Since 1995, when astronomers announced the discovery of a planet orbiting the star 51 Pegasi, exoplanets—which orbit stars other than the sun—have been a hot topic. I knew that dedicated amateurs could detect some of these exoplanets, but I thought it required expensive telescopes. Then I stumbled on the website of the KELT-North project at Ohio State University, in Columbus. The project’s astronomers find exoplanets not with a giant telescope but by combining a charge-coupled-device (CCD) detector with a Mamiya-Sekor lens originally designed for high-end cameras. That got me wondering: Might I be able to detect an exoplanet without a telescope or a research-grade CCD detector?
I discovered that one amateur astronomer had already posted online about how he had detected a known exoplanet using a digital single-lens reflex (DSLR) camera outfitted with a telephoto lens. He was able to discern the dip in the brightness of a star as an orbiting planet passed in front of it—a technique known as transit detection.
The exoplanet he chose to go after was a gas giant that belongs to a binary star system variously named HD 189733, HIP 98505, or V452 Vulpeculae, depending on the star catalog. It was the obvious choice because its parent star is relatively bright (although still invisible to the naked eye), and the star drops in apparent brightness during a transit by 2.6 percent, which is a lot as these things go. (Astronomers, who use a logarithmic scale to describe the magnitude of a star’s brightness, would call that a 28-millimagnitude difference.)
So I decided to follow this lead and went shopping for a telephoto lens for my Canon EOS Rebel XS DSLR. With old manual-focus lenses now useless to most photographers, I was able to acquire a 300-millimeter Nikon telephoto lens on eBay for a song (US $92, shipped), along with a Nikon-to-Canon adapter ($17 from Amazon).
Video: David Schneider
The next task was to figure out how to make the camera track a star during long exposures. I could have bought a commercial star tracker, but that would have put me back several hundred dollars. Instead I built a “barn door” tracker—essentially two pieces of plywood hinged together. Aligning the hinge to your hemisphere’s celestial pole allows you to track a star as the plywood “doors” separate at a constant rate.
The Long Shot: Sixty-three light-years away, HD 189733 is too dim to be seen with the naked eye. Finding it required the use of such waypoints as the Dumbbell Nebula (top). Once the star system is targeted, the Earth’s rotation causes the sky’s image to blur (second from top) during long exposures. An Arduino-controlled star tracker (bottom) compensated for this motion (third from top).Photos: David Schneider (4)
To drive the tracker, I pulled some gears out of a defunct inkjet printer, attaching one gear to a stepper motor and the other to a nut screwed onto a gently curved length of threaded rod. Rotating the nut pushes the doors of the tracker apart. The stepper motor is controlled, via a driver board, by an Arduino microprocessor that lets me set the rate at which the doors separate.
Initially, I mounted my tracker on a camera tripod. But I soon abandoned that as being too precarious and built a sturdy wooden platform. The final component of the tracker is a ball head ($18 on Amazon) bolted to the top, which allows me to orient the camera in any direction.
The trickiest step in the operation is getting the camera pointed at the target star. I aim my camera by first eyeballing things and then walking the field of view from star to star. A right-angle viewfinder attachment ($20, used) makes that easier, but it’s still a challenge. Some nights it has taken me 15 minutes or more to get the target star framed.
To take images, I used software that came with my Canon camera. It allows you to adjust the camera settings, take shots, record images directly to your computer, and program a sequence of timed exposures. I also purchased a $14 AC power adapter so that I could run my camera for hours without its battery giving out.
I took test sequences of images of HD 189733 for a few nights, settling on a routine of taking one 50-second exposure per minute. I figured that duration would minimize variations in brightness that come from scintillation—twinkling—and that it would also average over small periodic errors in tracking. With such long-duration exposures, I used a low ISO setting to avoid saturating the camera’s CMOS imaging sensor.
The hardest part of the whole project proved to be waiting for an opportunity to observe the transit of HD 189733’s exoplanet, which takes place once every 2.2 days. That sounds frequent, but transits that occur during daytime or are too close to the horizon are impossible to observe. (The Czech Astronomical Society provides a handy online resource for determining opportunities to observe this and other exoplanet transits.) And of course, I needed clear skies.
Finally, after weeks of waiting, an opportunity came in mid-October. I recorded images for almost 3 hours, beginning about a half-hour before the start of the 108-minute transit. That, I figured, would capture the transition from normal brightness to ever-so-slightly dimmed and back to normal again.
Of course, you can’t just look at the images to see the subtle effects of a transit: There are too many confounding influences, such as changes in the transparency of the atmosphere. And the response of a camera’s imaging sensor is seldom uniform: If the position of the target shifts in the field of view (which is hard to avoid over the course of an evening), the amount of light registered will also change, even if there is no actual change in brightness. To compensate, I used free software called Iris, which allowed me to perform the corrections needed to calculate the brightness of HD 189733, as well as four reference stars.
I loaded the results from Iris into Microsoft Excel to make differential-photometry calculations—that is, comparing HD 189733 with one of the four reference stars to compensate for changes in atmospheric conditions. The scatter in the final results was about the same size as the signal I was attempting to measure, but the general dip in brightness was easy enough to discern nevertheless. The average magnitude of the target star diminished and recovered just as the exoplanet’s transit began and ended.
And the shift in magnitude was very close to, if not precisely, the 28 mmag expected. So it seems my home-brew observatory diddetect an exoplanet—using little more than run-of-the-mill DSLR and a $92 eBay camera lens!
This article was updated on 02 December 2014.