Building Miniature Spacecraft at The Aerospace Corporation

David Hinkley and Siegfried Janson

Imagine flying a satellite with a technology freeze date that was only six months ago. Aerospace has completed 11 miniature satellites for technology demonstrations, with an average design, build, assemble, test, and deliver cycle of about a year.

During the last decade, researchers at The Aerospace Corporation have pioneered the development of nano satellites (1–10 kilograms) and picosatellites (0.1–1 kilograms). These ultra-small spacecraft can be fabricated quickly and launched into space for less than $100,000 as secondary payloads. They are ideal platforms for flight-testing micro- and nanotechnologies, new materials and sensors, and advanced spacecraft software. With expanded small satellite launch opportunities, the design, build, test, flight test, and redesign cycle can be shortened to six months, thus enabling an order-of-magnitude increase in the evolution of new spacecraft technology. This approach enables rapid technology development while providing practical, hands-on training for the students, engineers, space system development managers, and research scientists who are taking space systems into the 21st century.

The 250-gram PicoSats.

PicoSats

The Aerospace program in miniature satellites started in 1999, when a corporate research initiative on microtechnology led to a DARPA (Defense Advanced Research Projects Agency) grant to fly microelectromechanical systems (MEMS) in space. Within six months, Aerospace delivered a pair of 250-gram satellites measuring just 1 by 3 by 4 inches in dimension. Rockwell Science Center in Thousand Oaks, California, provided battlefield radio-node electronics (also funded by DARPA) consisting of a microprocessor and radio. Aerospace packaged, tested, and reprogrammed them as the command and control unit for these ultra-small satellites. This effort brought together corporate engineers from the thermal, mechanical, communications, and software disciplines. These so-called "PicoSats" had basic functionality and minimal mission requirements: survive launch, listen for a ground station, deploy a 100-foot tether (to simulate constellation flight), exercise MEMS radiofrequency (RF) switches, and transmit experiment and housekeeping data. They were released from the 23 kilogram OPAL (Orbiting Picosatellite Automated Launcher) satellite in February 2000.

The two largest concerns for mission success were satellite tracking and the communications link budget. Each satellite could only generate 64 milliwatts of RF power, and the circular orbit was 773 kilometers in altitude. To enhance tracking of the satellites by the Air Force Space Surveillance Network, the tether between them incorporated gold dipole threads to increase the radar cross section. To close the communications link, a huge 150-foot-diameter antenna at SRI International in Menlo Park, California, was required—but its very narrow beamwidth added to the tracking challenge. After ejecting from OPAL, the tethered PicoSats operated for about two and a half days. Tracking was good; data from the MEMS switch experiment and measurements of satellite temperature were downlinked, and new commands were uplinked. It was a successful mission, but with plenty of lessons learned.

The 150-foot-diameter ground station antenna needed to close the communications link to space.

An opportunity to eject another pair of tethered PicoSats presented itself several months after the OPAL delivery. A payload on the Air Force Research Laboratory's MightySat II.1 research satellite was demanifested late in the integration cycle, and a single OPAL-like launch tube would fit in the available volume. Aerospace built a copy of the OPAL launch tube and another pair of 1 by 3 by 4 inch PicoSats in six months, delivering in December 1999. The two 250-gram picosatellites would be tethered as before and would repeat the previous constellation exercise. They also would test an improved set of MEMS RF switches for DARPA, demonstrate the feasibility of storing miniature satellites onboard a host vehicle for a long period of time, and provide an exercise in integrating secondary satellites onto a high-value host to illuminate the practical mission assurance concerns.

In September 2001, the MightySat II.1 satellite released the tethered PicoSats after 15 months of storage on orbit. The SRI International ground station contacted both units and downlinked temperature and MEMS-switch resistance data; however, the communications link was worse than in the prior mission, and few contacts were made before the mission was declared over three days later. Nonetheless, the mission demonstrated the feasibility of including a daughtership on a mothership. This paved the way for the second generation of miniature satellites from Aerospace, the MEMS-Enabled PicoSatellite Inspector (MEPSI) series.

MEPSI

The MEPSI spacecraft was designed to test miniature system and subsystem technologies required for kilogram-class satellite inspectors and assistants. The concept included one or more satellite inspectors that would reside on a large host satellite—with minimal impact on mass and volume—to be ejected on command in the event of an on-orbit anomaly. They would photograph the host to provide high-resolution imagery of damaged or undeployed structures, or provide real-time imaging of complicated deployments and proximity operations directly to a ground station.

The MEPSI-class satellites had to be very capable spacecraft to carry out the inspection mission. Several functions—propulsion, closed-loop attitude control, ranging, and a radio downlink with sufficient bandwidth to transmit images to a ground station—were hitherto unheard of in a satellite of this size. Furthermore, the MEPSI had to be extremely reliable because it would orbit a high-value satellite, and a collision was unacceptable. For these reasons, a spiral development plan was chosen, whereby each successive MEPSI would be an improvement on the prior version.

For the first MEPSI mission, two 800-gram picosatellites and a space shuttle-qualified launcher were designed, built, tested, and delivered in two years. The loaded launcher was installed onto the sidewall of the cargo bay of the space shuttle Endeavour one month prior to liftoff of STS-113 in December 2002 (so close to the liftoff date that the orbiter was already in a vertical orientation). The two identical 4 by 4 by 5 inch picosatellites were powered with primary batteries and had a flight computer, radio, triaxial rate sensors, and triaxial accelerometers. They were tied together with a 50-foot tether with gold dipole wires woven along its length to increase the radar cross-section of the pair. The satellites were identical and they had no redundant subsystems, so having two of the same design provided the only redundancy against random defects and workmanship errors. On orbit, once released from the launcher, they turned on and started recording the accelerations and angular rates caused by the unspooling of the tether and the eventual rebounds caused by the end of the tether. The purpose of this exercise was to compare the performance of two different types of MEMS rate sensors installed in each satellite. The Jet Propulsion Laboratory (JPL) was Aerospace's partner on this mission. JPL integrated and characterized the MEMS inertial rate measurement unit on each satellite.

Photograph of the space shuttle Discovery taken seconds after ejection by a MEPSI spacecraft.

This mission was only a partial success. A new picosatellite launch system for the space shuttle and two new satellites had been designed, qualified, and delivered. The NASA photographs of the picosatellite ejection were exciting and dramatic, and they helped to promote this class of satellite. The Aerospace Corporation ground station team, with assistance from the USAF Space Surveillance Network, successfully tracked the satellites, and beacons were received that contained modest state-of-health data. Unfortunately, two-way communications were never established, and the acceleration and rotation rate mission data were not downloaded from either satellite because of a systematic problem with the satellite radio receivers. The error was known prior to delivery, but there was insufficient time to fix it (Rule 1: Primaries do not wait for secondaries!). The failure of this particular mission objective resulted in a mission assurance study in which the architecture of the satellite bus was found to have limited the designers' ability to react quickly and fix the problem in time. The picosatellite team redesigned the satellite bus to be more flexible and easy to debug.

The second MEPSI attempt had to wait for STS-116 in December 2006, mainly because of the orbiter disaster in 2003. It used the new bus architecture and new subsystems. Once again, a tethered pair of picosatellites was ejected from the space shuttle; this time, however, the dipole-laden tether was only 15 feet long to keep the two satellites in visual distance. The goal was to practice a visual inspection mission: one unit was configured as an "inspector," and the other was the "target." The two satellites were functionally identical except that the inspector had maneuverability. It used a five-thruster propulsion unit (invented and patented at The Aerospace Corporation) and three orthogonal reaction wheels to maneuver so that two tiny color 640 by 480 pixel resolution (VGA) cameras could take pictures of the target. The target had no attitude control, but had a suite of five color VGA cameras on different faces to take pictures of the inspector. Neither satellite had sensors for detecting the other—those subsystems were not ready in time. Therefore, all camera operations and attitude control changes had to be commanded while they were in contact with the ground station. Both satellites, as in all prior missions, could be commanded from the ground station independently.

The 1.4-kilogram inspector and 1.1-kilogram target were ejected from the space shuttle Discovery on the STS-116 mission in December 2006. Immediately upon release, they began preprogrammed operations that included taking photographs of the shuttle and recording satellite rotation rates and accelerations during tether deployment and subsequent rebounds. Tracking and ground operations were nominal, and the communications link was strong. The reaction wheels and cold-gas thrusters were successfully tested, but no pictures of the other picosatellite were successfully taken because, in sunlight, the tether between them was so bright, it overexposed the images. Nominal mission life was two weeks because the satellites used primary batteries, but the mission was actually shorter because of a memory overflow condition. (Both satellites suffered the same fate, which suggested a systematic error.) Nonetheless, the STS-116 picosatellites were a large step forward in demonstrating the capabilities required by a MEPSI vehicle.

CubeSats

At the same time that Aerospace was building the first MEPSI pair for STS-113 and the space shuttle picosatellite launcher, Stanford University was teaming with California Polytechnic State University (Cal Poly) to define a new standard of picosatellite called a CubeSat along with a deployment system that was compatible with expendable launch vehicles. In 2000, they jointly introduced the new specification: a cube-shaped satellite that was 10 centimeters on a side and weighed at most 1 kilogram. The interface control document was simply a single 11 by 17 inch drawing that defined the mechanical attributes of the standard. These terse requirements—along with an advertised integration and launch cost of $40,000 per CubeSat—resulted in a frenzy of development at universities around the world. In 2003, the first CubeSat launch placed six of these miniature satellites into a sun-synchronous orbit from a Russian launch vehicle. Other CubeSat cluster launches occurred in 2005 (3 satellites), 2006 (15 satellites), 2007 (7 satellites), and 2008 (6 satellites). All of these used a foreign launch vehicle except for one CubeSat in 2006 (NASA's GeneSat).

A picture of the Cal Poly CubeSat CP-4 taken by AeroCube-2—the first and, so far, only instance of one CubeSat photographing another.

The Aerospace Corporation picosatellite group began participating in this CubeSat community in 2004. At that time, the return-to-flight status of the space shuttle was still unknown, and it was important to routinely fly satellites to keep program office customers interested and to keep the picosatellite development team engaged. The team reserved a spot on the next Russian Dnepr flight and set to the task of developing AeroCube-1, its first CubeSat. The goal was to test MEPSI hardware and buy down risk; AeroCube-1 was a repacking of the MEPSI electronics. The two form factors were not too different: a CubeSat at 10 by 10 by 10 centimeters had a little less volume than a 4 by 4 by 5 inch MEPSI picosatellite. AeroCube-1 therefore featured the improvements and new capabilities of the MEPSI picosatellites on STS-116 except for propulsion and reaction wheels. Unfortunately, the AeroCube-1 satellite waited at the integrator for 15 months until the primary satellite was ready (Rule 2: Primaries can hold up secondaries!). To add insult to injury, the Dnepr vehicle failed and crashed back into Earth. However, the exercise of building the CubeSat proved beneficial when the time came to build the MEPSI flight articles a year later. The process had revealed important assembly issues, fostered the development of proper test procedures, and kept the team both together and in practice.

AeroCube-2 was an improvement over AeroCube-1 with better packaging and new capabilities. The new subsystems included the first rechargeable power system and a deorbit balloon. Four solar cells, one on each of four faces of the CubeSat would recharge the satellite's lithium ion batteries. The deorbit balloon was a 9 by 6 inch pillow-shaped Kapton bag that was inflated by a gas stored onboard in a system derived from the MEPSI mission thrusters. AeroCube-2 reached orbit in April 2007, but the rechargeable power system was not up to the task and it operated for less than a day. During that time, however, ground stations downloaded pictures and state-of-health data. Included was a picture of the Cal Poly CubeSat CP-4, the first and only picture taken of one CubeSat by another in space.

In 2008, Aerospace developed its third and most sophisticated CubeSat, AeroCube-3. It featured new technology including a 200-foot long tether, a tether cutter, a tether reel, a 30-inch nearly spherical deorbit balloon, a sun sensor, an Earth sensor and two new customer proprietary sensors. It also had a new rechargeable power system very conservatively sized. It was launched on a Minotaur launch vehicle with TacSat-3 as the primary payload in May 2009. The tether was intended to keep AeroCube-3 within camera distance to the upper stage. In the first part of the mission, it would take pictures of the upper stage in a MEPSI-like fashion. The tether reel would close the distance as needed and the tether cutter would free the researchers to perform the second part of the mission. In the second phase, a permanent magnet passively orients the free-flying spacecraft, creating North and South faces. A single miniature reaction wheel spins the spacecraft on an axis normal to the North and South faces. Two proprietary sensors and a color VGA camera sweep the surface of Earth at a rate determined by the reaction wheel, gathering data and snapping pictures. AeroCube-3 continues to be operational and 28 MB of data have been downlinked (1000 pictures and satellite health telemetry).

PSSC Testbed picture of the California coast, roughly from San Diego to Malibu.

PSSC Tested

The first Aerospace nanosatellite, the PicoSatellite Solar Cell (PSSC) Testbed, was launched in November 2008 from the space shuttle. Measuring 5 by 5 by 10 inches in dimension, the satellite's primary mission was to test two new types of solar cells in the harsh space environment. It was designed to serve as a pathfinder for a second satellite that will fly in geosynchronous transfer orbit to obtain accelerated space environment degradation data for advanced solar cells. The resulting data will provide insight into the actual performance of new solar cells before they are used to power a multimillion dollar national security spacecraft. In the past, space missions have been adversely affected by the degradation of solar cells, and attempts to collect actual exposure data for new cells have been delayed by several years due to the time required to build and launch conventional experiments. The PSSC Testbed solves that problem.

The PSSC Testbed bus includes a solar power system that can characterize new solar cells. Once it has been successfully demonstrated in space, it can be used as a standard testbed for any type of future solar cells with minimal modification. Ultimately, with a picosatellite launch capability on multiple EELV missions, a PSSC Testbed could be launched on demand, thus further reducing the time between initial production of new solar cell technology and the receipt of orbital performance data.

In addition to performing its primary mission, the pathfinder PSSC Testbed has been photographing Earth for more than 90 days. Operators have already downlinked more than 500 images and 18 megabytes of data.

Rapid Development

As these projects illustrate, speed and cost are two of the primary advantages of using small satellites for technology development. It typically takes about five STE (staff years of technical effort) to design and build an Aerospace picosatellite. In addition, purchased materials and parts reach about $100,000 when developing a new design. Each copy, however, is much less—about $10,000. Launch costs have ranged from $0 for shuttle flights sponsored by the Space Test Program to $40,000–$70,000 for an AeroCube through the CubeSat launch provider.

"Mass production" of PicoSat bodies.

"Mass production" of battery brackets.

A complex CubeSat such as AeroCube-3 has seven circuit boards. Ideally, each board requires three days to assemble, followed by two days for integration (i.e., harnessing), loading software, and testing. In practice, researchers have fabricated and flight-tested a picosatellite with minimal upgrades from previous designs in three months. The addition of new sensors and subsystems, however, can add significant nonrecurring development and testing time. The subsystems that require a long development time such as GPS and the advanced radio proceed in the background and are integrated on future flights as they become available.

Picosatellites have small, custom components that can be designed for rapid assembly and even mass production. The original DARPA sponsored 1 by 3 by 4 inch picosatellites were so small that a single CNC machine setup produced multiple copies of the satellite body and battery brackets. Furthermore, the same miniature satellite was packaged so that it could be snapped together using only a few fasteners. The more capable MEPSI, AeroCube, and PSSC Testbed satellites had more harnesses and took more time to assemble (Rule 3: Harnessing is the largest integration factor). Unlike satellites assembled from parts designed elsewhere (or worse, designed to the most versatile and therefore inefficient common interface), Aerospace picosatellites were completely designed and built in-house to optimize packaging efficiency. Commercial components were used exclusively because of their higher performance and because the cost and schedule impact of radiation-hard parts is not acceptable.

The 250-gram PicoSat just snaps together.

Flight software presents a significant cost and schedule risk for any space mission. In designing small satellites, Aerospace researchers opted to use several distributed processors, rather than one central processor. This approach breaks up the satellite programming into a number of parallel efforts. Each satellite function has a dedicated processor that is preprogrammed. Because each function is small and well defined, the program is easy to architect and does not interact with other "task" programs except through a common serial interface shared by the multiple microcontrollers. Each microcontroller (i.e., function) is contained within a 2.2 by 2.2 inch circuit board. When a satellite needs new functions, circuit boards that perform those functions are added to the stack by means of an expandable common backplane. Conversely, if development of a circuit board falls far enough behind to miss the flight, the stack becomes one board shorter, and most of the other boards are unaffected. The added capability of on-orbit programming will further mitigate the risk to schedule, or equivalently, speed up delivery.

The single greatest impediment to rapid access to space is the availability of flights. For a lightweight secondary payload, launch costs are reasonable. However, the satellites are subject to the schedule of the primary payload. This often means that the spacecraft developers deliver the hardware on time and then wait for the primary payload to be ready after a series of unanticipated delays. Often, the vehicle provider, the primary payload provider, and the customer of the primary satellite agree to unrealistic schedules, and when they fail to meet them, the schedules for the other stakeholders are pushed back as well. Meanwhile, the secondary payload sits in storage for months, ready to go. It often takes longer to go from delivery to liftoff than it takes to develop and build the next generation of miniature satellites.

Conclusion

Aerospace has been working to develop satellites with all the capabilities one would expect from a larger spacecraft. Systems that have flown so far include reliable communications, power generation, and command and data handling. Specific components have included sensors for measuring satellite rotation rates, accelerations, geomagnetic fields, and thermal infrared radiation from Earth as well as a 640 by 480 pixel visible/near-infrared camera. The 2008 PSSC Testbed included magnetic torque coils, a solar-cell power degradation experiment, a coarse sun sensor, an Earth sensor, and two customer payloads. The 2006 MEPSI flight had three orthogonal reaction wheels and a five-thruster cold-gas propulsion unit for attitude control. Additional spacecraft hardware in the near future will include an optimized link margin radio, a megapixel imaging camera, orbit-changing propulsion, and a space-qualified GPS receiver board; anticipated software will include closed-loop attitude control.

The end-to-end process of designing, building, and flying miniature satellites has provided numerous benefits to the staff at Aerospace. Researchers have come to understand the intricacies of designing ultra-small spacecraft systems, learned how contractors create space systems for government customers, and relearned the importance of qualification testing and mission assurance. In these projects, the corporation's usual role of contractor oversight was turned around by 180 degrees. Such an exercise powerfully illustrates the reasons for, and the psychological responses to, the standard space systems development process.

picosats and nanosats built at Aerospace

Since 1998, Aerospace has built 11 picosatellites and nanosatellites. Eight have been tethered pairs, and three were individual CubeSats. One overriding goal of these efforts has been to demonstrate that miniature satellites, launched as secondary payloads, can do a great deal to mitigate risk on much larger programs.

These miniature satellite efforts have trained more than 60 scientists and engineers and provided inexpensive, rapid flight tests of commercial and mission-specific hardware and software. Aerospace will continue these efforts for cost-effective development of ultra-small spacecraft and spacecraft systems to enable new mission applications and responsive space system architectures.

Acknowledgments

The Air Force Space and Missile Systems Center (SMC/XR) and the Missile Defense Agency (MDA/STSS) have funded picosatellite development at Aerospace since 2004. The U.S. Air Force Space Test Program has integrated the Aerospace picosatellites and a nanosatellite on four launch vehicles. The authors thank these groups for their steady support, which is necessary to invent, improve, and succeed.