A $5 million cubesat is definitely top of the line, but not when it is being developed to perform work similar to that underway on the $600 million Kepler planet-finder mission.

A group at the Massachusetts Institute of Technology (MIT) is developing a cubesat dubbed ExoplanetSat to evaluate whether any Earthlike planets found circling bright, relatively nearby stars have orbits that would permit spectral analysis of their atmospheres.

While the cost of the first planet-finding cubesat taking shape at MIT is high, its developers hope to be able to build enough of them to bring down the unit cost. Restrictions imposed by the tiny space available inside the 3U cubesat—measuring 30 X 10 X 10 cm—limit each to observing one exoplanet, so a “swarm” of “dozens” of spacecraft watching the same number of different stars would be needed, says Sara Seager, a professor of planetary science and professor of physics at MIT who is key to the work.

“Kepler is looking at faint stars that are by definition far away,” Seager says. “We're trying to look at the brightest, nearest stars. The bright stars are spread all around the sky, and that's why a Kepler wouldn't work for all the brighter stars, because Kepler only looks at one patch of the sky. We need one telescope per star.”

While Kepler is a survey instrument that stares at a tiny area of sky and measures the faint flicker that occurs when an orbiting exoplanet moves in front of it, ExoplanetSat will stare at a single star to gather as much data as possible from a transiting planet. Given the tiny change in the amount of light reaching the spacecraft's detector, keeping the light from a target star focused will require pointing accuracy at “the several-arcsecond level” if the noise level in the system is to be low enough to permit meaningful measurements.

Packing all that capability into a spacecraft only 30-cm long will require clever use of hardware—some of it off-the-shelf, as is typical of cubesats—and some really clever software. Gross pointing is achieved with miniature reaction wheels produced for the cubesat market by Maryland Aerospace Inc. of Crofton. These serve to point the spacecraft at the target star with an accuracy of 60-100 arcseconds, Seager says.

Light from the star is collected with a space-hardened off-the-shelf single-lens-reflex camera lens, and passed on to a detector that consists of a single charge-coupled device (CCD) surrounded by an array of several complementary metal oxide semiconductor (CMOS) detectors. Behind the detector plate is a piezoelectric actuator that moves it in the x and y axes (see schematic).

“We have a piezo stage; it's like an x/y control attached to the detector, which is where the focal plane of the telescope is,” Seager says. “And then we move that around and that gets us from that 60 arcseconds down to several arcseconds.”

The hardware is able to move the detector by microns in the two dimensions, but Seager notes that there is nothing new in the general approach of using starlight to guide a telescope.

“It's common,” she says. “That's basically how all telescopes are controlled. We're building upon things that we have done before. We're just trying to do it to a more precise level with smaller equipment.”

So far, the MIT team has tested the spacecraft's precision-pointing function with breadboard hardware on an air-bearing table. The camera and imaging-electronics board are also in hand, and have been tested both in the lab and outside against the night sky.

Although the work kicked off with a little astrobiology funding from NASA, the main financial support has come from Draper Laboratory, an MIT spinoff, and from MIT itself.

“That brought us about halfway in terms of the money spent, because we spent a lot of time on R&D,” says Seager. “We're still looking for more money to finish the project now.”

The MIT team has secured a launch, when the spacecraft is ready, via NASA's Educational Launch of Nanosatellites (ELaNa) program (see p. 44), and has a notional mission design and list of target stars. ELaNa payloads can't choose their orbits, but must follow the primary payload's route to space, so in general ExoplanetSat will go to an equatorial orbit in as low an inclination as possible, with an altitude that avoids the radiation belts to extend the lifetime of the detectors and other electronics.

“The field of exoplanets moves so quickly that by the time we launch, the list of targets will be different,” Seager says.

Those targets will be bright stars identified from the ground as having planetary systems. ExoplanetSat will determine if its target system includes a planet that transits the star, which could allow researchers to determine its size and fitness for study with larger and more expensive spacecraft.

“We ultimately want to do direct imaging from space, but that won't be done with cubesats unless you get them to self-assemble into something much bigger,” says Seager.