In order to characterize the behavior of the DM prior to and in flight, the payload was required to be capable of measuring the response of actuators at or below the level of typical MEMS DM manufacturing error, 10–20 nm root mean squared (RMS) (Evans et al., 2006; Morzinski et al., 2012). This requirement tests absolute actuation in open-loop operation.

The DeMi optical payload is designed to demonstrate closed-loop wavefront control in space, correcting thermally driven aberrations (e.g. focus and astigmatism), as well as spacecraft pointing errors. The DeMi payload is designed to correct both static and dynamic wavefront errors to less than 100 nm RMS. Closed-loop, in-space wavefront control with a MEMS DM is currently at a TRL of 5–6 (Cahoy et al., 2013a; Sirbu et al., 2015; Morgan et al., 2019; Prada et al., 2019), and DeMi is designed to raise this TRL using on-orbit performance data. This requirement tests closed-loop Adaptive Optics (AO) operation.

While many high-contrast mission designs include a Fast Steering Mirror (FSM) upstream of the DM to provide precision pointing correction (e.g. Roberts et al. (2013); Chakrabarti et al. (2015); Demers et al. (2015)), the volume requirements of a CubeSat platform suggest an all-in-one design where the DM provides both pointing and high-order correction. Continuous phase-sheet MEMS DM stroke depends on the displacement of adjacent actuators (Bifano et al., 1997); thus, a conservative margin was applied to determine the stroke required to correct for misalignments as well as spacecraft pointing errors using a Blue Canyon XACT attitude determination and control system (ADCS) which were assumed to be ≲ 10″ RMS (Mason et al., 2016).

In order to illuminate the deformable mirror and place it in a conjugate plane to the input pupil, an image of the input optical pupil must be formed on the DM. The size of this pupil in the system sets the magnification of the astronomical telescope. MEMS DM actuators are ∼300–500 microns across and the actuator stroke depends on the size of the actuators (smaller actuators having shorter stroke due to the increased actuation force required). Smaller actuators sizes also require a higher telescope magnification, increasing optomechanical alignment sensitivity. The telescope entrance aperture size and magnification was set to underfill the mirror by 10% to allow for payload shifts. The magnification and entrance aperture diameter were also constrained by alignment tolerances, limited by the ability of commercially available M2.5 x 0.20 adjustment screws to set the system to within <λ/4 wavefront error (as determined by Zemax ray-tracing), and ensure the stroke was not limited by edge actuators.

The pixel size of the cameras determined the accuracy of centroiding for the Shack-Hartmann Wavefront Sensor (SHWFS) (for basics of operation, see Tyson and Frazier (2004)) and the sensitivity of the image plane wavefront sensor. The DeMi concept of operations does not include on-orbit flat fielding, and radiation is expected to change the pixel responsivity; thus, photo response non-uniformity (PRNU) sets a lower limit on the noiseless SHWFS accuracy. To verify that the SHWFS system met wavefront sensitivity requirements, PRNU variation up to and exceeding the 5 kRad doses tested by Becker et al. (2008) were simulated with the realistic lenslet point spread functions (PSFs) (Generated in POPPY Perrin et al. (2016); Morgan (2020).). The center of mass of these PSFs were calculated as a function of increasing PRNU to show <0.1 pixel centroiding would be achieved even after irradiation and without flat-fielding (Douglas et al., 2017).

In-order to ensure mission success, emphasis was placed on redundancy at every level of the payload design, from optical paths to electronic subsystems: DeMi has two radio transmitters, two computers, two cameras, and two light sources (stars and an internal laser). The system was also required to allow reprogramming of the payload software while on orbit to test new control algorithms and fix any bugs that were not discovered on the ground.

In selecting a DM for the DeMi mission, a desire to maximize the number of actuators and improve the statistical power of the experiment was balanced against limitations in pupil size and controller electronics SWaP. A Boston Micromachines Multi-DM with 140 actuators and rated actuator stroke of 5.5 μm was selected for the mission. Actuator stroke is rated for 255 V and for four actuators together in piston. The mirror stroke, S was selected to provide sufficient tip-tilt wavefront control to correct for spacecraft pointing errors of <10″ after derating the stroke by the 40 degree-from-normal angle of incidence on the DM and the controller maximum voltage of 150 V (Morgan, 2019). Thus, the maximum tilt correction was calculated as (2 cos 40 deg S / D , where the factor of two is included because stroke is surface motion, not wavefront tilt. This predicted correction up to 11″, which provided little margin on 10″ pointing error. Thus, to further mitigate spacecraft pointing errors, offloading of pointing errors at 1 Hz or greater to the XACT was a requirement levied on the spacecraft by the payload (Figure 1). This is consistent with the slow, thermally driven behavior of large pointing excursions which as been observed on the XACT (Pong, 2018; Knapp et al., 2020).

The optical path is shown in Figure 2. A largely aluminum optical train was designed in order to minimize chromatic aberrations and coefficient of thermal expansion (CTE) mismatches. Reflective optics were all polished aluminum with SiO₂ protective coatings, and directly bolted to the 7075-alloy aluminum structure. Custom diamond turned off-axis parabola (OAP) mirrors, with focal lengths and bolt patterns matching commercial off-the-shelf (COTS) parts, allowed for flexibility and the rapid construction of a fully functional optical test bed and optically functional 3D printed models as described in Gubner (2018). As an approximation for launch loads, the optical train was designed to survive 100 g static loading. The optomechanical and thermal design of the payload are detailed in Allan et al. (2018); do Vale Pereira, (2020). Figure 4 shows the engineering model, which was built to flight-specifications including surface finishes and optical components, and was fully integrated with a functional deformable mirror and other sensors. The engineering model was used for mechanical and fit-testing, software development, and as a source of flight-spare components. The flight unit, Figure 4, was slightly refined to raise the first structural mode, as detailed in do Vale Pereira, (2020).

Pixel size was a key design driver to enable precision wavefront sensing in the CubeSat volume. A PixeLink CMOS PL-D775MU-BL MONO 1/2.5 in, with a monochromatic ON Semiconductor MT9P031 CMOS sensor, was selected for both the Shack Hartmann wavefront sensor and the image plane wavefront sensor. The MT9P031 has been radiation tested (Becker et al., 2008) and has extensive flight heritage as part of the Sinclair Interplanetary Star Tracker (Enright et al., 2011; Sinclair et al., 2015).

The MT9P031 has 2.2 micron pixels and this pixel size set the imaging lens (IL) focal length, since two pixels per λ/D resolution spot were desired to allow coarse image plane wavefront sensing (Codona, 2013; Allan, 2018). Each sensor is read out by a separate camera board connected to each of the two cross-strapped Raspberry Pi 3 modules that serve as payload computers. Each Raspberry Pi 3 contains two identical SD cards that can be used interchangeably in the event of radiation damage impacting memory or flight software.

The imaging system needed for PSF measurement provides a model of a typical in-space instrument whose operation would benefit from AO technology. Closed-loop wavefront control on a broadband stellar source fully demonstrates the feasibility of AO with MEMS DMs in space. The requirement also highlights potential usefulness of actively corrected telescopes like DeMi as tools for science, as accurate pointing correction on nanosatellites enables photometric observation of transiting exoplanets, as in (Knapp et al., 2020).

The optical system was aligned end-to-end using a Zygo Verifire interferometer. The optical payload was aligned in segments, and end-to-end system tests with the calibration laser inserted in the field mirror were conducted periodically throughout alignment to ensure the optical fiber laser signal was kept inside the imager FOV and within the SHWFS capture range. The end-to-end system was aligned to <0.25λ wavefront error with the DM in the unpowered position (all actuators commanded to 0 V). As described in do Vale Pereira, (2020), an engineering model of the payload was vibration tested. As seen in Figure 4, the aluminum optics were wrapped in polyimide tape to increase emissivity and thermal gap filler was added to minimize temperature gradients across the payload. he M1 mirror diameter reduced and the flight structure was also modified to increase stiffness, as described in do Vale Pereira, (2020). The aligned flight model of the payload was integrated into the spacecraft bus and was subjected to standard CubeSat thermal and mechanical cycling (described in do Vale Pereira, (2020)) prior to delivery to the launch services provider in December 2019.

DeMi was launched to the ISS on February 15, 2020 aboard the Cygnus NG-13 flight of Northrop Grumman from NASA’s Wallops Flight Facility. DeMi was deployed on July 13, 2020 as shown in Figure 5. The overall flow of signals through DeMi is shown in Figure 1. The upper right box is the payload subsystem described here, which passes data to and receives data from the spacecraft bus (lower left), which provides communication to ground segment at MIT and NASA Wallops. Initial checkout data from August 4, 2020 shows the DM illuminated by the internal laser diode and measured with the Shack Hartmann wavefront sensor. A reconstruction of this wavefront measurement compared to similar data from ground testing is shown in Figure 3. From October 8, 2020 UHF ground station mechanical issues at NASA Wallops Flight Facility prevented science data downlinks; however, a low-data-rate ground station located at the MIT campus was been used to monitor spacecraft health and downlink a small set of payload data. Following to Wallops return to service on March 22, 2021, DeMi commissioning has resumed and will be the subject of future publications.

Planned deformable mirror commisioning tests include: application and testing of pre-recorded Zernikes, image-plane wavefront sensing, and repeating the actuator poke test illustrated in Figure 3 for all actuators to monitor stability and function over time. Figure 3 shows a single actuator set to 60 V referenced to flat map. 60 V approximately corresponds to one wave of wavefront error at the test laser wavelength. The left panel shows a ground measurement and the right panel shows the flight measurement. The difference between the space and ground measurements of the same actuator is shown in the center bottom panel and has an RMS of 21 nm. This reported error includes all lenslets across the SHWFS, some of which are poorly illuminated, and contribute to excess error. For details on the accuracy and precision with which all actuators are calibrated and measured see Morgan et al. (2021).