On 19/20 October 1995 we used GRIM II to attempt spectroscopy of the 2.3 micron CO absorption band in stellar clusters. We used the f/10 optics and slit. On the whole, things worked well, although there are some peculiarities to be overcome. Here are some comments that may help other observers. Alan Watson & Jon Holtzman Throughput We obtained a peak signal of 500 DN/s/pixel (28 DN/s/A) and a total signal of 2500 DN/s/pixel (140 DN/s/A) from a K = 6.5 star under conditions of clear skies and 1.2ish arcsec seeing. Columns 128 and 256 Columns 128 and 256 seem to be offset down by one row. Read Noise Our typical backgrounds in 60s were about 100 DN (about 500 e), and so for faint objects our noise is utterly dominated by the read noise of about 110 e. We can think of three ways to improve this situation: integrate longer; reduce the read noise; or implement multiple reads. However, since we would have to integrate for 1500s for the background noise to equal the read noise, reducing the effective read noise seems very desirable. Implementing a quadruple read scheme or halving the read noise, neither of which are at all unreasonable goals, would increase the efficiency of spectroscopy by a factor of four. Bias Fluctuations In our object-sky subtractions we saw banding parallel to the rows in almost every exposure. The pattern of the banding was similar in each quadrant, suggesting fluctuations in the bias. About 20% of our exposures had banding with a peak-to-peak magnitude of about 100 DN in the low rows of each quadrant. In the remainder, the banding had a peak-to-peak magnitude of 15-30 DN and more variable structure. Since our primary targets (K about 12.5) had only about 100 DN peak signal in 60s, this banding was a very serious concern. We were able to significantly reduce the level of the banding by averaging portions of rows well away from the spectrum to determine a bias level for each row. Since the spectral dispersion is not perfectly parallel to the columns of the detector, this will only work well if the residuals from sky lines in object-sky subtractions are small. For stellar objects, a background aperture close to the object aperture may work well. Do people find this banding a problem in low background (i.e., narrow band) imaging? Flats The quartz lamps were too dim for flats. We used the incandescent lights which gave a few thousand DN in 10s. Wavelength Calibration Wavelength calibration of CO band spectroscopy is difficult because of the paucity of emission lines in the 2.3 micron region in each of the three standard sources of emission lines: planetary nebulae, HII regions, and the night sky. We attempted to obtain an Ar spectrum using the conventional set up: lamps shining onto the enclosure from behind the secondary. We detected four lines in a 300s lamps on/off pair, with peaks of about 100 DN in the brightest two. At the suggestion of Dan Long, we obtained much better spectra by removing the instrument from the telescope and placing a lamp on a step ladder directly in front of the dewar window; we obtained good signal on about a dozen lines spread across the K window in exposures shorter than 1 second. In retrospect, we think it might be possible to do this without removing the instrument from the telescope by pointing the telescope close to the horizon and placing the lamp between the instrument and the tertiary. Bruce and Karen are investigating Xe and Kr lamps, as these may turn out to have brighter lines and be suitable for use with the conventional set up. We can supply line lists for OH, Ar, Xe, and Kr. APO APO APO APO APO Apache Point Observatory 3.5m APO APO APO APO APO This is message 2 in the apo35-grim archive. You can find APO the archive on http://astro.princeton.edu:82/apo35-grim/INDEX.html APO To join/leave the list, send mail to apo35-request@astro.princeton.edu APO To post a message, mail it to apo35-grim@astro.princeton.edu APO APO APO APO APO APO APO APO APO APO APO APO APO APO APO APO APO