Minimize AAReST

AAReST (Autonomous Assembly Reconfigurable Space Telescope Flight Demonstrator)

Launch    Telescope Payload Design   Electromagnetic Docking System Design   References

In recent years, there has been a desire to develop space-based optical telescopes with very large primary apertures (over 20 m in diameter). Currently the largest primary aperture under development is that of the JWST (James Webb Space Telescope) with a diameter of 6.5 m. JWST represents a major shift in telescope design due to its use of a deployable primary mirror. However, its size is still ultimately limited by the use of a monolithic spacecraft which must fit within the diameter of its launch vehicle; a limitation for all current spaceborne telescopes. 1)

One method to overcome this obstacle would be to form the telescope in orbit by means of the autonomous assembly of small independent spacecraft, each with their own mirror. In this way, a telescope with a large, segmented primary mirror can be constructed.

Furthermore, if each of these mirrors is manufactured to have an identical initial shape, which can then be adjusted upon assembly, a substantial reduction in manufacturing costs can be realized. We believe that such an approach will be a key enabler for the practical and cost-effective realization of future large aperture (~100 m diameter) space telescopes, and will have broader application in other space missions which require on-orbit assembly.

To this end, over the last decade, the California Institute of Technology (Caltech) and the University of Surrey – Surrey Space Centre (SSC) have been working together in an effort to develop the key technologies and techniques needed for such a mission, and in 2016, expanded this collaboration to include the Indian Institute of Space Science and Technology (IIST), thus forming a truly global partnership.

Our work has been focussed on the design, manufacture and execution of a practical in-orbit demonstration (IOD) of an small astronomical telescope which uses autonomous assembly techniques: AAReST (Autonomous Assembly of a Reconfigurable Space Telescope).

As Figure 1 shows, the overall design concept for the spacecraft elements of AAReST has evolved over the years [2-4], generally moving to the increasing use of CubeSat technologies to form the spacecraft bus elements, however, the overall mission objectives have remained the same, that is to:

• Demonstrate all key aspects of autonomous assembly and reconfiguration of a space telescope based on multiple mirror elements – including the use of adaptive mirrors.

• Demonstrate the capability of providing high-quality (astronomical) images.

• Provide new opportunities for education in space engineering to undergraduate and post-graduate students and to foster academic collaboration between the project partners.

• Use this demonstration to provide outreach activities worldwide, to encourage participation of young people in science, technology, engineering and mathematics.

The AAReST project, which started in 2009, has the goal to be ready for launch in the timeframe2019-2020. 2) 3) 4) 5) 6) 7) 8) 9)


Figure 1: Evolution of the AAReST Spacecraft Design 2008-2018 (image credit: AAReST collaboration)

The current (2018) configuration for AAReST, which is one moving forward to flight manufacture, is based on three independent spacecraft elements: two "MirrorSats" and one "CoreSat".

The two MirrorSats are based on modified 3U CubeSat structures (one from SSC and one from IIST), each of which carry an electrically actuated DMP (Deformable Mirror Payload), designed by Caltech. The payloads are each housed in a cube-shaped "Mirror Box" mounted on the top end of each 3U structure.

Each MirrorSat is capable of autonomous undocking and re-docking with the central CoreSat using the optically guided, electro-magnetic docking system developed by SSC.


Figure 2: CAD view of AAReST in launch configuration (image credit: AAReST collaboration)

The CoreSat itself comprises a bespoke nanosatellite structure (designed by Caltech) approximating to three 3U CubeSats stacked side-by-side – i.e. 9U. This carries two fixed-shape RMPs (Reference Mirror Payloads) mounted in Mirror Boxes on top of the structure, similar to those on the MirrorSats. It also carries a deployable carbon-fiber reinforced polymer (CFRP) boom on which is mounted a focal plane assembly used for focussing and astronomical imaging CP (Camera Package). The RMPs, deployable boom and CP were also designed and developed by Caltech.

Mission Concept: The central premise of the mission is that the all three spacecraft elements are self-supporting, intelligent, and are capable of operating independently and cooperatively to provide systems autonomy. The MirrorSats must each be capable of un-docking, maneuvering and re-docking to the CoreSat to form different configurations, and the mirrors they carry must be able to move and change shape to enable focused images to appear on the camera's focal plane.

On orbit, the mission profile will firstly establish the imaging capability of the compound spacecraft before undocking – the so-called "narrow configuration". Here we require the CFRP boom to deploy, and position the CP such that it is able to "see" the four mirrors. The CoreSat will then maneuver to align the telescope axis with a chosen (bright) star and through closed-loop control, the four mirrors will be adjusted so that the star comes into focus and an image of it captured.

Once proven to be capable of producing images, one of the MirrorSats will then undock and hold its position in close proximity with the rest of the spacecraft, as the CoreSat executes a ~90º rotation. The deployed MirrorSat will then autonomously re-dock to the CoreSat, to occupy its new "wide configuration" position.

This will happen under the control of the SSC rendezvous and docking (RDV&D) systems (mounted in both the MirrorSats and the CoreSat). This uses electromagnetic forces only to carry out the maneuver and capture operations. Autonomous relative navigation is achieved using the MirrorSat's active lidar and passive machine vision sensors coupled to its advanced R-Pi processors. Thus, the maneuver will test the docking system, autonomous navigation and maneuver control technology.


Figure 3: CAD view of AAReST in (left) Narrow and (right) Wide Configurations (image credit: AAReST collaboration)

If successful, the next stage will see the second MirrorSat undock and re-dock to the CoreSat to form the wide linear formation which represents a large (but sparse) aperture for high resolution imaging. Celestial targets will be imaged.

Figure 3 shows a CAD representation of the spacecraft in narrow and wide configurations, and Figure 4 shows a schematic of the overall mission concept.


Figure 4: AAReST mission overview. Left: concept demonstrating the launch configuration; Middle: initial imaging configuration; Right: secondary imaging configuration (image credit: AAReST collaboration)

Currently, the AAReST Structure and Qualification Model (SQM) is under test and elements of the flight hardware are under construction. Launch is planned for 2019-2020 period, although no specific launch has been selected yet.

This paper builds upon previous work, and describes more recent test results and progress on the production of the flight hardware. 10)

Launch: Launch readyness of AAReST is planned for 2019.

Orbit: Current reference orbit: 650 km SSO.


Telescope payload design

The AAReST telescope payload represents a segmented sparse aperture comprising four circular mirror elements of 10 cm diameter each – two rigid and two deformable (Figure 5). It is a prime focus telescope, operating in the visible band (465-615 nm wavelength bandpass) and has a 0.34º FOV (Field of View) and 1.2 m focal length.


Figure 5: AAReST optical instrument configuration (image credit: AAReST collaboration)

In the narrow configuration, the four mirrors are encompassed within an aperture of 0.405 m, giving a focal length to aperture diameter (f/D) ratio (F-number) of 2.87. In the wide configuration, the mirrors are more widely spaced and fit within a 0.530 m aperture, giving an F-number of 2.19. In both cases the physical aperture and therefore the light gathering ability, is set by the mirrors, i.e. 4 x 78.5 cm2 = 314 cm2.

Each mirror is supported on a tip/tilt/piston mount allowing 3 rigid-body motions, so small shifts in mirror location due to deployment errors, thermal distortion or docking alignment errors, can be taken out (errors of order 10-3m). Actuation is via three picomotors. These can also provide fine scale adjustments (errors of order 10-6m). Mounting the mirrors such that they are held rigidly for launch, but are free to move in orbit, has been one of the greatest mechanical challenges, since we have wanted to avoid using deployable "lens caps" or risk damage to the mirror surfaces by pressing on them from the front. The mirrors therefore remain "open" to the environment during launch, and we have adopted a 3-sphere caged magnetic ball-bearing kinematic mount to provide retention, while still allowing movement. This is supported by a spring-release latching mechanism mounted behind the mirrors which holds the mirrors firmly during launch.

The deformable mirrors have proven to be challenging to manufacture to the correct tolerances, but, by means of a special slumped glass process developed in collaboration with the GSFC Goddard Space Flight Center), mirrors of the required quality and consistency have been produced.

The process starts with a sheet of 200 µm thick Schott D263 glass, which is heated over a fused silica mandrel protected by a boron nitride release layer (Figure 6).


Figure 6: DMP (Deformable Mirror Payload) mirror production method (image credit: AAReST collaboration)


Figure 7: Deformable mirror electrode pattern (image credit: AAReST collaboration)


Figure 8: Deformable mirror cross section and back view showing the electrode pattern (image credit: AAReST collaboration)

To enable co-focussing (but note that co-phasing of the mirrors is not carried out in this mission), the shape of the deformable mirrors has to change according to their differing positions in the imaging plane of the camera. Thus, these mirrors each have embedded within them 41 piezoelectric actuators driven by a ±500 V high voltage power supply. They are arranged in a pattern designed to optimize for astigmatism.

Figure 7 shows the electrode layout with an example voltage pattern applied. Figure 8 shows a cross section through the mirror, and a view of the back of the mirror. Imaging sensors and Shack-Hartmann Wavefront Sensors (SHWSs) in the CP (Camera Package) provide the feedback for active control and calibration of the telescope in flight (Figure 9).

The SHWS comprises a microlens array (Edmund Optics #64-476: 10 mm x 10 mm, 300 µm pitch, 5.1 µm focal length 33 x 33 lenslets in total) mounted in front of a CMOSIS CMV4000 CMOS imager with 2048 x 2048 pixels (5.5 µm pixel pitch) with an 11.26 mm x 11.26 mm active area and a peak quantum efficiency of 0.57 at 537 nm.

The actual imaging sensor is an Aptina MT9P031 5M pixel (2592 x 1944) 1/2.5" CMOS array with 2.2 µm x 2.2 µm pixel pitch mounted within a Ximea MU9PM CMOS camera. This was chosen for its convenience, low power consumption, low mass, good noise performance and high light sensitivity.


Figure 9: Active mirror adjustment feedback control (image credit: AAReST collaboration)

Figure 10 shows the internal layout of the CP (Camera Package). The optical design was aided by modelling in ZEMAX. 11) The pupil mask is motorized so that it can adjust for the narrow and wide configurations of the telescope.

The CP meets the design requirements of having a mass <4 kg, a volume < 10 cm x 10 cm x 35 cm and a power consumption < 5 W.


Figure 10: Camera Package optical design (image credit: AAReST collaboration)

Figure 12 shows the mechanical design of the CP and Figure 11 shows the mask mechanism.


Figure 11: CP (Camera Package) pupil mask mechanism (image credit: AAReST collaboration)


Figure 12: CP (Camera Package) mechanical design (image credit: AAReST collaboration)


Telescope manufacture and testing

The AAReST optical system has been subject to extensive modelling and testing over the years. Deformable mirrors have been produced for initial shape testing (GS-12), thermal testing (GS-13) and electrode testing (RS-14). Performance has been measured and compared to reference mirrors using a reverse Hartmann test-bed (Figure 13).

As a result, AAReST has been shown to be able to meet its science optical requirement of 80% encircled energy in a 25 µm radius. Flight versions of the mirrors are now in production.


Figure 13: Deformable mirror testing using the reverse Hartmann Test-Bed (image credit: AAReST collaboration)

The CP has been prototyped and is found to function well. This was tested using an auto-collimation technique to simulate the effect of a distant star (Figure 14). The masking mechanism works well, and the image sensors perform as expected. It is seen that for stars brighter than magnitude 2.0, both the SHWS and imaging sensors will attain sufficient signal-to-noise (SNR) ratio within the time limits imposed by the satellite's pointing stability (pointing accuracy ±0.05º in all axes, attitude stability requirement: <0.001º/s ). Further testing is planned on the protoflight model.

The Mirror Boxes and DMP (Deformable Mirror Payload) holding and tip/tilt/piston mechanisms have been built and tested for vibration and thermal performance. Initial results look good, but additional testing will need to take place with the boxes integrated with the spacecraft to provide full evaluation.

The algorithms needed to control the telescope, mirror positing and shape correction have been developed and tested. The next step is to verify performance with all the hardware in the loop.


Figure 14: Testing the Camera Package (image credit: AAReST collaboration)

The CP has to be deployed on the end of a 1.16 m long boom in order to place it at the focal point of the mirrors. The mechanical requirements are quite stringent – the boom must deploy reliably (after storage) and it must be both straight and stiff enough to resist bending and torsional flexing due to thermal cycling in space.

To this end Caltech developed a new "slotted hinge" boom of 38.8 mm diameter, 1.45 m length and 200 µm in thickness that has a mass of just 65 g. Its straightness is 5 mm in 1.45 m (~0.34%). A new lay-up and manufacturing process was developed which avoided the seam opening phenomena seen on earlier booms after multiple folding and unfolding. The boom was tested for outgassing and viscoeleasticity (to ensure correct performance after long-term storage), and a kinematic mount and burn-wire-based separation device was developed and tested (Figure 15).

A special rig was developed to offload gravity and to allow the boom to unfold in realistic dynamic conditions with a 4 kg CP and a 30 kg host satellite (Figure 16). The system was configured for a two stage deployment, and these were found to occur smoothly and reliably. Figure 17 shows the boom after the first stage deployment.

The boom meets our positional accuracy requirements, and the kinematic mount offers ample capability for fine adjustment. — Work is now progressing on the flight version.


Figure 15: Boom kinematic mount and deployment mechanism (image credit: AAReST collaboration)


Figure 16: Boom deployment test rig (image credit: AAReST collaboration)


Figure 17: Boom after first stage deployment – Note the "Dog-Bone" slotted-hinges developed by H.M.Y.C. Mallikarachchi and S. Pellegrino (2008-12), image credit: AAReST collaboration


Electromagnetic docking system design

The AAReST Rendezvous & Docking (RDV&D) system uses the SSC Electro-Magnetic (EM) Kelvin Clamp Docking mechanism, which comprises four pulse-width modulated (PWM), H-bridge-driven, dual polarity pure-iron cored electro-magnets, each of over 900 A-turns mounted in both the MirrorSat and CoreSat spacecraft. These are coupled to three "probe and drogue" (60º cone and 45º cup) type mechanical docking ports, arranged to form and extended area docking surface. Figure 18 shows the stainless steel "Probe" Docking Ports in the SSC MirrorSat (replicated in the IIST MirrorSat) and Figure 19 shows the receiving aluminum alloy "Drogues". Kinematic constraint is established using the Kelvin Clamp principle (3 spheres slotting into 3 V-grooves arranged at 120º), which allows for precision alignment, critical for the telescope mirror elements.


Figure 18: SSC MirrorSat showing the EM docking systems placed around the frangibolt interface so as to form a 3 Probe triangular arrangement (image credit: AAReST collaboration)

During launch, the MirrorSats are held rigidly onto the CoreSat via a frangibolt passing through an interface plate mounted on the outside of the MirrorSat structure (Figure 20).


Figure 19: AAReST CoreSat representation showing the position of the matching EM docking system "Drogues" set to receive the MirrorSats in wide configuration (image credit: AAReST collaboration)


Figure 20: CoreSat-MirrorSat frangibolt interface (image credit: AAReST collaboration)

Once in orbit, the MirrorSats remain held this way whilst the boom/CP is released and whilst initial imaging takes place in the narrow configuration.

Once the basic functioning of the telescope is established and the efficacy of the deformable mirror control system is confirmed, the docking port electro-magnets in the one of the MirrorSats will be activated, as will the matching docking port electro-magnets in the CoreSat, so as to give an attractive force.

The frangibolt joining the two is then activated, and the magnetic attraction between the MirrorSat and CoreSat should resist any tendency for the MirrorSat to recoil away from the CoreSat – at least in principal!

In practice we found that severing the frangibolt gave a much larger recoil force than we had expected, far too large to be resisted by the electro-magnets, the ends of which, for launch vibration reasons, had to physically spaced apart by at least 2 mm. We are therefore going to need to use burn-wires in addition to the magnets to resist this recoil. Once the burn-wires are severed (which is a recoilless action) the MirrorSat will only be held on to the CoreSat magnetically, but its position and orientation will remain fixed by the triangular-based pyramid structure which is part of the frangibolt interface (Figure 20). To save power, small permanent "latching" magnets are also mounted in the interface so that the electro-magnets can be turned off, and the spacecraft will still have sufficient attractive force to overcome any likely disturbance (such as a CoreSat attitude maneuver).

However, the force of attraction due to the permanent magnets can be overcome by switching both the CoreSat and MirrorSat electro-magnets into repulsion mode. This will cause the MirrorSat to undock and slowly move away. Air bearing table experiments indicate that the parting velocity will be around 3 cm/s, with a push off force of ~ 1 N.

As the magnetic force drops off very quickly with distance, the MirrorSat does not accelerate further once it is separated by a few cm.

By pulsing the electro-magnets on-and-off, and changing their polarities, the satellites' relative motion and orientation can be controlled within the docking port's extended "capture cone" force field. This has been simulated and experimentally verified to extend out to ~50 cm or so from the port's surface, with a half-cone angle of ~45º (Figure 21).

Thus, immediately separation occurs, an active RDV&D controller in the MirrorSat uses the MirrorSat's lidar and machine vision relative navigation sensors to detect the motion and switch the polarity of the docking port magnets so as to make the MirrorSat slow down and come to rest relative to the CoreSat.


Figure 21: Finite element modelling of the magnetic interaction between the docking port elements – verified in air bearing table tests (image credit: AAReST collaboration)

The design separation distance in the stationary state is ~25 cm ±5 cm – far enough to avoid the risk of contact while the CoreSat turns, but still close enough to remain within the capture cone of the docking port.

There is an ISL (Intersatellite Link) between the MirrorSat and CoreSat, which enables the RDV&D controllers in both spacecraft to set the strength and polarity of the Docking Port magnets appropriately. The maneuver however does not rely on this link being available: the CoreSat magnets can remain passively on, while the MirrorSat executes its control strategy alone.

Once the MirrorSat is stationary with respect to the CoreSat, the CoreSat executes a steady yaw rotation of ~90º. Thus, the CoreSat's side docking port, with the receiving drogues, will come into close proximity to the stationary MirrorSat's docking ports such that the docking ports are in their mutual controlling force field.

The MirrorSat can then re-activate its docking magnets to maneuver and dock with the CoreSat (Figure 22).

For space and mirror location reasons, the CoreSat drogues have to be placed such that there is no room to mount a second set of electro-magnets behind them. Instead, the CoreSat makes use of strong permanent magnets fixed in the aluminum triangular prism structures shown in Figures 23 and 24.

The reason why there are three probe-and-drogues but four magnets per docking system is that 3 sphere/V-groove structures gives 6 contacts and the necessary 6 kinematic constraints, but four magnets are needed to control the relative motion and orientation of the spacecraft during the RDV.


Figure 22: MirrorSat lower docking port (left) docked to CoreSat lower docking port (right), image credit: AAReST collaboration


Figure 23: CoreSat upper docking port showing the electro-magnets and drogues – permanent magnets mounted in the triangular prisms (image credit: AAReST collaboration)


Figure 24: CoreSat lower docking port showing the electro-magnets and single drogue – permanent magnets mounted in the triangular prisms (image credit: AAReST collaboration)

Once the MirrorSat has successfully docked to the CoreSat side port, images will be made in this L-shaped configuration (the other MirrorSat still being in the narrow configuration). After images have been acquired, the second MirrorSat will undergo an identical set of maneuvers to dock to its CoreSat side docking port. AAReST will then be in its full wide configuration mode, and further images will be taken.

Should the first MirrorSat not be able to successfully dock, the second MirrorSat provides a second opportunity to try. Thus, flying two MirrorSats gives a degree of redundancy. In any case, once the main imaging mission is over, there is the option for the IIST MirrorSat to undock and maneuver away from the CoreSat to carry out formation flying and other experiments. To this end, the IIST MirrorSat carries an IIST developed cold-gas (butane) propulsion system and a transceiver system to communicate with the CoreSat. The SSC MirrorSat is not intended to separate from the CoreSat once it has docked in the wide mode.

The docking ports allow the spacecraft to share electrical power, through 5 V, 1 A lines. The MirrorSats can also be activated and charged via the frangibolt interface.


Docking system manufacture and testing

Prototypes of the docking ports have been manufactured and tested on SSC's air bearing table using COTS 3U CubeSat structures mounted on top of compressed air carriers, which are used to provide near frictionless motion (Figure 25). The air table is a large, flat and level block of granite (Figure 26). The fly height is a few µm, and the gas cylinders allow ~15 minutes of flight time before needing to be recharged. SSC has developed a system of gimballed microporous carbon feet which allows the carriers to float without any disturbances from non-uniform air jets.


Figure 25: Prototype electro-magnetic RDV&D system under test at SSC (image credit: AAReST collaboration)

These experiments verified the docking port capture cone force field concept, and demonstrated that we could capture and dock, even under open loop control, from a distance of up to 50 cm. We could also undock, repel, stop and hold the MirrorSat in alignment at the required 25 cm distance, with the polarity of the electro-magnets first switching 2 s after separation, when the MirrorSat had departed to a distance of 6 cm.


Figure 26: Rendezvous, docking and undocking tests on SSC's large air bearing table (image credit: AAReST collaboration)

Force measurement results were compared to both our semi-analytical (Gilbert model) and finite element analysis (FEA) magnetic force models and were found to be in good agreement with the latter. The force drops off rapidly with distance as shown in Figure 27. We found that simple magnetic dipole models did not work unless scaled by empirical factors, however, such scaled simplified models were found to be useful for speeding up calculations in the on-board control loops.


Figure 27: Semi-log plot of measured force (red) compared to predicted force (blue). The force scale is 10-1 to 104 mN. Force at 50 cm = 1mN (image credit: AAReST collaboration)

FEA analysis also verified our experimentally determined capture cones and the effect of different magnetic polarity configurations (Figure 28).


Figure 28: 2D FEA modelling of different magnetic configurations: (top) effect of approach angle; (bottom) attract and repel modes (image credit: AAReST collaboration)


Relative Navigation Sensor System

Much experimentation has been made at SSC on using small COTS lidar systems for relative navigation. We started by using the Microsoft KINECTTM, but then moved on to the much more compact Softkinetic DS325. For AAReST, we have taken the Softkinetic DS325 Lidar/Camera system, stripped it down so that it can be mounted on a CubeSat and modified certain components to make it compatible with flight in space thermal-vacuum conditions (Figure 29). This will be used to monitor and control the rendezvous/docking process to the point of automatic capture.

The lidar projects a near-infra-red (NIR) speckle pattern via a laser diode which is picked up by a NIR sensitive camera (QVGA 320 x 240 pixels – 87º x 58º Field of View) for depth processing using PrimeSense system-on-chip (SoC) technology (60 fps). It has a quoted range of 15 to 100 cm. It also carries a full color (VGA) camera for machine vision.

In testing, we initially used the OpenNI2DS325 driver to determine range, however, we found it to be too inaccurate for our purposes. Therefore we developed new algorithms to convert raw sensor data into depth measurements, which led to much more accurate results (Figure 30).


Figure 29: The Softkinetic DS325 Lidar/Camera system (image credit: AAReST collaboration)


Figure 30: The Softkinetic DS325 Lidar/Camera system range estimation using original (black) and newly developed algorithms (blue),image credit: AAReST collaboration

We found that the lidar can be made to work out beyond 1m, and we have obtained good results out to 2.5 m using our new algorithms. However, there are limits on close range operation, with the system not really working well until the range goes beyond ~30 cm. This is too far for AAReST, and so we need to do further work to enable it to work over shorter ranges.

We also found that, while the system worked well in the subdued lighting conditions of the laboratory, it did not work well out in bright sunlight. As we cannot guarantee that the AAReST docking maneuvered won't need to take place in the full glare of the Sun, we have decided to add a second sensor system which can work even in full sunlight.

MVS (Passive Machine Vision) Sensor System: In our early MVS work, we made use of black-and-white glyph patterns and the POSIT algorithm to find the pose, range and speed of the target spacecraft using the VGA camera (Figure 31).


Figure 31: Using glyph patterns to identify the relative position, pose and motion of target spacecraft (image credit: AAReST collaboration)

Later we modified this technique to use glyph-like patterns of NIR light-emitting diodes (LEDs) (Vishay TSHG6400: 2.3V, 1A max., 700 mW/Sr - 850 nm wavelength), so that we could operate in darkness as well as in lit conditions (Figure 32). The detection and pose/range algorithms ran on a commercial R-Pi processor and typical update rates were ~1Hz. The total power consumption was < 1 W with the LEDs operated at 10% power (100 mA - 70 mW/Sr).


Figure 32: Using NIR LED patterns to identify relative position and pose (image credit: AAReST collaboration)

In further testing we found that, as with the lidar, this system was badly affected by the presence of the Sun when operated under bright sunlit conditions. We therefore adapted the technique to make it effectively "solar blind" by adding a narrow band NIR (850 nm, 10 nm full width half maximum) interference filter, placed over the camera. This was in addition to the camera's own built in NIR cut filter – thus reducing the camera's sensitivity to light dramatically (Figure 33). Note NIR interference filters also pass blue light due to the harmonic relationship of the wavelengths (425 nm is exactly half 850 nm). This can be removed by use of an additional blue filter.


Figure 33: MVS camera sensitivity and filter bands vs. wavelength (image credit: AAReST collaboration)

We found the system worked well in full bright sunlight – even with the Sun directly in the field of view. Figure 34 shows the camera's view of the Sun with the LED pattern still clearly visible at a distance of 1.4 m. Even if the Sun is directly behind the target spacecraft, the blocking effect of the body of the spacecraft still enables the LED pattern to be distinguished.


Figure 34: NIR LED pattern still visible with Sun directly in the FOV (image credit: AAReST collaboration)

Rendezvous and Docking System Processor: The MirrorSat and CoreSat RDV&D system is controlled by the Payload Interface Computer (PIC) developed by SSC – so called, as it provides the control communications for the telescope payload elements as well (via Wifi links).

The PIC (Figure 35) consists of two Raspberry Pi Compute Module 3s for processing and a XC95144XL configurable programmable logic device (CPLD), capable not only of detecting critical errors and taking specific actions to resolve them, but also able to directly communicate with other subsystems on the spacecraft.

The CPLD plays a key role on the PIC. It has several responsibilities, as it needs to be able to monitor the status of the active RPiC3M, the RT5370N wireless interface, and also two USB hubs. It also enables and disables all voltage regulators so that it can provide power sequencing for all devices on the board. It can toggle reset pins on the RPiCM3s, the USB hubs, the RT5370N wireless interface and switch the I2C and UART lines between the two RPiCM3s routing incoming and outgoing I2C requests from/to the Platform I2C and the Payload I2C busses. Power consumption data from Maxim Integrated ADCs is collected from multiple I2C data points.


Figure 35: AAReST PIC (Payload Interface Computer), image credit: AAReST collaboration

Since the RPiCM3 is the master of the payload I2C bus and a slave of the Platform I2C bus, it is mandatory for there to be a slave I2C core on the CPLD able to execute request from the RPiCM3 master. These requests would mainly involve USB routing and switching. Figure 36 shows the PIC system diagram.

By default, the BCM2837 have two UART interfaces; one full UART and one miniUART. On the Raspberry Pi Compute Module, there is no Bluetooth module, therefore it is the default UART interface mapped to the I/O pins named "ttyAMA0" with the default 115200 baud rate. The miniUART named "ttyS0" runs off the system clock, which by default is dynamic and therefore in order to use it, it is required to add the line "core_freq=250" to the "cmdline.txt" file in the "boot" partition. Since there are two Raspberry Pi Compute Modules, and one UART header, it is necessary to configure the CPLD to select one and connect it to the UART header, in order to enable external communications. The VHDL code needed to achieve this consists of a FSM statement and port assignments. The computer needs a UART to USB interface which can be provided by an external PL2303 module. Hardware and software flow control needs to be disabled in order for the UART to operate. We are able to access the RPiCM3s through the serial console (minicom) and fully operate them consuming 7 mA at 5 V and 73 mA at 3.3 V. Payload operations, which will consume larger computational loads, are to be profiled.


Figure 36: PIC (Payload Interface Computer) system diagram (image credit: AAReST collaboration)


Spacecraft Busses

AAReST MirrorSats: The MirrorSat requirements can be summarized as follows:

• Must support the Deformable Mirror Payload (DMP) mechanically and electrically via a 5 V 1 A supply (2 W continuous operational power) and telemetry/telecommand (TTC) via a USB 2.0 interface.

• Must be able to operate independently of other units.

• Must be able to communicate with the CoreSat out to 1km max distance. (via COTS Wi-Fi based ISL)

• Must be able to undock from the narrow configuration, rendezvous and re-dock in the wide configuration.

• Must have 3-axis control to ±2º precision.

• Must be rigidly attached to the CoreSat during launch, and must provide a low/zero power magnetic latch to hold in position on CoreSat in orbit.

• Must be able to safely enter the CoreSat Docking System's acceptance cone:

- 20-30 cm distance (mag. capture);

- ±45o full cone angle; < 5 cm offset;

- <±10º relative RPY error;

- < 1 cm/s closing velocity at 30 cm;

- < ±2º relative RPY error at first contact.

Given these requirements, SSC and IIST each developed their own design of MirrorSat, albeit both carrying a common set of mission critical systems:

• Deformable Mirror Payload (DMP) (Caltech)

• Payload Interface Computer (PIC) (SSC)

• Lidar/MVS Relative Navigation Sensor (SSC)

• Upper and Lower Docking Ports (SSC)

• Frangibolt Interface (Caltech)

SSC based its MirrorSat on an ISIS 3U structure.


Figure 37: SSC MirrorSat layout (image credit: SSC)

The SSC MirrorSat bus systems are a mixture COTS and bespoke hardware. In addition to the electro-magnetic docking ports, RDV&D Sensor, and PIC already described, there is a bespoke DMP Mirror interface plate, four bespoke solar panels (±X and ±Y facets) and power switch board, and an SSC developed single axis (Z) butane cold gas propulsion unit. This unit has the following specifications:

• 5 – 10 mN thrust range at ~ 80 s Isp.

• ~5-10 m/s ΔV

• Minimum Impulse bit = 10-20 µNs.

• System mass ~ 860 g (wet); with ~64 g butane.

Liquefied butane propellant is stored in the 120 ml volume two-part welded aluminum propellant tank (with anti-slosh baffle) at 2 bar pressure (0.53 g/cm3 density). The tank factor of safety is 12 (48 bar burst pressure). The butane is expelled at 0.5 to 1 bar via a pressure controlled plenum chamber into an electrically heated (1 W) expansion nozzle, which has an expansion ratio of 200:1 (0.1 mm throat diameter and 2 mm exit diameter).

Butane is chosen as a propellant as it has good density, good specific impulse and has no toxic or carcinogenic qualities. It self-pressurizes to modest pressures at the spacecraft's operational temperature range (e.g. 1 bar at 0ºC, 4 bar at 42ºC) and has a low freezing point (-138ºC), so no propellant heating is required. Figure 38 shows the layout of the system and Figure 39 shows the prototype system under test.


Figure 38: SSC butane cold gas propulsion system (image credit: SSC)


Figure 39: Prototype propulsion system under test (image credit: SSC)

The other systems are all COTS, including an ADCS (Attitude Determination and Control System) from Cubespace (Figure 40) comprising:

• CubeSense

• CubeControl

• CubeComputer

These include:

• CMOS Camera Digital Sun Sensor (fine Sun Sensor)

• CMOS Camera Digital Earth Sensor

• 6 Photodiode-based Course Sun Sensors

• MEMS (Micro-Electro-Mechanical-System) Gyro

• 3-Axis Magnetoresistive Magnetometer

• 3-Axis Magnetorquer (2 Rods + 1 Air Coil)

• Pitch-Axis Small Momentum Wheel (MW)

• Optional GPS Receiver (Novatel OEM615)

• Extended Kalman Filter (EKF) Control software +SGP4 Orbit Propagator.


Figure 40: QB50 ADCS unit (image credit: Cubespace)

Sensors and actuators



Error (RMS)


3-axis magnetoresistive

±60 µT

< 40 nT

Sun sensor

2-axis CMOS imager



Nadir sensor

2-axis CMOS imager



Coarse sun sensor

6 photodiodes

Full sphere

< 10º

Rate sensor

MEMS gyro


< 0.05º/s

Pitch momentum wheel

Brushless DC motor

±1.7 mNms

< 0.001 mNms


Ferromagnetic rods and air coil

±0.2 Am2

<0.0005 Am2 (remanence)

Table 1: ADCS unit specifications

The ADCS can also act as a basic OBC, but the main spacecraft control functions are implemented via the PIC. The EPS (Electrical Power System) is based on the GOMspace P31u EPS, with its integral 20 Whr battery. The EPS interfaces to the custom made solar panels, which are mounted with Azur Space triple-junction solar cells which have 28% efficiency.

Spacecraft communications are mediated through the CoreSat via the WiFi ISL (Intersatellite Link). There is no independent uplink/downlink capability.

The IIST MirrorSat bus is all bespoke, including its structure, EPS (with separate battery board), solar panels, ADCS, and OBC. The OBC consists of Microsemi Smart Fusion 2 SoC which includes an FPGA and cortex M3 as the processor. The FPGA provides flexibility for adding newer communication interfaces and parallel fast processing elements for future expansion. The OBC acts as both Command and Data Handling System and ADCS computer. The ADCS is being designed to satisfy the specifications similar to Cubespace ADCS.

The IIST MirrorSat also carries an IIST designed butane propulsion unit with similar capabilities to the SSC one. All these systems are currently in development. Figure 41 shows the IIST MirrorSat layout.


Figure 41: IIST MirrorSat layout (image credit: AAReST collaboration)

AAReST CoreSat:

The AAReST CoreSat is designed and developed by Caltech. It is the cornerstone of the mission, supporting key elements of the telescope payload, and providing precision guidance, suitable for star imaging. Its requirements are:

• Must be able to point accurately (< 0.1º 3 s error all axes)

• Must be stable in attitude (< 0.02º/s for 600 s) during payload operations.

• Must be able to slew at >3º/s for RDV maneuvers.

• Must be able to mechanically support 2 RPMs (Reference Mirror Payloads) and to supply them with 2 W power at 5 V.

• Must provide up to 5 W at 5 V power and I2C comms. to the "camera" (image data transfer only) and support boom.

• Must provide up to 5 W at 5 V power to both docked MirrorSats.

• Must be able to communicate with the MirrorSats via WiFi and to the ground via a VHF uplink (1.2 kbit/s) and UHF downlink (9.6 kbit/s).

• Must be able to operate with Sun >20º off the optical (Z) axis.

• Must provide hold-downs for MirrorSats, camera and boom during launch.

• Must provide the launcher interface.

The CoreSat comprises a bespoke 9U structure with bespoke solar panels, COTS EPS and Battery (P60 unit from GOMSpace), COTS ADCS Unit (3-axis unit from CubeSpace) with 3 magnetorquers, 3 reaction wheels, 2 magnetometers and a star tracker for precision pointing. There is also a VHF Uplink and UHF Downlink (He-100 from AstroDev).

The CoreSat provides the interface to the launch vehicle through a standard fitting compatible with the Indian PSLV rocket.

Figure 42 shows the CoreSat layout and Figures 43-47 show a CAD representation of the complete AAReST spacecraft.


Figure 42: Caltech CoreSat Layout (image credit: Caltech)


Figure 43: AAReST spacecraft – isometric view (image credit: AAReST collaboration)


Figure 44: AAReST spacecraft – side view (image credit: AAReST collaboration)


Figure 45: AAReST spacecraft – top view (image credit: AAReST collaboration)


Figure 46: AAReST spacecraft – bottom view (image credit: AAReST collaboration)


Figure 47: AAReST spacecraft – front view (image credit: AAReST collaboration)

AAReST Structural Qualification Model:

In June 2018, the Caltech, SSC and IIST teams each produced mass dummies of their AAReST spacecraft, and these were assembled into a SQM (Structural Qualification Model) for initial vibration testing (Figure 48).


Figure 48: AAReST SQM on vibration table (image credit: AAReST collaboration)

Unfortunately, the SQM did not pass the vibration test with the PSLV loads. During the sine vibration test, on the X axis, at 4.5 g and 10 Hz, the screws on the base of the CoreSat interfacing with the launch vehicle interface were seen to bend and shear off, forcing us to stop the test.

The Frangibolts held during the test, however, the interface plate on the MirrorSats deformed and this led to excessive displacements and impacts between the deformable mirrors and the rigid mirrors.

As a result, the design of certain aspects of the structure are being re-examined with a view of strengthening the areas where we saw distortion. We are also carrying out more detailed FEA to try to understand the responses we saw in the test.

In summary, the AAReST project has now been running for almost a decade. In that time, a great body of work has been carried out by the teams at Caltech, SSC and IIST. The project has proven to be technologically challenging, but a fruitful source of material for the education of students in our institutions. The international collaboration aspects have worked well, and all parties have benefitted in the coming together of different expertise.

The SQM test failure has set the project back a few months, but we still hope to deliver flight hardware in 2019, and look forward to a launch in the not too distant future.

1) Craig Underwood, Sergio Pellegrino, Hari Priyadarshan, Harsha Simha, Chris Bridges, Ashish Goel, Thibaud Talon, Antonio Pedivellano, Christophe Leclerc, Yuchen Wei, Fabien Royer, Serena Ferraro, Maria Sakovsky, Michael Marshall, Kathryn Jackson, Charles Sommer, Aravind Vaidhyanathan, Sooraj Vijayakumari Surendran Nair, John Baker, "AAReST Autonomous Assembly Reconfigurable Space Telescope Flight Demonstrator," Proceedings of the 69th IAC (International Astronautical Congress) Bremen, Germany, 1-5 October 2018, paper: IAC-18-B4.2.7, URL:

2) Craig Underwood, Sergio Pellegrino, Vaios J. Lappas, Chris Bridges, John Baker, "Using CubeSat/Micro-Satellite Technology to Demonstrate the Autonomous Assembly of a Reconfigurable Space Telescope (AAReST)," Proceedings of the 65th International Astronautical Congress (IAC 2014), Toronto, Canada, Sept. 29-Oct. 3, 2014, paper: IAC-14.B4.2.4

3) Craig Underwood, Vaios J. Lappas, Chris Bridges, "AAReST Spacecraft DDR (Detailed Design Review): Spacecraft Bus, Propulsion, RDV/Docking and Precision ADCS," 2014, URL:

4) Craig Underwood, Sergio Pellegrino, "Autonomous Assembly of a Reconfigurable Space Tele-scope (AAReST) for Astronomy and Earth Observation," 2011, URL:

5) Craig Underwood, Sergio Pellegrino, Vaios Lappas, Chris Bridges, Ben Taylor, Savan Chhaniyara, Theodoros Theodorou, Peter Shaw, Manan Arya, James Breckinridge, Kristina Hogstrom, Keith D. Patterson, John Steeves, Lee Wilson, Nadjim Horr, "Autonomous Assembly of a Reconfiguarble Space Telescope (AAReST) – A CubeSat/Microsatellite Based Technology Demonstrator," Proceedings of the 27th AIAA/USU Conference, Small Satellite Constellations, Logan, Utah, USA, Aug. 10-15, 2013, paper: SSC13-VI-5, URL:

6) Craig Underwood, Sergio Pellegrino, Ben Taylor, Savan Chhaniyara, Nadjim Horri"Autonomous Assembly of a Reconfigurable Space Telescope (AAReST) Rendezvous and Docking on a 2D Test-bed," 9th IAA Symposium on Small Satellites for Earth Observation Berlin, Germany, April 8 -12, 2013, IAA-B9-0508, URL:

7) Craig Underwood, Sergio Pellegrino, Vaios J. Lappas, Christopher P. Bridges, John Baker, "Using CubeSat/micro-satellite technology to demonstrate the Autonomous Assembly of a Reconfigurable Space Telescope (AAReST)," Acta Astronautica, Volume 114, September–October 2015, Pages 112–122

8) Sergio Pellegrino, "Autonomous Assembly of a Reconfigurable Space Telescope (AAReST)," 2015, URL:

9) Craig Underwood, Chris Bridges, "AAReST Spacecraft Update: Spacecraft Bus, Propulsion, ADCS, SSTL-50 CoreSat, RDV/Docking, OBDH and Comms," 2015, URL: .

10) Craig Underwood, Sergio Pellegrino, Vaios J. Lappas, Christopher P. Bridges, John Baker, "Using CubeSat/micro-satellite technology to demonstrate the Autonomous Assembly of a Reconfigurable Space Telescope (AAReST)," Acta Astronautica, Volume 114, September–October 2015, Pages 112-122,

11) ZEMAX Optical and Illumination Design Software, URL:

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 (

Space Segment    Launch    Telescope Architecture & Payload   Launch    Telescope Payload Design   Electromagnetic Docking System Design    Back to top