IR Channel Performance

In July 1999, the Ohio State team installed the IR array, a 1024x1024 HgCdTe HAWAII array from Rockwell. This permitted a checkout of the IR channel end-to-end. The internal mechanisms (dither/focus mirror and filter wheel) have been thoroughly exercised and appear quite reliable. The dither/focus motion in particular functions very well; the motions are accurate and repeatable, which should allow for straightforward assembly of dithered images. The IR channel is sufficiently parfocal with the optical filters that no internal refocus amongst the possible filter combinations is required (that is, the IR and CCD channels are "locked" together in focus, so only the telescope focus should ever be adjusted).

The array has very low read noise (~11 electrons) and is flat to ~20% (most of the variation is a gradient towards the right-hand-side or north edge of the array). There is a group of about 300 dead pixels and ~50 scattered dead/hot pixels over the array. The field of view is about 200" x 200" with a pixel scale of 0.2"/pixel. The system throughput is about as expected, suggesting the array has reasonable quantum efficiency for devices of this type. Measurements of a grid of standards spanning all four detector quadrants and ~15% of the area of the array show very good reproducibility (<1% rms in the photometry). These tests were limited by weather, and a larger grid covering more of the detector is planned.

Sensitivity of the infrared channel is as expected. A web page to to assist observation planning is under construction. A rough guide is that H=13 is easy (high S/N in a few minutes), H=15 is possible (high S/N in 30-40 minutes or so), H=17 is very tough (poor S/N in an hour), and anything fainter should be done elsewhere. Numbers at J & K are similar (a bit higher at J; a bit lower at K).

The JHK filters were made to follow the CIT system. Filter transmission curves have been determined by the OSU team. Note that the sky rates vary considerably (by factors of 2 or so) depending on conditions (temperature, humidity, cloud cover, etc.). Any exposure longer than 10 sec or so is over the read noise. Use 30-60 sec for K exposures. The telescope does not track well in open-loop. It is suggested that a guide star be found for exposures longer than 60 sec.

Preliminary use of the instrument over the first few nights suggests that only occasional sky measurements are required, which should improve observing efficiency. Optimal observing strategies (exposure times, co-adds, dither spacing, etc.) will be defined during July and August. All observing is done in queue-scheduled mode by a telescope operator/observer.

The dither pattern is done in groups of 7. The dither/focus mirror is held in place by tension against three actuators. The dither pattern starts at the original position. The next three dithers are made by moving each screw (while leaving the other two screws at 0) the requested number of "scale" steps (dither positions 2, 3, & 4). The final three dither positions are made by simultaneously moving two of the screws by the requestd "scale" steps (positions 5, 6, & 7). After exposures the screws are returned to their original position so motions are not cumulative. Because of the locations of the actuators on the mirror and the mirror's placement in the optical beam, one of the actuators moves a star more than the other two. All this together results in an asymmetrical pattern centered on the original position. If more than 7 dither positions are requested, the pattern is just repeated and the dithers go to the same postions. If you want to have unique dither positions, you must break your observations up into groups of 7, and move the telescope between each group by a fraction of the dither step.

The dither steps are not in arcseconds but in encoder units. The encoder units are almost exactly twice the number of arseconds moved. Thus a requested dither of 30 means that the dither move is about 15 arcseconds.