Datasets:

ai-aerospace

/

ams_data_full_2000-2020

like 0

Figure 4 Conceptual design of suspension (type A with active augmentation). The device consists of two main subassemblies: a moving carriage shown in shades of red and a fixed frame shown in shades of blue. The carriage moves vertically on four journal air bearings, two in each of the two cylindrical housings as shown. The carriage lifts the payload via a cable connected to its lower crossmember through a load cell as shown. In the present application, the load cells are used to monitor the force distribution among the devices when “parked”, i.e. when the air springs are deflated and the carriages rest against their lower travel stops. Uplift force against the upper crossmember of the carriage is supplied by the frictionless air piston through a connecting rod. The piston is actually a combination of a journal air bearing and a piston. A very thin, stable air film is produced around the piston skirt such that it never actually touches the cylinder, thus eliminating all friction. A small amount of air leakage is allowed with make-up air being supplied by the precision pressure regulator supplying the external accumulator tank. As noted earlier, the carriage lower crossmember is sealed against the bottom side of the baseplate by two flexible metal bellows seals which surround the two carriage rails as shown. Type A devices use four voice coil actuators operating in parallel with the air piston. Type P devices do not have voice coil actuators but have dual air cylinders for increased lifting capacity. Both Type A and Type P devices include LVDT displacement transducers to sense the carriage position, high-resolution pressure transducers for monitoring piston pressure, and load cells for monitoring the load distribution. Pistons are 146.0 mm (5.75 inches) in diameter for a bore area of 16,753 mm 2 (25.96 in2). This produces a gross uplift force of 13,890 N (3115 lbf) at a gauge piston pressure of 8.28 bar (120 psig). Type A devices are nominally rated at 11,147 N (2500 lbf) in vacuum which accounts for the weight of the moving carriage and the vacuum force pulling down on the carriage (the interior of the bell jar housing is at atmospheric pressure). The Type P devices use two pistons of this size and are rated at 24,524 N (5500 lbf) in vacuum. Vertical stroke between the travel bumpers is 3.8 cm (1.50 in) for both types. Voice coil actuators are driven by current-control analog power amplifiers using 16-bit digital commands. Force resolution of the active subsystem in a Type A device is approximately 0.36 N (0.08 lbf). The magnetic pendulum dampers (Figure 1) are built from the magnet bodies of the voice coil actuators with the wire coil replaced by a solid copper annulus. A flexure suspension is used to allow the conductor to move axially in the magnet body while remaining concentric to it. Details are in the next section. 320

Document

AMS_2008.pdf

Hardware Description Figure 5 Suspension devices. Type A, augmented (left), Type P, passive only (right). Figure 5 shows some of the delivered hardware. The Type A suspension device on the left is identified by its single piston and four voice coil actuators. The Type P on the right has dual pistons and no voice coils. The circular groove in the top side of the baseplates is for the O-ring that seals the bell jar flange to the baseplate. All air and electrical feedthroughs are located in the baseplate where the interfaces can be accurately machined. The devices are shown mounted on assembly and shipping stands. Figure 6 shows a bell jar being trial-fitted onto a Type A suspension device. Figure 7 shows a Type A during assembly. The carriage is assembled into the frame initially without the bellows seals. After the plumbing, wiring, and various small subassemblies are added, the lower crossmember is removed, the seals are mounted into place, and the crossmember is reinstalled. This allows the carriage motion to be checked for friction before the bellows are added and also protects the bellows, which are somewhat fragile, from damage during most of the assembly process. Figure 6 Trial-fitting of vacuum housing onto suspension device. 321

Document

AMS_2008.pdf

An example of the bellows seal is shown in Figure 8. It is formed bellows rather than edge-welded. While the latter type would have had the advantage of lower added stiffness, it would have been much more costly and would have had an unacceptably long delivery time. Since analysis had shown that the target suspension frequency could be met with formed bellows, that was the type used. The end flanges are machined with standard O-ring grooves as shown and then tungsten-inert gas welded to the formed bellows. The bellows are fabricated from 321 stainless steel because of its high ductility. The flanges are made from the same material for welding compatibility. Inside diameter of the bellows is 93.7 mm (3.69 in) and flange-to-flange length in the unstrained condition is 107.4 mm (4.23 in). Stiffness added by the bellows is about 2.90 N/mm (16.5 lbf/inch) per bellows. Figure 7 Partially assembled Type A device showing the carriage rails and lower crossmember. Bellows seals are installed later. An important consideration for internally pressurized, formed bellows is lateral (inchworm) buckling. Because the critical buckling pressure is always proportional to axial stiffness divided by length, there is a certain stiffness that must be tolerated. As usual, this controlled the design since the critical pressure had to be at least 1 atm. Measured buckling pressure was found to be in good agreement with theoretical predictions. Figure 9 shows two views of the passive magnetic pendulum dampers. The body of each damper was mounted to a temporary stanchion fixed to the floor of the vacuum chamber. The moving element of the damper was then connected to the paylo ad by a “stinger” having a flex joint at either end. The resemblance of the magnet bodies of the dampers to those of the voice coil actuator (Figure 5) is clear. While damping performance could have been improved by a custom design, the ad-hoc design based on the actuators was adequate so the ever-present schedule pressure dictated its use. Working stroke of the dampers was about 12.7 mm (0.5 inch). Overall length of the assembly as shown is 248.7 mm (9.79 in) and the envelope diameter is 279.4 mm (11.00 in). Weight is approximately 11.4 kg (25 lb) per damper. Figure 8 Flexible metal bellows seal. Experience in modal testing has shown that pendulum modes have practically zero inherent damping. The basic reason is that the potential energy is stored in geometric stiffness rather than elastic stiffness of a material. Hence, there is no material loss factor to dissipate energy. While relatively simple compared to the suspension devices, the passive pendulum dampers were essential to the overall success of the suspension system. 322

Document

AMS_2008.pdf

Figure 9 Passive magnetic dampers for payload pendulum modes Testing The aggressive development schedule for LSIS left time for only the most essential tests prior to shipping the hardware. These were friction and proof load tes ts of the suspension devices and leak rate tests of the devices and their accumulator tanks. Dynamic tests were performed after the fact on a spare damper for engineering purposes. This section reports methods and results for these tests. Friction tests of suspension devices. Figure 11 shows the setup for friction tests. A custom test stand (the white structure in the figure) was designed and built for the purpose. It provided a mounting location for the device under test and a means of moving the carriage up and down through a small stroke while the air spring and accumulator tank were fully pressurized to their maximum expected operating pressure. Force applied to the carriage to produce the motion and resulting displacement were transd uced and recorded. If friction was present, it would show up as a step in the force at the instant the carriage motion reversed direction. The amplitude of the step would be twice the friction force. Figure 10 shows a detail of the actuation device for moving the carriage. Initially a variable speed, DC elec tric motor was to be used to turn a crankshaft with a small, 1.27- Figure 10 Actuation mechanism for friction tests of suspension devices. 323

Document

AMS_2008.pdf

Figure 11 Test rig for measuring friction in suspension devices mm (0.050-inch) stroke. The motor speed was reduced to about 120 RPM at the crank by a two-stage timing belt drive. However, in initial trials, it was found that the SCR motor controller would not control the speed smoothly as the crank passed TDC and BDC and the motor transitioned from driving to retarding the load. Large spikes in the motor torque occurred which would have obscured the small transients due to friction, had such been present. The solution was actually a simplification to the rig. A hand lever was fabricated and mounted directly on the crankshaft to replace the motor drive. With practice, it was possible to use the lever to rotate the crank back and forth through about 180 degrees with very smooth transitions in rotational direction. This was the system used for the actual tests although the DC motor and most of the belt drive are still present in the photographs. Figure 12 and Figure 13 show typical results. Figure 13 is data obtained with a device operating correctly. The force trace, which has had its DC component removed, shows no sudden steps at the points where the displacement hits its extreme values and the direction of travel reverses. Figure 12 shows a case where there was friction. A step of about 5.3 N (1.2 lbf) occurs as the carriage reverses direction. This data was obtained in testing of a Type A device so the friction force (half the step amplitude) amounted to only about 0.019% of the average force. However, this exce eded the nominal specification for smaller devices of this type (<0.005%) so the offending part (the piston) was rejected. It was replaced with a spare unit that passed the test. Experience has shown that friction, if not due to some obvious rubbing of a visible moving part against a fixed part, is almost always due to the piston. Of the 22 pistons and cylinders fabricated, only one failed the friction test. This is actually a better yield that is typically obtained with smaller piston/cylinder sets. It appears that frictionless pistons are actually somewhat easier to build in larger sizes than in smaller. The force transducer used for the tests was a piezoelectric, charge mode type with a charge amplifier having a time constant of about 10 seconds. The extremely large dynamic range of this measurement system allowed detection of force transients well under 1 N in the presence of DC forces of 25,000 N. 324

Document

AMS_2008.pdf

Figure 12 Test result for suspension device showing friction between piston and cylinder. Figure 13 Typical test result for suspension device without friction 325

Document

AMS_2008.pdf

Leak rate tests Because both the suspension devices and their accumulator tanks were to be used in a vacuum chamber, their leak rate was obviously of concern. The usual method of leak testing, developed for vacuum chambers, involves helium bagging the devic e, pumping the interior down to vacuum, and analyzing the gases pumped out to see how much helium is present. Because the suspension devices contain air at several different pressures with vacuum outside the device, this standard method would not have been representative of operating conditions. Fortunately, the allowable leak rate was fairly generous, at least by the standards of the vacuum industry. The method used for the suspension devices was to pressurize the interior of the device to 1.01 bar (14.7 psi) above ambient, valve off the device with a low-leakage valve (Figure 14), and then record the interior pressure and temperature over several days. Mass loss from inside to outside was then calculated from the ideal gas law for air and compared to the allowable. The same method was used for the tanks except that they were pressurized to 3.45 bar (50 psig). Not surprisingly, it was found that sensing pressure alone was not adequate. Even with the device under test inside an insulated “doghouse”, pre ssure variation due to diur nal temperature variation was much larger than that due to leakage. Temperature compensation was essential to the method. Figure 14 Low-leakage vacuum valve used in leak rate tests Pressure was measured using the same ultra-high-resolution quartz crystal pressure transducers used in the ride-height control of the actual system. They use the change in natural frequency of a quartz crystal due to pressure as the sensing mechanism. A built-in digital frequency counter allows a pressure resolution of a few parts per million. A built-in serial interface allows transmission of data to the recorder or controller without loss of accuracy. While quite slow by normal analog standards, their update rate was adequate for both the leak rate tests and the ride height controller. Temperature wa s measured by RTDs with signal conditioning that included A/D converters and serial interface to the recorder. Figure 15 shows typical pressure and temperature time history data from leak testing of one of the accumulator tanks. Figure 16 shows the time history of air mass in the tank calculated from the measured temperature and pressure after subtracting the average value. Also shown in Figure 16 is a straight line whose slope represents the allowable leak rate. While the trace showing the measured mass loss is somewhat erratic, it is clear that the time-average leak rate is less than the specified maximum allowable. While some test articles failed the test initially and required some lea k chasing and debugging, all eventually passed in time for delivery per the schedule. While adequate for the purpose, the leak rate measurement method was slow and on ly practical because equipment was available to test more than one article at a time. It would not be practical for testing at the much smaller leak rates typically specified for vacuum chambers or for the bellows seals of the suspension devices. The latter were tested by standard helium mass spectrometer methods prior to assembly of the suspen sion devices. 326

Document

AMS_2008.pdf

Figure 15 Pressure (top) and temperature (bottom) of air inside an accumulator tank during leak rate test. Figure 16 Leak rate calculated from measured pressure and temperature 327

Document

AMS_2008.pdf

Proof load tests of suspension devices Each of the 14 devices was proof tested to twice its rated load capacity. A modified version of the friction test rig was used for this purpose. Shown in Figure 17, it substituted a hydraulic cylinder for the crankshaft mechanism of the friction tester. Force was determined from the hydraulic pressure applied by a hand pump. The proof loads produced severe deflection of the elastomeric travel stops built into the suspension devices (Figure 4) but they all survived and returned to normal after the load was removed. No failures occurred during proof load tests. Dynamic testing of dampers One of the passive magnetic dashpots was tested to verify its damping properties. Figure 18 shows the test rig. The body of the damper containing the magnets and back iron (the shiny object in the photo) is mounted to a bulkhead in the center of the aluminum frame of the rig. The conductor is mounted on a carriage rail that forms part of the rig. The rail runs on two journal air bearings to eliminate any friction. One end of the rail (the right end in the photo) protrudes through the end bulkhead of the test rig frame. It is driven axially by an electrodynamic shaker working through a piezoelectric load cell. The other end of the carriage rail mounts an aluminum disk that serves as a target for a non-contact eddy current displa cement sensor. The eddy current probe mounts to the left-end bulkhead of the test rig frame. The shaker is driven with a band-limited random current and the signals from the force and displacement sensors are input to a multi-channel digital Fourier analysis system. It computes a frequency response having carriage velocity as the input (denominator) quantity and force as the response (numerator) quantity. In the process, it removes from the force signal the portion that is due to the known inertia of the moving carriage. What remains is the drag force from the magnetic damper only. The result is a complex-valued function of frequency whose magnitude repre sents the damping force per unit of velocity (often called the dashpot constant, although it is anything but constant in this case). The phase of the complex function represents the phase of the damping force relative to the velocity. Figure 17 Setup for proof load test of suspension device Figure 18 Test rig for measuring complex stiffness of magnetic dampers. 328

Document

AMS_2008.pdf

Figure 19 Measured force/velocity ratio for the magnetic damper Figure 19 shows the magnitude of the measured frequency response function for three configurations of the conductor of the damper. The uppermost (red) curve is for the dampers as delivered and installed with the suspension system. These used a single, solid-copper annulus (also called a single shorted turn) as the conductor. The middle (green) curve is for a conductor formed from the same wire-wound coil used in the voice-coil actuators but with the wire ends shorted together. The blue curve is for the aluminum coil form (also called the bobbin) used in the voice-coil actuators but with no wire windings. Not surprisingly, the force/velocity ratio increases as the resistance in the induced current path is decreased, just as theory predicts. The back iron assembly of the voice-coil actuator (identical to that of the damper) was of a design that produced relatively high inductance in the coil. For the actuator, this was not important since it is used essentially only at zero frequency. For the damper, it was expected that this would cause the induced currents in the damper conductor (which cause the damping force) to drop off with increasing frequency. This would be valuable for the present application since it would allow the magnetic dashpots to damp the very low frequency pendulum mode (less than 1 Hz) but then disappear at higher frequency so as not to transmit disturbing forces into the payload at higher frequencies. The red curve of Figure 19 shows that the design was quite successful. At frequencies over about 50 Hz where the resonances of the payload begin, the force/velocity ratio is reduced by about an order of magnitude from its DC value. Conclusion Figure 20 shows eight of the fourteen suspension devices mounted in a framework, ready to be suspended from the roof of the vacuum chamber. These are the eight that support the Body C of Figure 1. While schedule constraints did not allow systematic tests, informal measurements of the vertical frequency by the end user indicated that it was, in fact, 0.50 Hz as designed. Most important, the degree of vibration isolation provided was sufficient to allow the functional tests of the spacecraft to be performed without limitations from ambient disturbances. The LSIS did its job. 329

Document

AMS_2008.pdf

The primary conclusion from the development was that the very low stiffness pneumatic suspension devices are scalable. In fact, anecdotal evidence suggests that, at least up to the largest sizes built to date, obtaining the critical zero-friction behavior is actually easier in larger devices (piston bores over about 76.2 mm (3 inches)) than in smaller ones. A system is now in preliminary design that will use multiple devices with piston bores over 203.2 mm (8 inches) to float a payload of approximately 45,454 kg (100,000 lb) in vacuum. Figure 20 Eight suspension devices with accumulator tanks mounted in frame for supporting Body C. Size is evident from the person in the lower right of picture. References 1. Kienholz, D.A. “Simulation of the Zero-Gravity Environment for Dynamic Testing of Structures,” Proc. 19th IEST Space Simulation Conference, Oct 28-31, 1998, Baltimore, MD 2. Sillls,Jr.J.W., Voorhees,C.R. “Characterization and Application of Pneumatic Suspension Devices for Vibration Disturbance Testing,” Pro ceedings of the 20th International Modal Analysis Conference, February, 2002. 330

Document

AMS_2008.pdf

Gas Strut Separation Alternative for Ares I Brian Floyd* and James Owens Abstract This paper presents a design alternative and the rationale for a stage separation system based on Metering Adiabatic Gas Struts (MAG Struts) for the Ares 1 launch vehicle. The MAG Strut separation system was proposed as an alternative to the current Ares 1 separation system, which relies on small solid rocket motors to provide the main separation force. This paper will describe technical issues that were addressed during the trade study and present a conceptual design of the strut system that best resolved the issues. Needed development testing and programmatic considerations will be addressed as part of the paper. Introduction Gas struts show promise as an efficient way to prov ide the separation force for launch vehicle staging. Strut systems are currently in use on a number of vehicles, but so far all have been unmanned. Several factors make the MAG Strut system unique. The struts are entirely self-contained. They are themselves pressure vessels, which are pre-charged with gas prior to launch. They require no additional actuation, but simply act as springs when the physical connection between stages is severed. Due to the mass properties of the separating sta ges, this system provides excellent no zzle clearance during fly-out in offnominal conditions. Consequently, safety and mission success objectives are enhanced. Since the struts are light weight relative to other separation systems capable of applying the same force, the separation timing can be adjusted to separate earlier during the assent trajectory, increasing payload lift capability. The proposed struts apply the separation force smoothly du ring release in order to minimize disturbance of the Upper Stage propellant and reduce the buckling loads applied to the upper stage aft skirt. The trade study also predicts significantly lower life-cycle-cost. Since the MAG Strut system is not in flight operation on any launch vehicle, development testing and system-qualification introduce some risk into the Ares program, which is a barrier to adopting the system. Background The Ares I launch vehicle will lift the Orion crew vehicle to low-earth orbit for manned missions to the International Space Station and to the moon. Ares I consists of two stages. The first stage is a modified Space Shuttle Solid Rocket Booster (SRB) with 5 solid motor segments instead of the 4 segments currently used for shuttle. The Ares I upper stage is a LOx / LH2 stage powered by a J-2X engine. The stages are connected by a cylindrical interstage and a conical frustum. The J-2X engine is housed in the compartment formed by the interstage and frustum. Figure 1 - Ares I In the current flight trajectory baseline, the first stage ascent phase ends when the first stage reaches 178 kN of residual thrust. Eight Booster Deceleration Motors (BDMs) fire to push the first stage aft. Eight Ullage Settling Motors (USMs) thrust forward to maintain positive acceleration on the upper stage. Once * NASA Marshall Space Flight Center, Huntsville, AL Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Space Flight Center, May 7-9, 2008 331

Document

AMS_2008.pdf

the USMs and BDMs are ignited, a pyrotechnic joint at the forward end of the interstage initiates and the vehicle begins to separate. Figure 1 shows the Ares I configuration with the BDMs mounted on the interstage. In the most recent configuration, they are relocated to the aft skirt of the first stage. The J-2X nozzle exit plane is 7.1 meters aft of the separation plane. With the current arrangement separation system, it takes approximately 1.7 seconds for the nozzle to pass the forward end of the interstage. Nozzle Clearance During Fly-Out Considerations Many factors affect the amount of radial clearance between the engine nozzle and the interstage wall during the fly-out. The most significant factor contributing to clearance issues for BDM separation is asymmetric plume impingement force on the first stage that can occur if one motor fails to fire. Secondly, since the first stage has 178 kN of residual thrust at the time of separation, significant pitching and yawing loads may be imposed on the stack before separation and on the first stage after separation due to thrust vector pointing uncertainties. With one BDM out, a worst-on-worst analysis of the separation shows contact between the interstage and the engine nozzle during fly-out. Monte Carlo analysis of this scenario shows that nozzle clearance can only be demonstrated to a 2.5-sigma level. The proposed MAG Strut system uses eight gas-charged struts mounted inside the interstage to force the two stages apart. The struts essentially act as alignment guides during separation. Figure 2 shows the relative position of the struts on the interstage to the USMs and the BDMs they will replace. Figure 2 - Interstage showing BDMs and Struts Although the struts extend above the separation plane, they provide superior clearance, even with one strut out. The primary reason for this superior pe rformance is that the mass-moment-of-inertia of the Ares I upper stage/crew vehicle is approximately ½ that of the mass-moment-of-inertia of the expended first stage, while the distance from the upper stage/crew vehicle center-of-gravity to the J2 nozzle exit plane is approximately ½ the distance of the center-of-gravity of the first stage to the separation plane. Figure 3 shows the relative positions of the centers-of-gravity of the separated stages to the nozzle exit plane and first stage separation plane. With a strut system, any disturbance force, regardless of its origin, is compensated for by the struts, forcing the separated stages to rotate in the opposite directions. The rate-of-rotation, W, induced on the two bodies in always close to 2/1 with the upper stage/crew vehicle rotating at twice the rate of that of the first stage. The rate of rotation of each body is small with the gas strut system. Distance D3 is considerably larger than distance D4 so some of the disturbance force coming from the first stage results in translating the upper stage in the same direction the interstage is moving. This translation effect, though beneficial, is not as significant as the rotational compensation. 332

Document

AMS_2008.pdf

Figure 3 - Comparison of Gas Strut Separation and BDM Separation Figure 4 - Ares I Separation Clearance using Gas Strut System Figure 4 shows the preliminary clearance results for the Ares I upper stage engine nozzle with one strut out. The WOW*1.5 curve represents a worst-on-worst assessment of the radial clearance with a margin of 50% added to account for unknowns in the analysis. Even in this conservative case, the nozzle clears the extended end of the strut by 45.7 cm. The dash lines represent WOW case clearances for different failed struts with different disturbance scenarios. Two seals must fail on the same strut to result in a 100% 333

Document

AMS_2008.pdf

pressure loss. Based on the analytical results, one strut failure cannot result in the loss of an Ares I mission due to nozzle contact. Consequently, the MAG Strut system is inherently two-fault tolerant. Plume Heating on Upper Stage At the Ares I System Definition Review, the vehicle was configured with BDMs mounted near the aft end of the interstage in four pods containing two motors each. The USMs were mounted on the upper stage aft skirt, also in four pods of two at the same angul ar positions around the cylinder. One problem with this configuration is the interaction of the USM and BDM plumes. Even though the nozzle exit planes were separated by over 4.5 meters axially, extreme heating was predicted in the upper stage engine compartment during separation because the BDM plumes deflect the USM plumes into the interior of the interstage. Also, debris generated by the separation pyrotechnics will likely be propelled into the engine compartment by the interacting plumes. The use of gas struts eliminates these debris and heating concerns. Relocating the BDMs to the first stage aft skirt would resolve this issue. Payload-to-Orbit Benefits Gas strut separation produces a significant increase in payload-to-orbit capability. This gain is a result of reduced aerodynamic drag, momentum transfer between the stages, and ascent trajectory optimization. The interstage-mounted BDM pods are the largest protrusions from the nominal outer moldline (OML) of the vehicle. As such they account for a total of a 110 to 120 kilogram payload pe nalty due to aerodynamic drag. The proximity of the BDMs to transition from the conical to cylindrical is a major factor in the high drag. Locating the struts inside the interstage eliminates all aerodynamic drag effects. For the baseline trajectory, the amou nt of residual first stage thrust at separation is limited by the capability of the BDMs. For an 8 BDM configuratio n with one motor out, separation must wait until first stage thrust drops to 178 kN. Because the struts have a better weight to performance ratio than BDMs, the trajectory can be optimized to improve performance. Figure 5 indicates the amount of payload that can be gained relative to the baseline flight profile. The steeper section of the curve indicates a significant payload improvement, but the strut system mass (including additional upper stage structural mass) begins to offset the benefit as residual thrust increases. Sepa ration at 356 kN of residual first stage thrust is thought to be optimum for Ares I. This results in approximately 90-kg additional payload due to improved trajectory performance. Figure 5 - Payload Delta from Baseline vs. Separation Thrust 334

Document

AMS_2008.pdf

During separation with gas struts, the first stage thrust continues to act on the upper stage until the end of the stroke. Initial calculations show that this momentum transfer adds payload performance at a rate of 8.93 kilograms for every meter per second of /g507V. Preliminary strut designs result in an increase in upper stage /g507V of 3 to 3.7 meters per second. This amounts to 27 to 33 kilograms of additional payload. Figure 6 shows the relative velocity gained by the upper stage for a separation with 356 kN of residual thrust. Figure 6 - Momentum Transfer Effects for a Separation at 356-kN SRB Thrust The mass of the struts and upper stage fittings for a 356 kN thrust separation are about half that of a BDM system that separates at 178 kN of residual thrust; however, because more of the mass remains with the upper stage, no additional payload advantage from the change in system mass is realized. Table 1 - Approximate Payload Benefit Reduced Drag 110 kg Earlier Separation 90 kg Momentum Transfer 27 kg Mass Delta Benefit 0 kg Total Payload Benefit 227 kg Cost Considerations The projected unit cost for each BDM is approximately $200,000. There are many reasons for this high cost. One of the most risky processes of solid rocket motor manufacturing is the casting and curing of the solid rocket propellant. The process is very hazardous and requires extensive risk mitigation to prevent inadvertent propellant ignition. The risk mitigation techniques are well known, and accidents are now rare, but the process is expensive. Additionally, post-casting inspection sometimes reveals defects in the cast propellant. If a defect is found, most often the motor is discarded. Per unit cost for gas struts should be significantly less than BDMs, since there is no hazardous material to procure and handle. Also each flight unit can be acceptance tested, so manufacturing will not require the strict process control necess ary for solid motors. If a defect is discovered during the acceptance testing, 335

Document

AMS_2008.pdf

in most cases the strut could be saved by simply reworking or replacing the defective parts. In addition, since the struts are inert until they are pressurized, ground handling hazards are eliminated, making handling a low-cost operation. Parametric cost modeling bases the estimated cost on weight and similarities to selected components for which cost are available. Since the struts are half the weight of BDMs, they would be half the cost assuming equal complexity. This is the only level of cost analysis that is possible given the maturity of the MAG Strut design. Actual per-unit cost would need to be reevaluated after developed units have been fabricated and the design finalized. MAG Strut Design The MAG Strut struts are designed to take advantage of the increase in payload to orbit by separating at 356-kN residual first stage thrust. To achieve this, a significant force is required. Consequently, the struts can place a substantial bendi ng moment into the edge of the aft skirt, increasing the potential for buckling during ascent. Also, sudden release of the energy stored in the struts could result in a significant jerk to the upper stage, which could affect propellant quality and tank pressure. The MAG Strut design is proposed in order to counter these effects. During ascent, only a low pressure acts against the upper stage aft skirt. At separation, the force applied increases gradually, which minimizes potential for skirt buckling and mitigates concerns about sloshing induced in the propellant tanks. The MAG Struts are designed with two chambers as shown in Figure 7. The low-pressure chamber is meant to provide the initial force requirement for separation. The initial force calculation for each strut would be as follows: N kPacm cmkPacm124,68 034,1**4) 62.7 78.17(342,10**4) 62.7(2 2 2 /g32 /g176/g191/g176/g190/g189 /g176/g175/g176/g174/g173 /g187 /g188/g186 /g171 /g172/g170 /g16/g14 /g176/g191/g176/g190/g189 /g176/g175/g176/g174/g173 /g187 /g188/g186 /g171 /g172/g170/g83 /g83 With 8 struts, the force of 545 kN is more than sufficient to overcome a SRM residual thrust of 356 kN and the transient oscillatory force from the SRM, and therefore preventing re-contact of the two stages during separation. (See Figure 8 for a plot of the transient oscillatory thrust of the Ares 1 first stage.) The high-pressure chamber is intended to store the gas needed for the main part of the strut stroke. After 40 cm of stroke, this force reaches 1,495 kN. This force is capable of driving the first stage and upper stage apart with sufficient velocity margin to achieve separation with a residual first stage thrust of 356 kN. Figure 7 – Schematic of the Proposed MAG Strut Design The metering rod has a pattern of holes that are exposed as the strut strokes, providing a gradual force buildup that will minimize impulse on the upper stage. Figure 9 shows a computer-aided design (CAD) rendering of the strut in the collapsed position. Figure 10 shows a CAD rendering of the strut in the extended position. Initially no holes are exposed. Once the strut has stroked 2.54 cm, 6 holes are exposed. Figure 11 shows the cumulative area for the exposed holes as a function of stroke. Every 2.5 cm of additional stroke exposes more holes to achieve the gradual force build-up. (The summation of 336

Document

AMS_2008.pdf

the total exposed hole-area for two different hole-sizes in shown at the bottom of the chart.) A large range of force profiles is possible with different hole-patterns. Holes larger than the “O” ring seal diameter would likely catch the seal, causing damage during stroking. A hole diameter of 3.96 mm would be the largest recommended hole size for a seal with a 4.83-mm diameter cross-section. Figure 8 – Average Thrust and Oscillatory Thrust Test Data for 5 Segment SRM Figure 9 - Strut Rendering (Collapsed) Figure 10 - Strut Rendering (Extended) 337

Document

AMS_2008.pdf

Figure 11 – Exposed Hole Area for Two Candidate Hole-Patterns If the low pressure chamber is allowed to be at ambient pressure by providing a very small hole to the exterior of the strut, the strut can operate with only one pr e-pressurized volume. This variation would make it possible to charge only on e chamber prior to launch, elimin ating some potential failure modes. A strut with a 9.208-cm diameter metering rod and with no pressure in the small chamber would provide slightly more initial separation force th an the strut shown in Figure 7. This strut variant opens up the possibility of designing a hermetically sealed strut or other point design. Since the desired thrust profile for the struts is based on requirements de rived from a fluids analysis of the hydrogen tank pressure, having a strut capable of accommodating a range of force profiles is preferable. For a -147 degree C initial ullage gas charge temperature, an acceleration rate of change of 2.5g per second is acceptable. A higher axial rate of change may be acceptable with the currently proposed -220 to -250 degree C pre-charge gas. Table 2 shows the predicted effect of lowering pre-charge gas temperature on the make-up gas required to recover from an ullage collapse. A change out of metering rods could adapt a set of struts to revis ed ullage requirements. Sloshing risk increases as the axial acceleration of the rocket diminishes. Surface tension and vibration force the fluid in the tank up the tank walls as shown in Figure 12. Stage separation with 356 kN of residual thrust assures that the average axial acceleration never drops below .12g. This is enough acceleration to force the ullage gas to remain in a hemispherical shape bubble. The MAG Strut system further mitigates the risk of ullage collapse by limiting the axial acceleration rate of change. Table 2 – Hydrogen Tank Recovery Gas Requirements Initial tanked He assumptions: T=-250 C; P=22,00 kpa Supply assumption: Isentropic Blowdown P=6,895 kpa H2 pre-press temp Mass for ullage recovery Initial storage density Final storage density Delta density Storage volume requiredBottle massTotal loaded He mass Total Mass 19 C -181 C -220 C -250 C 226.9 kg 115.7 kg 0.00.0192.38 kg/m 3 192.38 kg/m3 192.38 kg/m3 192.38 kg/m3144.81 kg/m3 144.81 kg/m3 144.81 kg/m3 144.81 kg/m347.58 kg/m3 47.58 kg/m3 47.58 kg/m3 47.58 kg/m3.486 m2 .248 m2 0.00 0.001,390.0 kg 708.6 kg 0.00.0917.2 kg 467.5 kg 0.00.03,642.2 kg 1,176.5 kg 0.00.0 338

Document

AMS_2008.pdf

Figure 12 - Ullage Gas Collapse Mitigation Proposals Real fluid analytical tools show that the smaller holes produce a force-profi le that does not exceed 8,896.4 kN per second level as shown in Figure 13. The force-profile has some irregularities that can be eliminated through further refinement of the hole-pattern. T he force spike at .4 seconds indicates that a few more holes are needed in the last 7.62 cm of sto ke for the 3.18-mm diameter holes. If the first row of holes were exposed after 1.27 cm of stroke rather than 2.54 cm of stroke, more energy could be recovered from the expanding gas. If a few less holes were exposed in the middle part of the metering rod, the rate of change peak could be lowered. For Ares I, the 3.18-mm diameter holes shown in this plot meet a 2.5 g/sec jerk requirement if the decay of the thrust of the SRB is considered. Figure 13 – Force Rate of Change Plot 339

Document

AMS_2008.pdf

Figure 14 and Figure 15 show the force profile analytical results for the same two hole-patterns as a function of stroke as well as a function of time respectively. Figure 14 – Strut Force as a Function of Stroke Figure 15 – Force as a Function of Time 340

Document

AMS_2008.pdf

Development Program Goals and Objectives Since gas struts have not been used for separation on a manned vehicle, development testing is needed to mitigate risk. The risk falls into three categories; performance related risk, reliability related risk, and programmatic risk. Programmatic risk is in some ways a sub-set of the stated technical risk because technical issues that arise in the strut development program could threaten the schedule for the launch of Ares I flight tests. This concern is one of the chief objections to this technology. A realistic approach to address this programmatic risk is to carry both BDMs and struts in the program until struts have demonstrated their capability. The struts are a bolt-on technology, using the existing hole patterns on the upper part of the Ares I interstage attach ring and a direct bolt through on the upper stage aft skirt, so they can be installed with little impact on other systems. The recurring cost of the struts will not likely increase because of the development program. Because of development testing, the qualification program cost for a strut separation system will be substantially reduced. Programmatic-risks are addressed in this paper by eliminating technical risk throug h a robust development test program. Resolving Performance Related Risk The metering function of the MAG Strut system is determined by the size and pattern of holes along the metering rod. Development testing is required to characterize the strut performance with different metering rods under different co nditions that simulate nominal operations and potential failures. Mathematical models provide solid indications of the flow rates for struts with various metering rods; however, their accuracy is not good enough to use for qualification by analysis. The development testing would provide data that would validate the analytical flow models. The best way to establish the force vs. distance performance characteristics of the struts is to test them with several different metering rods moving different masses. A range of pressures could also be investigated to establish the performance characteristics of the struts under nominal and degraded performance scenarios. A relatively simple test set-up as shown in Figure 16 is required to perform the development testing. In this performance test, a mass of approximately 22,680 kg is released to be pushed by the strut. It will accelerate to approximately 6.17 meters per second and then disengage from the fitting mounted on the mass. After disengagement, the moving mass must be stopped by a snubber. Side forces acting against the fitting will be simulated by attaching a spring to the mass that ap plies a side force as it rolls down the track on its metal wheels. High-speed video recording will measure any twang or motion oscillations. Figure 16 – MAG Strut Performance Development Test Set-up The development program would seek to characterize the performance of the struts for several separate side force profiles that would represent a range of operational possibilities and off nominal load cases. The strut has Teflon slides on the piston and in the rod housing. If sufficient side force was present, a strut that was pressurized to less than 10% of the design pressure may bind at some point during the stroke of the strut. The mating conical interface of the rod fitting and the spike fitting on the upper stage is intended to gradually relieve side force as the struts disengage. If binding occurred on a partially charged strut, this side load relief action is intended to preclude disengagement of the strut from the fitting while pressurized. Figure 17 shows the strut rod fitting and the spike fitting that is mounted to the upper stage. Because no failure scenario has been identified that indicates that binding is a problem, development 341

Document

AMS_2008.pdf

testing will establish the amount of side loading required to cause the strut to bind such that the load relief action from the conical interfaces will not be adequate to relieve it. Figure 17 – Strut Fitting Aft Skirt Mounted Spike Fitting Resolving Reliability Related Concerns The safety of the struts must be demonstrated by test. The struts are designed to leak before burst; however, only testing can demonstrate this. If the leak before burst design is proven prior to qualification, the potential for a costly redesign and schedule slip is avoided. After completion of testing, one or more of the test struts would be subjected to extreme pressure until leakage or burst occurred. This burst test would be done with an oil or water charge to avoid the explosive hazards associated with gas. All elastomeric seals leak a minute amount of gas because of permeation of the seal material. The expected performance of each seal must be bounded in order to establish launch commit requirements and pad operations. Nominal leak rates of the seals could be established without assembly into the struts by using a test fixture as shown in Figure 18. Different elastomer compounds could be evaluated for gas permeability at the pressures used in the strut. With this data the struts could be pressurized taking into account the number of days before launch. The low pressure chamber would gain a very small amount of pressure due to seal permeation during pad operations but not enough to exceed its required operating range. Pressurizing the large volume chamber while leaving the low volume chamber at ambient pressure as discussed in the performance section of this paper would also be an option to eliminate uncertainties about rate of leakage into the low pressure chamber from the high pressure chamber. Figure 18 shows potential test configurations for two different seals. Testing 50 seals of each type would provide a large enough sample size to characterize the nature of the seals under ambient conditions. Temperature extremes could also be evaluated by placing the small seal test fixture in a thermal chamber. Analysis Needed Prior to System Testing An analysis of the integrated system would be required to establish the overall capability of the MAG Strut system to achieve separation under all potential operational scenarios. Initial analysis shows startling results with large positive clea rance margins for th e nozzle during separation. Revisiting this analysis is required prior to system testing to assure that an undiscovered disturbance force acting in the system will not cause the results to degrade. To recover the first stage, the interstage with the extended struts must be separated from the first stage. However, no analysis has been done to establish the clearance between the first stage and the interstage. The struts extend about 2.44 meters from the interstage. Consequently, their presence will make it more difficult to gain adequate clearance between the first stage and the interstage after separation of the interstage from the first stage. 342

Document

AMS_2008.pdf

Figure 18 - Proposed Seal Test Configurations Stress analysis of the second stage aft skirt interface with the spike fitting would provide a better understanding of the threat of buckling with a failed strut. If the high-pressure seal fails on a strut, the good strut will apply 68 kN of load to the structure while the failed strut will apply 236 kN of load. The safety factor is 1 for analyzing a failure case. However, the safety factor is 1.65 for buckling without a failure. Showing sufficient margin under all conditions is required prior to approving a final design configuration. A stress analysis using finite element models of the struts themselves is required to assure adequate margin exists for all components. This analysis would allow for weight optimization of the strut prior to finalizing the design. Integrated System Testing Testing the integrated system has the decisive advantage of establishing the validity of the analytical models used to evaluate separation dynamics. A close match between the development testing and the analytical models will make it possible to qualify the separation dynamics by analysis, avoiding an expensive flight test dedicated to qualifying the separation system. Actually simulating the flight conditions is not practical considering the cost and complexity of such a test set up. A test setup that is capable of simulating any flight condition in one plane could be used to demonstrate the system incrementally. Figure 19 shows a proposed test setup that would be capable of simulating all of the most relevant conditions in the horizontal plane. Figure 19- System demonstration test set up Different asymmetric strut cases could be combined with various simulated thrust conditions. The simulations could be accomplished by placing many support points at the center of gravity of each of the mass simulators. The brake rod would have a ball joint attachment at the center-of-gravity and the brake body would be free to rotate on a pivot arrangement. When the separation joint is activated, the brakes would simulate the effects of the SRB thrust and the relevant component of gravity acting on the vehicle. This set up would simulate the mass and the mass-moment-of-inertia of each of the stages. Thrust vector side loads would be simulated by springs acting between the rod coming from the brake and the end of the first stage. The brakes would also arrest the motion of the two bodies after separation was 343

Document

AMS_2008.pdf

demonstrated. The axial thrust oscillation could be simulated by 2 large asymmetric counter-rotating masses near the center of gravity of the first stage. Demonstrating the ability to prevent re-contact after initial separation is a critical part of any separation qualification program. If the thrust oscillation was to slam the two stages back together after initial separation, impact loads would be transmitted to the sensitive avionics boxes on the aft skirt. Also, the structure of the aft skirt near the contact location could fail locally and unpredictable separation dynamics would be present. MAG Strut Qualification Qualifying the strut separation system will be a rel atively quick, low cost pr ogram if a well-designed development test program is completed before hand. The separation dynamics will be qualified by analysis. The struts could be stru cturally qualified by analysis with the end fittings being considered qualified by test assuming that the qualification strut was pressurize with fluid that would generate sufficient force to subject the fitting to 1.4 times the limit load. Since the strut is designed with a safety factor of 2 for static pressure containments and a safety factor of 2.5 for dynamic pressure containment, the end fittings could be subjected to the limit loads without subjecting the struts to pressures that would yield the structure. The structure of the aft skirt and the interstage could be qualified by analysis. The development test would provide the data to validate the analytical models for both the struts and the structure. If some design changes were made to the flight struts that were not reflected in the development test articles, the qualification testing could be done using the same test set up used for development testing. Conclusion The MAG Struts are the ideal separation system for Ares I. No other separation system has the capability to separate with 356 kN of residual thrust on the first stage. This capability increases the Ares I payload lift capability significantly over a BDM separation system. Secondly, the MAG Strut system is mounted internally minimizing aerodynamic drag. Finally the MAG Strut system pushes the first stage and the second stage apart increasing the momentum transfer between the stages. The struts reduce the potential for ullage collapse in two ways. Separating with 356 kN of residual thrust mitigates the potential for ullage collapse because the liquid hydrogen does not have the have the tendency to climb the walls of the tank as is possible when operating at very low levels of acceleration. The MAG Strut limits the amount of ac celeration the vehicle experience to less than 2.5 g per second decreasing the potential to agitate the liquid hydrogen. The MAG Strut limits the amount of load applied to the aft skirt during assent to 68 kN while they have the capability of stroking with a peak force 187 kN each. The MAG Struts produce superior nozzle clearance under all conditions including one strut out cases. This means that the struts are inherently two-fault tolerant against pressure bleed down. The struts also greatly mitigate the effects of the SRB nozzle pointing accuracy and any other disturbances coming from another source because of the matching of the mass properties of the two separated stages. Although struts have not been used on a manned vehicle, the struts can be brought up in design maturity in time to support later Ares I test la unches assuming that the developme nt test program is conducted concurrently with other Ares I development programs. Doing the development program facilitates the inclusion of the struts at a later date in the Ares program. The authors of this paper would like to acknowledge the contributions of the following people: Acknowledgement 1: Young Kim of Marshall Space Flight Center for fly-out analysis Acknowledgement 2: Mike Hannan of Marshall Space Flight Center for performance analysis Acknowledgement 3: Mike Martin of Marshall Space Flight Center for gas flow and pressure analysis Acknowledgement 4: Mike Perry of Marshall Space Flight Center for computer aided design integration 344

Document

AMS_2008.pdf

Evaluation of Separation Mechanism Design for the Orion/Ares Launch Vehicle Kevin E. Konno*, Daniel A. Catalano* and Thomas M. Krivanek* Abstract As a part of the preliminary design work being performed for the Orion vehicle, the Orion to Spacecraft Adaptor (SA) separation mechanism was analyzed and sized, with findings presented here. Sizing is based on worst case abort condition as a result of an anomaly driving the launch vehicle engine thrust vector control hard-over causing a severe vehicle pitch over. This worst-case scenario occurs just before Upper Stage Main Engine Cut-Off when the vehicle is the lightest and the damping effect due to propellant slosh has been reduced to a minimum. To address this scenario and others, two modeling approaches were invoked. The first approach was a detailed Simulink model to quickly assess the Service Module Engine nozzle to SA clearance for a given separation mechanism. The second approach involved the generation of an Automatic Dynamic Analysi s of Mechanical Systems (ADAMS) model to assess secondary effects due to mass centers of gr avity that were slightly off the vehicle centerline. It also captured any interference between the Solar Arrays and the Spacecraft Adapter. A comparison of modeling results and accuracy are discu ssed. Most notably, incorporating a larger SA flange diameter allowed for a natural separation of the Orion and it s engine nozzle even at relatively large pitch rates minimizing the kickoff force. Advantages and disadvantages of the Simulink model vs. a full geometric ADAMS model are discussed as well. Introduction A component of the Vision for Space Exploration, Orion will be capable of carrying crew and cargo to the ISS, or rendezvous with a lunar landing module and an Earth departure stage in low-Earth orbit to carry crews to the moon and, one day to Mars-bound vehicles assembled in low-Earth orbit. Orion borrows its shape from the capsules of the past, but it takes advantage of 21 st century technology in computers, electronics, life-support, propulsion, and heat protection systems. Orion will be launched into low-Earth orbit by the Ares I Crew Launch Vehicle. To maximize the crew’s safety, Orion and its abort system will be placed at the top of the Ares I rocket. Other means of abort are available after the Launch Abort System is jettisoned at ~75 km (250,000 ft). The Orion vehicle will be able to remain docked to ISS for up to six months and have the ability to stay in lunar orbit untended for the duration of a lunar surface visit that could be up to six months. A separation mechanism design is being developed to assure clearance between Orion (Crew Exploration Vehicle and Service Module) and the Spacecraft Adapter (SA), which stays fixed to the Ares upper stage as the two vehicle elements separate from each other during both normal post-launch staging or in an abort event. Figure 1 depicts the Ares/Orion stack configuration prior to separation. The preliminary design of the sep aration mechanism requires the balan cing of several competing design parameters most notably sufficient kickoff forces to ensure separation, highly reliable components, limited space to house these mechanisms, and a requirement to keep the mechanisms lightweight due to tight mass budgets. The abort case will typically drive the size of the separation mechanism design for a crewed vehicle. Activation of the separation mechanism cannot occur until the thrust levels of the Ares I Upper Stage (US) are significantly reduced. The potential hard-over gimbal abort case can induce a severe pitch over rate (often referred to as “dump rate”) of up to 35 degrees/second on the stack if it were to occur just before Upper Stage Main Engine Cut-Off when the vehicle is the lightest and the damping effect due to propellant slosh has been reduced to a minimum. The transients of the controls and engine thrust tail-off (~ 3 sec total from hard-over to low thrust) are the main reason a large dump rate can be induced. A Residual Engine Thrust also continues to induce a small force at 5 degrees off the vehicle * NASA Glenn Research Center, Cleveland, OH Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Space Flight Center, May 7-9, 2008 345

Document

AMS_2008.pdf

centerline resulting in an applied moment to the Upper Stage after separation. This defines the worstcase environment that the separation mechanisms must overcome. Recontact during either an abort separation, or a nominal separation, can be catastrophic resulting in a Loss of Mission or a Loss of Crew event. A recent example of this type of detrimental recontact was observed during the March 2007 SpaceX Corporation Falcon I launch first stage separation event in which the first stage adapter inner wall contacted the second stage engine nozzle and induced a propellant slosh in the second stage tanks, prematurely shutting down the second stage engine before reaching the proper orbit [1]. The Orion separation system must be adequately sized to reliably separate the crew and vehicle safely for all design cases. Multiple types of mechanisms were evaluated including spring actuators, constant pressure pneumatic actuators, and pyrotechnic-actuated gas thrusters. Crew Exploration Vehicle (CEV) Solar Arrays (2) Spacecraft Adapter (SA) Ares 1 Launch Vehicle J-2X engine & remainder of vehicle shown on leftService Module (SM) Service Module Engine NozzleAres I Upper Stage (CLV) J-2X EngineAux Engines (8 total) Crew Exploration Vehicle (CEV) Solar Arrays (2) Spacecraft Adapter (SA) Ares 1 Launch Vehicle J-2X engine & remainder of vehicle shown on leftService Module (SM) Service Module Engine NozzleAres I Upper Stage (CLV) J-2X EngineAux Engines (8 total) Crew Exploration Vehicle (CEV) Solar Arrays (2) Spacecraft Adapter (SA) Ares 1 Launch Vehicle J-2X engine & remainder of vehicle shown on leftService Module (SM) Service Module Engine NozzleAres I Upper Stage (CLV) J-2X EngineAux Engines (8 total) Figure 1. Ares I Upper Stage/Orion Spacecraft Conf iguration (Lockheed Martin Concept) The Point of Departure separation system is shown in Figure 2. Unlike Apollo’s Service Module, which was bolted to the top of a four piece faring and severed from it via a circumferential linear shape charge, this system incorporates compression kickoff springs and pyrotechnic separation bolts to join the Orion Vehicle to the SA. Separation is triggered by firing the pyrotechnic retention bolts, which allows the compression springs to push the SM away from the Upper Stage. The spring force must be sufficient to accelerate the separated bodies away from each other while maintaining a minimum clearance throughout separation. 346

Document

AMS_2008.pdf

Figure 2. Separation Spring Concept Separation System Hardware The force needed for separation can be generated from various competed technologies, including mechanical springs, pneumatic actuators or gas thrusters. The other components of interest in the staging mechanism are the pyrotechnic fasteners through which the launch loads are transmitted. Mechanical Springs: For this study open coiled, helical, compression springs were the preferred form of kickoff devices if the required energy level was low enough to warrant their use. Spring kickoff devices were incorporated in most of the models because the separation environment did not require large kickoff forces. Obviously, mechanical springs are used in countless terrestrial applications as well as space. They are highly reliable and when designed correctly can handle millions of cycles. Several papers detail the use of compression springs in spacecraft staging mechanisms [2], [3], [4], [5], [6], [7], [8]. For spacecraft mechanisms mechani cal kickoff springs must be designed with the resistive force (F r) and the force required for acceleration (F a) of the bodies in mind. Where redundant springs are used instead of a backup mechanism, they should be designed to provide adequate force for a one-spring-out case [5]. Additionally, the spring system should have a 100% positive Margin of Safety on drive force over resistive force, as measured at acceptance or qualification testing. It is prudent to carry additional margin prior to testing. Also, spring systems are required, when practical, to have a dynamic force margin of safety over the required force F a of 25%, as tested. Additionally, when sizing mechanical springs, spring material stress relaxation and resid ual stresses must be factored in, for which some test data exists [9]. These effects decrease the driving force a spring is capable of after prolonged storage, and can vary by greater than an order of magnitude depending on the material. 302 SS, a common aerospace spring material, can go through 3-5% stress/preload relaxation in 1000 hours of storage time. The dynamic modeling of springs in this system always considered six compression springs located equidistant around the circumference of the SA interface flange. If incorporated in the final flight design, equivalent redundant pairs of springs will be used to improve reliability. It is important when designing mechanical redundancy to do so wisely as it has been shown that some redundancy can actually decrease overall system reliability, even in spring actuator designs [10], [11]. 347

Document

AMS_2008.pdf

Pneumatic Actuators: Pneumatic actuators have extensive spaceflight experience, most notably on the Delta launch vehicle stage separation system. Pneumatic actuators possess larger specific force capability (N/kg) than mechanical springs, giving 4 to 5 times the kickoff force of springs of the same mass. Higher part count and pressuri zed components leads to pot entially lower reliability than the simpler mechanical springs, making them less attractive for a crewed mission. Gas thrusters: While compressed gas thrusters have aerospace flight heritage in solid rocket booster separation (Figure 3) they have no known experience as a spacecraft or payload separation device to this author’s knowledge. Their benefit is in producing a large specific force giving very high drive capability, even greater than 10 times that of mechanical springs for the same mass. Where high kickoff forces are not required, their greater complexity and potentially lower reliability may make them less attractive. Gas thrusters are currently being traded against pneumatic actuators for the Ares 1 launch vehicle staging mechanism as well. Figure 3. Pyrotechnic Gas Thruster/Actuator (left), Pneumatic Actuator System (right) (Used with permission of Scot, Inc.) Modeling Approach To address the mechanism design sizing, two modeling approaches were invoked. Each method allowed for easy evaluations as vehicle configuration changes occurred. The first approach was a simplified Simulink model to quickly assess the critical clearance between the Orion Engine nozzle and the SA. The second approach involved the generation of an ADAMS 3D geometric model to assess secondary effects due to offset mass centers of gravity, off-diagonal moment of inertia terms and other out of plane effects. It also captured any interference due to potential contact between other parts of the Orion Vehicle and the SA. Simulink Model Approach The Simulink model approach [12] for the abort simulation was based on the translation and rotational equations of motion, which are integrated through the time step function to determine the relative positions of the Ares Upper Stage and Orion. Figu re 4 depicts the full Simulink model. The separation force is applied as either a constant pressure (as from a gas thruster) or a variable force (mechanical spring) over the length of the actuation. Assumptions of planar motion for the location of element centers of gravity and constant component masses for the duration of the separation event are incorporated. Capability has been added to the model to include the residual J-2X engine thrust acting on the Ares I US after separation and the contribution of Reaction Control System thrust to the separation acceleration. 348

Document

AMS_2008.pdf

1 w , deg/s pedestal threshold sw412:34 run time, sec .5 radius..5radiusp1mc/mo1/12 in to ft -K-gcenergy*x/t .5 a1 time OrionSp Wrksp211 W pedestal height, inIn1 In2 In3 In4 In5 In6 In7Out1 Out2 Out3 Out4 Out5 Out6 Vertical VelocityIn1 In2 In3 In4 In5 In6 In7Out1 Out2 Out3 Velocity Calculation 0 Thrus t, RCS, lbf 1 TVC angle, degIn1 In2 In3 In4 In5Out1 Out2 Out3 Subsystem1In1 In2 In3 In4 In5 In6 In7 In8Out1 Out2 Subsystem1 Spring limit, in Spr Sep Clr 111Residual Engine Thrust, lbfRel ative Clearance, inR2D Rads to Deg In1 In2 In3 In4In5Out1 Out2RET System PertubationIn1 In2 In3Out1RCS Effects In1 In2 In3 In4 In5 In6 In7 In8 In9 In10Out1 Out2 Out3 Out4 Output Condition 111Orion length, in Cg to OMS engine bell0 Orion Rotation deg/s - 11 Orion OMS Diameter, in1e8Orion Moment of Inertia about "z", lbm in211111 Orion Mass, lb 1111 Orion Cg, in from CLV engine baseIn1 In2 In3 In4 In5 In6Out1 Out2 Moment Subsystem100 K s pring, lbf/in1 In1 In2 In3 In4 In5 In6 In7 In8 In9 In10 In11 In12Out1 Out2 Out3 Out4 Out5 Geometry Separation Velocity SubsystemDivD2R Degr to Rads CLV w/ped length total, in CLV to Orion Cg inCLV system inertia 111 CLV length, in00 CLV Rotation deg/s - 111CLV Pedestal ID, in1e9 CLV Moment of Inertia about "z", lbm in211111CLV Mass, l b CLV Length total, in1111 CLV Cg, in from engine base In1 In2 In3 In4 In5 In6 In7 In8 In9 In10Out1 Out2 Out3Angular VelocityIn1 In2 In3 In4 In5 In6 In7Out1 Out2 Out3 Out4Actuator Force Subsystem u2 u21time Figure 4. Simulink Separation Model The basis of the Simulink analysis utilizes the conservation of momentum and kinetic energy equations shown below: /g11/g12 o2oo o c2cc c stack r mI rmI I /g90/g184/g185/g183/g168/g169/g167/g14/g14/g90/g184/g185/g183/g168/g169/g167/g14/g32/g90 and /g184/g184 /g185/g183 /g168/g168 /g169/g167/g90/g14/g14/g184/g184 /g185/g183 /g168/g168 /g169/g167/g90/g14/g32/g184/g184 /g185/g183 /g168/g168 /g169/g167/g90 2 I v m 2I v m 2 I2oo2oo2cc2cc2stack Where the variables are defined as: Istack = Stack (Orion + Ares 1 US) moment of inertia, kg-m2 Ic = Ares 1 US moment of inertia, kg-m2 Io = Orion moment of inertia, kg-m2 /g90 = body stack rate of rotation, deg/s /g90c= Ares 1 US rate of rotation, deg/s /g90O= Orion rate of rotation, deg/s mc= Ares 1 US mass, kg 349

Document

AMS_2008.pdf

mO= Orion mass, kg rc = Ares 1 US c g moment arm to system cg, m r0 = Orion c g moment arm to system cg, m vc = Ares 1 US relative velocity to system, m/s vO = Orion relative velocity to system, m/s The results of solving these equations are that t he rotation rate of the separated components is maintained at the same rate as the sta ck rotation prior to separation. The addition of a residual engine thrust post separation does induce an additional mom ent onto the Upper Stage and results in an angular acceleration, reducing the clearan ce during separation. ADAMS Approach The SA/Upper Stage and Orion vehicles were also modeled using the ADAMS dynamic software code [13]. This is a motion simulation code that allows the user to create a mechanism model and then solves the simultaneous equations for kinematic, static, quasi-static, and dynamic simulations. Figure 5 depicts the ADAMS separation model during a simulation as Orion clears the SA. For the purposes of this study the Upper Stage and Orio n were modeled as rigid bodies, each with six degrees of freedom. The SA was modeled as rigidly linked to the Up per Stage since it never separates fr om it. However the geometry of the SA, particularly the top flange was important for this analysis since the Orion engine nozzle needs to be extracted from this cavity and translate beyond the top flange of the SA without impact. The Crew Module and SM were modeled as a single rigid body (i.e. Orion) since, again, they never separate in this analysis. The modeling of the separation systems for the Orion to Ares 1 US included the single axis springs or actuators located around the SA top flange, between the SM and SA. These compression springs produce a translational motio n when released. Once their free len gth is achieved they no longer impart any force onto the vehicle. The Ares 1 US’s J-2X engine thrust is modeled at the bottom of the Upper Stage. Auxiliary engine thrust is also accounted for on several design studies and these are also modeled as point forces located at the current auxiliary engine locations near the separation plane. Vehicle “dump” or pitch rate is applied as an initial velocity condition to the Ares 1 US/Orion at the combined vehicle stack center of gravity (CG). Joints were added to the model as follows: a fixed joint was created between the Upper Stage and the CEV to allow them to pitch together at the start of the simulation (time = 0 sec.) and release when the separation event began arbitrarily at time = 1.0 sec.; a hinge joint was created at each solar array anchor point under the avionics ring to allow for flaring of the arrays in order to investigate different launch configurations to optimize array clearance from the outer fairings at launch as well as avoid impact with the SA upon separation. Figure 5. ADAMS Separation Model 350

Document

AMS_2008.pdf

Results The clearance requirement for this system was to provide a 5.1-cm (2-in) minimum clearance between the Orion vehicle (engine nozzle, solar arrays, and other protrusions) and the SA. Several analyses were completed to size the actuator forces, determine practical separation times and optimize the vehicle geometry (engine size, solar array placement, SA diameter, etc.). Figure 6 plots the time that the minimum clearance is reached versus the vehicle dump rate for a gas actuator system and for the case of no actuator forces, using the Simulink model. From this plot the two systems are seen to coincide at the higher body rates where the actuation force needed for the 5.1-cm (2-in) clearance is diminishing as the system approaches the no-force required condition. 0123456 0 1 02 03 04 05 06 Vehicle Dump Rate, deg/sTime to Minimum Clearance, sec 00.051m Clear Constant Force No actuator force Figure 6. Clearance Time as a Function of Vehicle Body Rate The initial actuation force required to meet the clearance requirement at the baseline 5 deg/s dump rate is shown in Figure 7. The curves in the figure are the Orion radial clearance (red), the Orion axial separation distance (green), and the actuator force line (blue) which shows when the actuator force is terminated (0.33 sec) and is reflected in the slope of the Orion separation distance curve which tends to be more linear after this force is removed. The jog in the clearance curve occurs when the Orion vehicle clears the SA at approximately 4.8 seconds, which is when the change in clearance becomes a positively sloped line as the vehicles move further apart from each other with no chance of contact. 351

Document

AMS_2008.pdf

00.20.40.60.811.21.41.61.82 00 . 511 . 522 . 533 . 544 . 55 time, secDistance/Clearance, mActuator Force ClearanceAxial Separation Figure 7. Time Required for Separation at 5 deg/s, (Simulink Model) Figure 8 is a plot of the Simulink model separation force required to provide clearance at an initial dump rate of 5 deg/s for both types of actuators, springs and gas (or pneumatic) thrusters. In the spring type actuator the force is a function of axial displacement where the initial force is very high and then decreases along a power curve as a function of time. In the gas thruster system, which can be modeled as constant pressure through its full stroke, the force is maintained throughout the action time at a constant level. Both systems were sized to separate the vehicle with exactly 5.1-cm (2-in) clearance maintained at the engine nozzle. Duration of the const ant pressure actuator is determined by the 10.2-cm (4-in) actuator stroke. The calculated time necessary to provide this clearance is shown to be independent of the actuation method, since the vehicles are rotating and translating at a rate as a function of dump rate and the residual engine thrust on the Upper Stage. The 5.1-cm (2-in) minimum clearance point in space or “gate” is reached at the same point in time, which varies from 24.0 deg (4.8 s) to 23.4 deg (0.39 s) of vehicle rotation for the 5 deg/s to 60 deg/s body rate respectively. There is a very slight difference in the final velocity induced by the actuators since to make the clearance gate time, the spring actuator provides a higher initial acceleration and then coasts at the resulting velocity of 0.477 m/s (1.56 ft/s) with an acceleration rate of 0.144 g’s, while the constant pressure actuator provides an acceleration rate of approximately 0.118 g’s over a longer time period resulting in a higher 0.485 m/s (1.59 ft/s) final velocity imparted on the Orion system. Therefore, less energy is required for the spring actuator compared to the gas thruster mechanism. 352

Document

AMS_2008.pdf

05001000150020002500 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5time, secForce, kgSpring Force Gas Thruster1095 J (808 ft-lbf) 1144 J (844 ft-lbf)Total impulse for Spring is 1095 J vs Gas thruster @ 1144 J Average accel for Spring is 0.026 g's greater Actuator induced velocity for Spring is 0.477m/s vs 0.485 m/s Both achieve clearance at 4.83 sec Figure 8. Separation Force Input for Spring vs. Gas Thruster For the ADAMS model several cases and design studies have been run to determine the optimal separation system. The two critical considerations in these dynamics analyses are to ensure that the SM engine bell can get extracted from within the SA without bumping (avoiding a Falcon 1 type of hazard), and that the solar arrays, which are mounted down the sides can clear the SA without interfering from the outside. Using the given mass properties the resulting spring stiffness case results are shown in Table 1 as well as Figure 9. All cases are assuming there is no separation assistance from the Orion SM main engine. Table 1. Summary of Parameters Analyzed case #dump rate (deg/s)J-2X Residual ThrustCEV RCS Thrust kg (lbf)Spacecraft Adapter flange ID m (in)Spring stiffness kg/m (lb/in)Spring stroke length m (in)min clearance, ADAMS model m (in)min clearance, Simulink model m (in)model delta m (in) 1 35 yes 0 3.4 (135) 0 0.102 (4) 0.14 (5.6) 0.11 (4.5) 0.03 (1.1) 2 0 yes 0 3.4 (135) 10,724 (600) 0.102 (4) 0.80 (31.5) 0.72 (28.3) 0.08 (3.3) 3 10 yes 0 3.4 (135) 178 (10) 0.102 (4) 0.07 (2.6) 0.08 (3.0) -0.01 (-0.4) 4 10 no 0 3.4 (135) 178 (10) 0.102 (4) 0.15 (6.0) 0.13 (5.2) 0.02 (0.8) 5 20 no 0 3.4 (135) 178 (10) 0.102 (4) 0.15 (5.9) 0.13 (5.1) 0.02 (0.8) 6 20 yes 0 3.4 (135) 178 (10) 0.102 (4) 0.13 (5.2) 0.10 (4.1) 0.03 (1.1) 7 5 yes 0 3.15 (124) 4,147 (232) 0.102 (4) -0.08 (-3.3) -0.05 (-1.9) -0.04 (-1.4) 8 35 no 0 3.15 (124) 0 0.102 (4) 0.01 (0.2) 0.01 (0.2) 0 9 5 yes 366 (808) 3.4 (135) 0 0.102 (4) 0.18 (7.2) 0.11 (4.5) 0.07 (2.7) 10 5 yes 0 3.4 (135) 4,147 (232) 0.102 (4) 0.05 (2.0) 0.08 (3.2) -0.03 (-1.2) 353

Document

AMS_2008.pdf

00.050.10.150.20.250.30.350.40.45 2500 4500 6500 8500 10500 12500 spring stiffness, kg/mclearance, m3.08m (121.6") flange dia.- array 3.82m (150.4") flange dia.- nozzle 3.08m (121.6") flange dia.- nozzle3.82m (150.4") flange dia.- array3.34m (131.2") flange dia.- array 3.34m (131.2") flange dia.- nozzle3.58m (140.8") flange dia.- nozzle 3.58m (140.8") flange dia.- arrayallowable Figure 9. Engine bell and 6-m Array clearance for 5o/sec dump rate Figure 9 depicts the results of several ADAMS cases using different SA flange diameters to assess clearing the solar arrays on the outside versus clearing the engine nozzle on the inside. Previous Orion designs incorporated a longer but narrower vehicle with a likewise narrower SA. The redesign of the vehicle allowed for a wider SA flange. As can be seen in the plot, the arrays have adequate clearance (>5.1 cm) for any SA inner diameter of 3.58 m (140.8 in) or less, while the engine nozzle will have adequate clearance for a SA diameter of 3.34 m (131.2 in) or greater. Thus an inside diameter of 3.34 – 3.58 m (131.2 – 140.8 in) satisfies both. In these cases it is assumed that the J-2X engine residual thrust is active and the flange width is 0.343 m (13.5 in) radially. A nominal separation system would include springs located at SA nodes as shown in Figure 2 outboard of the separation pyrotechnic device with a 10.2-cm (4 in) stroke and 3,842-kg/m (215-lb/in) stiffness. Six standard 1.27-cm (1/2 in) separation bolts located directly inboard of the push o ff springs at each node will transmit launch loads through the structure. Figure 10 is a plot of the separation clearance as a function of time for the different models used for the baselined configuration at a body rate of 5 deg/s. The ADAMS model includes the Solar Arrays for additional clearance studies while the Simulink model only considers the clearance for the Engine Nozzle to the SA, which becomes the limiting parameter for both models after approximately 2.4 seconds. 354

Document

AMS_2008.pdf

00.20.40.60.81 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 time, secSeparation Clearance, mADAMS Global - Nozzle Clr ADAMS Global - Solar Array Clr Simulink Global - Nozzle Clr Figure 10. Separation Clearance at 5 deg/s Body Rate Figure 11 depicts the clearance achieved as a function of actuator force for the Simulink and ADAMS models at a 5 deg/s dump rate. The interference at the low actuator force is due to the influence of the applied Residual Engine Thrust moment and to the longer separation time required for the low dump rates (<10 deg/s). This slower separation time allows the induced moment on the Upper Stage to rotate the SA reducing the clearance below the required limit. From the plot it is evident that an actuator force of greater than 3500 kg/m (196 lb f/in) is required to assure the clearance is achieved and that use of a grossly oversized actuator has diminishing returns since the actual clearance is not a linear function of spring stiffness. 00.10.20.30.40.5 3500 5000 6500 8000 9500 11000 12500 14000 actuator spring stiffness, kg/mclearance, mAdams Simulink 0.051m Clearance Re quired Figure 11. Comparison of Clearance vs. Actuator Force 355

Document

AMS_2008.pdf

Natural Separation It has been learned th rough this analysis that for two bodies rigidly fixed together and undergoing rigid body rotational and translational motion which then separate, if body 1 has a protruding feature (like an engine bell) tucked inside a recessed area of body 2 (such as the SA cavity) that, due to the centrifugal forces naturally propelling them apa rt, there exists a relationship between the diameter and length of the protruding feature and the mating clearing radius Y B1 of the recess whereby for a recess radius greater than Y B1, the vehicles will separate without collision at any dump rate with no additional kickoff force required nominally. Figure 12 depicts the geometry definitions used. Body 2 w/recessBody 1 w/protrusion Figure 12. Ares I US/Orion Vehicle Geometry and Definition RB*COS(/g84Bt) = X A1+XAtrans+XArot ; where X A1is the initial axial distance from Body 2 CG to Protrusion point A, X Atransand X Arotdepict the axial & rotation motion components of point A, which represents the outermost point of the protrusion. RB = R A *(SIN(/g84At)/SIN(/g84Bt)) SIN(/g84A1) = YA 1/RA RB = SQRT[(X B1)^2 + (Y B1)^2] XCG1A+XA1= CG off; CG offis the distance between Body 1 and 2 CG’s prior to separation . XAtrans = /g90*Tcoll* CG off ;Tcoll is the time needed for pt. A to separate and pass thru pt. B at X. XArot = R a *[COS(/g84A1) - COS(/g84At)] XBt = R B *COS(/g84Bt) XBt+XAtA = CG off +X Atrans /g84Bt = /g84B1-/g90*Tcoll /g84At = /g84A1+/g90*Tcoll Thus, the minimal recess radius for natural separation, Y B1, can be solved for easily, if X A1, YA1 (protrusion radius), and lengths X B1 (recess radius) and CG off(distance between CG’s) are known. This analysis 356

Document

AMS_2008.pdf

assumes the protrusion and recess are modeled as straight cylinder sections. This finding gives a designer a useful preliminary size for the vehicle recess diameter (such as in a spacecraft adaptor cone flange) or protruding diameter (such as an engine bell), and is independent of the vehicle’s dump rate at separation. While this does not account for secondary effe cts like residual engine th rusts and separation event side loads, tank slosh or friction which can either help or hurt this clearance, these effects are typically secondary to the overall conic area that the bodies follow dynamically upon separation. For our case, assuming no external forces, the separation event has been determined to occur naturally for the approximately 28.5 degrees of rotation needed and will provide a minimum clearance of 0.173 m (6.8 in) for all significant body rates. This is primarily due to the location of the Vehicle Stack system Cg, which is located very close to the separation plane and is therefore very sensitive to any changes in that location. Taking this concept further, separation cases were run (see Figure 13) in which the same conditions were applied to a vehicle with an adequately large recess diameter (SA flange diameter) and to an undersized flange diameter (124 in). As can be seen, cases were run with the US engine on or off for comparison. As the lowest curve shows for a smaller SA flange of 3.15 m (124 in), spring force is dependent on dump rate as the higher dump rates require much larger spring stiffness to clear upon separation, while for a SA flange ID of 3.4 m (top curve) no springs are required at any dump rate even with the US engine on (2nd curve). The vehicle separates naturally without help of any kickoff device. -0.1-0.0500.050.10.150.2 10 15 20 25 30 35 vehicle dump rate, deg/sclearance, m Contact Positive Clearacne3.4m (135") flange dia., No actuation or J-2X .051 m Re quired Clearance 4,893 kg/m 8,018 kg/m 6,250 kg/m 10,358 kg/m 3.4m (135") flange dia. , No actuation w/J-2X 3.15m (124") flange dia., No actuation w/J-2Xadded spring stiffness to meet req'd clearance Figure 13. Engine Nozzle Clearance with No Separation Forces (no springs or actuators) Lessons Learned One of the lessons learned was that intelligent preliminary sizing of spacecraft geometry can greatly improve reliability and save on vehicle weight. Also for two connected bodies rotating at a fixed dump rate, that same angular rate will be maintained by each body after separation, as angular momentum is conserved. Another important lesson learned was that separation mechanism component mass can be 357

Document

AMS_2008.pdf

minimized and reliability maximized if the geometry is dim ensioned to allow for “natural separation” concepts. However, the need for a controlled separation event necessitates the use of applied force actuators to overcome any potential external forces. The la st lesson learned was that there exists a very steep curve between separation clearance and the vehicle main parameters of mass, inertia and geometry with that sensitivity often resulting in inadequate clearance dynamics. Summary The simulations conducted indicate that a low fidelity, 2-D equations of motion model can be useful in separation mechanism design. It provides insight into separation events and the many parameters and their relative sensitivities. A more detailed 3-D geometric dynamics model is also required to clearly define the actuator requirements while accounting for all factors in three dimensions and can also identify interferences due to other hardware on the vehicle. The overall design conclusions drawn are that a simple, dependable spring system can be used for the Orion crewed vehicle separation system. Minimizing the actuator force is preferred in terms of mass, reliability, and cost. However, ensuring the separation system controls the event and all potential external forces is still paramount. This is especially true in an abort scenario. Additional effort needs to be invested to assure second order effects due to propellant slosh or thruster imbalance does not violate the design criteria used in the analysis. Acknowledgments The authors would like to acknowledge the contributions, advice, and suggestions of Keith Schlagel and Lance Lininger of Lockheed Martin Corporation who aided in the development and compilation of this work. References 1.http://www.spacex.com/media.php?page=57 2. Onoda, J. “The Development of Staging Mechanisms for the Japanese Launcher Mu-3SII,” 19th Aerospace Mechanisms Symposium, NASA Ames Research Center, August, 1985. 3. Harrington, T.G. “Compression Spring Separation Mechanisms,” First Aerospace Mechanisms Symposium, University of Santa Clara, Santa Clara, California, May 19-20, 1966. 4. Abdul Majeed, M. K., Matarajan, K., Krishnankutty, V. K. “Separation and Staging Mechanisms for the Indian SLV-3 Launch Vehicle,” 18th Aerospace Mechanisms Symposium, NASA Goddard Space Flight Center, May 1984. 5. AIAA-S-114-2005, Movinq Mechanical Assemblies Standard for Space and Launch Vehicles , American Institute of Aeronautics and Astronautics standard, July 2005. 6. Conley, Peter L. Space Vehicle Mechanisms, New York, John Wiley & Sons, Inc. 1998. 7. Brennan, Paul C., NASA Space Mechanisms Handbook, July 1999. 8. Purdy, W., Hurley, H., “The Clementine Mechanisms,” 29th Aerospace Mechanisms Symposium, NASA Johnson Space Center, May 1995. 9. Hanna, W. D., Chang, R. S., Sheckel, G. L., “Stress Relaxation of Spring Materials,” Fortieth Anniversary: Pioneering the Future, May 1998. 10. Chew, M. “On the Danger of Redundancies in Some Aerospace Mechanisms,” 22nd Aerospace Mechanisms Symposium, NASA Langley Research Center, May 1988. 11. Holmanns, W., Gibbons, D., “Misconceptions in Mechanical Reliability,” 34th Aerospace Mechanisms Symposium, NASA Goddard Spaceflight Center, May 2000. 12. Simulink program Ver.7.0.1.29704, The Math Works Corporation, September 2004. 13. ADAMS dynamic software code Ver. 2005 r2.0, MSC Software Corporation, August 2005. 358

Document

AMS_2008.pdf

Focus Mechanism for Kepler Mission Kraig Koski* Abstract The Focus Mechanism built for the primary mirror on the Kepler mission provides a method for adjustment of the mirror position for the duration of the mission. The Focus Mechanism also provides structural support for the 87 kg primary mirror. The Kepler mission requirements provided some interesting and difficult design tasks for the Focus Mechanism. This paper will describe the development, design, function and testing of the Focus Mechanism. Introduction The goal of the Kepler mission is to survey our region of the Milky Way Galaxy to detect and characterize hundreds of earth-size and smaller planets near the habitable zone. The habitable zone encompasses the distances from a star where liquid water can exist on a planet’s surface. The transit method will be used for detecting extrasolar planets. A transit is when a planet crosses in front of its star as viewed by an observer, resulting in a small change in the star’s brightne ss for a repeatable amo unt of time. Once detected, the planet’s orbital size and mass can be calculated using Kepler’s Third Law of planetary motion (T 2 = R3). The size of the planet is found from the depth of the transit (how much the brightness of the star drops) along with the size of the planet’s star. From this information, the planet’s characteristic temperature can be calculated. Figure 1. Kepler Flight Segment [1]. * Ball Aerospace & Techno logies Corp., Boulder, CO Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Space Flight Center, May 7-9, 2008 359

Document

AMS_2008.pdf

Kepler Mission Design For a planet to create a transit visible from our solar system, the orbit must be li ned up edgewise to us. The probability for a planet in an Earth-like orbit around a solar-like star to be properly aligned is 0.5%. For this reason, one must look at thousands of stars to determine if Earth-like planets are common or rare. The time between transits for planets in the habi table zone is around 1 year and to reliably detect a sequence, four transits are required. The Kepler instrument, called a photometer, has a large field of view (12° diameter) in order to observe more than 100,000 stars in the Cygnus Region of the Milky Way continuously for the entire 3.5 year mission as shown in Figure 2. Kepler is scheduled to be launched in February, 2009 on a Delta-II rocket into an Earth-trailing heliocentric orbit with a period of 372.5 days which provides the optimum Sun-Earth -Moon avoidance criteria [1]. Figure 2. Kepler field of view in Milky Way Galaxy [2]. Kepler Flight Segment The Kepler flight segment, which was designed and fabricated at Ball Aerospace & Technologies Corp., consists of the Photometer mounted onto a Spacecra ft as shown in Figure 1. The Spacecraft provides power, pointing and telemetry for the Photometer. Pointing at a single group of stars for the entire mission greatly increases the photometric stability and simplifies the Spacecraft design. The Photometer, shown in Figure 3, is a specially designed Schmidt telescope with a 0.95-meter diameter aperture and an array of 42 CCD detectors. Each 50x25 mm CCD has 2200x1024 pixels, which are read every three seconds to prevent saturation. The CCD’s are not used to take pictures and the images are intentionally defocused to 10 arc seconds to improve the photometric precision. The instrument has a spectral bandpass from 400 nm to 850 nm. Data is stored on the spacecraft and transmitted to the grou nd once per week [1]. 360

Document

AMS_2008.pdf

Figure 3. Kepler Photometer, Section View. Primary Mirror Three Focus Mechanism units are used to support and focus the primary mirror as shown in Figures 4 & 5. Each Focus Mechanism attaches to the primary mirror with two flexured struts resulting in a hexapod configuration. The low CTE FRIT bonded primary mirror is made from ULE Titanium Silicate and is 1.45 meters in diameter with a mass of 87 kg. The three Focus Mechanisms are situated at 120º to each other and mounted to invar inserts embedded in a composite bulkhead. Figure 4. Solid Model of Kepler Primary Mirror/Focus Mechanisms/Aft Bulkhead Assembly. 361

Document

AMS_2008.pdf

Figure 5. Kepler Primary Mirror Mounted onto the Focus Mechanisms. Focus Mechanism Driving Requirements Table 1 summarizes the requirements for the Focus Mechanism and shows the performance based on flight unit testing. The focus mech anism has the ability to tip and tilt the primary mirror, but this is not a mission requirement. Only axial focus displacement of the PMA is required. Table 1. Focus Mechanism Key Driving Requirements. Requirement Value Performance Axial position knowledge ±1.4 μm 0.5 μm Operating Temperature Range -55C to +35C Meets Survival Temperature Range -65C to +45C Meets Focus mech Motor Case Operational Temperature Range -55C to +60C Meets Focus Mech Motor Case Survival Temperature Range -65C to +70C Meets Range of Travel ±762 μm ±900 μm Smallest increment of Travel /g1481.5 μm 0.4 μm Unidirectional Repeatability /g1481.25 μm Meets Focus Mechanisms shall not require power during launch n/a Meets 4 Year Lifetime equates to mechanism cycles 112 cycles Meets Mass (1 Focus Mechanism) /g1484.9 Kg 4.13 Kg 1st Mode Frequency, Axial >50 Hz 61.2 Hz 1st Mode Frequency, Lateral >50 Hz 70.0 Hz Maximum Launch Acceleration 41 g Meets Focus Mechanism Design Overview A unique design was incorporated in order to meet the difficult resolution requirement of less than 1.5 microns over a range of at least 1500 microns. This design includes a variety of the classical moving mechanical components that were integrated into an elegant, robust and efficient system. Some of the key components of the Focus Mechanism include a stepper motor, gears, ball screw, bearings, lever arm and several flexures. Figures 6 and 7 show the front and aft views of one of the three flight Focus 362

Document

AMS_2008.pdf

Mechanisms. Redundant LVDT (Linear Variable Differential Transformer) sensors are mounted on both sides of the Primary Mirror Interface shelf and used for position sensing. The Stepper Motor, which drives the entire system, also has primary and secondary windings for redundancy. Figure 6. Kepler Focus Mechanism, Front View. Figure 7. Kepler Focus Mechanism, Aft View. Function The Focus Mechanism is driven by a 30° stepper motor with an integral 100:1 gear head. The output shaft of the gearmotor attaches to a 2:1 pinion-spur gear reduction. The spur gear is attached to the shaft of a 2-mm pitch ball screw which is supported by duplex bearings at the base and a radial bearing on top. A titanium ‘wishbone’ bracket with flexures attaches the ball nut to the end of a titanium lever arm. As the ball nut moves up or down, the lever arm pivots on an integral pivot flexure while a 2 nd drive flexure, also integral to the lever arm, moves the primary mirror interface shelf up or down. A parallel double bladed flexure provides nearly perfect up and down movement of the interface shelf. The 8.75:1 lever arm provides mechanical advantage by reducing load requirements and also allows finer displacements. Redundant LVDT sensors, mounted on both sides of the primary mirror interface, measure 363

Document

AMS_2008.pdf

displacements. Upper and lower hard stops are attached to the ball screw to prevent damage to the mechanism. Figure 8 shows a section view of the mechanism solid model. Figure 8. Section View taken from Focus Mechanism solid model. Focus Mechanism Design Background The initial conceptual development of the Focus Mechanism began at Ball Aerospace in September, 2002 by Robert Warden. The design con cept (Figure 9) was based on the components of the Spitzer Space Telescope (formerly SIRTF) Secondary Mirror Cryogenic Focus Mechanism. Figure 9. Initial Concept for Focus Mechanism in 2002. Numerous developments and improvements were made to the mechanism from late 2003 to 2006. Some of these developments include: 1. The parallel double bladed flexure was made more compliant to decrease the torque requirements and stresses on the drive and pivot flexures. 2. The main flexure interface ‘shelf’ was widened to accommodate the interface to the primary mirror strut brackets. 3. The geometry of the lever arm was optimized to minimize bending and maximize mechanical advantage. 4. The geometry of the lever arm pivot and drive flexures was optimized to minimize stresses. 5. The ball nut flexure ‘wishbone’ design was optimized to eliminate ‘wind-up’ and minimize stresses. 6. Redundant LVDT sensors were ad ded for 16 bit position sensing. 364

Document

AMS_2008.pdf

Design Discussion Motor The Focus Mechanism is driven by a 2-phase driven, bipolar, permanent magnet, redundantly wound, 30° stepper motor. The motor possesses a detent torque which prevents backdriving of the mechanism during launch. Integral to the motor is a 100:1 gear head which enhances mechanical advantage and helps meet the strict Kepler resolution requirements. The detent torque combined with the large gear ratio eliminates the need for a launch lock in the mechanism. The Focus Mechanism is open-loop controlled by commanding the motor to move a specified number of steps. During actuation a ‘commanded step’ from the control electronics moves the motor rotor four 30° steps. This equates to 0.38 μm of axial movement at the primary mirror. The ‘commanded steps’ are used to start and stop the motor rotor in the same phase which increases step count accuracy. The resulting position of the primary mirror is verified with the LVDT sensors. Transfer Gears Both the 40-tooth pinion gear and 80-tooth spur gear are made from custom 455 stainless steel heat treated to condition H1050 to give adequate strength margins. Other than the special material callout, the gears are standard off-the-shelf parts. There is a small amount of backlash in the system due to the transfer gears and 100:1 motor gear head which results is a maximum vertical motion of 0.43 μm at the primary mirror interface shelf. This is well below the 1.25 μm repeatability requirement. Ball Screw The ball screw has an 8-mm diameter and 2-mm pitch and is preloaded with oversized balls to eliminate backlash. The balls, shaft, nut and deflectors are fabricated from 440C passivated stainless steel. The ball screw was sized to result in positive stress margins. Hard Stops The upper and lower hard stops are non-jamming and use a radial face-to-face design which minimizes stress on the ball screw. Figure 10 shows a close up view of the mechanism model with the lower hard stop engaged. The hard stops were fabricated from Titanium and designed to insure that the maximum torque delivered by the gear motor did not over stress them. The hard stops were positioned onto the ball screw during the build up of the focus mechanism so that they stop movement at approximately ±890 μm, well beyond the required ±762 μm range requirement. After the hard stops were correctly positioned, the ball screw was removed from the assem bly and the hard stops were match drilled to it. The threads of the ball screw were completely covered to prevent contamination during this operation. The ball screw and hard stops were then reassembled into the mechanism after the match drilling operation. Wishbone Bracket The wishbone bracket attaches the ball nut to the lever arm. It contains two flexures that extend down on both sides of the ball nut. The wishbone flexures are designed such that they are compliant enough to bend without binding when the Ball Screw rotates and t hey translate rotary motion to linear motion. At the same time, the wishbone flexures are stiff enough to prevent ‘wind- up’ when the ball screw rotates. If the wishbone flexures are too complia nt, the bracket will rotate along with the ball screw and then at an unexpected time it will ‘unwind’ causing unwanted movement and unfavorably affecting the step size and repeatability of the system. Figure 11 shows the titanium wishbone bracket and the tension joint attachment to the lever arm. 365

Document

AMS_2008.pdf

Figure 10. View of lower hard stop engaged. Figure 11. Wishbone Flexures/Bracket. 366

Document

AMS_2008.pdf

Bearings The bearings used to support the ball screw are a duplex pair at the base and a radial ball bearing on top. The duplex pair uses face-to-face mounting to allow for slight misalignment and also to eliminate axial and radial play in the system. The preload in the duplex pair is 13.3-26.7 N (3-6 lb). As shown in Figure 12, the ball screw shaft is supported and rotates with the inner races of the duplex pair. The sleeve and retainer disk also rotate with the ball screw. The ball screw shaft is allowed to float axially on the upper radial bearing which prevents possible binding due to CTE mismatch. There is a 0.2-mm radial gap in the mounting design for both bearings to provide a labyrinth seal which keeps dust out and prevents the lube from migrating. Figure 12. Section view of Bearings/Ball Screw Lever Arm The lever arm provides a mechanical advantage of 8.75:1 which helps achieve the required resolution and torque margin in the system. The shape of the lever arm resulted in analysis to optimize the design for minimum bending at the extents of travel. The bending forces on the lever arm increase as the mechanism approaches the hard stop positions due to bending of the flexures. This results in a system that is not perfectly linear. An identical amount of motor steps will result in approximately 5% less movement of the primary mirror at the extents of travel versus at the null position. This non-linearity is mitigated by the LVDT measurements. The lever arm was fabricated from titanium, like most of the 367

Document

AMS_2008.pdf

components in the focus mechanism, to minimize CTE mismatch and for optimum mechanical properties for the pivot and drive flexures which are integral to the lever arm. Figure 13. Lever Arm Flexures There are 4 flexures in the focus mechanism design. The pivot flexure, drive flexure, parallel double bladed flexure and wishbone flexure. All four ar e fabricated from Titanium 6AL-4V and have been analyzed and designed to have positive stress margins for launch loads and actuation cycles. A fatigue analysis showed that the flexures will survive 86 lifetimes of actuations. The parallel double bladed flexures cause the primary interface shelf to rotate a very small amount during actuation of the focus me chanism. The movement of the interface shelf is not perfectly vertical. At the extents of travel, the maximum rotation of the interface shelf is 0.21° and the maximum out of plane displacement is 7.4 μm. This non-axial motion is mitigated with the compliant flexures in the struts that attach the focus mechanisms to the primary mirror. Lubrication The lubricated elements associated with the Focus Mechanism assembly are the motor-gearhead assembly, transfer gears, ball screw, and the ball screw support bearings. All elements were lubricated at Ball Aerospace, or in the case of the motor gears, under specific instructions from Ball. The selection of lubrication is highly dependent on operating temperature. The minimum operational temperature requirement for the Focus Mechanism originally was -75C. This made lubrication selection difficult because at this temperature, the wet lube becomes more visco us, increasing drag a nd friction in the system, which results in a lower torque margin. Dry lube is an alternative at extremely low temperatures, but restrictions in the operating environment for dry lube increase the cost and test set-up complexity. For this reason Braycote 601EF wet lube was chosen for all of the lubricated elements in the Focus Mechanism. A torque test was performed on the ball scre w and support bearings usin g the chosen lube at a temperature of -75C. The resulting rotating torques were dete rmined to be acceptable and used in initial torque margin calculations. Initially the Focus Mechanism design called for I 2R heating of the motor windings to warm the lube in the motor components, which decreases viscosity, if it was required at cold temperatures. Thermal analysis on the motor driver board later revealed that the driver chips on this b oard would heat up too quickly and reach dangerous levels. As a re sult of this, the design changed from I2R heating to a foil heater wrapped around the motor body. Subsequent drag torque testing at the motor vendor revealed that heating the gearmotor with a foil heater at -75C resulted in ample torque margin. Ultimately the Kepler thermal model was refined and the minimum operational temperature requirement was increased from -75C to -55C, which resulted in even higher torque margins in the Focus Mechanism. Redundancy There are two LVDT sensors mounted on either side of the primary mirror interface shelf. Only one is required for position feedback. Figure 14 shows one of the LVDT sensors on the focus mechanism. 368

Document

AMS_2008.pdf

The stepper motor contains 2 sets of 2 phase windings, each capable of independently meeting the performance requirements. The motor heater foil contains two sets of coils, only one is required for normal operation. Figure 14. Focus Mechanism LVDT Sensor and Motor Heater. Analysis Detailed finite element models of the focus mechanism were produced to predict structural and thermal stresses along with modes during launch loads. As a result of this analysis, several features of the focus mechanism were adjusted as the design matured to produce positive safety margins. A surrogate primary mirror was mounted onto the three completed focus mechanisms and vibration tested. The finite element models were correlated to the vibration test data to more accurately predict the axial, lateral and rocking modes of the focus mechanism/primary mirror assembly during launch loads. Testing Static Load Test On April 18, 2006 a static load test was performed to measure the stiffness of the focus mechanism in all 3 axes. An MTS Sintech Extensometer was used to apply a force on the focus mechanism in the area of the primary mirror interface and measure the resulting displacement. The forces applied were within the elastic limit of the unit and the data was used to correlate the finite element model of the focus mechanism. Figure 15 shows the Y-axis test. Required Torque Test In March, 2007 a required torque test at cold temperatu re was performed on the focus mechanism to determine the torque margin at the worst case temperature. The test was run in a climatic chamber that was cooled with liquid nitrogen. The motor was removed from the focus mechanism and replaced with a bearing/pinion gear/shaft assembly that was heated with a foil heater wrapped around the perimeter of the cylindrical housing. The torque to actuate the mechanism was measured with a torque watch. The results were within 20% of predicted values a nd the torque margin was acceptable. 369

Document

AMS_2008.pdf

Figure 15. Focus Mechanism Static Load Test, Y-axis. Functional Test All three of the focus mechanisms underwent and passed extensive functional testing in the spring of 2007. Most of these tests were performed to show that the focus mechanism meets the requirements shown in Table 1. Some of these tests included range of motion, unidirectional repeatability, axial position knowledge and smalle st increment of travel. One issue that somewhat slowed testing was the requirement for the motor ca se maximum operational temperature. This requirement states that the maximum temperature that the motor case can operate at is +60C. The reason for this is to avoid warming the lubricant in the bearings. If the lube becomes too warm, the viscosity decreases and it can migrate out of the bearings. During testing the motor could only run for approximately one minute at a time before the case temperature became too high. Temperature sensors on the motor case were monitored closely to insure that this limit was not exceeded. Figure 16 shows actual data from Focus Mechanism functional testing. The graph shows absolute position vs. LVDT counts for the entire range of motion. The X-axis is the absolute position, measured with a Ziess CMM (Coordinate Measuring Machine) and the Y-axis shows the corresponding primary and redundant LVDT counts. Each mechanism was fully characterized during this testing. Vibration Testing In April, 2007, all 3 focus mechanisms passed vibration testing with a surrogate primary mirror mounted to them. Figure 17 shows the test setup for one of the axes. 370

Document

AMS_2008.pdf

Figure 16. Range of motion test. Figure 16. Vibration Test Thermal Vacuum Testing In May, 2007, all three focus mechanisms passed thermal vacuum testing. All three units were tested simultaneously as shown in Figure 17. Mass simulators, 1/3 the mass of the primary mirror, were mounted to the interface shelf to simulate the correct loads. Each focus mechanism was mounted to an Invar base plate to simulate the CTE of the composite aft bulkhead. Cooling boxes were mounted over the motor cover box to increase the cooling rate at the motors and speed up functional testing 371

Document

AMS_2008.pdf

Figure 17. Thermal Vacuum Test Conclusion The design, fabrication, assembly and testing of the focus mechanism was successful due to the combined efforts of dozens of engineers, machinists and technicians at Ball Aerospace. It is impossible to list all of the lessons learned during this effort. One issue that caused a slight schedule delay involved the ball screw. When the ball screws were disassembled and lubricated at Ball Aerospace it was discovered that the ball sizes were oversized and the pre-load was too high. New balls needed to be ordered from the manufacturer which delayed schedu le. It is recommended up front to order a spare set of balls, 2.5 & 5.0 /g541m (0.0001 & 0.0002”) larger and smaller than the standard ball size to avoid delays. The focus mechanism has successfully met all design and schedule requirements that it has faced and is currently operating flawlessly in the Kepler Telescope. Acknowledgements The author would like to thank all of the Kepler machinists, technicians, analysts, engineers and managers that played a role in developing the focus mechanism. The skill, hard work, passion and dedication of this team of people were vital to the success of this effort. References 1. Kepler mission overview. Kepler/NASA website. http://kepler.nasa.gov/ 2. Illustration of Milky Way Galaxy copyright Jon Lomberg. http://www.jonlomberg.com 372

Document

AMS_2008.pdf

Precision Linear Actuator for Space Interferometry Mission (SIM) Siderostat Pointing Brant Cook*, David Braun*, Steve Hankins*, John Koenig* and Don Moore* Abstract “SIM PlanetQuest will exploit the classical measuring tool of astrometry (interferometry) with unprecedented precision to make dramatic advances in many areas of astronomy and astrophysics” (1). In order to obtain interferometric data two large steerable mirrors, or Siderostats, are used to direct starlight into the interferometer. A gimbaled mechanism actuated by linear actuators is chosen to meet the unprecedented pointing and angle tracking requirements of SIM. A group of JPL engineers designed, built, and tested a linear ballscrew actuator capable of performing submicron incremental steps for 10 years of continuous operation. Precise, zero backlash, closed loop pointing control requirements, lead the team to implement a ballscrew actuator with a direct drive DC motor and a precision piezo brake. Motor control commutation using feedback from a precision linear encoder on the ballscrew output produced an unexpected incremental step size of 20 nm over a range of 120 mm, yielding a dynamic range of 6,000,000:1. The results prove linear nanometer positioning requires no gears, levers, or hydraulic converters. Along the way many lessons have been learned and will subsequently be shared. Introduction SIM will improve “our understanding of the physical properties of stars, determining the mass, including the dark matter component, and its distribution in our Galaxy, observing the motions of the Milky Way’s companions in the Local Group, and probing the behavior of supermassive black holes in other galaxies” (1). Using three interferometers, 1 science and 2 guides, SIM will deliver a dramatically more accurate mapping of our universe as well as a better understanding of the formation and evolution of other planetary systems outside our own. Accurate mapping using interferometers requires high precision actuation, pushing the limit of both the mechanical positioning realm and the electronics/control realms. The required lifetime of SIM is also extremely challenging with a need to meet performance requirements for no less than 5.5 (with a goal of 10) years of continuous science observation. This is approximately 3 million large angle gimbal moves. Figure 1: SIM PlanetQuest While many precision mechanisms are required for SIM to deliver its science data, the Siderostat is the initial pointing mechanism used to direct starlight into the instrument. The Siderostat tips and tilts a 304.5-mm clear aperture optic across a 15-degree Field of Regard (on the sky) via a two-axis hexfoil flexured gimbal mechanism, with a required coarse accuracy of 1arc-second (as) and a fine accuracy of * Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Space Flight Center, May 7-9, 2008 373

Document

AMS_2008.pdf

5 milli-arc-second (mas) on the sky. Implementing a pair of linear actuators approximately 300 mm from the gimbal axis yields a required linear coarse accuracy of approximately 1 micron and a fine linear accuracy of approximately 5 nanometers. See Figure 2, Potential Siderostat Configurations. Figure 2: Potential Siderostat Configurations This paper will discuss the design, build, and test of a linear actuator capable of repeatedly, over long periods of operation, performing submicron positioning maneuvers. The design successfully pushes the limits of mechanical positioning while remaining true to the JPL principles of heritage, simplicity, and robustness. Along the way interesting lessons were learned and will be put forth for the reader’s benefit. Mechanical Design Generally speaking, design is a continuous balancing act to meet competing requirements. It requires the designer/engineer to carefully balance the design process between many conflicting requirements. In actuator design, if mutually exclusive positioning requirements are equally important a two-stage mechanism is often the solution. Unfortunately, multiple stages add to complexity, mass and cost. In our design, the need for large fast moves, trump the needs for fine positioning, and vice versa. It was originally thought that no single actuator could meet the large stroke, high speed, small incremental step size (actuator resolution), required by the Siderostat. The original intention of our Precision Linear Actuator, Direct Drive (PLADD) was to fill the role of a coarse actuator. The design approach taken with PLADD is one of simplicity, heritage, and robustness. The initial design required at least 2000 incremental positions per revolution of the nut/motor in order to obtain a linear incremental step size of 1 micron. Figure 3 shows a cross section of the actuator. Figure 3: SID PLADD Cross Section View 374

Document

AMS_2008.pdf

The bearing housing, as shown in Figure 4, is the center of the actuator, housing the back-to-back duplex angular contact bearings, brake, and motor. Both sets of bellows are suspended from the bearing housing, while the housing bolts directly to the carbon fiber outer tube for thermal stability. Figure 4: Bearing Housing The angular contact bearings are assembled into the bearing housing from opposing ends of the housing. The housing is stepped, creating a bearing spacer against which the bearings can be preloaded. The bearings are thin section MPB (Miniature Precision Bearings) angular contact ball bearings with a phenolic retainer ball spacer. The bearings have 440C balls and races. The bearings are shown in Figure 5. Figure 5: Timken MPB Angular Contact Ball Bearing The bearing inner races are mated to the motor drive link. As depicted in Figure 6, the inner race of the inner bearing seats on a non-sliding preload flexure. The DC brushless motor rotor is assembled to the rear of the drive link via set screws. The ballnut is mounted to the front of the drivelink via bolts. Figure 6: Bearing Preload Flexure and Rotor mounted to Drive Link The motor is coupled to the NSK ballnut by way of an in-house designed and built drive link. The drive link is shown in Figure 7. The complexity of the mechanism is reduced and backlash of a gear train is eliminated via a direct drive approach. By eliminating mechanical backlash in the actuator the control system is able to actively eliminate any mechanical imperfections in the actuator using feedback from the linear glass scale 375

Document

AMS_2008.pdf

encoder mounted to the actuators’ output. The direct drive approach also greatly simplifies manufacturing and assembly of the actuator by lowering part count and complexity. Figure 7: Motor Drive Link The heart of the actuator is a 12-mm diameter, 2-mm lead, NSK ballscrew mounted onto an NSK Double nut utilizing a spring preload of 130 newtons. The combination can be seen in Figure 8. Figure 8: NSK Ballnut The ballnut is connected directly to the drive link via 8 bolts. Figure 9: Ballnut bolted to Drive Link The ballnut drive link combination is assembled into the bearing housing with the angular contact bearings, preload flexure, and finally the piezo brake as pictured in Figure 10. 376

Document

AMS_2008.pdf

Figure 10: Housing Assembly The piezo brake consists of a set of flexured levers, preload springs, and guide flexures machined out of a single piece of titanium. The drive-link is radially compressed by the machined springs upon power-off. This constrains the drive link, and thus the ballnut, in rotation. The power must be applied in order to release the brake. The braking force is applied by the compressed preload springs via a wedged preload shim. The preload force is eliminated by the application of power to the piezo stacks. Figure 11: Piezo Brake The actuators’ driving torque is delivered via an off the shelf DC brushless motor from BEI/Kimco Magnetics as pictured below. Figure 12: BEI DC Brushless Motor Long life requirements along with a high correlation between lubricant consumption and life limitation led the team to immerse the mechanical drivetrain of PLADD in oil via a set of hermetically sealed bellows mounted to each end of the ballscrew. The bellows create a constant volume to house Brayco 815Z oil, and subjects the mechanical system to a continuous oil flush. The bellows additionally are used to hold the screw in rotation, thus allowing the rotating ballnut to produce linear motion at the screw. The continuous motion of oil through the mechanical system is conjectured to eliminate many failure modes. This hypothesis is being tested via life tests that are in progress. The motor windings, piezo brake, and inner bellows are bolted to the outer housing via thru bolts. 377

Document

AMS_2008.pdf

Figure 13: Inner Titanium Bellows Figure 14: Motor Windings Mounted A MicroE linear glass scale is connected to the end screw via a set of parallel motion flexures. These flexures provide a thermally stable mount for the glass scale via a low CTE metering rod. The scale is allowed to float relative to the inner tip of the ballscrew, minimizing the coupling of thermal drifts in the parts of the actuator not in the position path. The use of metering rods helps minimize false delta readings at the sensor head. The glass scale is mounted on a carbon fiber rectangular rod. A dummy glass scale is mounted opposite the operating scale to help eliminate bending in the scale due to changes in bulk temperature. Figure 15: Glass Scale Encoder Mount via Flexures The sensor head is mounted via a set of parallel motion flexures and the composite tube that acts as a metering rod. This completes the thermally stable sensor mount. 378

Document

AMS_2008.pdf

Figure 16: Sensor Mount All component alignment is achieved via sets of self aligning slip fit radial diameter lips between mating parts. All components are thus axially self aligning to within the tolerances of the slip fits. O-rings are used at all interfaces to eliminate oil leaks. The actuator connects to ground via a coaxial orthogonal flexure blade flange. The metering rods are hard mounted to each end joint. Figure 17: Flexured End Joint Results The initial resolution requirements for the ballscrew actuator was ±1 micron which was not clearly achievable based on vendor data for ballscrew position as a function of rotation angle. The selection of the Micro-E linear encoder capable of 5-nm position feedback was expected to give position sensing feedback well beyond that which was needed to determine the limits of ballscrew positionability. The direct drive architecture and a preloaded ballnut minimized actuator windup and backlash that enabled maximal position control. Additionally, rather than control the motor commutation based solely on the hall effect sensor as with typical motor control schemes the team also utilized the output sensor data. Since a 3-phase DC brushless motor is being driven, a PID controller reads the linear encoder and outputs 3-phased sinusoidal voltage commands to the 3 motor windings. Typical trapezoidal commutation uses only the Hall Effect sensors to apply the voltages in a rough way (24 steps per motor rotation). These 3 discrete sensors decode to 6 distinct points per pole, and since there are 4 poles per motor revolution that provides 24 points per revolution. Sinusoidal commutation makes use of the encoder data to fill in the gaps between the Hall Effect transitions and increase the points per revolution from 24 to 409,600. This allows the sinusoidal shape of the back-EMF waveform to be matched much more precisely. The geometry of the motor is such that sinusoidal commutation converts all of the applied current into useful torque (in the ideal case), rather than into undesirable radial forces. Since all the current is being converted into torque, the variation in torque is minimized and thus torque ripple is reduced compared to 379

Document

AMS_2008.pdf

trapezoidal commutation. This is critical for the SIM project because torque ripple induces excess vibrations in the structure that affect other sensitive components of the interferometer. The results of this architecture were surprising! The PID control loop utilizes 4 position sensor steps to start and stop the screw resulting in repeatable incremental step sizes of 20 nm! This means the DC brushless motor has to be capable of positioning to 100,000 distinct points around a revolution and the mechanical imperfections are beneath the control loop sensitivity. Furthermore, PLADD is able to track a prescribed profile over its entire range with a lag of approximately 16 ± 1.5 microns. However, with correct sequencing or control logic PLADD has an effective positional tracking error of ~3 microns peak to peak at a rate of approximately 3 mm/sec. Below is a short review of the results for a 10-mm move, a 1-micron move, and a 20-nm move. Also included is a discussion of the precision piezo brake effects, future work/improvements, and lessons learned. 10-mm Slew-Nominal Gains Nominal gains are the gains chosen for well-rounded performance characteristics for a given move size. There is a nominal set of gains for both high resolution (less than 1 micron) moves and low resolution (greater than 1 micron) moves. The nominal gains were set based entirely by trial and error, and are thus not necessarily optimal in any specific sense. Encoder resolution is increased for moves larger than 1 micron, down to 80 nm due to 22-bit electronics counting limitation. The electronics limitation is a designed in counter size on the FPGA that was not realized to be problematic until late in the build. It is in no way related to the commonly discussed 14-bit analog noise floor in most space electronics. This allows PLADD to cover the entire 120-mm range regardless of the initial zero point location chosen by the operator. PLADD is able to track a prescribed profile over its entire range with a lag of approximately 16 ± 1.5 microns. However, with correct sequencing or control logic PLADD has an effective positional tracking error of ~3 microns peak to peak at a rate of approximately 3 mm/sec. In Figure 18, the position command profile is created using constant acceleration to get up to speed, constant velocity for the main portion of the slew, and constant deceleration to approach the command position. Figure 18: 10-mm Slew (Nominal Gains) - Entire Move Figure 19 highlights the 16 micron lag during the constant velocity portion of the slew. It also shows the ~3 um error, peak to peak about the 16 micron offset. The sinusoidal commutation is also clearly displayed in this view. Figure 19: 10 mm Slew (Nominal Gains) - During Slew Figure 20 illustrates the end of the slew and the initiation of the integral gain to reduce the steady-state error. The proportional gain primarily contributes to slew performance. However, it can only reduce the error down to ~5 um using the nominal gain value. Once within this 5-micron zone the controller’s integral term integrates the error and creates the final step to its commanded position. 380

Document

AMS_2008.pdf

Figure 20: 10-mm Slew (Nominal Gains) -End of Slew Finally, Figure 21 shows the final result of the integral action. There is a small overshoot, which is then reduced to within ±80 nm of the command position. This is the limit of the capability of low-resolution mode, since the PID controller does not see any motion less than 80 nm (due to the previously discussed 22-bit legacy hardware limitation). The actuator could now be commanded with a new set of high-resolution gains allowing the zeroing in of the actuator down to ±5 nm of the commanded position. It is important to note that for large moves, although the actuator can move to within ±5 nm of the commanded position, in reality the overall accuracy of the move is affected by many other contributors, and are not discussed in this paper. The contributors can be, but are not limited to, thermal drifts in the unmeasured structure during the move, grating accuracy of the glass scale, and inter-grating accuracy. Large move accuracies are extremely complex comprising many error sources and are beyond the scope of this paper. Figure 21: 10-mm Slew (Nominal Gains) - Steady-state Error 1-Micron Step-Nominal Gains Full encoder resolution of 5 nm can be used for small moves of less than 20 mm. However, actuator velocity requirements generally lead to the use of high-resolution mode for moves only less than 1 micron. Using a different control resolution changes the scaling of the PID controller and thus a different set of gains are used in high-resolution mode versus the low-resolution mode used for large slews. Figure 22 shows a 1-um move using nominal gains. Figure 23 focuses in on the last part of the move to show the steady-state error. The nominal gains provide a fast response with minimal overshoot. The steady-state error is improved over the low-resolution mode because the PID controller is able to respond to deviations from the commanded position as small as 5 nm. The background noise in the plot will be further addressed in the Noise and Vibration section of this paper. Figure 22: 1-um Step (Nominal Gains) - Entire Move 381

Document

AMS_2008.pdf

Figure 23: 1-um Step (Nominal Gains) - Steady-state Error 20-Nanometer Step-Nominal Gains Due to a 4X minimum encoder resolution command limitation within the control system (a legacy piece of hardware originally intended for another use) 20 nm is the smallest incremental step that can be taken. The following series of plots will show 20-nm steps using the same nominal gains as the 1-um steps. Figure 24 displays a 20-nm move. Overshoot is non-existent, with a move time of roughly one second. At this nanometer level of position control, the noise of the encoder position signal is significant. It has a noise of approximately ±5 to ±10 nm, with occasional spikes of ±20 nm. The nature of this noise is not fully understood, but is conjectured to be from a combination of mechanical (background vibration) and electrical (sensor and wiring noise) sources. Environmental noise and its potential elimination are discussed in the Noise and Vibration, Future Work, and Lessons Learned sections of this paper. Figure 24: 20-nm Step (nominal Gains) Noise and Vibration Nanometer level work requires an extremely quite environment. Vibration and external noise sources can greatly affect test results. All test data was taken in Bldg 318 (Space Interferometry Test Lab), which is specifically designed to be a low noise test facility. The test was performed on an air table separated from the building via an isolation pad. In order to better understand the test noise environment, we sampled 1 second of high rate position data (approx. 1240 samples/second) with no power applied to the actuator. The measurement below is unfiltered data. It is clear that there is a noise disturbance identical to the noise seen when the actuator reaches its commanded position at steady state. This suggests that the noise seen in our position graphs is either environmental mechanical or the electrical noise of the sensor. 382

Document

AMS_2008.pdf

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-4985-4980-4975-4970-4965-4960-4955POSITION TIME (s)POSITION (nm) Figure 25: Vibration Measurement - Test Environment and Sensor Noise Figure 26 shows a 30-point moving average of the steady state error of the actuator with the actuator powered on and attempting to hold a position. Averaging decreases the steady-state error down to ±4nm. The average position is -0.2 nm with a standard deviation of 1.14 nm. Figure 26: Steady State Error, 30-Point Moving Average Piezo-Brake Effect Creating a brake for an actuator performing nanometer size moves is not an easy job. In order to achieve high braking forces, PLADD is equipped with a piezo-actuated brake mechanism. This is required in a coarse stage/fine stage configuration where the coarse stage is turned off and locked while the fine stage operates. A brake is also a necessity when the actuator moves periodically. Through the use of a custom piezo brake and correct power shut off sequencing, the SID team was able to brake the actuator with minimum disturbance. The correct sequencing of the brake and actuator control power off must be performed in order to achieve the best results. Figure 27 shows a 200-nm step at one second with the brake engaged at the third second. Since the PID controller remains on during the slow braking, it recognizes and corrects for the brake-induced drift, bringing the actuator back to its commanded position. At the ninth second, the power is removed from the motor. The current brake design degrades the relative accuracy of the move and also requires much longer times to reach a position and stop. The SID team believes that some design alterations could create a brake with much less induced drift. Also, if the actuator were to be used as a fine actuator it could potentially always be powered on, thus requiring no brake. 383

Document

AMS_2008.pdf

Figure 27: Brake Test (Good Sequencing) Future Work To potentially reduce the incremental step size of PLADD down into the 2-5 nm range, we plan on taking the following steps: Mechanical Modifications MicroE Mercury II Encoder— Micro E recently introduced its next generation Mercury II encoder capable of obtaining 1.2-nm LSB. Installing this improved read head would allow the controller to attempt moves as small as 4.8 nm. Mechanical Disturbance Isolation— While the current test setup utilizes an extremely quiet building, an isolation pad, and an air table there are potentially further layers of isolation that could be added to the test setup. Potential gain can be had by installing an isolation layer of foam padding between the test fixture and the table. The row of actuators is also currently mounted along one edge of the air table. Control Modifications PLADD’s control logic is derived from code developed by JPL’s Ted Kopf for the Mars Exploration Rovers. Many elements of the code can potentially be improved to optimize PLADD’s performance. The following is a list of potential adjustments. Filter Encoder Count— The value read from the encoder has high frequency noise of +/- several bits. Currently, this value goes directly into the PID controller. Placing a filter in front of the controller would clean up the signal so that the PID command does not react to small noise variations. This could improve performance for very small step sizes.Adjust Rate Filter— There is already a filter inside of the PID controller on the internally generated rate signal. However, the time constant of this filter was set for an entirely different application (MER motor control) that used very different encoders. Adjusting this filter could help make the derivative gain useful for this application. Adjust Controller Scaling— The PID controller scaling was set for a very different application (MER motor control) and is not ideal for the SID ballscrew. Adjusting this scaling could result in performance benefits due to finer tuning of the PID gains. Add Bits to PID Controller Position Value— The PID controller currently uses 22 bits for the position value. For this application, the limited number of bits forces a choice between high resolution and large range. Increasing the bits allocated for the position value would allow high resolution control over the entire ballscrew range. This modification is tied in with the scaling modification listed above, as the scaling is effected by the number of bits. Control Down to Encoder LSB— The PID controller does not allow delta position commands less than 4x the encoder resolution. It might be worth investigating the reason for this in detail to figure out if it is feasible to remove that limitation. The idea is to find out just how far down we can push the incremental command resolution. Reduce Encoder Noise— The encoder noise is probably due to a combination of mechanical and electrical sources. Investigation into the nature of this noise might lead to ways to reduce it. One idea is to improve the shielding of the cable between the encoder read head and the encoder interpolation module. This cable transmits sensitive analog signals that are susceptible to noise. 384

Document

AMS_2008.pdf

Gain Calibration— By calibrating gains for many different move sizes and adjusting the command software to generate such commands on a per move basis PLADD could improve many performance parameters such as response time, overshoot, lag, etc. Lessons Learned 1.Nanometers are really, really small. When working in small numbers it is easy to forget how small nanometers truly are. In real life mechanical terms a nanometer is approximately 1/50,000 of the diameter of a human hair. When dealing with numbers of this size, deflections of mechanical components can play a large role in performance. Movements may not be actual position changes, like one is familiar with, but instead a “wind up” or deflection of the mechanical components in the drive-train. A well-designed system should minimize these effects. The remaining system deflections can be actively eliminated by employing an encoder in a location such that it measures the output of the actuator that one is interested in controlling. Also, environmental noises, especially when not in a “quite” building may completely overwhelm any experimental results 2.Nanometer incremental step size is achievable with no gears, levers, or hydraulic reducers. This one speaks for itself. It also was something I never expected possible. 3.Sensor technology is the limiting factor for fine positioning when proper mechanical design considerations are taken. Sensor technology is a fast growing field. As the sensor technology progresses one should not hesitate to take advantage of technology. This especially applies to space-flight applications where a fear of the unknown can often be crippling, especially when attempting to actuate mechanisms with finer resolutions and more precision. 4.A clear path for oil circulation/pumping shall be designed into the system to avoid excessive driving loads. In designing our actuator a clear path for the oil to pass was not considered. Upon complete assembly and operation it was discovered that much of our reserve torque was required to force the oil lubricant through the motor, screw, nut, bearing assembly. This leads to a reduced maximum velocity and an increased input of heat into to the actuator under continuous operation. 5.Full alignment adjustability makes everyone’s job easier. Access to the sensor without disassembly is vital to technician (or assembling engineers’) sanity. The actuator’s design requires that the sensor is housed inside a carbon fiber tube. The sensor cannot be tested for functionality or alignment until the actuator is assembled. There is also no way to easily adjust the sensor once it is determined it is not aligned properly (inevitable, as it is practically impossible to blind assemble the sensor in perfect alignment). This requires that the actuator be slightly disassembled, shims guessed on based on previous experience and the whole procedure repeated. This can be extremely frustrating and time consuming. This problem may be eliminated as sensor technology improves and alignment tolerances grow. However, until then be sure to design in alignment adjustability and allow for active alignment post-assembly. 6.Instrument thin section bearing should be used with caution as they are extremely easy to damage. We ran into a few issues with the handling of the thin section bearings during disassembly of the actuators. More than one bearing was lost as a result of improper handling by technicians. Take caution with your technician choice and your handling procedures. 7.Obtain external mechanical noise measurements to isolate electronics noise and environment noise. If external sensors are not used to measure the mechanical environmental noise it is impossible to troubleshoot noise issues between the mechanical and electrical systems. The use of a local mechanical 385

Document

AMS_2008.pdf

noise sensor can help in this process greatly. This is something we did not foresee, and thus we are still uncertain as to the root of our environmental noises. 8. Be aware that bellows buckle in proportion to their linear stiffness. Be cautious when using them to react torque over a large stroke. PLADD’s design uses titanium bellows to react out the rotational torque in the screw. Our bellows are extremely soft as they must travel 120 mm full stroke. The first buckling mode of the bellows occurs when a torque is applied. The bellows collapse down as it buckles. The torque required to induce buckling is directly related to the linear stiffness of the bellows. This phenomenon was ignored during design. Fortunately our actuator works at the bitter edge of buckling, but does not buckle. Rotational alignment of the bellows is also critical to buckling, as a misalignment, or initial rotational load lowers the buckling torque drastically. Conclusion The SIM Siderostat team successfully outperformed our initial goal of less than 1-micron repeatable incremental step size by a factor of 50. We strongly believe we will attain incremental step sizes of 2-5 nm for large ranges given the current sensor technology. Through the elimination of gears and their inherent backlash we have been able to show that the mechanical system is not the limiting factor in fine positioning applications, if appropriate design considerations are taken into account. We have realized that sensor technology more than any other factor is the limiter with regards to fine positioning over large ranges. We hope to be able to further develop this novel linear actuator concept and to eventually use it on SIM in our PLANETQUEST, and many other future projects. Acknowledgements The research described in this (publication or paper) was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. I would like to thank all involved with this effort for their hard work and dedication towards achieving our goals. Those involved beyond the authors were Rob Calvet, Yutao He, Bruce Scardina and Don Benson. I would finally like to thank Caltech and NASA for providing the great freedom to explore the unknown. References [1] Edberg, Stephen J, “SIM PlanetQuest: A Mission for Astrophysics and Planet Finding”, NASA/JPL White Paper, May 2005 [2] Koenig, John, “SIM Siderostat Ballscrew Testbed Electronics Report”, JPL, Feb 2007 [3] Moore, Don, “Actuator Logs and Assembly Notes”, JPL, 1999 386

Document

AMS_2008.pdf

Development of a Low-Cost Fine Steering Mirror Steven R. Wassom and Morgan Davidson* Abstract The Space Dynamics Laboratory has used intern al funds to develop a prototype low-cost two-axis fine steering mirror (FSM) for space-based and airborne applications. The FSM has a lightweight 75 mm-by-150-mm high-reflectance mirror, high angular deflection capability for along-track ground motion compensation and cros s-track pointing, and a 70-Hertz bandwid th for small amplitudes to help cancel unwanted jitter. It makes use of off-the-shelf components as much as possible. Key performance parameters are: Clear aperture, 75 mm; elevation angle, ±15 deg (mechanical); azimuth angle, ±60 deg (mechanical); slew rate, greater than 75 deg/sec; bandwidth, 70 Hz; steady-state average error, about 1 arcsec; average power dissipation, 0.4 Watts; mirror surface, figure, <0.1 waves RMS; and total mechanical mass, 1 kg. Key components for the elevation axis include a rotary voice coil and a unique patent-pending non-contact feedback sensor. The azimuth axis features a brushless DC motor and a high-resolution optical encoder. Rapid prototyping, autocoding, and real-time hardware-in-t he-loop (HIL) testing were used to develop the control algorithms. Additional accomplishments include temperature mapping of the feedback sensor, inventing a successful passive launch lock, launch vibration testing, and subjecting the system to a space-like environment at pressures down to 1e-7 torr and temperatures down to 164 K. Introduction Over the past two years, Space Dynamics Laboratory (SDL) has used internal funds to develop a prototype low-cost two-axis fine steering mirror (FSM). This mechanism is intended to have a broad flexibility for both spacebased and airborne applications. Figure 1. Prototype FSMThe main purpose of this effort was to reduce the cost of a normally expensive technology while maintainin g performance, thus enabling this capability for more applications. A $2 million price tag has not been uncommon. The first year’s development effort required about $200,000 and resulted in a successful working prototype (Ref. 1). This report will review the accomplishments of the first year and describe the additional progress of the second year that includes: /g120 Subjecting the system to a space-like environment at pressures down to 1e-7 torr and temperatures down to 164 K /g120 Verifying that the system will survive a typical launch vibration test profile /g120 Performing temperature mapping of the patent-pending feedback sensor /g120 Inventing a successful passive launch lock /g120 Mounting the system on a portable cart for demonstrations * Space Dynamics Laboratory / Utah State University Research Foundation, North Logan, Utah Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Space Flight Center, May 7-9, 2008 387

Document

AMS_2008.pdf

General Requirements Although this FSM was not intended fo r a specific program, some general top-level requirements and performance goals were established. An angular position accuracy of about 1-2 arcsec was desired to minimize jitter. A clear aperture of 75 mm was chosen as a good starting point, with the intention that the design be scalable to larger or smaller apertures. An elevation axis deflection of ±15 deg (mechanical) was chosen to perform ground motion compensation (GMC) along the flight path, and the azimuth axis deflection was chosen to be ±60 deg to achieve both off-track pointing and the ability to rotate far en ough to view an on-board calibration source. A high bandwidth approaching 70-100 Hz was desired to enable jitter control during pointing and scanning. The final performance parameters achieved are tabulated in the section below describing the test results. Design Description Mirror Design To meet the aperture and bandwidth goals, the mirror needed to be lightweight and low cost. Mirror trades included several materials (Be, Al, SiC) and shaft connections (bonded, bolted, or integral). The final selection was a lightweighted Al mirror to keep costs low, and a bolted connection to facilitate future design changes. The flat mirror geometry provides a 75-mm aperture when scanned to a 60-deg optical angle (30-deg mechanical angle). The flat mirror is elliptical with a 75-mm-by-150-mm clear aperture. Thermal considerations coupled with the light-weighting requirement for high bandwidth necessitated the open back form shown in Figure 2, which provides a 60% reduction in mass compared with a solid mirror. Total mirror mass is 0.16 kg corresponding with a 17 kg/m 2 optical surface area mirror density. Triangular cells were selected over hexagonal or square cells as they are considered to be the optimum cell geometry (Ref. 2). Triangular cells also work well in providing a uniform distribution of ribs across the part and structure for mounting features. The mirror is mounted on four flexure-isolated pads in the center of the mirror back. The mounting pads require flexure isolation to reduce mounting distortion of the optical surface; however, the stiffness of the flexures must balance mounting distortion isolation and mirror modal response. Finite element analysis was used to find the optimum performance balan ce for the scan mirror application. Mirror Surface Finish The mirror surface is fabricated using single point diamond turning (SPDT) combined with post-polishing. An RMS surface roughness better than 20 angstroms is obtained utilizing a post-polishing method termed VQ for “Visible Quality”. This process aids in the removal of the grooves typical from SPDT. Table 1. Mirror Optical Performance Specifications Description Specification Mirror Mass 0.16 kg Operating Temperature 210K Optical Aperture @60 deg Optical Angle 75 mm RMS Surface Figure < 1/10 waves HeNe RMS Surface Roughness < 20 angstroms Triangular Cell Open-Back Design Triangular Cell Open-Back Design Figure 2. Back View of Mirror Showing Triangular Rib Structure The RMS surface figure of the mounted mirror was measured to be near 1/12 wave HeNe at 210K. This shows the capability of this mirror to hold good optical figure. Test and analysis results are discussed in detail later in the paper. Table 1 summarizes the optical performance specifications. 388

Document

AMS_2008.pdf

Elevation Axis For the elevation axis bearing, options considered were flexures, flexural pivots, and conventional ball bearings. Flexural pivots were chosen based on their high deflection capability, negligible friction, and SDL’s extensive experience with these devices in their Michelson interferometer. Actuator choices for the elevation axis included piezo actuators, magnetostrictive actuators, voice coils, brushless DC motors, and stepper motors. A rotary voice coil was chosen primarily for its favorable ratio of rotation angle to package size, since only a limited rotation was required. Other advantages to the rotary voice coil are simplicity, low cost, and high bandwidth potential. The moving coil of the elevation axis and the mirror are bolted to a mirror mount, which is supported on flexural pivots. The pivots and the voice coil magnets are mounted to a compact U-shaped yoke. The mounting scheme for the pivots is the same that is used in SDL’s Michelson interferometers. The pivots were sized to support the rotating mass without failure during a representative launch load and to ensure virtually infinite life during on-orbit duty cycles. Figure 3. Close-up of Wedge Sensor, Showing Precision Wedge and Displacement Transducers The feedback sensor for the elevation axis was selected from the following options: non-contact proximity transducers (inductive or capacitive), strain gages, optical encoders, resolvers, and inductosyns. Inductive non-contact sensors were cho sen for their low weight, compa ct packaging, low cost, high resolution, and SDL’s related experience. The feedback device for the elevation axis is unique and patent-pending. In this angular measurement device (called the wedge sensor), two non-contact inductive displacement transducers are arranged in opposition facing a moving target wed ge made of aluminum (see Figure 3). These transducers exhibit excellent resolution and repeatability over a limited range. The precision wedge extends the range of the non-contact transducers by converting angular motion to limited linear motion. Using two opposed transducers minimizes the sensitivity of the device to relative displacements out of the plane of symmetry of the wedge (wobble, vibration, misalignments, etc). Nonlinearity in the wedge sensor over the large deflection angles was mapped using a theodolite. This map was incorporated into the elevation axis control algorithm. Azimuth Axis The azimuth axis requires much larger deflections and needs to move a larger inertia. Consequently, the bearing trades were more limited. Ball b earings were chosen due to their extensive heritage in space and SDL’s experience base. Actuator options were brushless DC motors and stepper motors. Brushless DC was chosen for its high bandwidth and accuracy capabilities and typically long service life. Feedback device options were encoders, resolvers, and inductosyns. An optical encoder was chosen for high accuracy, low mass, and related experience. The output shaft of the motor is connected to the yoke supporting the elevation axis. The optical encoder has a resolution of 4.5 arcsec per count. Although this drive system is not designed for space, it uses the same type of components as another SDL scan mirro r drive system, which has been operating flawlessly for over six years in orbit. Upgrading the azimuth axis to space-worthiness has low risk. Electronics Two independent sets of electronics are used to drive the azimuth and elevation motors. The azimuth motor is driven using a commercial linear brushless servo amplifier. The elevation motor uses a simple amplifier circuit based on an operational amplifier. 389

Document

AMS_2008.pdf

Structural/Thermal Analysis Mirror Modal Analysis Finite Element Analysis (FEA) was used to predict the modes of the mirror. The high control system bandwidth near 70 Hz required the mirror modes to be considerably higher. Figure 4. Mirror Mode 1, Torsional about ZAxis, 270 Hz A normal modes analysis was run. Table 2 lists the mode shape descriptions. Figure 4 depicts the first mode: a torsional mode about the Z-axis at 270 Hz. The majority of the deformation and strain energy is in the flexure mount. Mirror Mounting Distortion Predictions Some wave-front error (WFE) is caused when non-flat surfaces are mated to gether with a bolt preload, producing strain in the mated parts. FEA was again used to predict optical surface displacement. The analysis assumed that the combined worst-case flatness tolerance of the two mating surfaces produces a Z-axis forced displacement of the mounting surface. The mounting pads on the mirror have a coplanarity requirement of 2 μm. The mating surface is assumed to have a coplanarity or flatness of 12.7 μm. Conventional machining can get surfaces as good as 7.6 μm in flatness. This approach conservatively assumes that all the strain occurs in the mirror as if it were mated to an infinitely stiff part.Table 2. Mirror Modes Summary Mode Shape Number/Description Frequency (Hz) 1 Mirror torsional, about Z-axis 270 2 Mirror rocking, about Y-axis 390 3 Mirror vertical or Z-direction translation, with mirror bending (saddle mode) 1230 4 Complex X-direction side to side translation with mirror surface bending/waves 2700 5 Complex Y-direction side-to-side translation with mirror surface bending/waves 2900 6 Second saddle mode 3100 Optical surface deformations from each of the enforced displacements are then transformed into surface normal interferometric space using Zernike polyn omial fitting. Piston and tilt can then be subtracted, resulting in WFE due to mounting distortion. The results show that distortion in the mirror is dominated by astigmatism and tetrafoil. Predicted surface normal displacements are shown in Figure 5 with the dominant aberrations identified in Table 3. Mirror Thermal Elastic Predictions Thermal simulation of the FSM in a representative orbit was used to evaluate expected performance. For the simulation, the representative orbit was a nadir-pointing low-Earth orbit. A typical baffle geometry was assumed. Standard minimum and maximum values were used for long-wave infrared heat from the earth and for reflected sunshine. Diurnal temperature profiles for the system as well as a detailed mapping of temperature variation across the face of the mirror were the principle outputs from the thermal simulation. Table 3. Zernike Polynomial Coefficients Ordered by Highest Contributor Standard Zernike No. Zernike Description RMS WFE (waves HeNe) 14 Second order Astigmatism y 0.018 6 First order Astigmatism y 0.017 27 Second order Tetrafoil y 0.005 26 Third order Astigmatism y 0.005 390

Document

AMS_2008.pdf

Table 4. Results of Thermal Analysis Units Case 1 Case 2 Case 3 Azimuth motor / yoke isolation W/K 0.019 0.007 0.007 Baffle temperature K 160 160 180 Average mirror temperature K 210 201 210 Mirror face gradientK 0.11 0.08 0.07 Mirror face distortion Z-dir*nm 43 27 22 Figure 5. Predicted Surface Normal Deformations Thermal analysis results showed the two most sensitive elements of the thermal design were the level of thermal isolation between the yoke and the azimuth motor and the baffle temperature. These parameters were the principle drivers in determining the operational temperature of the mirror. Because the gradients across the mirror are affected by the overall mirror temperature, the optical distortion was weakly coupled to the same parameters, showing small changes as the mirror temperature and baffle temperature varied. Table 4 shows various amounts of isolation and baffle temperatures and the resulting mirror temperature and optical distortion. A typical diurnal temperature profile during one orbit and the resulting thermal gradients across the mirror face are shown in Figure 6. Mirror Temperature Diurnal Variation - One Orbit 209.6209.8210.0210.2210.4210.6 0 1 02 03 04 05 06 07 08 09 0 Time (minutes)Temperature (K) Figure 6. Typical diurnal temperature profile (left) and thermal gradients across mirror face (right) FEA was again used to predict elastic distortion caused by thermal gradients across the mirror during operation in space. Coefficient of thermal expansion (CTE) mismatch at the mounting interfaces has no impact given that the mating materials are aluminum. The thermal elastic analysis predicts a peak-to-valley sag (Z axis on a flat mirror) of 22 nm or 0.034-wave HeNe (see Figure 7). This distortion is very small an d should not be a factor in maintaining a good optical figure in operation. 391

Document

AMS_2008.pdf

Figure 7. Sag (meters) of Optical Surface due to Thermal Elastic Deformation Control System Development Instead of following a traditional waterfall or serial development process for the elevation and azimuth closed-loop control algorith ms, SDL used the Rapid Prototyping Methodology (RPM) spiral development process shown in Figure 8. This process can be quickly iterated due to its software-driven nature. Customized features for line-of-sight (LOS) optical pointing systems include: integrated end-to-end modeling, co-simulation, and 3-D visualization of structural and system dynamics, controls, and optics; automatic C-code synthesis from block diagrams; real-time hardware-in-the-loop (HIL) testing; dynamic automated ray tracing, and a customizable GUI to monitor testing and change control parameters “on the fly.” Eventually, the block diagram transforms from a virtual model of the system to the complete integrated assembly. The control algorithm is then embedded in the actual computer. The advantages of RPM include: /g120 virtually no software written by hand (except for occasional device drivers) /g120 substantial savings of time and money in code generation /g120 short iteration cycles that result in early problem identification and solution /g120 changes can be made early in the design cycle at the component level The left side of Figure 9 shows the multi-body dynamic model of the FSM. This high-fidelity model includes flexible- and rigid-body representations of all major components (mirror, mirror flexure, flex pivots, actuators, etc.). The right side of Figure 9 shows the top-level hierarchy of the block diagram simulation of the FSM control system with the dynamic model embedded for co-6. Full System Verification via Embedded Code Testing5. Subsystem Verification via Hardware-in-the-Loop Simulation/Testing4. Automatic Code Generation, Real-Time Simulation & Visualization1. LOS Control Problem & Requirements Definition2. Physical & Empirical Plant Modeling & Visualization3. Control Algorithm Design, Analysis, Simulation & Visualization ITERATIONS & REFINEMENTS 6. Full System Verification via Embedded Code Testing5. Subsystem Verification via Hardware-in-the-Loop Simulation/Testing4. Automatic Code Generation, Real-Time Simulation & Visualization1. LOS Control Problem & Requirements Definition2. Physical & Empirical Plant Modeling & Visualization3. Control Algorithm Design, Analysis, Simulation & Visualization ITERATIONS & REFINEMENTS Figure 8. SDL’s RPM for Control Systems 392

Document

AMS_2008.pdf

simulation. Each of the blocks shown has detailed algorithms and calculations within it. Analytical tools in the software were used to determine the control gains, shape the open- and closed-loop responses, and simulate the step response and frequency response. Figure 9. FSM Dynamic Model Test Results and Model Validation Mirror Surface Figure Testing Surface figure tests were performed on the mirror to better understand the optical performance and validate the predictions. Mounting distortion and thermal elastic distortion at cryogenic temperatures were tested using a commercial Zygo Fizeau interferometer. The test setup (see Figure 10) consisted of mounting the mirror in a bell jar test chamber on an optical bench. The mirror is insulated and baffled to reduce thermal gradients during testing. The mirror looks out the bell jar window. The interferometer is set up to view the test mirror through a fold mirror. The fold mirror is necessary to adjust tilt on the mirror, as it is very difficult to move the large bell jar or the interferometer in fine increments. It should be noted when viewing the test results that the interferometer only spans 100 mm of the 150 mm elliptical test mirro r aperture. The test mirror aperture outside the 100-mm center was tested separately with the results showing a more behaved WFE. Also, WFE from the fold mirror was not removed from any of the results. Room temperature tests were performed with various mounting torques applied to the mirror. The interferometric results show very little distortion from mounting. Figure 11 shows the WFE after applying 0.68 N·m to the number 4 screws. The RMS WFE error is essentially 0.025 waves HeNe with or with out the applied torque. The circular grooves noticeable in the interferogram are residual SPDT marks in the Figure 10. Mirror Surface Figure Test Setup 393

Document

AMS_2008.pdf

mirror. The polishing house is confident that these marks can be polished out of future mirrors. Figure 11. Mounting Distortion Test, 0.68 N·m Torque Applied The mirror was then cooled with liquid nitrogen. Testing the mirror over temperature showed that WFE is dominated by the CTE mismatch of the VQ coating and the aluminum substrate producing bending in the mirror at the low cryogenic temperatures. The aluminum shrinks more at cold temperatures due to its larger CTE producing a slightly convex mirror. This bimetallic bending is much less than is typically seen with nickel-plated mirrors. Figure 12 shows how the mirror shape changed from concave at 292K to convex at 186K. RMS WFE error measured at 292K and 186K were 0.039 and 0.106 waves HeNe respectively. Figure 12. Interferometric Measurements at 186K Figure 13 shows the RMS, PV, and Power WFE from 292K to 141K. Using this data, the RMS WFE at the required operating temperature of 210K is linearly interpolated to be nearly 0.083-wave HeNe. The mirror retained a very good figure at 141K with a deformation shape nearly identical to the 186K measurement and a 0.125-wave RMS WFE. At temperatures lo wer than 141K, the WFE did not increase noticeably. Wave front measurements were measured as low as 92K with very similar performance as seen at 141K. Transient thermal gradients affected the accuracy of this data. Quilting or rib print-through was not detectable in the testing, even at temperatures as low as 92K. The mirror does however show substantial high order WFE. 394

Document

AMS_2008.pdf

The mirror surface figure exhibits great stability under mounting and cryogenic temperatures with some expected and very small bimetallic bending of the optical surface. At the 210K operating temperature the RMS WFE is near 0.083-wave HeNe. Control System Performance Testing Considerable testing was performed to quantify the control system performance. The real-time HIL tools of the RPM process were used to test and optimize the control laws for both the elevation and azimuth control loops. A graphical block diagram was used to monitor the performance and change the control parameters “on the fly” as the tests were running. Tests performed included large and small steps, and small-angle frequency respon se. The frequency response tests were performed using a control systems analyzer, which excited each axis with a sine sweep of 100 arcsec amplitude and a frequency range from 1 Hz to 1000 Hz, and analyzed the resulting waveforms for gain and phase shift. The analyzer was also used to test individual components and subsystems and thus obtain their open-loop transfer functions, which were then implemented in the model.-0.6-0.5-0.4-0.3-0.2-0.100.10.20.30.40.50.6 0 50 100 150 200 250 300 350 Temperature (K)WFE Wa HeNe (633nm) ves PV RMS Power210 K Operating Temperature Figure 13. Mirror WFE at Cryogenic Temperatures As an example of model validation, Figure 14 shows the response of the azimuth axis to a 30-deg step and compares it to the prediction from the simulation. The response is well behaved with little overshoot. The average slew rate from 0 to 100% of the commanded angle is 164 deg/sec. Performance Summary Table 5 summarizes the performance of the final prototype in key areas. The mass total is only for the mechanism and does not include the electronics, since the electronics are only breadboard at this stage of the development. The average power was determined from the simulation by commanding both axes to perform continuous simultaneous slow scans, as would be used for GMC. The peak power was obtained by commanding both axes to do a large-angle step simultaneously. 0 0.1 0.2 0.3 0.4 0.5 -505101520253035 Time (sec)muth Angle (deg)Azicommand data simulation Figure 14. Azimuth Axis Large-Angle Step Response 395

Document

AMS_2008.pdf

The error mean and standard deviations were calculated using the statistics of the angular position data at the end of a 10-second step-and-hold for both large and small steps. This is not absolute mirror pointing accuracy, but rather the tracking error between the commanded angle and the angle measured by the encoder for azimuth and the wedge sensor for elevation. Absolute pointing accuracy, as measured with the theodolite, was about 0.03 deg at room temperature. The azimuth error is largely due to the limited counts of the encoder, 288000 counts over 360 deg of rotation, which results in 4.5 arcsec/count. The encoder toggles by one count during steady state, resulting in the error. The elevation axis is extremely repeatable, as shown by the small mean error; however, the noise, represented by the error standard deviation, is larger than desired. The noise is attributed mainly to the 16-bit A/D computer card’s peak-to-peak noise of over 15 counts, as claimed in the card manufacturer’s specification sheet. This equates to about 25 arcsec of peak-to-peak noise, when the 16-bit resolution is applied to the full 30-deg mechanical angle of the elevation axis. Some noise is also gener ated by the wedge sensor electronics box. A digital moving average filter was applied to the wedge sensor feedback, which helped considerably. The control algorithm also contains a derivative term in the forward loop, which tends to enhance the noise. The closed-loop bandwidth for each axis, as measured using the control systems analyzer, was about 70 Hz at the -3 dB point. The gain margin of 6 dB was verified by doubling the control algorithm gains in each axis and performing a step response test to determine that the system remained stable, which it did. FSM Portable Demonstration Figure 15 shows the FSM portable demonstration system, which includes t he FSM, the breadboard electronics, the host and target computers, and a laser for “light show” demonstrations. The system was programmed to project a Lissajous figure and an “S” figure on the ceiling at high frequencies, and also to do slow scans for audience visualization. Wedge Sensor Temperature Mapping The current wedge senso r for feedback experiences drift with temperature, so a mapping was performed to show feasibility of achieving high absolute pointing accuracy. One possible application of the FSM requires a ±50-arcsec absolute pointing accuracy over a 0-40°C temperature range. The angle was measured using an autocollimator and mapped versus sensor voltage and temperature. Figure 16 shows that the results appear predictable and highly repeatable, verifying feasibility of mapping. Table 5. Performance Specifications of Final Prototype Specification Value Aperture (mm) 75 Mass (electronics not included) (kg) 1.0 Avg. Power (W) 0.4 Peak Power (W) 30 Azimuth Rot. (mechanical deg) ±60 Elevation Rot. (mechanical deg) ±15 Azimuth Error Mean (arcsec) <1 Azimuth Error Std. Dev. (arcsec) <3 Elevation Error Mean (arcsec) <0.05 Elevation Error Std. Dev. (arcsec) <6.5 Azimuth Slew Rate (0-100%) (deg/sec) 160 Elevation Slew Rate (0-100%) (deg/sec) 75 Bandwidth (Hz, 100 arcsec amplitude) 70 Gain Margin (dB) 6 Figure 15. Portable Demonstration 396

Document

AMS_2008.pdf

-1200-1000-800-600-400-200020040060080010001200 11.4 11.5 11.6 11.7 11.8 11.9 12 12.1 12.2 12.3 Sensor Voltage (A+B, volts)Angle (arcsec )20.9 deg C 21.3 C 22.0 deg C 22.7 deg C 23.9 deg C 24.1 deg C 26.4 deg C 29.9 deg C 35.6 deg C 40.0 deg C Figure 16. Temperature Mapping Launch Lock Launch lock options included a permanent magnet, a shape-memory alloy pin-puller, and a wax actuator. The magnetic concept was chosen based on simplicity and a 6-year on-orbit heritage with another SDL scan mirror. Figure 17. Magnetic Launch Lock The patent-pending launch lock is shown in Figure 17. The combination of a magnet with a cone-shaped receptacle provides restraint in all directions. The restraint force is adjustable by changing the air gap, which is done by screwing the cup holding the magnet in or out and locking it with the set screw. The holding force is high enough to restrain motion during external loads, but is low enough that the actuator can overcome the holding force to release the lock. For this particular prototype, the latch force can be about 2.2 N maximum, since it is limited by the maximum current (~1 amp) and torque of the voice coil actuator. Analysis was performed to validate the design by applying the launch vibration spectrum to the dynamic model of the FSM shown in Figure 9. Vibration Testing A vibration test was conducted to verify the performance of the launch lock and the survivability of the FSM. The 3-minute test was based on proto-qualification levels for the planned STP-SIV spacecraft, Table 6. STP-SIV Spacecraft Vibration Spectrum Frequency (Hz) PSD – Acceptance (g2/Hz) PSD – Protoqualification (g2/Hz) 20 0.005 0.01 100 0.012 0.024 800 0.012 0.024 2000 0.005 0.01 Overall (grms) 4.27 6.04 397

Document

AMS_2008.pdf

shown in Table 6, with the first minute at reduced power and two minutes at full power. The test consisted of the following duty cycle: /g120 -12 db for 15 sec /g120 -9 db for 15 sec /g120 -6 db for 15 sec /g120 -3 db for 15 sec /g120 0 db for 2 min. The launch lock magnetic force was set to about 1.5 N for the first vibe test. The launch lock released after 105 se c., but no violent mirror motions were observed. Measurement of the latch force after the test showed it to be about 1.4 N, so the set screw may have been loose. The latch force was increased to about 2.2 N for the second test. This test was successful. Before and after each test, the FSM was powered up and exercised with its nominal duty cycle of a Lissajous figure, an “S” figure, and slow scans. The performance appeared nominal before and after each test. The FSM was also repeatedly commanded to latch and unlatch positions, verifying that the voice coil could overcome the latch force. Figure 18. FSM Mounted in Thermal Vacuum Chamber Thermal-Vacuum Testing Although the FSM prototype was not intended for harsh te sting, it was mounted in one of SDL’s thermalvacuum chambers (Figure 18) and subjected to 55 duty cycles with pressures as low as 1e-7 torr and temperatures to 164 K (the lower limit of the chamber’s capability). The azimuth axis worked for 26 runs as the temperature dropped, until the thermocouple on the az imuth motor housing reached 227 K. This axis then ceased working as expected due to the bearing lube freezing up, since its published range is 233 K to 423 K. The elevation axis operated for all runs, down to 164 K. Figure 19 shows how the elevation axis “rings” at cold temperatures due to a change in the plant resulting in improper gains. The feedthrough cables add considerable noise, as shown for small angles in Figure 20. The azimuth noise was noticeably worse than the previous year, possibly due to bearing abuse during the vibe test, or loss of lubricant during vacuum testing. The encoder optics could also be impaired due to outgassing on the lens. Figure 21 shows how the “before” and “after” performance of the elevation axis at ambient for large angles is nearly identical. Some outgassing was evident from the residue collected in the chamber’s cold trap, although it was not analyzed after testing. 9 9.2 9.4 9.6 9.8 10 10.2 10.4 10.6-6-4-20246 Time (sec)Elevation Angle (deg) Command Case 45, 1.2e-7 torr, 254 K Case 43, 1.1e-7 torr, 165 K Figure 19. Effect of Temperature on Elevation Step Response 398

Document

AMS_2008.pdf

Lessons Learned and Future Work The FSM prototype has been shown through testing to have the potential for robust performance in the harsh environment of space. The elevation axis operates at hard vacuum and cold temperatures down to 164 K. The “before” and “after” ambient performance of the elevation axis is virtually the same. The azimuth axis works down to 233 K until the current bearing lube freezes up. The launch lock successfully handles a typical launch environment. Obviously, some materials changes would need to be made for space and/or cryogenic applications. The azimuth lubricant would need to be changed. Also, the gains for both axes should be scheduled based on temperature. The cause and corrective action for the cable noise in both axes would need to be identified. 1.95 2 2.05 2.1 2.15 2.2 2.25-100-50050100150200250300350400 Time (sec)Elevation Angle (arcsec)Command Original 2006 data, ambient Case 52, ambient, original cables Case 51, ambient, feedthrough cables Case 35, 1.e-6 torr, 198 K, pump off Figure 20. Effect of Cable Noise on Elevation SmallAngle Response The elevation axis circuit design could be modified to allow a slightly higher current limit of about 2 amps to provide additional latch force and stall torque margins. The focus of future efforts are being centered more on the development of space-type electronics, as described in the following paragraphs. The control algorithms and filters need to be further refined and optimized to reduce the noise, especially in the elevation axis. This may be done by moving the derivative term to the feedback loop, using optimal state-space methods with observers, etc. The algorithms need to be implemented in low-cost space-qualified processing hardware, such as an FPGA, DSP, or microprocessor. This may necessitate conversion to fixed-point math, as was done for the radiation-hardened fixed-point microprocessor used to control a FSM in another SDL space payload (Ref. 3). A protocol or set of operational codes needs to be defined to communicate digitally with the FSM, such as commanding it to move to a position, querying it for the current position, etc. The D/As, A/Ds, wedge sensor electronics, and quadrature encoder functionality need to be implemented using space-rated components. 1.95 2 2.05 2.1 2.15 2.2 2.2501234567891011 Time (sec)Elevation Angle (deg)Command Original 2006 data, ambient Case 33, 1.9e-7 torr, 198 K Case 53, ambient, original cables Case 50, ambient, feedthrough cables Figure 21. Ambient Data for Elevation Angle LargeStep Response The autocoding feature of the RPM process needs to be evaluated to determine how much of the code automatically generated for the control algorithms can be used in the final product. 399

Document

AMS_2008.pdf

The input power interface needs to be defined to either use the spacecraft power to drive the motors or else have a customer supply the motor power. Off-the-shelf space-rated three-p hase motor amplifiers need to be evaluated. It would also be desirable to develop an automated procedure for mapping the wedge sensor output against temperature and angle using optical techniques and precise temperature control. Summary A low-cost two-axis FSM has been successfully developed and demonstrated for air- and space-based sensors. The 75-mm aperture mirror is lightweight, isolated by flexures, and maintains a surface figure of less than 0.1-wave RMS down to a temperature of 210 K. The drive system uses an innovative combination of off-the-shelf components to achieve large angles, high slew rates, high bandwidth, and relative position error less than 1 arcsec. A rapid prototyping methodology has been used to develop the control laws. Temperature mapping has been employed to improve the absolute pointing accuracy. A passive patent-pending launch lock has been developed and demonstrated. Vibration testing has been performed to show the survivability of the FSM and the successful operation of the launch lock. The mechanism has been subjected to many duty cycles in a hard vacuum and at low temperatures. Areas for improvement include a lower-temperature azimuth lubricant, gains scheduled with temperature, noise reduction in the cables, a higher elevation axis current limit, and development of space-worthy electronics. Acknowledgments The authors express their sincere appreciation to the following individuals for their significant contributions: Richard Sanders and Melissa Draper, mechanical design; Trent Newswander, mirror design, analysis, and testing; Brent Jensen, structural analysis; Brian Thompson, trade studies and conceptual design; Steve Dans ie, Andrew Little, and Scott Schicker, me chanical technologists; Duane Miles, optical technologist; Jeff Blakeley, analog and power electronics; James Cook and Zach Casper, electronics and software; Adam Shelley and Quinn Young, thermal design; Aaron Gilchrist, temperature mapping; and Jim Herrick, vacuum testing. The authors also appreciate the funding and support from the Research Division of SDL, directed by Dr. J. Steven Hansen and Dr. Scott Jensen. References 1. Steven R. Wassom, Morgan Davidson, Trent Newswander, James Cook, Zach Casper, Adam Shelley, “Fine Steering Mirror for Smallsat Pointing and Stabilization,” 20th Annual AIAA/USU Conference on Small Satellites, Paper #SSC06-VIII-7, 17 August 2006. 2. Ahmad, Anees, Optomechanical Engineering Handbook , CRC Press, 1999. 3. Steven R. Wassom, Chad Fish, Mitch Whiteley, Dave Russak, Joel Nelsen, Brian Thompson, Glen Hansen, Jason Wooden, Larry Gordley, John Burton, Mark Hervig, Paul Cucchiaro, Dan Hammerle, “SOFIE Pointing Control System,” SPIE Proceedings Vol. 6297, Infrared Spaceborne Remote Sensing XIV , 7 Sep. 2006. 400

Document

AMS_2008.pdf

Cryogenic Focus Mechanism for the Spitzer Space Telescope William C. Schade* Abstract A new focus mechanism was developed, tested, and flown for the Spitzer Space Telescope (“Spitzer”), one of NASA’s “Great Observatories”. Figure 1 shows the Flight Focus Mechanism (FLT-FM), now in Spitzer. The mechanism uniquely provides robust support and precise focus adjustment for the Spitzer secondary mirror, from 300 K to a 5 K cryogenic environment. This paper summaries the requirements, performance, description, and testing of the focus mechanism, including key component level tests of a geared-stepper motor and ball screw. Also, a secondary mirror mount is described that minimizes mirror distortion and supports high loads. Several design and test challenges were overcome and lessons l earned from this successful development include: /g120 Titanium is useful as a flexure material to liquid helium temperatures. /g120 Adhesive bonds at cryo-temperatures should be well understood and / or tested. /g120 Geared-stepper motor and ball screw components were simply modified to work to < 5 K. Figure 1. Flight Focus Mechanism (FLT-FM) * Ball Aerospace & Technologies Corp. (BATC), Boulder, CO Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Space Flight Center, May 7-9, 2008 401

Document

AMS_2008.pdf

Introduction The Spitzer observatory shown in Figure 2 includes the Cryogenic Telescope Assembly (CTA) that directs infrared signals to various instruments by means of a beryllium primary and secondary mirror. Ball Aerospace & Technologies Corporation (BATC) supplied the CTA as shown in Figure 3, with funding and oversight provided by NASA-JPL. Early in the program it was decided a focusing capability was desired, providing the cost and complexity to achieve it was reaso nable and providing it could be developed on time. Fortunately these objectives were met, so the focus mechanism is included in the CTA as shown, mounted on a beryllium metering structure. Spitzer was launched on August 25 th, 2003 with the focus mechanism and other key telescope components at ambient temperature. The telescope components were later cooled in space to less than 5.5 K. This space-assisted cooling approach was beneficial in helping to preserve Spitzer’s cryogen. Consequently, the observatory has surpassed its expected 2.5-year life and is approaching a 5 year life goal. Additional Spitzer facts are given on the next page, for reference. 0.85 m dia PrimaryMirrorFocus Mechanism (& Secondary Mirror) Instrument Chamber Cryostat (Liquid Helium) Metering Structure Dust Cover (ejected after deployment) Solar Panel and Sun ShieldCryogenic Telescope Assembly (CTA) Spacecraft Bus Figure 2. The Spitzer Observatory Figure 3. Cryogenic Telescope Assembly (CTA) 402

Document

AMS_2008.pdf

Spitzer “Fast Facts” For reference, this page includes Spitzer “Fast Facts” from the California Institute of Technology web site http://www.spitzer.caltech.edu/about/fastfacts.shtml . The Spitzer Space Telescope is a space-bor ne, cryogenically-cooled infrare d observatory capable of studying objects ranging from our Solar System to the distant reaches of the Universe. Spitzer is the final element in NASA's, Great Observatories Program and an important sc ientific and technical cornerstone of the Astronomical Search for Origins Program. Launch Date: 25 August 2003 Launch Vehicle/Site: Delta 7920H ELV / Cape Canaveral, Florida Estimated Lifetime: 2.5 years (minimum); 5+ years (goal) Orbit: Earth-trailing, Heliocentric Wavelength Coverage:3 - 180 microns Telescope: 85-cm diameter (33.5 inches), f/12 lightweight Beryllium, cooled to less 5.5 K Diffraction Limit: 6.5 microns Science Capabilities:Imaging / Photometry, 3-180 microns Spectroscopy, 5-40 microns Spectrophotometry, 50-100 microns Planetary Tracking: 1 arcsec / sec Cryogen / Volume: Liquid Helium / 360 liters (95 Gallons) Launch Mass: 950 kg (2094 lb) [Observatory: 851.5 kg, Cover: 6.0 kg, Helium: 50.4 kg, Nitrogen Propellant: 15.6 kg] Major Innovations /g120 Choice of Orbit /g120 Warm-Launch Architecture /g120 New Generation of Large-Format Detector Arrays /g120 Lightweight, cryogenic optics The Spitzer Team /g120 Jet Propulsion Laboratory /g120 Spitzer Science Center, California Institute of Technology /g120 Ball Aerospace and Technologies Corporation /g120 Lockheed Martin Space System Company /g120 Smithsonian Astrophysical Observatory /g120 NASA-Goddard Space Flight Center /g120 Cornell University /g120 University of Arizona 403

Document

AMS_2008.pdf

Focus Mechanism Key Requirements and Performance The purpose of the focus mechanism is to maintain the Spitzer’s secondary mirror position and, if desired, move it along the CTA optical axis for focus adju stment. As a single axis device, it must maintain mirror alignment with minimal de-center or tilt motion of the mirror due to launch, cool-down, or focus operation. This basic functionality is reflected in the key requirements and tested performance of the mechanism, as summarized in Table 1. And, since it must be controlled reliably and remotely, redundancy is required. The performance results are for ambient and cryogenic temperatures, before and after exposure to launch vibration levels, and after life testing of a Focu s Engineering Model (FM-EM) to 1X life. The FMEM was built and tested to reduce risk and for life testing up to 4X life, prior to making the Flight Focus Mechanism (FLT-FM). The tests for the FLT-FM were more abbreviated, but performan ce was similar. Table 1. Focus Mechanism Key Requirements and Performance Key Requirements Tested Performance (Meets all) Range /g116/g114 0.25 mm /g100/g114 0.50 mm /g116/g114 0.25 mm (soft limits) /g100/g114 0.33 mm (hard stops) Step size /g100 2.5 /g80m /g100 1.3 /g80ma Repeatability/g100/g114 1.25 /g80m (unidirectional) /g100/g114 0.81 /g80ma (bi-directional) b De-center over range c/g100/g114 5.0 /g80m /g100/g114 3.61 /g80m Tilt over range c/g100/g114 58 /g80rad /g100/g114 41 /g80rad Shift after launch d /g100/g114 12.5 /g80m de-center /g100/g114 116 /g80rad tilt /g100/g114 9.3 /g80m de-center /g100/g114 110 /g80rad tilt Operating temperature 300 to 2.5 K Meets Operating pressure Ambient to 10-6 Torr Meets Launch acceleration 70 G lateral 125 G axial Meets First mode frequency > 150 Hz 330 Hz Clear aperture of SM /g135 120 mm min /g135 123.8 mm min Mass (with mirror) /g100 4 kg Meets Notes pertaining to Table 1 : a) Mean and standard deviation values for these are given in the assembly test section. b) Repeatability was met bi-directionally, while only unidirectional was required. c) De-center and tilt over range refer to mirror motion al ong and rotation about any axis normal to the optical axis, respectively, over the mechanism’s full range. d)Shift after launch refers to mirror motion due to temperature repeatability and launch vibration combined. The former is any difference in mirror position from ground alignment at 5 K and after cooling again on orbit. The latter is any permanent shift due to launch. Aside from the requirements, a goal was set to limit the focus shift of the mirror after launch to /g100 2.5 /g80m. This was a goal only since the mechanism can correct for focus after launch and it was anticipated performance might slightly exceed this. Performance for this was measured at /g100 3.2 /g80m, which was just over the goal, as was expected might happen. The FM-EM also demonstrated a viable flight mechanism could be made on time. The FM-EM design was started in November 1997, it was built, and cryo-testing was started by June 1998, as required. This success paved the way for the build of the FLT-FM. 404

Document

AMS_2008.pdf

Focus Mechanism Description Concept Stepper Motor SM Carrier An initial challenge was to select a concept that would efficiently constrain the mirror in all but one direction, yet provide for precise focus control. The concept in Figure 4 achieves this. The secondary mirror (SM) is held on a carrier and tube suspended on two diaphragm flexures, which are stiff in all but the focus direction. Focus is precisely controlled by a stepper motor acting though a lead screw, lever arm, and flexure system, as shown. Focus step size and position are simply determined by counting motor steps. Advantages of this concept are summarized as follows: Lead Screw Lever Arm Pivot & Drive Flexures SM /g120 Simplicity – Minimizes cost /g120 Effectiveness – Meets requirements /g120 Robustness – Carries high launch loads /g120 No free play in flexures – Repeatable /g120 Low actuation force – Low screw loads /g120 Symmetry of major supports – S table over temperature Focus Mechanism Engineering Model (FM-EM) The FM-EM shown in Figure 5 was built and tested to mitigate risk early. Key challenges then were to find motor and lead screw components and a material suitable for flexures, for operation to 5 K. A 2-phase stepper motor (Figures 8) was selected with a gearhead for further reduction to achieve the desired output step size. Motor detent torque and the overall mechanical reduction also prevent backdriving during launch. A ball screw (Figure 9) was selected to provide for a lo w friction, precision lead screw. Both these components were tested for operation near 5 K, as described in their testing sections. Titanium was selected for all the flexures and major structure. Beyond its other desirable properties, it nearly matches the coefficient of thermal expansion (CTE) of the SM and metering tower (and flexures in the SM mount and main housing legs compensate for the small mismatch). The FM-EM used 6Al-4V, due to schedule. For flight, extra low interstitial titanium 6Al-4V (ELI) was chosen since it is generally tougher. There was concern titanium may become too brittle at 5 K and not flex well over life. This was resolved by determining its plane strain fracture toughness (K IC) was sufficient, as shown in the lesson learned below and with fatigue analysis. Cryo-life testing of the FM-EM to 4X life further alleviated this concern. Lesson learned #1: Titanium is useful as a flexure material at cryogenic temperatures. 6Al-4V (ELI) predicted K IC at 4 K is 54.9 MPa* (50 ksi* ) per the NASA/FLAGRO manual, JSC-22267A. This was more than sufficient to meet the focus mechanism require ments for flexure operations below 5 K. Additionally, 6Al-4V flexures were life tested in the FM-EM at 35 K to 4X life. Other development included: For modularity of the SM mount and carrier, the carrier mounting screws are accessible from the cover side. A SM and carrier mass simulator (shown in the photograph) was made for FM-EM testing. The drive flexure passes through the lever arm before attaching to it, to reduce its deflection. Ball screw nut flexures provide compliance to prevent binding at the screw, yet are stiff in the drive axis and prevent excessive rotational windup of the nut. And non-jamming hard stops on the screw provide for range limits. Housing Carrier TubeDiaphragm FlexureCover Focus Direction Figure 4. Focus Mechanism Concept in m 405

Document

AMS_2008.pdf

SM and carrier mass simulator NutFlexure Hard Stop (1 of 2) LeverArm SM Cruciform Flexure Drive Flexure PivotFlexure LeverArmFM-EM Hardware (with cover) Drive Housing SM Flexure (not flight design) Stepper MotorGearhead Ball Screw and Nut Main Housing Figure 5. Focus Mechanism Engineering Model (FM-EM) 406

Document

AMS_2008.pdf

Flight Focus Mechanism (FLT-FM) Figure 6 shows the Flight Focus Mechanism, which is much like the FM-EM except for some notable developments. Most notable is the addition of the flight secondary mirror and its mount, shown in the next section. Also, the motor is dual wound for redundancy. The drive and pivot flexures were also significantly strengthened because the axial (focus) load requirement was changed from 70 G to 125 G and a “low risk fracture part” methodology was voluntarily imposed to enhance reliability. This methodology is defined in a JSC memo (June 1992), TA-92-013 and included designing the flexures to /g116 10X their required fatigue life (including all vibe cycles). Other developments included: The addition of the mirror mask and stray light baffle. A larger motor was provided for more torque margin. And Variable Impedance Transducer (VIT) sensors were added for “soft” limits. VIT (1 of 2) SM Mask / Stray Light Baffle SM Figure 6. Flight Focus Mechanism (FLT-FM) Secondary Mirror Mount The secondary mirror mount approach is shown in Figure 7. The mirror is a convex hyperbolic front surface on a beryllium substrate. Beryllium spacers are bonded to the substrate and titanium bi-pod flexures are then conventionally attached. This approac h helps to minimize mirror distortion and provides high load capacity. It also presented challenges due the bonds being used at cryogenic temperatures. 407

Document

AMS_2008.pdf

Titanium Bi-Pod Flexure attaches to spacer Titanium Spacer SM Carrier Secondary Mirror (SM) Beryllium Spacer bonded to SM (The small post in middle of bond area is not a load-carrying feature) Figure 7. Secondary Mirror Mount The beryllium spacers match the CTE of the mirror to minimize thermally induced strain in the bondline and to maintain mirror figure. The bondline also pr ovides some isolation from flexure fastener preloads, which were light, to further minimize mirror distortion. Titanium spacers at the carrier interface were machined in thickness and wedge to achieve ideal flexure alignment and for optimal mirror figure at 5K. High load capacity was achieved with epoxy bonds to beryllium (> three times better vs. titanium). Beyond strength, a specific epoxy was also chosen for its desirable CTE and modulus of elasticity vs. other adhesives, based on tests at room temperature to 5 K. A challenge arose when finite element analysis (FEA) predicted high stress in the bond due to the epoxy shrinkage when cycled to 5 K. The high stress occurred around the exposed periphery of the bond, essentially independent of bond area. This raised concern micro-cracking could occur around the periphery and cumulatively degrade bond strength over multiple 5 K cycles. However, we realized the analysis could be too pessimistic in predicting the adhesive stresses, due to assumed linearity and stress infinity conditions at the bond edges often inherent in such FEA [1]. And BATC had successful use of adhesives to 5 K, which further suggested the FEA did not provide for sufficient understanding of this issue. Testing was done to address this concern. Beryllium coupons were bonded and cycled ten times from room temperature to 4 K. They were then shear and leverage strength tested at room temperature to determine if any degradation resulted. There was essentially no difference in results between cycled and un-cycled samples. In fact, all the cycled sample actually showed slightly higher strengths (a pleasant surprise). Lesson learned #2 : Adhesive bonds at cryo-temperatures should be well understood and / or tested. For the secondary mirror mou nt, FEA showed high stresses that suggested the bond joint strength could cumulatively degrade by thermal cycling to 5 K. To better understand if this was an issue or not, tests were done as described above. These tests showed no degradation in bond strength after thermal cycling to 5 K and provided the understanding we needed to proceed. 408

Document

AMS_2008.pdf

Stepper Motor Component Testing A stepper motor-gearhead assembly was tested early, as a component, because it was considered a significant risk item to work at 5 K. We had one in house from a vendor that does a g ood job of matching the CTE of materials in their units and that would have a good chance of working. The following briefly summarizes the test of this motor. Test Description Measure power-off backdriving torque and operating output torques at room and cryogenic temperatures. Figure 8. Stepper MotorGearhead Motor Modifications The stepper motor-gearhead is shown in Figure 8. This unit was disassembled and all its key bearing and gear surfaces were dry lubricated with a BATC proprietary process. The unit also had non-metallic parts that were replaced with metallic ones, as supplied by the vendor in another version of this unit. It was baked out to remove moisture and dry purged for testing. Test Set-Up See Figure 10 on the next page for the set-up description (also used for ball screw component testing). Success Criteria /g120 Operate near 5 K. /g120 Show no big increase in backdriving to rque (exhibit no to low binding). /g120 Show starting (pull in) torque /g116 353 mN*m (50 oz*in). Results The unit met all the success criteria, as shown by the results in Table 2. Stall (pull out) torque and winding resistance were also measured for reference. Table 2. Stepper Motor Component Test Results Temperature of Unit Operated? Backdriving Torque (mN*m) Starting (pull in) Torque (mN*m) Stall (pull out) Torque (mN*m) Room (pre-test) Yes 212 - 233 466 - 480 508 - 530 7-14 K Yes 247 - 268 395 - 565 424 - 706 Room (post-test) Yes 212 - 247 438 - 480 480 - 530 (Tests were done at less than peak po wer, CW / CCW, and at multiple step rates) An approximately 100:1 Relative Resistivity Ratio (RRR) was measured at 5 K, the RRR being the ratio of winding resistance at room temperature vs. 5 K. This lower winding resistance at cryogenic temperatures was not an issue since the motor was current limited by its drive electronics. The flight motor was similarly tested to higher acceptance torque values, since it is larger. Lesson learned #3(a) :The stepper motor-gearhead needed only simple modification to work near 5 K. The modification was primarily dry lubric ation, using a BATC proprietary process. 409

Document

AMS_2008.pdf

Ball Screw Component Testing Ball screw testing was performed early because there was concern its running friction torque might change appreciably when rotating at 5 K. The following briefly summarizes the testing of the ball screw. Test Description Measure starting and running torque at room temperature and near 5 K. Two ball screws were tested, a smaller, preloaded, carbon steel unit, shown in Figure 9, and a larger un-preloaded 440C version. Ball Screw Modifications The ball screws were disassembled and dry lubricated with a BATC proprietary process. Some internal plastic parts were changed to metal parts, for flight but not for this testing. The units were baked out to remove moisture and dry purged for testing. Figure 9. Ball Screw Success Criteria Max torque of 35.3 mN*m (5 oz-in), but a goal of 17.6 mN*m (2.5 oz-in) is preferred. Results The test was a success, as shown in Table 3. The larger, un-preloaded scre w met the torque limit and was near to the preferred goal. The preloaded units torque was shown to be undesirable. So, a smaller un-preloaded screw was chosen for the FM-EM, which ultimately met the preferred goal. Table 3. Ball Screw Component Test Results Temperature at test Smaller, preloaded screw max torque* (mN*m) Larger, un-preloaded screw max torque (mN*m) Room (pre-test) 21.2 - 35.3 8.8 At 4.8 – 5.2 K 35.3 - 67.1 22.0 Room (post-test) 21.2 - 28.2 11.3 * 6 balls missing (2 per track), lost in lubrication process Dynamometer (shown) or torque watch LHe LevelItemunder test Dewar Tubes Test Set-Up (used for ball screw and motor) The items were tested in a dewar with liquid helium, per Figure 10. They were held with a low thermal conductance outer tube and turned with or by another tube. A torque watch was used for ball screw testing, while the dynamometer shown was used for the motor testing. Lesson learned #3(b) : The ball screw needed only minor modification to work near 5 K. The modification was primarily dry lubrication, with a BATC proprietary process. (Dewar not shown) Figure 10. Component Test Set-Up 410

Document

AMS_2008.pdf

Focus Mechanism Assembly Testing The FM-EM and FLT-FM assemblies were tested as described in this section, to evaluate how well the focus mechanism met its requirements over ambient and cryogenic temperatures, before and after vibration testing, and after life testing of the FM-EM. Challenges overcame included precisely measuring the “mirror” (simulator) motion in a cryogenic environment. Also, the flight mechanism survived a severe over-test condition, as described herein. And relevant items that contributed to successful testing are also noted. Test Description All the key requirements per Table 1 were evaluated, and are summarized as follows: /g120 Focus range, step size, and repeatability /g120 De-center and tilt over the focus range /g120 De-center, tilt, and focus shift after launch vibe /g120 Vibration testing and determination of mode frequencies /g120 Life testing to 1X life and for up to 4X life FM-EM performance testing was don e in vacuum at cryogenic temperatures and in ambient conditions, both before and after vibration testing, and after life testing. FLT-FM performance testing was done at ambient conditions, but it was later cryo-tested at the telescope assembly level. The FM-EM and FLT-FM are similar, except for the notable differences shown in Table 4. Table 4. Differences between the FM-EM and FLT-FM FM-EM FLT-FM SM mass simulator Flight SM and mount Smaller motor and single wound Larger motor and dual wound Flexures sized for 70 G in all axes Flexures sized to 125 G in focus axis Carbon steel ball screw 440C induction hardened ball screw No limit sensors Limit sensors Test Set-Up Description The FM-EM was tested in the dewar shown in Figure 11. The test measured mirror motion relative to the mechanism interface, using a mirror simulator. A brief description of the test set-up is as follows: /g120 The mechanism was mounted on a stable reference plate that was used for cryogenic or ambient testing and for vibration testing. It had position sensors on it that monitored the mirror simulator motion relative to the plate, hence, the mechanism interface. /g120 Legs (not shown) mounted the reference plate to the dewar and prov ided a conductive path to the dewar cold plate. The legs were removed so the plate could be used in vibration testing too. /g120 VITs (Variable Impedance Transducers) monitored the mirror simulator motion along the focus axis, in two axes of tilt, and in two lateral axes for decenter. The VITs were initially calibrated at room temperature and to 80 K, but not to lower temperatures. /g120 Anticipating the VIT calibration could change below 80 K, a secondary focus measurement technique was provided by using two laser interferometers on a stable base. One measured mirror simulator motion and the other measured the reference plate motion. The difference between these measurements represented the desired focus motion, relative to the plate. 411

Document

AMS_2008.pdf

/g120 Likewise, another secondary measurement technique was provided for tilt of the mirror simulator in two axes (tip / tilt) using two quad-cell autocollimators, on a stable base. One measured mirror simulator tip/tilt motion and the other measured t he reference plate tip/tilt motion. The difference between these measurements represented the desired tip / tilt motion, relative to the plate. /g120 The dewar has two stages, one for liquid nitrogen and one for liquid he lium. It was modified to fit windows for the secondary measurement techniques. /g120 Thermocouples were used to monitor temperature at various locations. Success Criteria Meet the key performance requirements, as shown in Table 1, over ambient and cryogenic temperatures, before and after vibration testing, and over 1X life for the FM-EM, and as a goal for up to 4X life. Results The FM-EM met the above success criteria. Table 1 shows the FM-EM performance before and after launch vibration and over 1X life. Performance did not change appreciably for up to 4X life, meeting the goal to provide significant life margin. Table 1 shows maximum values for step size and repeatability. The mean and standard deviation values for these over 1X life are provided in Table 5. Table 5. Mean and Standard Deviation for FM-EM Step Size and Repeatability Parameter Mean Standard Deviation Step size 0.42/g80m 0.17 /g80m Repeatability (bi-directional) 0.0005/g80m 0.29 /g80m The FLT-FM performance was similar, based on abbreviated testing. A difficulty occurred during FLT-FM testing when it was erroneously vibration tested to 170 G in the lateral direction, due to a faulty control of the vibration table. This was far beyond the required 70 G. The effects of the 170 G over-test were carefully evaluated. Fortunately, due to the robustness and conservatism in the design, there were relatively few issue s. However, the ball scre w was loaded beyond its rating. So we loaded a spare flight ball screw to the same over-test level and well beyond. Based on inspections of the overloaded spare ball screw, we concluded the flight ball screw was still acceptable for use. This 170 G over-test also reduced the fatigue life of the flexures from 10X to 6X. But this was deemed acceptable, with customer approval. Notable Items Contributing to Successful Testing : /g120 The FM-EM was tested at 30 K to 49 K since the dewar had only one, liquid nitrogen shroud and due to parasitic radiation from the optical windows. This was accepted since the motor and ball screw components were tested near 5 K. Also, most material shrinkage occurs by these temperatures and the predicted toughness (K IC) for titanium at 5 K was more than acceptable. /g120 Including secondary measurement techniques proved to be a prudent decision. As anticipated might happen, the VIT calibration changed below 80 K. Using interferometer and autocollimator data, it was found the slope of the calibration changed but the VITs retained good linearity. After accounting for this, the data between all the measure ment methods agreed well. This also allowed for interpretation of VIT data measuring the lateral motions, which did not have the secondary measurements. /g120 Vibration testing without force limiting showed high magnification factors at resonant frequencies that would have resulted in over-testing the mechanisms capability, if vibrated to the required input levels on a standard shaker. As recommended by JPL, using force limiting [2] resolved this problem and it was successfully tested to its full levels. (Force limiting is now more routinely used at BATC) 412

Document

AMS_2008.pdf

Dewar Laser Interferometers Dewar with Laser Interferometers (Quad-Cell Autocollimators not shown) FM-EM Shrouded in Dewar (Looking up at ref plate side) FM-EM on Reference Plate (Without legs) Figure 11. FM-EM Test Set-Up 413

Document

AMS_2008.pdf

Conclusion The new focus mechanism was completed and shown to provide repeatable focus positioning to 5 K. Its design and test challenges were successfully overcome, on schedule. The mechanisms concept and implementation have been sh own to be simple, effective, and very robust. It is currently in service in the Spitzer Space Telescope, one of NASA’s “Great Observatories”. After launch on August 25 th, 2003, the new focus mechanism has performed as planned in flight. The observatory has surpassed its expected 2.5-year life and is now approaching a 5-year life goal. (The mechanism only needed operation twice during this time; once to verify its successful operation, and then to achieve the final focus position that is currently in use.) Testing of the stepper motor and ball screw components significantly mitigated risk early in the program. The build and test of the mechanism engineering model further mitigated risk and allowed for verification of performance over and above the required life. Lessons learned, resulting from this successful mechanism development, are summarized below. Lessons Learned Summary 1. Titanium 6Al-4V (ELI) is useful as a flexure material to liquid helium temperature s since it has sufficient plane strain fracture toughness (K IC) to < 5 K. 2. Adhesive bonds at cryo-temperatures should be well understood and / or tested. For the mirror mount, FEA did not provide sufficient understanding of the bond. Testing with coupons was required and showed the bond strength was acceptable after multiple cycles to 5 K. 3. The geared-stepper motor and ball screw components needed only slight modification to work at 5 K, which was primarily dry lubrication. Subsequent Developments This mechanism has provided heritage for othe r programs at Ball Aerospace, as follows: /g120 The James Webb Space Telescope (JWST): Similar geared-motors are planned for use in the cryogenic nano-actuators used to position its primary mirror segments. /g120 Kepler (another space-borne telescope): A derivative of this mechanism is planned to move the primary mirror for focus adjustment. Acknowledgements This device was developed with funding and oversight from JPL under contract 960669. Thanks to Robert M. Warden (BATC) for developing the mechanism conce pt and to the BATC team and suppliers that supported this effort. And in memory of Mike Rice, an outstanding technician on the team. References 1. MIL-HDBK-17-3F, section 6.2.3.6. 2. T. D. Scharton. Force Limited Vibration Testing Monograph. NASA Reference Publication RP-1403 (May 1997). 414

Document

AMS_2008.pdf

Development of a Spacecraft Antenna Pointing Gimbal Charles Monroe* and Peter Rossoni* Abstract The development of the pointing gimbal in the high-gain antenna system (HGAS) of the Solar Dynamics Observatory spacecraft is described. The gimbal was designed for 5 years of service in Geo-Synchronous orbit. The hardware incorp orates multiple levels of redundancy, allows harnessing and waveguide along its full length across its two axes of rotation and points with an accuracy of better than 0.065°. Significant issues with actuator alignment, Electrical Contact Ring noise, pointing budget, and waveguide failures are described, along with their respective resolutions. Introduction This paper outlines requirements, design and development activities of the SDO gimbal. Several hardware anomalies and their resolution are described. The critical reliability level was a driver for most of the issues uncovered during the gimbal development. Significant design areas include the actuator and contact-ring mechanisms and waveguide. Unique events and lessons-learned include the encoder alignment to the actuators, noise during component-level testing, replacing flex waveguide and accommodating the harness. Background The Solar Dynamics Observatory (SDO), shown in Figure 1, is a NASA spacecraft that will collect data from the Sun during its 5-year life. The spacecraft was designed by and is being integrated at NASA Goddard Space Flight Center in Greenbelt, MD. Universities and industry provide its science instruments. This observatory transfers 150Mbps (millions of bits per second) of solar imagery (with overhead) per day from its 28.5° inclination, geosynchronous orbit at 36,000 km (22,400 mile) altitude, to the ground station in White Sands, New Mexico. The gimbal geometry that is most conducive to this end is a two-axis azimuth/elevation configuration. The azimuth axis will rotate once per orbit (once per day), and the elevation axis will rotate up to ±65 degrees to allow the antenna to point to the desired Earth coordinates at the SDO Ground Station. To avoid exce ssive spacecraft roll maneuver s, a dual HGAS approach was taken, with antenna systems on opposite sides of the spacecraft, allowing selection of the optimum gimbal for downlink via scheduled hand-offs. * NASA Goddard Space Flight Center, Greenbelt, MD Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Space Flight Center, May 7-9, 2008 415

Document

AMS_2008.pdf

Figure 1. Solar Dynamics Observatory with One of Two High-Gain Antennas and Gimbals Circled Driving Requirements An important driver for the gimbal system is the downlink requirement. A rate of 150 MB/second is needed with 99.99 percent reliability over a 99 percent duty cycle. The 99.99 percent reliability requirement reflects the transmission error rate. To achieve this percentage during periods of transmission, the azimuth axis must be able to rotate continuously without downtime for “rewinding” of the harness about the axis of rotation. This drives the need for an electrical contact ring assembly (ECRA), a slip ring or roll ring, to pass power and signal through the axis. The 99 percent duty cycle addresses periods of fog and rain at the ground station. The spacecraft itself will occasionally occlude the view from a single antenna. A continuous downlink capability dictates two antennas—one on either side of the spacecraft. During portions of the year, a daily hand-off between antennas will be required. These requirements lead to a highly reliable, 100% duty-cycle design, with no planned datalink interruption. The characteristics of the antennas and the power av ailable for transmission drive the need to keep RF throughput loss low—the gimbal itself was allocated a loss of less than 1.45 dB. To meet this requirement, an all-waveguide RF system was selected as opposed to the simpler coaxial cable approach. This necessitated waveguide rotary joints at each axis of rotation. Also, minimizing loss drives the need for having a pointing capability of + 0.30° to the ground station for all error sources, including spacecraft position and orientation. Of this amount, there is 0.14° budgeted for random and calibration errors of the gimbal. 416

Document

AMS_2008.pdf

Gimbal Design The overall configuration is shown in Figure 2. The azimuth axis has unlimited rotation and the elevation axis has a + 69° range. The continuous azimuth rotation is made possible by having the power and signal transferred through the ECRA. For the elevation rotation, a rotary cable wrap wherein the cable is carefully spiraled through the center of the elevation actuator manages the harness. There are two rotating sections of waveguide on this two-axis gimbal. An azimuth section rotates with the azimuth axis and extends from the azimuth actuator up to the elevation axis. An elevation section rotates with the elevation axis and extends from the elevation actuator up to the antenna. Elevation Axis High Gain Antenna Azimuth RF Slip Joint Elevation RF Rotary Joint Electrical Contact Ring Assembly (ECRA) Azimuth Axis Azimuth Waveguide Azimuth Rotary Actuator Azimuth RF Rotary Joint Figure 2. Gimbal Cross-section 417

Document

AMS_2008.pdf

Rotary Actuators The rotary actuators for azimuth and elevation are identical except for hard stops on the elevation actuators limiting travel to + 69 degrees. Each commercial actuator consists of a stepper motor, hybrid optical encoder, and harmonic drive gear reducer in a titanium housing. Once delivered, the units were tested for compatibility with the GSFC-designed control ele ctronics, and characterized for settling time, torque margin, encoder output and alignment. Then thermal hardware was applied, as shown in Figure 3, and the harnessing prepared for integra tion to flight hardware. Figure 3. Rotary Actuator Shown During Thermal Hardware Application Actuator Description The three-phase stepper motors are redundantly wound. There are two redundant encoders: a coarse encoder on the output that determines hemisphere and home, and a fine encoder on the input that counts each step taken. The actuator details are summarized below: Actuator Parameter Value Output step 0.0075 degree Harmonic drive gear ratio 200:1 Motor step 1.5 degrees Unpowered detent torque 34 N-m (300 inch-pounds) Max required slew speed (under the following conditions at qualification temperatures) 30 degrees/min (66.7 pulses/sec) Driven inertia 2 kg-m2 Driven offset load 28 N-m (250 inch-pound) Driven friction load 2.5 N-m (22 inch-pound) Step Settling of Bearings On orbit, a motor step will be taken roughly every two seconds. For the actuator life test, time constraints drive the need for more frequent steps, but the period between steps should be no less than the time required for bearing balls to settle. It was decided that after the ball motion decreased to a point where the magnitude of the oscillations is less than the width of the Hertzian contact patch, the bearings would be considered essentially settled. We believe this settling criterion to be consistent with ball-pass analysis for lubricant tribo-degradation. The time required to reach this point is 35 msec, so the actuator life test could be run at ~28 pulses per sec, which is an acceleration factor of ~57. Actuator Encoder Alignment Actuator position is determined by internal optical encoders. Alignment of the encoder is inferred from its output. Even though settling time and torques were within requirements, so me actuators had marginal 418

Document

AMS_2008.pdf

alignment of their encoder discs to the step detents. The easiest way to test for alignment is by the quality of the encoder output during settling in its detent. A well-aligned encoder disc would have its LightEmitting Diode (LED) centered in the disc aperture. Even at the beginning of settle, when the oscillations are highest, little or no light would be occulted by either edge, as shown in Figure 4a (each window is two steps wide). Encoders that are not optimally aligned would shadow a portion of the light while settling in a step detent. Of 10 potential flight actuators, 6 were aligned with less than optimal performance, as shown in Figure 4b by the encoder light output “hash” during characterization testing. The actuator specification called out static alignment only; as a result, all 10 flight actuators satisfied the specification requirements. Just prior to delivery to the spacecraft, two of the flight actuators were damaged beyond repair by excessive heat in a Goddard thermal vacuum chamber. Two of the less-than-optimal actuators were brought to flight status. These passed the static alignment specification but during a high-rate slew operation could incur positioning errors of one step. This error is reset in the control electronics when the actuator passes through the “home” position, so the condition is tolerable during slews. Under normal tracking, there is adequate time for the encoder output to settle and the output to be verified. The only remaining issue is diode output over the mission life. As the diode response decays due to radiation effects and the normal degradation due to operation, the partial occulting could reduce margin on the encoder output. Figure 4a. Good Encoder Alignment – Two-Step Window Figure 4b. Less than Optimal Encoder Alignment 419

Document

AMS_2008.pdf