The lesson learned here is that because this application did not require high force dyno testing across the full stroke length , the spring box alternative should have been selected. Figure 17 Fligh t Sealing Station

Document

AMS_2020.pdf

150 Conclusion In this paper the two mechanisms are presented to document the hardware and the function they fulfil l as a part of the Mars 2020 rover, launching later this year. The Ram Mechanism is a compact, high force linear actuator, while the Gripper Mechanism is a lower torque system featuring a unique flexurized output for interacting with the Sample Tube. The Ram Mechanism uses a PRS rarely used in spacecraft mechanisms and is not rated for any amount of cross -moment loading. While extensive characterization testing has not been performed in this body of work, we s how that the PRS indeed has capacity for cross -moment loads, making it a more robust component than previously understood. Also used in the Ram Mechanism are cylinder and tapered roller bearings, again, rarely used in spaceflight mechanisms. We describe the method used for installing a low preload vi a its threaded retainer by characterizing the drag to preload relationship prior to the installation of the retainer. Despite the low preload, the locking nature of the retainer holds it securely. The relative step size of the magnetic detent brake to t he mechanism hardstop resulted in conditions that would not allow for preload to be maintained against the hardstop. The presented understanding of the interaction allows for the problem to be avoided. We present the linear dyno, a mechanism for a high f orce linear thermal dyno testing , that worked moderately well for its intended purpose. Because of the low stiffness of friction pad material, the mechanism was more difficult to operate than intended. The simpler alternative, a spring box, could have been used in this application to save time and effort during schedule sensitive thermal dyno testing. Acknowledgements The authors would like to thank the entire SCS team and all the other unsung heroes at JPL. A special thanks goes to the ACA I&T team and SCS Leadership. Big, complex missions don’t fly without wonderful , talented motivated people like you all. Semper Gumby. The work was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. References 1. Silverman, Milo, Justin Lin. “ Mars 2020 Adaptive Caching Assembly: So Many Challenges ” Proceedings of the 45th Aerospace Mechanisms Symposium, ( 2020). 2. LoSchiavo, Michael , Robin Phillips, Rebecca Mikhaylov, Lynn Braunschweig. “ Mars 2020 Maxon Commercial Motor Development from Commercial -Off-the-Shelf to Flight -Qualified Motors, Gearboxes, and Detent Brakes: Overcoming Issues and Lessons Learned” Proceedings of the 45th Aerospace Mechanisms Symposium, (2020). 3. SKF Roller Screw C atalog. < https://www.skf.com/binary/82 -153959/14489- EN---Roller -screw - catalogue.pdf > accessed Jan 24, 2020 4. Nook Precision Screw Catalog, <http://www.nookindustries.com/Content/media/NOOK -Screw - Catalog.pdf > accessed Jan uary 24, 2020 5. Rolvis Roller Screw Catalog < http://www.nookindustries.com/Content/media/NOOK -Screw - Catalog.pdf > accessed Jan uary 24, 2020 6. J.V. Poplawski & Associates, C OBRA AHS bearing analysis software <http://www.bearingspecialists.com/software.asp > accessed January 24, 2020

Document

AMS_2020.pdf

151 Design and Development of a R obust Chuck Mechanism for the Mars 2020 Coring Drill Anthony Barletta* Abstract The Sampling and Caching Subsystem (SCS) onboard the Mars 2020 rover includes a coring drill that can perform coring, abrasion, and regolith collection operations. The coring drill requires a chuck mechanism to allow for the changing of the bits needed to perform sampling. The chuck mechanism must be capable of securely connecting a bit under drilling, rover slip, and pullout loads , but also readily release a bit after application of these loads . The Mars 2020 chuck consists of a ball lock mechanism that allows for robust engagement and release of bit assemblies that are preloaded to ensure a secure, stiff connection between the bit assembly and the rest of the corer . To prevent jamming, the chuck incorporates rollers to preload the ball lock. The rollers are mounted onto a flexured cam that minimizes changes in i nternal mechanism loads that result from thermally induced dimensional changes , further mitigating the potential for jamming. The chuck mechanism is moved between engaged, disengaged, and loose chuck positions by rotating the cam with a gearmotor that incorporates a magnetic detent brake that prevents inadvertent mechanism motion under design loads. Ongoing testing is occurring on the qualification model in a dirty environment while coring and abrading rocks in a M ars representative environment. Introduction and Previ ous Flight Heritage Mars 2020 is a Mars surface mission planned for launch in the summer of 2020, with a planned landing date of February 2021 in Jezero Crater. A major objective of the mission is to collect core and regolith samples for potential return to Earth by future missions as part of a Mars sample return campaign. The Mars 2020 rover will carry a coring drill, also known simply as the corer, to collect these samples. More details concerning the corer can be found in [ 1] and [ 2]. The chuck mechanism acts as the interface with the various bit assemblies needed to perform coring, abrasion, and regolith sample collection operations. The location of the chuck mechanism within the overall Mars 2020 coring drill as well as views of its internal components can be seen in Figure 2. The Mars 2020 chuck can be described as a preloaded ball lock mechanism that interfaces with sleeves on the bit assemblies and is able to accept and engage these bit assemblies regardless of their relative clocking about the central axis of the chuck and corer. Design and development were carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration (80NM0018D0 004). Figure 1: Mars 2020 Coring Drill Chuc k with coring bit installed. * Jet Propulsion Laboratory, Pasadena, CA; anthony.j.barletta@jpl.nasa.gov Proceedings of the 45th Aerospace Mechanisms Symposium, NASA Johnson Space Center, 2020

Document

AMS_2020.pdf

152 Figure 2: Location of chuck mechanism within overall Mars 2020 coring drill as well as cross sectional view of major components. Ball lock mechanisms have been used previously in spaceflight applications, most notably in the Mars Science Laboratory (MSL) chuck mechanism (Figure 3). The c huck on the MSL drill also utilized balls to grab onto a dr ill bit assembly. A one-piece cam, which was rotated by the chuck actuator, was used to move these balls radially outwards into pockets on the inside of the drill bit assembly. This design provided a connection between the drill and the bit assembly that w as sufficiently stiff enough to allow for the collection of the crushed regolith samples created during drilling. However, when the MSL chuck is in the engaged state, there is still a small amount of radial play (0.3 mm) at the balls. This amount of play w as allowed for in the design in order to make the design tolerant to dust and to avoid the possibility of the balls jamming inside the pockets in the bit assembly. The radial play at the chuck balls in MSL was acceptable since the drill was capable of coll ecting acceptable cutting samples even if the bit had a small amount of play. However, for Mars 2020, it was found after prototype testing with a brass board unit that a bit with significant play resulted in core samples of unacceptable quality. Therefore, a stiffer connection between bit and chuck was needed. The ball lock on the Mars 2020 chuck mechanism is unique in that the balls are preloaded in order to provide a very stiff connection between the bit and corer assembly. Other flight ball lock designs , including ones used elsewhere on the Mars 2020 Sampling and Caching Subsystem (SCS), avoid preloading balls due to fears of potentially jamming the mechanism. On the Mars 2020 chuck this jamming potential is eliminated by using rollers mounted onto a flexured cam to provide a stiff but tolerant and robust interface with the bit assembly.

Document

AMS_2020.pdf

153 Figure 3: Overview of chuck ball lock design used on the Mars Science Laboratory (MSL). The MSL chuck allowed for a small amount of play of the bit assembly even when fully chucked. This play resulted from ~0.3 mm of radial ball travel that was permitted when the bit was engaged. This play was intentionally included so as to allow for dust accumulation and to accommodate part dimensional and positional deviations, and was permissible for MSL since a stiff bit -to-drill connection was not a requirement. Design Requirements The Mars 2020 corer chuck mechanism had several requirements that led to a design unique from those used on previous missions and applications. The chuck needs to allow for the exchange of bit assemblies for different sampling activities and to increase overall sampling possible during the mission. It also needs to be able to retain a bit under various high load cases, including rover slip and bit pullout. To ensure high quality samples, especially during coring, the chuck needs to provide a stiff restraint of bit assemblies to the corer to minimize excessive play of the bit during sampling. At the same time, some degree of compliance is needed to accommodate bit exchange operations. While providing a secure, stiff connection between the bit and coring drill, the chuck also needs to be resistant to jamming, which could prevent removal of the bit and severely hamper mission operations. Another requirement was the ability to fully release a bit without the application of an external load from another part or mechanism. This requirement is needed since the corer will launch with a bit already installed in the chuck. After landing, the chuck needs to be able to disengage, release, and discard this bit onto the surface. The chuck had to satisfy all these requirements over a range of temperatures and be tolerant to deviations in part and assembly dimensions.

Document

AMS_2020.pdf

154 Mechanism Architecture For the Mars 2020 chuck design, it was decided to still use a variation of a ball lock. Ball locks such as that used in MSL have proven flight heritage, and such a design is compact, s imple, strong, and places few restrictions on bit orientation during the mating and demating operations. Each bit assembly consists of a bit contained within a sleeve (see Figure 4). The chuck mechanism holds onto the bit vi a this sleeve. B alls in the chuck assembly are preloaded onto a concave surface on the bit sleeve. Due to the contact angle at the chuck balls, this preloading causes the sleeve to be pulled further into the chuck until contact occurs between the front of the chuck housing and a spherical surface at the front of the bit sleeve. Preload exists both between the chuck balls and sleeve and between the chuck housing and sleeve, stiffly restraining the sleeve and therefore bit within the chuck . The preload torque to apply onto the cam is chosen based off anticipated side loads during sampling operations, these anticipated loads being informed by measured loads from prototype testing. Figure 4: Fully assembled flight chuck mechanism (left) with major components indicated and a model illustrating the interface between the chuck mechanism and a bit. The chuck is a ball lock mechanism that holds onto a bit via preload ing of a sleeve on the bit assembly against balls and the front of the chuck housing. The chuck is unique on Mars 2020 in that it is the only ball lock mechanism on the subsystem where the balls are purposely preloaded. Elsewhere on the Mars 2020 SCS, ball locks are designed so as not to load the balls in order to avoid jamming. Rather than not loading the balls, which would result in an unacceptably loose bit -to-chuck connection, the corer chuck mitigates jamming potential by using rollers and flexures . Preloading of the chuck balls and bit sleeve is accomplished via rollers mounted on a rotating cam. When the cam is rotated in the proper direction, the rollers will make contact wi th the balls, pushing the balls down towards the chuck centerline and onto the bit sleeve. To release a bit, the cam is rotated in the opposite direction, separating the rollers from the balls and removing preload from the sleeve. After sufficient cam rotation, the balls are free to move radially out of their holes in the housing enough to allow the bit sleeve to be removed from the chuck. Cam internal geometry is such that in the chuck’s disengaged position the balls are free to move out of their holes eno ugh to allow for bit removal but are still sufficiently captured to prevent them from falling completely out of their respective holes. Rotation of the chuck cam is accomplished via an actuator acting through a segmented gear mesh (Figure 5 and Figure 6). The chuck actuator incorporates a magnetic detent brake acting through a staged planetary

Document

AMS_2020.pdf

155 gear train. The magnetic detent brake resists rotation of the actuator output under an externally applied torque, thereby resisting cam rotation under externally applied loads on the chuck and bit. More details on the chuck actuator design and implementation can be found in [ 3]. Figure 5: (Left) CAD view of chuck actuator interfacing with mechanism via segmented gear. (Right) View of chuck mechanism engaged with bit sleeve. Note the exposed gear mesh between the chuck actuator output pinion and segmented gear. Segmented gear is fastened to the cam. Figure 6: Detail view of mesh between chuck actuator output pinion and segmented gear. Note the hardstop surfaces on either side of the segmented gear which serve to limit angular motion of the cam relative to the rest of the corer. A housing cover part (not shown) is installed over the volume occupied by this gear mesh. Cam Rollers To allow for preloading of the chuck balls while avoiding the possi bility of jamming, the balls are pushed down onto the bit sleeve using rollers mounted on the chuck cam. The use of cam rollers allows for a rolling contact that effectively reduces the friction that develops at the cam and ball interface. The use of rolli ng contacts is common in mechanisms to avoid the potential for jamming or binding which might otherwise occur for sliding contacts.

Document

AMS_2020.pdf

156 Rollers were chosen primarily to mitigate the potential for jamming. Jamming will occur if the torque applied by the actuator cannot overcome the frictional forces developed within the chuck mechanism in its preloaded, engaged state. The use of rollers significantly reduces the friction present at the interface with the chuck balls. The rollers are mounted onto flexures machine d into the cam. These flexures provide a compliant connection between the rollers and cam, allowing for better load distribution amongst the balls as well as reducing changes in preload due to temperature induced dimensional changes and part and assembly dimensional deviations. Figure 7: Cross sectional detail view s of cam roller s, both in the engaged and disengaged states . Flexures and Load Distribution Another concern with preloading balls is the potential for uneven load distribution amongst the balls. The number of balls in the chuck was determined by performing an analysis for contact stress under off -nominal load cases such as rover slip and bit pull out. The goal of the analysis was to determine the minimum number of balls needed to avoid brinelling at the balls and interfacing parts. The analysis indicates that six balls is the minimum needed to react all design loads without yielding assuming a fairly even and predictable load distribution. However, once machining and other errors are accounted for, the load distribution is unlikely to be even or predictable and could result in very high loads being applied to a small number of balls. To mitigate high ball loading, the cam rollers were mounted onto flexures cut into t he cam. The flexures allow for a small amount of compliance at the cam rollers, allow them to take up any difference in relative radial position due to machining and assembly errors. This helps to spread the load more evenly amongst

Document

AMS_2020.pdf

157 the balls, resulting in lower maximum ball loads. The flexures were sized so as to provide enough compliance to spread out loads amongst the balls but be stiff enough to allow for a very stiff bit -to-chuck connection to obtain good quality cores. The cam also includes hardstop p ins which serve to limit total flexure travel. Under pure tensile axial loading on the bit, the cam rollers will move radially out under the action of the balls being pushed outward by the bit sleeve. The hardstop pins serve to limit the radial deflection of the flexures for such cases so as to prevent plastic deformation. These pins are captured in the cam by cap that is bolted onto the cam front. Figure 8: Detailed CAD view of the flexures on the Mars 2020 chuck cam. The cam rollers used to preload the balls that restrain the bit assembly are mounted onto the flexures via roller pins. The flexures are sized so as to provide a stiff bit -to-corer connection at drilling loads while allowing the bit assembly to contact an internal load shunt at high off nominal side loads. Flexure deflections are limited by hardstop pins to prevent yielding under pullout loading. Figure 9: View of the chuck cam with integral flexures. The cam is machined as one piece using standard manufacturing processes including milling, drilling, turning, and band saw cutting. Despite the complex shape of the flexures, there was no need to utilize electrical disc harge machining (EDM).

Document

AMS_2020.pdf

158 An analysis was performed to determine the torque needed to disengage the chuck given a preload torque and loads applied on the bit during the unchucking operation (Figure 10). This analysis was performed for various coefficients of friction. It was determined that even if the coefficient of friction between the roller and shaft was 1, the chuck actuator had suffic ient torque to disengage the chuck. If one defines jamming as a condition where two parts cannot be separated given a certain force or torque, the analysis indicates t hat the chuck is resistant to jamming. Figure 10: Plots of c alculated engagement and disengagement torques as a function of coefficient of friction. Given the specified chuck actuator torque capabilities, the chuck is capable of being disengaged even if the coefficient of friction at all sliding surfaces is 1. Mechanism Functioning The chuck is typically actuated to one of three positions: engaged, disengaged, and loose chucked as shown in Figure 11. The chuck holds onto a bit in the engaged state and releases it in the disengaged state. The loose chucked state is an intermediate position where the chuck still captures a bit but allows it to have a small degree of angular , axial, and lateral play. This loose chucked state is important for bit exchange operations where a certain amount of play is needed to accommodate positional uncertainties between the coring drill and bit holders. To mov e to either the engaged or disengaged states, the chuck cam is rotated by the chuck actuator until a previously defined current limit is reached, indicating that contact with a hardstop has been made. When moving to the disengaged state, reaching the current limit indicates that the segmented gear on the chuck has contacted a hardstop surface. When moving to the engaged state, reaching the current limit indicates either that the gear has contacted another hardstop surface within the

Document

AMS_2020.pdf

159 corer or that the cam rollers are preloaded onto the chuck balls, depending on whether a bit is inserted into the chuck. During flight, no operations are planned to occur that would involve actuating the chuck to the engaged state without a bit present within the mechanism. In addition to these three states, the chuck can be commanded to any intermediate position. For example, during bakeout, the chuck was in an “almost engaged” position with a bit installed. This position was chosen so as to avoid preloading the bushings in the c am rollers during the high temperatures (~114° C) of bakeout, which could result in creep in the bushing material. The use of intermediate chuck positions has also been explored for use in recovery operations for a bit stuck in rock. Figure 11: Illustration of three major chuck states : 1) engaged, 2) loose, and 3) disengaged. A bit assembly is fully restrained and preloadd within the chuck in the engaged state, this is the position the chuck is in during sampling operations. In the loose chucked state, the bit assembly is still restrained within the chuck but is not preloaded and has some lateral, angular, and axial play. This play in the bit assembly is important in allowing for robust and reliable bit exchange that can accommodate for dimensional and positional deviations. The bit assembly is fully released when the chuck is actuated to the disengaged state. Testing and Results Several iterations of the chuck mechanism were assembled and tested, including three flightworthy units. Various tests to characterize chuck performance and verify capabilities were conducted at the mechanism and corer level. Some of more noteworthy tests conducted and their results are summarized here. Loose chuck bit ass embly play measurement As previously mentioned, some degree of axial, angular, and lateral play is needed in the bit assembly during bit exchange operations while in the loose chuck state. Measurement of this play in the assembled flight units was accompli shed using a test bit with an affixed laser pointer. By measuring how much the light from this laser moved on a screen when the chuck was actuated from engaged to loose, it was possible to calculate what play was allowed. Subsequent bit exchange tests on a qualification unit confirmed that the play allowed for by the chuck enabled nominal functionality.

Document

AMS_2020.pdf

160 Bit Pullout Static Test When sizing the components within the chuck, one of the major driving load cases was bit pullout. This covers a situation where a bit is stuck in the rock and the full margined pullout load capability of the corer feed mechanism is applied on the bit. When including anticipated pullout load needed to free a bit, force margin, and potential feed actuator torque overages, this pullout l oad sums up to 6514 N. To verify the capability to react this high load, one of the flight chuck units was subjected to a static test to verify that it could retain a bit under high pullout loads without damage. Figure 12 includes plots of the applied forces during this static test. After the static test, the chuck mechanism retained full functionality with no detrimental yielding on any parts observed. Figure 12: Plots of force data during a static test of a Mars 2020 coring drill chuck. The applied forces represented the worst case loading for the chuck, which is pull out on a bit with maximum applied load from the coring dr ill feed mechanism. The lower plot is a magnified view of the force data with further detail of the loads actually reached. After the static test, the chuck mechanism retained full functionality with no detrimental yielding on any parts observed.

Document

AMS_2020.pdf

161 Qualifica tion Model Dirty Testing (QMDT) The qualification model of the entire Mars 2020 SCS, which includes the chuck, is currently undergoing environmental testing as part of the Qualification Model Dirty Testing (QMDT) campaign currently ongoing at the time of writing. As the name implies, QMDT involves dirty testing, with the qualification model of the SCS being used to core, abrade and collect regolith from test rocks in a representative thermal environment. This testing is being used to verify numerous subsystem and mechanism capabilities, including ability to function in a dirty environ ment , and is informed by previous dirty testing performed using prototype units [4]. Ability to function in a dirty Martian environment drove several aspects of the chuck design, including the use of rollers and flexures, the decision not to cover the outside of the cam in order to facilitate the flow of dust and debris out of the mechanism, and exposed part fitments. QMDT thus far has demonstrated the ability of the chuck to nominally operate in a Mars representative dirty environment, including the ability to restrain a bit assembly to enable quality sample collection and the ability to actuate the various required states. Actuator current draw and position when driv ing to engaged, disengaged, and loose states have been the major metrics used to evaluate chuck performance during tests. Example plots of this data can be seen in Figure 13. Figure 13: Example plots of chuck cam angular position (blue plots on the top row) and chuck actuator current draw (red plots on bottom row). In this particular test, the chuck was actuated to intermediate positions betwe en fully engaged ( -0.46 rad) and loose (-0.2 rad) . High current draws at the start and end of motion are a result of the chuck actuator driving through its magnetic detent brake.

Document

AMS_2020.pdf

162 Challenges and Lessons Learned Several challenges were encountered during the design of the chuck mechanism as summarized here. Hertzian contact stresses Hertzian contact stresses generated during high load cases were a major driving factor in material selection and component design. These high contact stresses required the selection of high strength materials , such as Maraging 300 steel, and platings , such as electroless nickel, for parts such as the chuck balls, rollers, and bit sleeve. Contact stresses also played a role in the selection of the number and size of chuck balls. A higher number of balls would lessen the amount of load reacted per ball but would also increase contact stresses due to the smaller radi us of each ball. High contact stresses also drove the material and surface treatment selection in the bit sleeve assemblies. The geometry of the cam rollers was also optimized so as to minimize contact stresses, specifically by adding a concave race to the central portion of the rollers which preload against the balls. Figure 14: Detail view of cam rollers and concave central races . The rollers preload ag ainst the balls at these races. A concave race was added to each roller so as to minimize the Hertzian contact stresses generated under worst case loads such as would occur during rover slip or bit pullout. Flexure design The design of the cam flexure mounts for the rollers was a highly iterative process, requiring multiple finite element analyses and design changes to optimize stresses and deflections in the flexures. Constraints on flexure geometry included limitations imposed by manufacturing processes, total size envelope allocated for the overall chuck mechanism, number of balls needed in chuck, and stresses developed under worst case loads. Contamination control accommodations Contami nation control requirements placed restrictions on the type of bearings and lubricants that could be used in the mechanism. Fortunately, since no high -speed rotations occur in the chuck mechanism, the use of ball bearings and grease lubricants was avoided. Instead, plain bearings were used throughout. To accommodate dimensional changes due to temperature and material differences, the larger plain bearings had a split cut into them to minimize changes in fitment. Chuck actuator design and implementation Challenges were also encountered in the design and implementation of the actuator, particularly regarding the use of a magnetic detent brake and the issues this caused in controlling actuator torque output. The great advantage of a magnetic detent brake is that it is a passive brake. Whereas an active mechanism such as a solenoid or toothed brake would introduce an addi tional potential point of failure, a passive detent brake does not pose such an issue. However, when applying torque, the actuator must overcome this detent in addition to producing the needed torque output. This complicates the controlled application of torque into the mechanism and results in potentially high overages in the torque delivered versus what is desired.

Document

AMS_2020.pdf

163 Accommodations for these overages were made in the chuck mechanism by designing components for the worst case torques and mini mizing actuator output speed. Lessons learned Many key lessons were learned in the design and development of the Mars 2020 chuck mec hanism. Chief among these are that accommodations should be made for compliance in mechanisms whenever possible to increase robustness and to decrease the need for overly tight part and assembly tolerances. Initial prototypes for the chuck mechanism utilized a one- piece cam, similar to what was used on MSL. Although ambient testing demonstrated the functionality of using a one-piece cam, no operations were attempted at temperature using this prototype chuck. Furthermore, no testing at worst case pullout or rover slip loads was performed. Analysis indicated that large internal loads could be generated for a chuck with a one-piece cam due to thermally induced dimensional changes. Such a design would also be very sensitive to dimensional deviations and would likely require very tight tolerances and assembly controls to ensure equitable load sharing amongst parts. The use of rollers and flexures in the chuck eliminates the need for such tight controls while also resulting in a more robust design. Another important lesson is to design for large uncertainties in the torque output of the driving actuator, especially if the actuator mak es use of a magnetic detent brake. Such a brake requires that the actuator overcome the magnetic detent in addition to providing output torque, resulting in substantial uncertainties in applied torque. Early in the design, it was assumed that actuator torque output could be fairly well controlled (±25%), and therefore the preload torque applied to the chuck could be known within a reasonable certainty. However, later analysis and test data for the actuator indicated that much larger torque overages (±150%) were possible. This required designing parts of the chuck to be able to react these higher torques even if a much lower torque is commanded. However, the much increased reliability of a passive magnetic detent brake likely outweighs the increased uncertainty and overages in torque output. Conclusion s A novel preloaded ball lock mechanism has been designed and implemented for use as the chuck in the Mars 2020 coring drill. Through the use of rollers and a flexured cam, the Mars 2020 coring drill chuck is capable of securing bit assemblies with a preload to provide a stiff connection to the corer and ensure high quality samples while also preventing potential j amming. This chuck mechanism is capable of operating across the required proto- flight temperatures and in the dirty environment expected on Mars. Testing has demonstrated that the chuck mechanism is capable of reacting and surviving worst case design loads and provides the functionality needed for sampling and bit exchange operations. Future missions requiring a chuck, end effector, release mechanism, or any other assembly that needs to readily, reliably, and repeatedly secure and release components can mak e use of the features in this chuck mechanism design. References 1. K. Chrystal, “Percussion Mechanism for the Mars 2020 Coring Drill” Aerospace Mechanisms Symposium, Houston, TX, 2020. 2. T. Szwarc, J. Parker, and J. Kreuser, “STIG: A Two- Speed Transmission A board the Mars 2020 Coring Drill” Aerospace Mechanisms Symposium, Houston, TX, 2020. 3. M. Loschiavo, R. Phillips, R. Mikhaylov and L. Braunschweig, “ Mars 2020 maxon Commercial Motor Development from Commercial -Off-the-Shelf to Flight -Qualified Motors, Gearboxes, and Detent Brakes: Overcoming Issues and Lessons Learned” Aerospace Mechanisms Symposium, Houston, TX, 2020. 4. L. E. Chu, K. M. Brown and K. Kriechbaum, " Mars 2020 sampling and caching subsystem environmental development testing and preliminary results, " 2017 IEEE Aerospace Conference, Big Sky, MT, 2017, pp. 1- 10. © 2020. California Institute of Technology. Government sponsorship acknowledged.

Document

AMS_2020.pdf

165 Percussion Mechanism for the Mars 2020 Coring Drill Kyle Chrystal* Abstract The Mars 2020 rover includes a Sampling & Caching Subsystem that is required to acquire core samples of hard rocks and to abrade rocky surfaces flat and smooth in order to facilitate surface science . Hard, high strength rocks exhibit brittle failure modes, and can be drilled and chiseled efficiently using percussive impacts . The coring drill percussion mechanism was designed to provide those impacts with variable output force and to be used with various drill bits . A unique hammering mechanism, using a base- driven hammer, powered by a conventional rotary actuator driving a Scotch yoke, was designed to robustly provide the required function in a Martian environment for a life exceeding 2 million impacts. Introduction The coring drill, also called the “corer ,” is the sampling tool on the turret at the end of the rover’s robotic arm. The primary functions of the corer are to drill into the surface with a coring bit to generate a rock core sample and to abrade the surface with an abrasion bit . The abrasion bit creat es a spot-faced surface that allows other turret -mounted instruments to perform contact surface science. Hard rocks are not generally worked by cutting. They are usually either fractured apart (a brittle failure of the rock) by some form of chiseling or spalling or they are ground down when abraded by a harder material than the rock . A practical method for fracturing chips from a rocky surface is to use a percussive impact to chisel chips from the parent rock . Such impact s would deliver high force, short duration impulses to the rock . The force magnitude of the impulse must be large enough to cause a large enough stress to break the rock apart and the short duration of the impulse precludes the entire drill from having to react large forces . By breaking the rock with many short duration impulses, large forces can be reacted by only a very small subset of hardware and the time- average reaction required to hold the drill bit against the rock can be on the order of 100 N or less. This method is also practical for spa ce flight applications because it enables low powered actuators (order 50 W to 10 0 W) to produce loads sufficient to fracture hard rocks . The percussion mechanism uses an oscillating hammer to produce the impacts required for the corer to drill and abrade hard rocks effectively . The mechanism is situated near the center of the drill, above an anvil that combines rotational motion and percussive impact forces and transfers them both to an interchangeable drill bit . The bi t is held into the front of the drill by a chuck and it is free to spin and translate along the drill axis to do rotary -percussive drilling. When the hammer contacts the anvil , typically at velocities up to 3 m/s , the mutual deformation of the contacting surfaces creates a local compress ive stress in each part . The stress gradient in the anvil causes a wave to travel at the speed of sound down the axis of the anvil, into the bit, and finally from the carbide bit teeth into the rock, causing portions of the rock to be chipped off from the parent rock . The subject of this paper is the mechanism that creates this hammer motion and performs the percussion function. *Jet Propulsion Laboratory, California Institute of Technology ; kyle.chrystal@jpl.nasa.gov Proceedings of the 45th Aerospace Mechanisms Symposium, NASA Johnson Space Center, 2020

Document

AMS_2020.pdf

166 Figure 1. (a) Corer in turret structure (b) “Corer body” (c) Percussion mechanism in context (d) Flight percussion mechanism und er test Mechanism Architecture Principle of Mechanism Operation The basic principle of operation of the hammer in the percussion mechanism is a classical base- drive, single degree of freedom vibration system. The hammer is mounted through springs to a base that moves with fixed amplitude, but variable frequency . As the base oscillates sinusoidally at increasing frequenc y, the hammer and base begin to oscillate out of phase, energy is stored in the spring, and the motion of the hammer is amplified - provided that the base drive frequency does not exceed the natural frequency of the system . The percussion mechanism adds a fixed stop to this system (the anvil) that the hammer will contact once a certain amplitude of motion is reached. After the initial contact between the hammer and anvil, the system has a new, non- linear, dynamic characteristic , that can be ful ly described using vibro- impact theory [1]. As the base is driven to higher frequencies beyond the point of initial hammer anvil contact, the hammer will impact the anvil with greater velocity, producing larger percussive forces for the drill . Therefore, the system has a practical range of frequencies at which it will percuss and those frequencies have associated varying levels of impact energy. This system can be parametrized non- dimensionally by describing th e drive frequency (𝜔𝜔) as a fraction of the spring/mass natural frequency (�𝑘𝑘/𝑚𝑚) and the static gap between the hammer and anvil before the system was put into motion (Δ) as a fraction of amplitude of base motion (𝐴𝐴𝑑𝑑) – see Figure 2. Proper selection of these non- dimensional parameters will ensure that operation will occur in a regime where a stable vibro- impact process will be maintained and that the system will always operate with an impact frequency equal to the base oscillation frequency , as opposed to regimes where impacts occur as integer multiples of the base oscillation frequency.

Document

AMS_2020.pdf

167 Figure 2. (a) Block diagram for the percussion hammer mechanism, a base driven spring mass with a stop (b) The physical components represented in the bloc k diagram (c) A plot of hammer and driver motion from a simulation. Note that the hammer amplitude grows as driver frequency increases until first hammer/anvil contact at 25 Hx (indicated by red asterisk on the plot). Subsequent impacts at higher frequenci es will have higher velocity and will generate more force. In the case of the Mars 2020 percussion mechanism, the hammer mass is 200 g, the total spring stiffness is 11 N/mm, the single- sided amplitude of motion at the base (i.e., crank shaft throw) is 5 mm and the static gap to the anvil is approximately 9 mm. The result is a system that first impacts with base motion (and impact frequency) of approximately 25 Hz and the corer typically operates the percussion mechanism between 25 Hz and 40 Hz . Impact velocities are typically between 0.5 m/s and 3 m/s but will depend on the effective coefficient of restitution of the system and other factors.

Document

AMS_2020.pdf

168 Figure 3. Percussion mechanism architecture – rotating and oscillating elements

Document

AMS_2020.pdf

169 The percussion mechanism it self can be divided into the rotary portion of the mechanism and the oscillating portion, with a Scotch yoke performing the conversion from rotational motion to translation. See Figure 3 for a mechanism cross -section with the rotary and oscillating portions clearly delineated. Rotary Portion of the Mechanism – actuator, gears , and bearings The r otating components of the mechanism consist of a gear motor which drives a gear train in the mechanism with the final element being the crank shafts that push the diver plate up and down. The actuator output has an attached pinion spur gear that turns a mating gear . The mating gear is screwed to the end of a part that combines a hollow shaft and a bevel gear and this shaft is supported in the mechanism housing by an angular contact bearing pair, arranged back -to-back, with spacers . The bevel gear meshes with three synchroniz ed bevel pinions which are attached to the crank shafts . Each bevel pinion / crank s haft assembly is also supported by an angular contact bearing pair, arranged back -to-back. Transition from Rotary to Oscillatory - three Scotch yokes The t ransition between the rotary and oscillating parts of the mechanism is accomplished by the three Scotch yokes – see Figure 4 for details . A spherical roller, which is a 3/8-in (9.5-mm) diameter Grade 10 440C ball with a precision hole added, is placed on each crank shaft and is free to spin on the crank shaft pin in a plain bearing arrangement . The crank pin to roller ball interface is grease filled with Castrol Braycote Micronic 600 EF . As the crank turns, the spherical roller spins on the crank pin and also rolls back and forth in a cylindrical groove on the sid e of the driver plate while pushing the driver plate up and down . Three cranks arranged 120° apart from one another fully define the height and attitude of the driver plate, except for rotation about the axis of hammering motion, so no overturning moment i s carried by the roller and the driver plate needs no additional linear guidance. This arrangement balances loads and minimizes sources of friction and drag. Anti-rotation rollers consist of solid Vespel crowned rollers that ride in a plain- bearing arrange ment directly on a stub shaft that is integrally machined into the sides of the driver plate in between each groove . These provide very loose guidance by preventing the driver plate from rotating in- plane, which would be an incidental motion. Figure 4. (a) A conventional Scotch yoke which is the means of converting rotation into translation in the percussion mechanism (b) Vertical section view through bevel pinion, crank shaft, roller ball, and driver plate (c) Horizontal section through divider plate sh owing three cranks and anti -rotation rollers in guides Oscillating Portion of the Mechanism – driver plate / springs / hammer The oscillating portion of the mechanism consists of the driver plate, springs, hammer shaft, hammer, hammer bushings and the guide shaft – see again Figure 3. The driver plate only pushes against the springs as it oscillates, but it has a large clearance hole in the center to allow it to move up and down along the axis of hammer motion without contacting the hammer shaft . Two springs are arranged one on top of and the other below the driver plate and are captured and preloaded between the hammer and hammer shaft, so

Document

AMS_2020.pdf

170 that both springs are always in compression throughout the full range of driver and hammer motion. Assembly of the springs and driver plate onto the hammer is possible because the hammer shaft is a separate part that is threaded into the hammer . The hammer rides on split -bushings made from Vespel SP-3 (polyimide doped with MoS 2) that slide without additional lubricant against a guide shaft made of 15-5PH H1025 steel with a fine ground finish. Figure 5. (a) Hammer assembly (b) Single bearing, bevel pinion, crank shaft, and roller ball The Percussive Chain as a System – Hammer / Anvil / Bit The effectiveness of the percussion mechanism at the drill level depends heavily on the design of the downstream components in the percussive chain, specifically the anvil and drill bit . To transfer the impact loads efficiently, the parts must be axially stiff , ideally with a solid column of metal running directly form under the hammer head down to the rock . The anvil and bit have numerous other functions besides transferring impact loads , so their design must be a compromise. In general, designs that transfer impact loads well will avoid abrupt changes in the cross section s normal to the drilling axis (i.e., gentle tapers, not deep sharp- cornered grooves) and will avoid highly compliant features that extend out radially, as these will parasitically sap percussive energy when set ringing by an impact (i.e., no thin, large radius flanges) . Although the anvil, bit, and ground should all be in firm contact before an impact from the hammer is delivered, so as not to waste energy sending parts into rigid body motion, testing on the Mars 2020 corer has shown that approximately 80 N to 100 N was sufficient weight -on-bit preload to achieve efficient force transfer and that additional preload did not have a benefit for that system. Contact surfaces pairs (face of hammer, back to anvil for example) were made curved, to minimize sensitivity to misalignment as parts came in and out of contact, and also conformal (sphere in cup) , in the hopes of creating a more uniform stress wave through the part cross section.

Document

AMS_2020.pdf

171 Requi rements Development The percussion mechanism functional requirements are derived from the planned set of sampling operations which specify operation type and number (i.e., 43 cores, 74 abrasions ) and the distribution of those operations across a range of representative rocks (the “rock suite”) of varying hardness . The percussion mechanism must provide enough impact force to allow the drill bits to penetrate these materials and must have enough life to complete the mission and both output force and life must be sufficiently margined . At the same time, excessive conservatism is not possible due to tight constraints on available actuator power and the volume and mass that can be accommodated by a largely heritage robotic arm. There is an interesti ng force- life entanglement that emerges when the additional requirement of drill bit lifetime is brought into consideration . Impact force, Rate-Of-Penetration (ROP) into the rock (i.e. , “speed” of drilling), and drill bit lifetime are all closely related b ecause insufficient impact force will not allow the drill bit teeth to chisel the rock effectively . This in turn causes the bit to do too much rotating in place without much rock being chipped away (low ROP) and with weight -on-bit applied during this rotation, the bit teeth will be continually ground against the rock and will wear down quickly . Therefore, in order to allow drill bits to meet their required lifetime, the percussion mechanism must output a sufficient impact force to k eep the bits effectively chiseling at all times (when drilling hard rocks ). This is measured by proxy using ROP . The result was that the percussion mechanism nominal required output force level (“1x” force ) was set by finding the minimum percussion output level (frequency) that would allow a drill bit to successfully complete a 2x life test without falling below a minimum ROP . Then percussion output forces were measured across the operational frequenc y range using a “percussion dyno” (see next section) . Force margins could now be computed by comparing the output force levels measured on the dyno at the max operating frequency and at the frequency previously established by the bit life test as “1x” force (i.e., the lowest percussion frequency that could pass a 2x bit life test ). Finally, a life distribution across percussion output levels was derived based on taking all the planned sampling operations and assigning each of them a percussion level that should ac hieve a reasonable ROP, and therefore an associated duration for each activity . The result of this was a total 1x percussion mechanism life of 2.3 million impacts distributed across five percussion frequencies/impact force levels . This nominal lifetime is equivalent to 2 1.5 hours of continuous percussing. Due to the mechanism containing a dry -sliding interface (hammer bushing on hammer guide shaft ) and JPL Design principles, a 3X life test was conducted and completed successfully with over 7.2 million impacts performed in the test . The approach to deriving force and life requirements described above was designed to be appropriately conservative and philosophically in- line with normal NASA and JPL mechanism design practices, but several non- linear affects complicate the true interpretation of the margins . Drill bit life and drilling efficiency are highly non- linear responses to varying impact force, due to the nature of brittle failure and rock material properties . Simply stated, doubling impact forces will not produce a doubling of drill bit lifetime, and in fact probably results in a greater than linear increase in bit life to a point at which effective chiseling is no longer the dominant factor in bit wear. Development Program – Combining Theory , Analysis and Testing A combination of tools and techniques were used to respond to the unusual mechanism requirements and difficulty in sizing the output . Several generations of prototypes were built in the course of the development and these proved to be invaluable for developing the requirements, the design of the mechanism, validating analyses, and for interacting with other teams and specialists who needed particular test data. The basic parameters for the mechanism were initially based on iterative use of a numerical simulation (Simulink) to achieve a target impact velocity and impact energy . Subsequent research found mathematical models of vibro- impact systems that provided tremendous insight into the non -linear dynamics of the mechanism [1]. These models allowed the design to be altered in response to other changes to the Sampling & Caching Subsystem in a predictable way without relying exclusively on guess and check

Document

AMS_2020.pdf

172 simulation. For example, when a change to the core sample tube dimensions rippled through the bit and then the entire corer, the hammer needed to grow in diameter to a degree that maintaining its mass was not possible . The entire system was successfully rescaled with some other parameters changed to reduce the sensitivity of first impact frequency to anvil position using the referenced theory. With basic design parameters such as hammer mass, spring stiffness, driver amplitude and frequency, and the static gap to the stop determined, three dimensional FEA was now needed to determine the level of force that impacts of a given velocity would produce. LS-Dyna is an FEA tool with an explicit solver that is capable of accurately simulating dynamic stresses and large- scale motions and it was used extensively in the design of this system [2]. Parts in the impact chain (hammer, anvil, bit) have many required features that are detrimental to efficient transfer of force to the rock . In the Mars 2020 Corer, peak force under the hammer is reduced by 50% by the time it transits the percussive chain and reaches the end of the bit . This knockdown would be far worse without design changes and compromises made possible by LS -Dyna simulations . LS-Dyna provided tremendous general insight that primar ily facilitated general changes in direction (i.e. , this area thicker or thinner, this par heavier or lighter) but in many cases it was also used as for stress analysis of parts that were driven primarily by operational percussion loads – although there was an expectation that these parts had large margins (e.g. unlimited fatigue life). Beginning early in the project, hardware testing was used to improve mechanism performance and durability , and to validate the simulations . Impacts from the percussion mechanism have a peak force magnitude exceeding 40 kN and a typical impulse duration is 50 microseconds . A “percussion dyno” was built using a piezoelectric load washer (Kistler 9041A) and charge amplifier that had the necessary stiffness and fr equency bandwidth to measure these impacts . A sampling rate of about 200 kHz was used to ensure an accurate measurement of peak impact force could be made. Extensive functional testing was also done using the percussion mechanism in a complete coring drill . Multiple generations of percussion mechanisms cored and abraded hundreds of rocks. The data gathered from this testing was used to correlate percussion output level to the rate of penetration in various rocks . That test data, combined with requirements f or drill bit life, was used to create the force output and life duration margin story . A combination of testing in rocks, dyno measurements, and LS -Dyna analysis were used to verify and validate the design. Mechanism Function Mechanism operation consists of simply driving the actuator to the required speed to achieve the intended percussion frequency and force level output . The hammer will first begin to impact the anvil around 25 Hz, depending on the precise anvil position†. Impact peak force will increase as the actuator speeds up and the hammer impact frequency is increased. The maximum output frequency is 40 Hz and it is primarily intended for special use. The mechanism functions the same with the actuator rotating in eit her direction, and nominal operation calls for the life to be accumulated approximately equally in both directions for the purposes of wear leveling, particularly the flanks of the gear teeth . Vibration produced by hammer and shock loads due to hammer/anvil impacts have a few uses for Mars 2020 beyond coring and abrading. Vibrations from the percussion mechanism can be used to help transport powder that is collected by special regolith bits and to free the root piece of a core that is improperly broken off from the parent rock (so called “mushrooms” that extend below the mouth of the sample tube) . The most extreme special use of percussion is to “self -clean” the turret . Because the drill is designed to safely allow the hammer to impact the anvil with no dr ill bit present (or with a bit in the chuck but no weight -on-bit applied) the percussion will be used at the 40 Hz max output setting to apply an enveloping dynamic environment for the turret post -landing that should either remove any Earth spores still pr esent on the turret, or ensure that they will not be able to be dislodged during future sampling activities . This guarantees these spores would not contaminate a Martian surface sample with Earth life. † The anvil is separate from the drill body by a spring that is intended to isolate the corer (particularly nearby spindle bearings) from the rebound shock loads from impacts – therefore the anvil to hammer static gap will depend slightly on weight -on-bit preload.

Document

AMS_2020.pdf

173 When the mechanism is operating, the oscillating hammer creates a sine- vibration environment for other parts of the coring drill and other turret mounted instruments, even when no impacting is occurring. This environment becomes much more pronounced during impacting , when a maximum of 6.2 g’s of acceleration between 34 and 42 Hz is experienced by neighboring actuators and parts on the turret . Mechanism Advantages The Mars 2020 percussion mechanism has several unique features and advantages compared to similar extraterrestrial sampling technologies. Variable Output Force The ability to drill with a wide range of impact force makes it unique from the most common and highest heritage space drill percussion mechanisms which are typically of a cam and spring type. These mechanisms often use a helical cam to push a hammer back against a spring until the end of the cam profile is reached and then the hammer is shot forward by the spring and into the anvil or bit . Examples of mechanism that use this type of percussion are t he lunar dri lls used by the Apollo astronauts [3] and the HP3 instrument on the InSight mission to Mars [4]. Also, the vast majority of handheld commercial rotary hammer drills also have a single fixed level of percussion output . On Mars 2020, variable output force was a derived requirement that flowed from a science requirement specifying the quality for core samples (essentially how intact the collected cores should be) . Variable percussion output enables soft rocks to be cored with the minimum required force, thereby improving core quality, without foregoing the ability to penetrate hard rocks using higher levels of force. Variable output force has the additional benefit s including improving the power efficiency of the corer by not using only the power required for a given operation , lessening the harshness of the operational environment and therefore accumulated damage/fatigue for neighboring turret components , and genera lly creating a more flexible tool for the overall sampling system by providing ancillary functions described in the Mechani sm Function section. The MSL SA -SPaH drill also had a percussion mechanism capable of variable output force, but it was a very different type of mechanism that used a voice- coil actuator to provide hammering functi on. Although both systems have similar capabilities, there are also some advantages to the Mars 2020 percussion mechanism design, as compared to the MSL voice coil percussion. High Level Output Compared to MSL It was understood early in the project that creating a Mars 2020- sized core by cutting an annulus into a rock was going to require a more powerful drill than MSL used simply due to shear volume of rock to be removed and face area of the cut . Force dyno testing from MSL show impact forces with a maximum value of 6 kN under the drill bit . By contra st, the Mars 2020 drill can deliver more than 15 kN under the bit . Partially this is due to a more powerful actuator that was used for the Mars 2020 Percussion, but there is also reason to believe that the improvements in the cr oss section of the percussive part chain have also contributed to the higher force output . Measurements taken directly under the hammer of the Mars 2020 percussion show peak impact force values of approximately 40 kN as measured by a Kistler 9041A load was her. Due to the higher output forces achieved by the Mars 2020 percussion, significantly larger ROP values (roughly a factor of 5 or greater) were achieved when comparing each system in equal strength rocks – this should enable significantly longer bit lif e. Simple, Robust Parts Bear Impact Loads The MSL voice coil uses a very unique and highly efficient actuator to directly create linear hammer motion. However, with the MSL heritage architecture of a wound wire coil and permanent magnet pair there is a challenging durability problem where the coil must be connected to a driver by wires that must move with the striker and therefore subjected to high-cycle bending fatigue . The Mars 2020 percussion system has the advantage that the hammer and its assembly are essentially simple steel lumps and springs which are generally easier to design for tolerance to repeated shock loads and to meet the life requirement by fatigue analysis . This was a major reason for opting to develop a different style mechanism for Mars 2020 as opposed to sim ply re -sizing the MSL mechanism.

Document

AMS_2020.pdf

174 Conventional Actuator and Avionics Another distinction between the MSL percussion and Mars 2020 is the use of a conventional brushless DC motor/gearbox actuator as compared to a voice coil . This allows the Mars 2020 percussion to be driven at a range of speeds using very conventional avioni cs, software, and control system techniques . Essentially, the mechanism could leverage the whole system up to an actuator output shaft that was already in place for the numerous other actuators on board the rover . This more common actuator and avionics arc hitecture might also make the Mars 2020 style percussion mechanism more accessible to other projects. Testing Results For the Mars 2020 Verification & Vallidation campaign, a dedicated Life Test Unit percussion mechanism was built and successfully passed a 3x life test at the mechanism level . In the mechanism- only arrangement, the mechanism is placed on the percussion dyno and is tested across the qualification temperature range ( -70°C to +70°C) and percussion life frequency distribution . In addition, a percussion mechanism in a complete Qualification M odel (QM) corer has passed a 1x accelerated life test and is now in the process of completing a comprehensive life test performing sampling operations on rocks in the qualification environment . All units undergo a mechanism- level test for approximately one hour in the dyno configuration to ensure workmanship. At the end of the test, force data is collected across the operational frequency range for characterization purposes . Figure 6 show s example data from this type of dyno test, taken on the Life Test Unit. The mechanism is driven at speeds from 28 Hz to 40 Hz in 1 Hz increments and held at each speed for 3 seconds, thereby gathering approximately 80 to 120 consecutive impacts at each speed . This process is repeated for both actuator rotation directions . Each impact produces a complex force- time profile starting when the init ial stress wave arrives and continuing for several impact durations as the load cell and surrogate anvil, which are preloaded to a rigid base plate, ring down. The peak value of force from the initial blow is taken as the “peak force” and statistics for each operational speed/frequency are based on those values. Challenges and Lessons Learned The requirements of a long -life percussion mechanism for use on Mars present major design challenges . High wear sliding interfaces have the extra complication that the precise loads experienced are difficult to define and understand and although conservatism of design can solve the problem to a degree, gaining insight into the actual capabilities of the design through extensive prototype testing proved invaluable. Fatigue failures were also a challenge for the hammer assembly parts, owing to the large number of harsh shock load cycles the hammer assembly must cope with. The long time it takes to complete life tests in relevant environments underscore the general lesson – take great care in designing features such a sliding wear interfaces or points susceptible t o fatigue failure that can only be proven out by lengthy life tests . Design of these areas calls for extra conservatism, extra time spent reducing uncertainty in the loads, and early and environmentally relevant testing that is possible only with hardware rich development.

Document

AMS_2020.pdf

175 Figure 5. Typical results from a force dyno test performed at the mechanism level (force measured under the hammer). (a) Peak impact force statistics gathered from 3 seconds of operation at each frequency from 28 Hz to 40 Hz. (b) By us ing the time stamp from each peak force point, the frequency of impacts can be calculated independent of actuator telemetry. Results show that impact occur at a very consistent rate given that the hammer is driven through a spring and that each individual impact has a number of uncontrolled parameters in terms of the motion and orientation of the contacting parts. Challenge: Achieving Durability in the Crank Shaft and R oller The interface between the crank shaft pin and roller ball is a grease filled plain bearing that will experience as many cycles as there are hammer impacts . The estimation of the loads experienced by the interface is complicated by the fact that the load is shared by three crank shaft s with an unknown load distribution and that i t bears a cyclic load due to the oscillating inertia of the hammer assembly combined with hammer impact loads, although the impact loads should be smoothed by the springs . In retrospect, more attention should have been paid early in the design better understand the loads at this interface and thus properly

Document

AMS_2020.pdf

176 size the members and select materials, finishes, and clearances to better reflect tribological best practices . Early prototypes experienced a variety of failures at this interface, both with roller bores wearing and enlarging to unacceptable sizes and with crank pins wearing severely and, in some cases, break ing off completely as a result of the drastic reduction to their cross section by wear . A redesign attempted to reduce the bearing load associated with the highest percussion output level to below 3.45 MPa ( 500 psi ). Very often such plain bearing pairs use a hard shaft in a bearing that is made from a copper -based, chemically dissimilar metal and usuall y the bearing is softer and will preferentially wear . Such systems are good when a number of common assumptions are true that do not apply to this mechanism – specifically that the parts are serviceable and that petroleum- based lubricants will be used. The strategy adopted for this design was to si mply create the longest lasting (slow est wearing) possible design by making both parts as hard as possible, reasonably similar in hardness, and also by reducing the starting clearance between the parts to the smal lest comfortable value, to minimize impact loads between the roller and shaft when the driver plate changes direction. The final design increased the overall size of the parts from 3.175 mm (.125 in) shaft to 4.7625 mm (.1875 in) and 6.35 mm (.25 in) roller ball size to 9.525 mm (.375 in) . The shaft was made significantly harder by the addition of a 75 µm (.003 in) thick nitride case approximately 65 to 68 HRC so as to balance the high hardness of the ball, which would be made from a Grade 10 440C ball, 58 HRC minimum . Shaft to bore clearances of approximately 7.6 µm ( .0003 in) on diameter were achieved by manufacturing balls with several graded bore sizes . The shaft and bore both had 0.2 µm (8 µin) surface finishes specified . As before, the bearing was filled with Castrol Micronic Braycote 600EF grease and disassembly at the end of a 3x life test showed some grease was still present between the parts. Challenge: Hammer bushing wear and retention Development prototypes failed to achieve acceptable levels of wear life in the hammer bushings and a threaded retainer that was used to capture the upper bushing catastrophically failed shortly after 1x life. Many brute force attempts were made to improve the retainer – to pin it in place, to drill point it and drive set screws through the hammer shaft into the retainer, and to peen it into place . LS-Dyna analysis showed that the loads in this region were very high and, more importantly, the back of the hammer shaft experiences a large load reversal w hen the traveling compression wave from the impact encounters the end of the part and starts a tensile wave in the reverse direction. The small volume of the region did not lend itself to robust fastening features, perhaps besides welding, which was not at tempted . The solution was to eliminate any sort of bushing retainer altogether and capture the bushing in a blind groove. To enable this simplification, a change in bushing material had to happen at the same time, so that the bushing could be designed to be installed into this groove. A split bushing made from DuPont Vespel® SP-3 (MoS 2 impregnated) was sufficiently flexible that it could be wound in on itself and inserted past the shoulder of the groove (see Figure 8) and as it turned out the Vespel® SP-3 was also slightly slower wearing than the prior metal/polymer composite bushing, a Glacier Garlock Bearings (now “GGB”) DP4®. A key take- away for other percussion mechanisms is to reduce or eliminate small parts whenever possible that might not easily be able to grow to accommodate high loads , and particularly to expect high reversing l oads at the back end of the hammer or striker. Challenge: High drag during cold operation The initial flight design called for 10% to 15% grease fill in all ball bearings . All the bearings have a phenolic retainer impregnated with Castrol Brayco 815Z oil and use Castrol Braycote 600EF grease. Although much early life testing was done on prototype units, including some cold testing, none had gone below -40°C . The first tests at the min imum qualification temperature of -70°C showed that the mechanism cou ld not reliably start from cold and the actuator would sometimes stall at the 10 A current limit, equivalent to 0.75 N·m [6.6 lbf·in] of torque at the actuator output shaft .

Document

AMS_2020.pdf

177 Figure 6. (a, b) Heavily worn crank shaft pins from early prototype. Wear was so severe in some cases that the pin became too thin and broke off. (c) Comparison of the design after improvements were introduced to reduce wear between the roller ball and crank shaft pin After a comprehensive investigation that eliminated all other sources of drag, the bearings were found to be the cause of the large increase in drag at - 70°C . The solution was to drastically reduce the grease fill to approximately 3%, and th us ultimately reduce the drag to acceptable levels . Improved warm up routines were also devised that eliminated starts and stops until the mechanism had gradually accelerated and then run for about 1 minute. The mechanism life tests mentioned above were all conducted using this reduced grease fill and at the end of the test there was no sign of bearing performance degradation. The lesson here is well worn, but worth repeating: early life testing in a relevant environment is invaluable and the increase in the viscosity of Braycote 600EF grease is extremely non- linear in the - 50°C t o -70°C range. A large number of bearings and relatively low -torque actuator (relative to power) in this mechanism exacerbated the sensitivity to the thickening of the grease. Conclusions A percussion mechanism that will enable the Mars 2020 coring drill to core and abrade hard rocks in the Martian environment was successfully designed, built, and tested and is scheduled to arrive on the surface of Mars in February 2021 and begin operation soon thereafter . The novel mechanism, with a conventional rotary actuator drive, a base -driven hammer, and triple Scotch yoke architecture made it possible to perform variable output force percussing whilst subjecting only a simple and robust subset of mechanism hardware to the large, dynamic impact loads. Acknowledgement The research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration (80NM0018D0004).

Document

AMS_2020.pdf

178 Figure 7. Hammer bushing and retainer. (a) Origi nal GGB DP4 bushing wore quickly and (b) Bushing retainer failed catastrophically soon after 1X life. (c) LS -Dyna analysis revealed the severity of compression / tension reversing loads at the top end of the hammer shaft. (d) The best retainer was no retai ner but instead to keep the bushing in a blind groove, which was enabled by (e) A Vespel SP -3 split bushing that could be installed into the blind groove and exhibited better wear performance. References [1] V. I. Babitsky, Theory of Vibro -Impact Systems and Applications, Berlin: Springer -Verlag, 1998. [2] A. Siddens, "Validation and Predictions of Explicit Dynamics Simulations for the M2020 Percussion Drill," in Spacecraft Launch Vehicle Dynamic Environments Workshop , El Segundo, 2018. [3] Y. Bar -Cohen and K. Zacny, Drilling in Extreme Evironments, Darms tadt: Wiley, 2008. [4] T. G. M. S. S. e. a. Spohn, "The Heat Flow and Physical Properties Package (HP3) for the InSight Mission," Space Science Review, no. 214, 2018.

Document

AMS_2020.pdf

179 STIG: A Two -Speed Transmission Aboard the Mars 2020 Coring Drill Timothy Szwarc*, Jonathan Parker ** and Johannes Kreuser+ The coring drill, part of the Sampling & Caching Subsystem (SCS) aboard the Mars 2020 rover, demands a wide range of drill bit torque and speed capabilities during sample acquisition operations. The two driving operating points are high speed , low torque for rotary -percussive coring, as well as low speed , high torque for separating the rock core sample from its parent rock . The spindle twin- input transmission (STIG) allows these and other operating points to be reached with an actuator of subst antially less peak power and maximum current draw than that of a single -speed actuator . Rather than containing gearing of its own, the transmission interfaces to an actuator with two outputs of different gear ratios , allowing the transmission to select one of the two of the outputs to be coupled to the drill bit . This paper describes the design, capabilities, and challenges associated with the transmission and the dual-output actuator. Introduction The Mars 2020 sampling tool, commonly referred to as the “ corer ” and shown in Figure 1, possesses three methods of surface interaction. T he corer takes its name from its ability to core and retain acquired rock core samples 13 mm in diameter and 76 mm in length. An abrading bit allows for the removal of the several mm of weathered exterior surface from rock, in patches 50 mm in diameter. Operation of the abrading bit requires similar torque to that of coring, although the bit is rotated at a much lower speed. Finally, a regolith bit al lows for collecting and caching surface dirt. This bit operates at low torque and low speed to allow powder to flow into the bit from an opening on the bit’s side. Generating cores typically requires speeds a nd torques at the bit of 200 rpm and 2.5 Nm, respectively. Following bit penetration into the rock, core breakoff and retention are performed by mis aligning the eccentric sample tube and the drill bit in which the tube is contained [ 1]. This operation, which fractures rock in shear, requ ires as much as 28 Nm in the test rock suite used by the mission, although the speed of core break can be arbitrarily low. Providing margined capability at the core breakoff and coring operating points in the absence of a transmission would require an actu ator with a 400 W peak operating point. Assuming a gearhead with a single ratio capable of providing coring speed, the driving electronics would need to provide about 500% more electrical current than is currently possible on the rover in order to provide sufficient torque for core breakoff . An actuator coupled to a transmission with suitable ratios would require a peak power of 200 W and no alterations to electronics. Therefore, a transmission was determined to be the preferred solution. The resulting spindle twin- input gearing (STIG) transmission and associated dual output actuator allow for two modes of operation that differ by a gear ratio of 16. Mechanism Architecture The dual-output spindle actuator provides motion at its low - and high- gearing output simultaneously . These outputs, which are offset in different planes along the actuator output axis, are shown in Figure 2. More detail on this actuator is provided i n a later section of this paper. STIG interfaces with one actuator output at a time and couples that motion to the coring, abrading, or regolith bit. The STIG interface to the actuator, including a specific STIG part named the selector which translates axi ally to access one of the two actuator outputs, is shown in F igure 3. The completed ST IG assembly is shown in Figure 4. * Jet Propulsion Laboratory, California I nst. of Technology, Pasadena, CA; timothy.j.szwarc@jpl.nasa.gov ** Sierra Nevada Corp., Durham, NC; jonathan.parker@sncorp.com + CEROBEAR GmbH, Herzogenrath, Germany; johannes.kreuser@cerobear.com Proceedings of the 45th Aerospace Mechanisms Symposium, NASA J ohnson Space Center, 2020

Document

AMS_2020.pdf

180 Figure 1. The STIG mechanism, spindle actuator, and Core Break Lockout (CBLO) actuator are shown in the context of the Mars 2020 corer. Figure 2. The output interfaces of the dual -output gearhead , part of the spindle actuator . The large, outer internal spline provides high torque. The smaller, external spline undergoes fewer planetary gearing stages and outputs at a faster speed. The high speed output is recessed deeper into the actuator than the high torque output . This difference in depth is essential to the function of STIG .

Document

AMS_2020.pdf

181 Figure 3. The STIG input interfaces . The large, three -tooth external spline interfaces with the actuator’s high torque output when the selector is retracted into STIG . The smaller, ten- toothed spline interfaces with the high speed output when the selector is extended further into the actuator . The gear in view is invol ved in positioning the selector, but it is not part of the torque- carrying path between actuator and drill bit. Figure 4. Completed STIG flight assembly . In this image, the left side of the mechanism outputs torque via the large gear to the spindle gear train, and ultimately to the drill bit . The right side of the image interfaces to the dual -output spindle actuator as well as to a separate gear train that determines gear shifts . Also visible in the image are a heater, two platinu m resistance thermometers, and their associated cabling . Overall STIG dimens ions are approximately 120 mm in length and 60 mm in diameter, with some flanges extending further to serve as covers for other corer components.

Document

AMS_2020.pdf

182 Rather than having an additional, dedicated actuator that determines the axial position of the selector, STIG references the position of a separate mechanism that is positioned according to whether drilling or core break operations are occurring. This aforementioned mechanism is named c ore break lockout (CBLO). The CBLO mechanism’s actuator output is attached to a single piece part containing both a lead screw and a spur gear . More detail about this actuator can be found in [2]. The lead screw/ gear is shown in Figure 5. Figure 5. CBLO actuator and combined lead screw/spur gear output piece part . The lead screw drives a mechanism that interact s with the sample tube. The gear (via an idler) connects to STIG, allowing the selector to interface to the proper dual -output actuator interf ace. The CBLO lead nut is extended during core break operations, allowing the CBLO mechanism to interact with the sample tube and fix it rotationally, enabling shear of the rock core sample as the drill bit rotates with high torque. During other operations such as drilling, CBLO is retracted to avoid contact with the sample tube. As the lead screw operates during CBLO motion, the spur gear on the same piece part mounted to the actuator output rotates, which is an input to STIG. Figure 6 provides an overall schematic for the STIG architecture . This gear does not provide torque to the drill bit, rather, it acts in translating the STIG selector axially so that the selector interfaces with the desired output of the spindle actuator. The task of STIG is to use knowledge of the CBLO mechanism’s position to place the selector at the appropriate output of the spindle actuator, in order to provide the appropriate torque and speed to the bit .

Document

AMS_2020.pdf

183 Figure 6. STIG utilizes the positi on of the CBLO actuator to deter mine which output of the spindle is connected to the coring bit . Other bits may be used as well . During core break, CBLO extends into the sample tube, fixing it rotationally . Core break torque is applied t o the bit, creating eccentric motion and shear forc es in the rock core. Mechanism Function Figure 7 shows the STIG design in detail, with several piece parts and features labeled. STIG’s lead screw is not directly attached to the selector. Rather, the lead screw and selector are connected by a spring that is preloaded in a state of compression. This use of this pr eloaded spring has several advantages . First when the selector is moved to a given position, there is 37 N of force holding the selector in place. This force is sufficient to prevent relative motion between the selector and other drill components during the roughly 6 G accelerations produced by rotary -percussive coring and abrading, keeping the selector fully engaged with the intended spindle actuator output . More detail about the corer’s percus sion mechanism is available in [3]. This preload also allows the selector to be held in place while in the launch stow configuration. In addition, the spring can be further compressed past its preload once the selector reaches the end of its travel in either direction. This allows CBLO to be commandable to four different positions, with two positions mapping to each of the two spindle actuator outputs. For example, CBLO can be posit ioned in the drilling state, which positions the selector in the high speed mode. But if CBLO continues to retract in order to perform a homing operation, whereby a non- rotating part of the CBLO mechanism seats into a groove in a spindle piece part and locks out motion, the STIG lead screw can be further retracted. The STIG spring will be further compressed, but the selector remains in place and mated with the high speed actuator output . Conversely, while the sel ector remains fully en gaged with the spindle actuator high torque interface, CBLO can either be fully extended to interface with the sample tube during core break operations, or only partially extended in order to provide breakoff torque to the bit without locking out the tube, which could be useful in recovering from a situation where a bit becomes stuck in the rock.

Document

AMS_2020.pdf

184 Figure 7. STIG parts in detail . Red parts translate axiall y (and in some cases, rotate) . Magenta parts rotate without translating. The selector (dark blue) is supported on its right side by the spindle actuator (not shown) and on its left side by a bushing mounted to corer structure (not shown) . The right end of the output shaft is supported by splines tha t ride on the outside of the sel ector and on its left side by corer structure (not shown). Finally, the spring provides the ability to shift gears while being agnostic to particular clockings of the selector and actua tor outputs. As the STIG lead sc rew retracts, in the example of a shift from high torque to high speed, the selector initially tracks the motion of the lead screw due to the preloaded spring. This simultaneous motion continues until the selector’s high speed splines contact the splines of the actuator high speed output. Due to the differences in acutat or output gearing ratios and differences in backlash, no assumptions are made about the relative clocking of the splines of the two output interfaces before a shift. The lead screw continues to retract while the selector is prevented from moving, further compressing the spring. Once the lead screw reaches its intended positon, the spindle is slowly rotated until the splines align, allowing the selector to translate under the force of the sprin g preload into the fully -mated state. It is possible that friction between the contacting axial faces of the actuator and selector is capable of sustaining torque, so the selector may rotate with the actuator instead of remaining in place rotationally and waiting for spline alignment . In order to guarantee successful mating, a previous benefit of the STIG spring is utilized, namely, the ability for CBLO to travel to two locations for a given STIG gear. By driving CBLO into its homing state and locking the STIG/spindle gear train, relative motion can be enforced between STIG and the actuator. When the splines align and the selector moves axially into the mated position, a stall will be detected by the actuator as it attempt s to drive a locked gear train. At that point, the gear shift is guaranteed to be complete and CBLO can be extended to unlock the gear train without moving STIG out of the high speed gear . The motion of the spring and a shifting example are shown step- by-step in Figure 8. A similar sequence of events is used to attain high torque gearing, whereby CBLO locks the tube rotationally to initiate a stall between the sp indle actuator and the gear train.

Document

AMS_2020.pdf

185 Figure 8. STIG is shown shifting from high torque to high speed. A: STIG in in high torque mode, with the the selector interfacing with the spindle actuator’s high torque output . Note that some corer components that provide support to stig components are not shown in the images. B: The lead screw and attached components begin retracting (moving right) . The selector follows. C: The lead screw and attached components continue to their final position, but are agnostic to the fact that the selector had bottomed out on the high speed spindle output and stopped translating. A spring is compressed during this difference in axial travel between the lead screw and selector. D: The actuator is rotated. At some point, the high speed splines align. The dotted line represents the extent of the splines in this cross section before actuator rotation. E: The selector is immediately pushed toward, and interfaces with, the high speed output . The spring partially relaxes . The corer is now in the homing configuration and STIG is in high speed.

Document

AMS_2020.pdf

186 Mechanism Advantages The STIG mechanism has several features that make it well -suited for its function within the coring drill. Simple and Robust Shifting Algorithm Initiating a gear shift with STIG is essentially two steps. The STIG lead screw is pos itioned according to the desired dual-output actuato r ratio. Then the spindle actuator is rotated until a stall is sensed. Because a spring is used in series between the lead screw and the selector, there is no need for alignment of the spindle actuator out puts before initiating a shift. In other words, it is not necessary to ensure that the selector can shift out of one gear and directly into the other. Instead, the spring preload force pulls the selector out of one interface and into a neutral zone, and li ghtly loads the selector against the desired actuator output until alignment occurs at a later time. Without the spring, loss of position knowledge during a rover fault would be extremely problematic, as no force sensor is in place to allow a hunt -and-peck style of shifting . No Need for a Dedicated Gear Selection Actuator Because of the connection between the STIG and CBLO lead screws, the STIG gear cannot be chosen independently of the corer’s CBLO position and independently of operations related to the sample tube. This interconnectiv ity allows for STIG to be incorporated into the corer without the need for a dedicated gear selection actuator. Although it may initially appear that this would limit the number of corer configurations possible (one CBLO pos ition while STIG interfaces with the high speed spindle output, one CBLO position while STIG interfaces with the high torque spindle output), two STIG configurations map to four CBLO positions. The STIG spring allows this mapping. Full Functionality in B oth Output Rotational Directions STIG retains the ability to provide motion in either rotational direction to the bit at either torque level by avoiding the use of ratchets or impact drivers . This ability greatly benefits sampling operation algorithms and fault recoveries. For example, the bit can be rotated in the non- drilling direction while being retracted from a borehole, reducing the chance of becoming stuck. Additionally, if the bit becomes stuck during high torque breakoff, the bit can be rotated with high torque in the opposi te direction to alleviate a jam. No Sacrificial Components A pressure of single digit MPa is all that is sustained by the axial faces of the selector and actuator during a shift . Contact geometry at the interface is planar. Because of the low contact pressure and presence of wet lubrication (Braycote 600EF), there is no need for a clutch. Mechanism life is not expected to be driven by any sliding interface. As mentioned earlier, an actuator stall indicates the completion of a gear shift. So during a shift and prior to said stall, spindle speeds may be kept very low. Because of these low speeds and the low axial pressure applied to the parts, no debris generation is expected. The slow spindle speeds used while seeking a stall are necessary before commanding high speed motion that could otherwise damage the splines without full engagement . Dual -Output Spindle Actuator The dual -output actuator consists of a four-stage planetary gearbox, with a unique concentric output shaft arrangement. A cross section can be seen in Figure 9. Several challenging requirements included a bit - seizure fault load case of 68 Nm on the high speed shaft, which is approximately 10X the n ominal operating torque load , while minimizing mass, overall envelope size and maintaining low overall drag torque over a wide operational temperature range of - 70°C to +70°C. A typical gearbox design would use ball bearings to support the planet gears in the first stage and increment up to higher capacity bearings as the stages progress to the output stage. The spindle gearbox used a ball bearing first stage, while the second and fourth stages required a larger size ceramic hybrid needle roller bearing to handle the bit -seizure case and the high loads of the low -speed output shaft. The third stage utilized a smaller sized roller bearing to minimize mass and overall size. Traditional approaches to support

Document

AMS_2020.pdf

187 a highly loaded planet, a double- row ball bearing or porous bronze bushing, would have resulted in an unacceptable increase to the gearbox volume and mass . Figure 9. Cross Section of Dual -Output Actuator . Credit: Sierra Nevada Corporation While the hybrid ceramic needle roller was show n to be capable of withstanding the demanding load requirements, another concern early in the design phase was the roller bearing’s impact on drag torque. Its location in the second stage of the gear train required rotation at approximately 1000 rpm during drilling operations, at a temperature extreme of - 70°C. Braycote 600EF grease was selected as the baseline grease lubricant for the gearbox due to its performance at low temperatures and widespread space heritage. In order to validate the bearing selection and lubrication scheme, a bearing torque test over temperature was necessary. A single bearing expedited from the vendor and lubricated in- house with a 10-15% fill by volume of grease was tested. The breakaway torque as well as the running torque at maximum speed of 1000 rpm at - 67 °C measured approximately 17.6 mNm. This test was performed such that the bearing was operated at the hot operational temperature extreme of 70 °C to distribute the grease and the n taken to the cold temperature extreme in a static condition to avoid disturbing the grease. Then, the breakaway torque was measured prior to measuring running torque. Gear train drag torques reflected back to the motor were calculated and found to be wi thin the bounds of the motor performance specification. Key to meeting the bearing performance was the use of advanced materials in the bearing design. The roller bearings utilized Cronidur X30 raceways, PEEK cages , and Silicon Nitride rollers. Sierra Nevada Corporation had successfully implemented these material combinations on a prior program but they lacked widespread space heritage, especially given the operating temperature extremes. As such, the results of the life test and outcome of the teardown ins pection were of great interest. The loads in the bearings resulted in several operating points that were within the high risk region for Bray grease. The cumulative degradation factor (CDF) however was <1.0 for each of the three roller bearings used in the design [4]. Figure 10 plots the cumulative stress cycles for a 2X life of each bearing relative to the CDF factors of 2 and 8 in addition against guidelines established by NASA CP- 3062 [5]. For the purposes of this figure, a stress cycle is calculated si milar to a ‘ball pass’ in a ball bearing.

Document

AMS_2020.pdf

188 Figure 10: Mean Compressive Stress Cycles . Credit: Sierra Nevada Corporation Folloiwng the qualif ication test program, the actuator met all performance specifications. The testing included static loading, which subjected the roller bearing to an estimated 3476 MPa ( 504 ksi) max hertzian contact stress prior to any operation, random vibration testing, shock testing, and conclud ed with a 2X lif e test. The life test consisted of 10 thermal cycles, the first five cycles tested the high- speed output, while the next five cycles tested the low -speed output. The number of revolutions at each torque level were distributed evenly among four different temperatures ( -70°C, -55°C, +25°C , +70°C ). Figure 11: (Left) Inner r ace of roller bearing (X30) post life test. (Right) cage and rollers, post life test. Credit: Sierra Nevada Corporation

Document

AMS_2020.pdf