Telescope Goes "Semi-Automatic"

Andrew Howard is a postdoctoral fellow at Harvard working on the Optical SETI telescope. He is responsible for designing and building much of the electronic hardware that went into the telescope's camera and control mechanism, and he also conducts some of the observations. He updates us on the progress to date:

So far we've made 176 observations spanning 89 hours over 16 nights. These observations have covered about 5% of the full sky, and about 7.5% of the Northern sky. We've had 14 triggers (in 5 episodes), but no events. We define a "trigger" to be any time that the instrument detects coincident electrical pulses during an observation. Not all triggers are genuine flashes. The subset of triggers that pass all of our tests of what a real signal should look like are called "events." So far we have not seen any of these during observations. (We see then all the time during testing.) All 14 triggers were detected in bursts (episodes) on spatially separated pixels and are therefore not classified as events. An instrumental noise source (such as corona breakdown) seems a much more likely explanation for two or more signals simultaneously received from widely separated parts of the sky.

It's probably more accurate to call our observations "semi-automated." When the experiment first launched, observations were totally manual, and were done out at Oak Ridge. The observer had to push lots of buttons on the web interface to move the roof and telescope, and program the various electronics in the camera. We had a checklist to make sure that nothing was forgotten. Now, with the semi-automated observations, large groups of tasks in the checklist are encapsulated in macro functions. The observer just has to click on buttons labeled "Ready Observatory," "Power On Electronics," and "Program Clock/DACs," and hundreds of things happen. The other important macro function is "Semi-automated Observations," which causes the instrument to make a set number of observations, with testing and reprogramming in between, and then automatically lower the telescope, close the roof, and turn off the electronics. After installing a daylight sensor that automatically closes the roof (and pushes down the telescope, if it's up) in an emergency, we gained enough confidence to control the experiment from Cambridge over the web. The observer can start the observations from home, go to bed, and check the results in the morning.

Our goal for fully automated observations is to remove the observer completely. Currently the observer has to verify that the observing conditions are adequate, start up the experiment, verify that the instrument is functioning, and tell it how many observations to make until dawn or the weather is likely to be bad. To this end, we just bought an all-sky CCD camera that we'll use to look for clouds. That should be installed in a week or so. We're also in the process of writing software for the on-site weather station. With these additions, the control computer will have all of the environmental data necessary for automated observations. We just have to program in the smarts so that it can automatically verify that the instrument is functioning properly and safely make observations during good conditions.