M3 / M-Cubed
M3 / M-Cubed (Michigan Multipurpose Minisat)
M3 (or M-Cubed, also written as MCubed) is a CubeSat student initiative of the University of Michigan (UM), Ann Arbor, Michigan. The functional objective of M-Cubed is to demonstrate the highest resolution color imagery to date (< 200 m) of Earth's surface with at least 60% land mass and a maximum of 20% cloud coverage from a single CubeSat platform. Michigan's S3FL (Student Space Systems Fabrication Lab) is also developing the M-Cubed bus with the intention of making it a heritage design, thus allowing for future missions to be flown on the same bus.
The project was started in the summer of 2007. The primary educational objective is to provide students with a design-build-test-fly learning environment for training the future space systems workforce. 1) 2) 3) 4) 5) 6)
Figure 1: Photo of the M-Cubed CubeSat flight model (image credit: UM)
The satellite bus is composed of in-house, student-developed spacecraft subsystems built from COTS (Commercial Off-The-Shelf) hardware. The spacecraft corresponds to the 1U CubeSat form factor standard in size (10 cm x 10 cm x 10 cm) and mass (≤ 1kg). The main structure is composed of six rectangular isogrid panels attached to four rails at each corner. The isogrid panels provide rigidity while being lower in mass than a solid panel. The rails to which these panels are attached will be hollowed out from the bottom face to reduce mass as well as providing a channel through which the power and electrical subsystem can access the spring-loaded plunger necessary to indicate the release of M-Cubed from the P-POD (Poly Picosatellite Orbital Deployer).
Figure 2: Basic structure of the M-Cubed spacecraft (image credit: UM)
ACS (Attitude Control Subsystem): M-Cubed utilizes a passive magnetic attitude control subsystem to achieve a proper orientation for Earth-imaging. The assembly consists of a single permanent magnet aligned on one CubeSat body axis, along with additional magnetic hysteresis materials aligned on each additional perpendicular body axis. In this configuration, the permanent magnet aligns one body axis of the CubeSat with the local Earth magnetic field direction. Since the magnet still permits CubeSat rotation about this single axis, the hysteresis materials are added to dampen unwanted rotation. Chosen for their high magnetic permeability, the HuMy80 hysteresis materials create internal current as they are rotated through the local magnetic field. This dissipates rotational energy as heat, effectively damping the rotational motion of the CubeSat.
In practice, this passive attitude control system will allow for Earth-imaging throughout only a designated portion of the M-Cubed orbit. Ideally, the camera will continuously point in the nadir direction or straight down towards Earth. Since the camera is aligned along the permanent magnet axis, however, its direction relative to nadir is dictated by the CubeSat’s orbital position. Figure 3 illustrates the shape of the Earth magnetic field lines, as well as the camera orientation at each point in a typical polar orbit.
Although limited in performance, this type of passive control system was chosen for several reasons. When compared with active attitude control systems, such as magnetic torque coils, passive systems of this type require less mass and no power consumption. Furthermore, passive attitude systems offer a robust, simple control strategy that boasts extensive flight heritage in similar Earth-imaging CubeSat missions.
Figure 3: Schematic of the Earth magnetic field lines and camera orientation throughout polar orbit (image credit: UM)
EPS (Electric Power Subsystem): M-Cubed utilizes solar arrays for power generation which are surface mounted on the outside of the CubeSat. The power collected by the solar cells is sent through current and voltage sensors connected to the microcontroller and then through a 5 V voltage converter, where is it distributed between active buses controlled by switches (during discharge mode) or to the battery charger (during charge mode). When M-Cubed is in discharge mode, the solar cell power is supplemented by the battery power through either a 3.3 V or 12 V voltage converter to power other buses. The satellite is expected to require 1.2 W of average power and 4.7 W of peak power.
Figure 4: Block diagram of the EPS (image credit: UM)
C&DH (Command & Data Handling) subsystem: The C&DH is in charge of all onboard monitoring and control functions. This includes also the control of the payload camera operations and the compression of the imagery (Figure 5).
The Stamp9G20 microcontroller from Taskit is used as OBC. This is a computer on module device, meaning it contains all the vital parts of a computer on a single circuit board. The Stamp9G20 has 400 MHz ARM9 core, 54 MB of SDRAM, 128 MB of NAND Flash, and an extensive set of connectivity ports. Running a real-time Linux operating system, the Stamp9G20 is extensible for future missions. Initial tests show successful operation even in thermal vacuum conditions.
The C&DH subsystem is providing the command interface for the payload interface module, the JPL payload (MSPI OBP algorithm), and the telemetry subsystem. The microcontroller is powered on at all times and is monitored by an external watchdog timer. The system operates at 8 MHz on a 3.3 V bus and can communicate over SPI (Serial Peripheral Interface), I2C (Inter-Integrated Communication), and USART (Universal Synchronous/Asynchronous Receiver Transmitter) protocols. USART is used to communicate with and control the radio transceivers. An I2C bus is implemented to communicate with the payload PXA270 and the EPS health data.
The C&DH subsystem is required to control the camera, and to sample pictures to reject images of space, clouds, or open ocean. Once an acceptable picture is selected, the C&DH subsystem will compress the image for transmission. Additionally, the C&DH subsystem needs to collect system voltage and current health status to transmit to ground.
Figure 5: Block diagram of the C&DHS (image credit: UM)
RF communications: The communication subsystem’s main objective is to transmit the data from onboard M-Cubed to the ground station. Using a 144 MHz (VHF) uplink and a 430 MHz (UHF) downlink, amateur radio bands will be used to control and receive data from the satellite. A basic beacon signal containing satellite health data will be transmitted intermittently throughout operations. The Li-1 (Lithium-1) radio CubeSat Kit of Astronautical Development LLC, Sunnyvale, CA is used for UHF/VHF communications.
A dedicated receiver will operate at all times, while the dedicated transmitter will be operated only to send a beacon signal or transmit picture data. Both receiver and transmitter are the same component, Analog Devices ADF7020-1, hardwired to their independent tasks to save development time and costs. From the transmitter, the signal will be amplified to 1 W, the calculated necessary transmit power. A 16 cm monopole and a 40 cm monopole are the respective antennas for downlink and uplink. Data and commands will be transmitted using the AX.25 protocol, a standard in amateur radio data transmission.
Figure 6: Photo of the AD7020-1 radio (image credit: UM)
Figure 7: M-Cubed computer aided drawing with JPL FPGA payload (image credit: JPL)
Launch: M-Cubed was launched as a secondary payload to NASA's NPP (NPOESS Preparatory Project) spacecraft on October 28, 2011. The launch site was VAFB, CA. The launch vehicle was the Delta-2-7920-10 of Boeing, and the launch provider was ULA (United Launch Alliance).
Secondary payloads: The secondary payloads were part of NASA's ElaNa-3 (Educational Launch of Nanosatellites) initiative. All secondary payloads were deployed from standard P-PODs (Poly Picosatellite Orbital Deployer). 7)
• DICE (Dynamic Ionosphere CubeSat Experiment), two nanosatellites of the DICE consortium (Utah State University, Logan, UT) with a total mass of 4 kg.
• E1P-2 (Explorer-1 PRIME-2), a CubeSat mission of MSU (Montana State University), Bozeman, MT, USA.
• AubieSat-1, a 1U CubeSat of AUSSP (Auburn University Student Space Program). Auburn University is located in Auburn, ALA, USA.
• RAX-2 (Radio Aurora eXplorer-2), an NSF-sponsored 3U CubeSat of the University of Michigan, Ann Arbor, MI.
• M-Cubed (Michigan Multipurpose Minisat), a 1U CubeSat of the University of Michigan, Ann Arbor, MI.
Orbit of the secondary payloads: After the deployment of the NPP primary mission, the launch vehicle transfers all secondary payloads into an elliptical orbit for subsequent deployment. This is to meet the CubeSat standard of a 25 year de-orbit lifetime as well as the science requirements of the payloads riding on this rocket. The rocket will take care of the maneuvering and when it reaches the correct orbit, it will deploy all of the secondary payloads, into an orbit of ~ 830 km x ~ 350 km, inclination of ~ 99º.
Orbit: Sun-synchronous near-circular orbit of the primary mission, altitude = 824 km, inclination = 98.7º, period = 101 minutes, LTDN (Local Time on Descending Node) at 10:30 hours.
M-Cubed antenna deployment: M-Cubed requires a deployment mechanism for both antennas needed for communication with the ground station. The 130 MHz receiving antenna is a 0.33 m long dipole while the 435 MHz transmitting antenna is a 0.5 m long monopole. Both antennas are made of copper tape shaped to maintain a straight profile. The monopole antenna is fastened at one end while the dipole is fastened at its midpoint. Both antennas will be wrapped around the structure perpendicular to each other. The remaining ends are then tied down to the inside of the structure using nylon string with nichrome (NiCr) wire wrapped around it. When the antennas need to be deployed, a current will run through the nichrome wire, melting the nylon string and allowing for the antennas to release to their straight unfurled position.
Figure 8: Photo of AubieSat-1, E1P-2, and M-Cubed (from left) along with the P-POD prior to launch at VAFB (image credit: MSU/SSEL)
Status of the mission:
• March 2013: Since launch, there is still no separation between the two CubeSats MCubed and HRBE [Hiscock Radiation Belt Explorer, formerly known as E1P-2 (Explorer-1 PRIME-2) of MSU] after on-orbit deployment. — The MCubed project team of the University of Michigan completed an extensive investigation of the on-orbit conjunction between the MCubed and HRBE CubeSats. The simulations have shown that magnetic conjunction between MCubed and HRBE is possible. 8)
Note: The conjunction is mainly a problem for MCubed. MCubed is still alive and transmitting, but the project team hasn't been able to command the spacecraft. The hypothesis is, that this is due to a de-tuned antenna given the close proximity of HRBE — perhaps HRBE is stuck to the antenna. The team's attempts to command the spacecraft have mostly been through the UHF radio. The project is planning to do more testing to get through via VHF. The HRBE team of MSU has been able to perform their own mission and they don't have any sensor data for further insight into the nature of the conjunction. 9)
Table 1: This table represents a condensed abstract of the investigations performed by the MCubed project team (Ref. 8)
• In April 2012, the situation with M-Cubed remained unchanged - in spite of numerous attempts to contact the spacecraft, with lots of outside help, and elaborate recovery analysis. 10)
The information available shows some unusual behavior of the spacecraft:
1) M-Cubed had some unusual and concerning telemetry
- Two solar panels developing potential, but generating little to no current
- Received Signal Strength Indicator (RSSI) off the scale (> -30 dBm)
- Reset count showed spacecraft was resetting frequently
2) Every fourth beacon came in much stronger than the proceeding three, and was overlapping with E1P beacon transmissions
3) Joint Space Operations Center did NOT observe any other objects related to the NPP launch since first acquisition of the other secondary spacecraft.
How could this happen ??
- Several possibilities were investigated (ex: antenna entanglement)
- Strongest evidence currently available suggests magnetic conjunction
- E1P & M-Cubed both used relatively strong magnets compared to other 1Us for passive attitude control
- The magnets used by both satellites were NOT facing toward each other in the P-POD – conjunction had to occur AFTER deployment from the P-POD
- Inoperable solar panels on M-Cubed correspond to magnet axis (telemetry).
Recovery operations at SRI:
- Unable to command M-Cubed from Michigan ground station due to high noise floor created by M-Cubed electronics on UHF band
- Calculations showed that the project needed a much greater EIRP than available at UMich for uplink
- Fortunately, the project was granted access to SRI’s 18 meter dish
- However, all uplink attempts made on nearly every pass over a 3-day period were without success despite having sufficient margin over the noise floor.
The RAX-2 nanosatellite continues to remain power positive even with two inoperative solar panels.
In the meantime, NASA has approved funds to build a second M-Cubed! (Ref. 10)
• Dec. 6, 2011. The project is suspecting that perhaps M-Cubed is still stuck to the E1P-2 (Explorer-1 Prime-2) CubeSat of MSU. After studying the Doppler shifts of both satellites, the project noticed that a consistent correspondence between the two satellites is far more than a coincidence. Also, NORAD has not picked up any further satellites between the two objects believed to be M-Cubed and E1P-2 (Ref. 11).
• Nov. 10, 2011. The project received numerous updates regarding M-Cubed beaconing around the world and have heard M-Cubed on several occasions when it is passing Ann Arbor. However, the project has not seen any proof that it has received any of the commands transmitted. The project has numerous theories that are tried in the lab; there is hope that this issue will be resolved soon. In the meantime, the project continues to send M-Cubed various commands while it is over Ann Arbor. 11)
Sensor complement: (Camera, COVE)
M-Cubed’s primary payload is an OmniVision 2 Mpixel CMOS camera chip (OV2655) that will take quality color images of the Earth from LEO (Low Earth Orbit) at a resolution of < 200 m. This requires a data transfer of 1600 x 1200 pixels (3.76 MB/image) from the camera to the M-Cubed flight computer using the I2C (Inter -Integrated Circuit) and ISI (Image Sensor Interface) protocols.
The CMOS color camera with a 9.6 mm EFL (Effective Focal Length) Plano-convex lens (Figure 9). The camera will take an image and save it to a Colibri PXA270 microprocessor at a resolution of 1280 x 1024 pixels, each pixel with a size of 3.6 µm x 3.6 µm. This allows for moderate to high-resolution images of the Earth after postprocessing. Even with the lens positioned at the correct focal length, the whole camera payload subsystem is fairly small and takes up only 55 cm3 of volume (Ref. 16). The CMOS camera is also referred to as µEye.
Figure 9: Photo of the CMOS camera (left) and the Colibri microprocessor (right), image credit: UM, JPL)
The M-Cubed project selected the Stamp9G20 microprocessor of Taskit GmbH, Berlin, Germany to save and process the imagery.
The Stamp9G20 is a small and compact CPU module. All components, such as the processor or the memory unit, are integrated onto a board measuring just 53 mm x 38 mm x 6 mm. It was conceived for industrial applications and is well-suited as a basis for developing mobile devices and embedded computing solutions. Typical application domains for the Stamp9G20 include, for example, data logging, log conversion or measurement systems of all types.
Stamp9G20 comes with the Linux open-source operating system and the U-Boot boot loader pre-installed. Linux offers all the functionalities of a modern operating system. Multitasking, virtual memory management and dynamically loaded libraries make Linux ideal for many fields of application.
Figure 10: Schematic of M-Cubed with the COVE and Camera boards and other elements (image credit: UM)
COVE (CubeSat On-board processing Validation Experiment)
A collaboration between the University of Michigan and JPL (Jet Propulsion Laboratory) was approved for funding by NASA's ESTO (Earth Science Technology Office) ATI (Advanced Technology Initiative) in May 2010 in order to advance the TRL (Technology Readiness Level) of the new Virtex-5 SIRF device implemented with the MSPI (Multiangle SpectroPolarimetric Imager) OBP (On-Board Processing) algorithm. - A key technology development needed for MSPI, a payload on the Decadal Survey Aerosol-Cloud-Ecosystem (ACE) mission, is the OBP (On-Board Processor) to calculate the polarimetry data as imaged by each of the 9 cameras forming the instrument. 12) 13) 14) 15) 16) 17) 18) 19)
This new task is called COVE (CubeSat On-board processing Validation Experiment). The COVE task is an 18-month effort to develop the flight-ready M-Cubed CubeSat of UM with the integrated JPL OBP payload. The targeted completion date is September 2011.
Since the Virtex-5QV FPGA is not yet space-flight qualified; this in-flight validation of the technology on a pre-cursor CubeSat mission (M-Cubed) is extremely valuable toward advancing the technology readiness levels of both the FPGA and the polarimetric OBP algorithm for MSPI and the ACE mission.
Figure 11: MSPI design. MSPI will measure cloud and aerosol properties with 1 fixed and 8 gimbaled cameras (image credit: NASA/JPL)
Figure 12: Photo of the COVE FPGA processor board (image credit: NASA/JPL)
The COVE board is designed to allow for image data to be transferred to shared on-board memory without turning on the FPGA (the most power-consuming device on the board). The FPGA is only powered on when necessary to process already acquired image data; however, a direct data transfer mode does exist for circumstances when it is desirable to extend the FPGA’s operation in the space environment (after primary mission objectives have been met). Nominally, the FPGA will be powered on only long enough to process the image data and transfer its results.
In order to fit comfortably inside the M-Cubed satellite frame, the COVE board was designed to measure 90 mm x 90 mm. The footprint of the FPGA alone is 42.5 mm x 42.5 mm. The remaining board space had to accommodate memory, configuration PROMs (Programmable Read-Only Memory), regulators, ADCs (Analog-to-Digital Converters) and other support components.
Power interface: M-Cubed provides raw battery voltage in the range of 6.0-8.4 V to the COVE payload. The COVE board regulates secondary voltages required for all of its components, as indicated below:
• FPGA: 3.3 V, 2.5 V, 1.0 V
• Configuration PROM: 3.3 V, 1.8 V
• Magnetoresistive RAM: 3.3 V
• Flash memory: 3.3 V
• ADCs: 3.3 V, 1.25 V
• Buffers, muxes: 3.3 V.
M-Cubed additionally made available to COVE 3.3 V and 5 V regulated power with current limits of 2 A on each line. While the COVE design for MSPI OBP could work with this dual input power supply interface (as demonstrated in the EM design), there were two reasons why the project elected to use the single unregulated power bus input for the final FM (Flight Model) design.
- The first reason is that by not limiting the input power to 2 A, future FPGA designs that consume more power than the MSPI OBP design can be accommodated (via upload reconfiguration).
- Secondly, also to accommodate future COVE payload applications, interface requirements are eased and testing is simplified by requiring only a single supply input of a wide voltage range (the project selected dc-dc power converters that can run off any input voltage in the range of 5.5 V to 26.5 V).
The COVE payload architecture, presented in Figure 13, can be subdivided into the following logical subsystems:
• FPGA, PROMs, clock resources
• Shared memory
• Non-shared memory
• System health measurement
• Isolation buffers.
Unlike the commercial-grade Virtex-5 FPGA, the V5QV does not have an embedded PowerPC processor. The MSPI OBP algorithm requires the use of a processor which supervises and controls the algorithm running in the FPGA fabric. On the V5QV, the project utilizes the MicroBlaze soft-core processor to replace the PowerPC.
Since the V5QV is an SRAM-based volatile FPGA, it requires non-volatile configuration PROMs in order to load a bitstream on power-on. The project uses two extended temperature range XQF32P PROMs to store one FPGA configuration bitstream. These PROMs are reprogrammable via a JTAG interface, accessible on the ground only. Before final integration with M-Cubed, the project programs these PROMs with the latest FPGA firmware and MicroBlaze software.
Only a single clock source is required on the COVE board. The project uses a 100 MHz oscillator to provide the main system clock for the FPGA. Internally in the FPGA, PLL (Phase Lock Loop) and DCM (Digital Clock Manager) blocks perform all the necessary clock distribution.
Shared memory: The primary method of transferring image data to the FPGA and processed results to the M-Cubed is via a shared flash memory device. A secondary data transfer mechanism exists via an SPI bus transfer directly from M-Cubed to FPGA. In this mode, the FPGA must stay powered on the the data transfer (and subsequent image processing). The Numonyx P5QPCM flash is accessible via an SPI bus. On the COVE board, the flash SPI bus is multiplexed between the FPGA and M-Cubed, allowing either device to have independent access to the shared memory. When accessed by M-Cubed, the FPGA can be powered off.
Non-shard memory: The COVE board is also populated with non-volatile MRAM (Magnetoresistive Random Access Memory), intended for use as additional instruction and data storage for the MicroBlaze processor and FPGA. Although the OBP algorithm fits entirely within on-chip FPGA memory (BRAM) having additional external RAM enables future memory-demanding designs to be uploaded to the COVE board.
System health measurement: Two Analog Devices AD7714 ADCs and a Maxim MAX6627 temperature sensor measure board voltage, current, and operating temperature of the FPGA. All three devices operate over independent SPI buses tied directly to the FPGA. Data from the ADCs and the temperature sensor is recorded by the FPGA and reported to M-Cubed along with the image processing results.
Figure 13: Block diagram of COVE (image credit: JPL)
Communication interface: COVE communicates with M-Cubed over a 26-pin ribbon cable that is shared with other CubeSat subsystems. Out of a total of 26 I/O lines, 15 are allocated for communication with COVE.
Power: Two Linear Technology LTM4619 dual 4-amp DC/DC converters are supplied by the unregulated M-Cubed battery voltage and output 3.3 V (VIF), 3.3 V (VCC), 2.5 V, and 1.0 V secondary voltages. A single Linear Technology LT3029 dual low-dropout (LDO) regulator generates the remaining low-current secondary voltages by first regulating the 2.5 V rail down to 1.8 V and then 1.8 V down to 1.25 V. Enable pins on these parts are used to switch power on/off eliminating the bulky MOSFET based switching circuits used on the EM board. Secondary supply currents are monitored with Maxim MAX9634 current-sense amplifiers driving the ADC.
Isolation buffers: Since the FPGA on the COVE board is not powered on at all times, special care must be taken not to exercise I/O lines tied to the FPGA while it is in the off state. We utilize tri-state buffers and inverters to isolate the FPGA from M-Cubed and other board components that may be in a powered state when the FPGA is off. Additionally, these buffers are capable of the bus-holdiv functionality, ensuring that all I/O to other devices on the COVE board is held at a non-floating logic level.
Figure 14: Photo of the assembled EM COVE payload (image credit: JPL)
COVE concept of operations: The COVE board is designed with the intent to minimize power consumption while offering the full data processing capability of the V5QV FPGA. This is achieved by selectively powering on/off the interface and the FPGA. In the section below, parentheses indicate the specific component in the M-Cubed/COVE design, but can be generalized for other CubeSat applications.
COVE primary mission: During most of M-Cubed operation, the COVE board is powered off. Although the voltage regulators receive power from the battery, their control lines are set to disable voltage output.
When M-Cubed has acquired an image (taken with its on-board OmniVision camera) and is ready for onboard processing, a command is sent to COVE to turn on its SPI interface and flash memory. M-Cubed then writes the raw image data to a predetermined address in the shared flash memory. This process can take nearly 3 minutes for a standard 5.76 MB image because the (Stamp9G20) microprocessor on M-Cubed is limited in its SPI transfer rates to approximately 30 seconds per MB. The power consumption in this step, however, is very low (~0.25 W) as only a few devices on the COVE board are in the on state.
Once the full image has been transferred to the shared flash memory, the (Stamp9G20) microprocessor gives up control of the shared flash memory and commands COVE to turn on the FPGA. The V5QV FPGA loads the configuration bitstream from the PROMs and begins to read and process the data from the shared flash memory. The results of the (MSPI algorithm) data processing are written back to the shared flash memory into a predefined address space.
The FPGA notifies the (Stamp9G20) microcontroller when it is done processing data. The microcontroller then acquires control of the shared flash memory and commands the FPGA to power off. The processed data is then read back by the (Stamp9G20) microcontroller and the COVE payload activity is complete. As the final step, the microcontroller commands the COVE payload to power off its SPI interface and flash memory.
Future use of COVE: Once the primary mission is complete, the COVE board could be made available to other customers to demonstrate their FPGA-based technology on the V5QV FPGA. To accommodate future uses of the COVE payload, the board is designed with the ability to upload new configurations to the FPGA while M-Cubed is on orbit. In addition to the XQF32P PROMs that store the “golden” bitstream (non-updateable on orbit), the FPGA is also able to configure from the shared flash memory. A region of the shared flash memory is reserved for storing a secondary configuration bitstream. This bitstream may be uploaded to the flash memory from the microcontroller in the same manner image data is transferred to this memory.
COVE also provides non-volatile MRAMs for external code and data storage as may be required by future memory intensive MicroBlaze applications. This use case is enabled by first uploading a new configuration bitstream into the shared flash memory, including additional content for insertion into the MRAMs. Upon configuration, the FPGA can be instructed to copy certain contents of the shared flash memory into the MRAMs and then commence MicroBlaze processor boot.
2) Kiril A. Dontchev, Kartik Ghorakavi, Cameron E. Haag, Thomas M. Liu., Rafael Ramos, “M-Cubed: University of Michigan Multipurpose MiniSatellite with Optical Imager Payload,” AIAA paper, URL: http://www-personal.umich.edu/~mjregan/MCubed/Pages/Documents/M-CubedAIAAPaper.pdf
3) Kiril Dontchev, “Michigan Multipurpose MiniSat M-Cubed,” Summer CubeSat Workshop: Aug. 8-9, 2009, San Luis Obispo, CA, USA, URL: http://mstl.atl.calpoly.edu/~jfoley/Summer2009/Sun_1015_M-Cubed%20Cal%20Poly%20Summer%2009%20Talk.pdf
4) “M-Cubed (Michigan Multipurpose MiniSat),” CubeSat Developers Workshop, CalPoly, San Luis Obispo, CA, USA, April 8-9, 2008, URL: http://mstl.atl.calpoly.edu/%7Ebklofas/Presentations/DevelopersWorkshop2008/session1/3-MCubed-Kiril_Dontchev.pdf
5) “Trying to build a satellite on the cheap,” The Michigan Daily, April 10, 2008, URL: http://www.michigandaily.com/content/trying-build-satellite-cheap
7) “CubeSat ELaNa III Launch on NPP Mission,” NASA, October 2011, URL: http://www.nasa.gov/pdf/598567main_65121-2011-CA000-NPP_CubeSat_Factsheet_FINAL.pdf
8) John C. Springmann, Andrew Bertino-Reibstein, James W. Cutler, “Investigation of the On-Orbit Conjunction Between the MCubed and HRBE CubeSats,” 978-1-4577-0557-1/13/$31:00 © 2013, IEEE, URL: http://exploration.engin.umich.edu/blog/wp-content/uploads/2013/03/MCubed-E1p-IEEE-main.pdf
9) Information provided by John C. Springmann, Ph.D. candidate in Aerospace Engineering at the University of Michigan, Ann Arbor, MI.
10) Matt Bennett, Andrew Bertino, James Cutler, Charles Norton, Paula Pingree, John Springmann, Scott Tripp, “The M-Cubed/COVE Mission,” 9th Annual Spring CubeSat Developer's Workshop, Cal Poly State University, San Luis Obispo, CA, USA, April 18-20, 2012, URL: http://mstl.atl.calpoly.edu/~bklofas/Presentations/DevelopersWorkshop2012/Bennett_M-Cubed_Docked.pdf
12) Paula J. Pingree, “Technology Readiness Level (TRL) Advancement of the MSPI On-Board Processing Platform for the ACE Decadal Survey Mission,” ESTF 2011 (Earth Science Technology Forum 2011), Pasadena, CA, USA, June 21-23, 2011, URL: http://esto.nasa.gov/conferences/estf2011/papers/Pingree_ESTF2011.pdf
13) Paula J. Pingree, Dmitriy L. Bekker, Thomas A. Werne, Thor O. Wilson, “The Prototype Development Phase of the CubeSat On-Board Processing Validation Experiment,” 2011 IEEE Aerospace Conference, Big Sky, MT, USA, March 5-12, 2011, paper: 2.0402
14) Thomas A. Werne, Dmitriy L. Bekker, Paula J. Pingree, “Real-Time Data Processing for an Advanced Imaging System Using the Xilinx Virtex-5 FPGA,” Proceedings of the 2010 IEEE Aerospace Conference, Big Sky, MT, USA, March 6-13, 2010
15) Paula J. Pingree, “Real-Time On-Board Processing Validation of MSPI Ground Camera Images,” ESTF (Earth Science Technology Forum) Arlington, VA, UAS, June 22- 24, 2010, URL: http://esto.nasa.gov/conferences/estf2010/presentations/Pingree_ESTF2010_a4P3.pdf
16) Dmitriy L. Bekker, Thomas A. Werne, Thor O. Wilson, Paula J. Pingree, Kiril Dontchev, Michael Heywood, Rafael Ramos, Brad Freyberg, Fernando Saca, Brian Gilchrist, Alec Gallimore, James Cutler, “A CubeSat Design to Validate the Virtex-5 FPGA for Spaceborne Image Processing,” Proceedings of the 2010 IEEE Aerospace Conference, Big Sky, MT, USA, March 6-13, 2010
17) Dmitriy L. Bekker, Paula J.Pingree, Thomas A. Werne, Thor O. Wilson, Brian R. Franklin, “The COVE Payload – A Reconfigurable FPGA-Based Processor for CubeSats,” Proceedings of the 25th Annual AIAA/USU Conference on Small Satellites, Logan, UT, USA, Aug. 8-11, 2011, paper: SSC11-I-2
18) “NASA Launches JPL-Built Earth Science Experiment,” Space Daily, Nov. 2, 2011, URL: http://www.spacedaily.com/reports/NASA_Launches_JPL_Built_Earth_Science_Experiment_999.html
19) Charles D. Norton, Michael P. Pasciuto, Paula Pingree, Steve Chien, David Rider, “Spaceborne flight validation of NASA ESTO technologies,” Proceedings of IGARSS (International Geoscience and Remote Sensing Symposium), Munich, Germany, July 22-27, 2012
The information compiled and edited in this article was provided by Herbert J. Kramer from his documentation of: ”Observation of the Earth and Its Environment: Survey of Missions and Sensors” (Springer Verlag) as well as many other sources after the publication of the 4th edition in 2002. - Comments and corrections to this article are always welcome for further updates.