Phoenix has been out of service for the past six months to correct a number of problems that have been present since first light. These included astigmatism in the collimator, high slit losses, and balky, slow performance of the mechanisms. We are pleased to report that all the instrumental problems have been solved. Overall system throughput including slit losses in the H band is now 10%, a high efficiency for an echelle spectrograph and an improvement of a factor of three from the original instrument performance.
The astigmatism in the collimator was traced to a warped collimator mirror mount. The problem was solved by installing three mounting pads made from a piece of aluminum foil! The width of the spectrum on the array is now the result of image scaling and seeing. Using the 2 pixel slit, a best FWHM resolution of 70,000 is possible. This resolution is consistent with optical models of the collimator.
The infrared imaging and spectroscopy modes now share a common focus. At the focus determined from infrared imaging, the spectrograph has both highest throughput and best spatial focus. Slit losses are those expected from slit geometry. In July the 4-m was producing 0.7" images in the near IR and the spectrograph throughput including slit losses at 1.6 µm was 10%. We find the best throughput is possible on the 4-m using the off-axis guide probes on faint guide stars. Because of size and weight restrictions on the 2.1-m, Phoenix cannot be used there with the off-axis guider.
The maximum time to convert from spectroscopy to imaging modes is now 20s. The new Ethernet controlled mechanisms allow the slit and viewer wheels to operate simultaneously and much faster than previously possible. A number of small improvements also were made to the individual mechanisms resulting in much improved reliability.
While a number of spectra have been taken at both 3.3 µm and 4.6 µm, a complete set of calibration exposures has not been possible due to bad weather during both the evaluation runs this spring. As a result the limiting magnitudes in the thermal infrared are approximate. Also in the thermal infrared it is important to remember that the limiting magnitude depends largely on background radiation and thus on weather conditions.
The detector in Phoenix is a two-quadrant section of 1024 × 1024 Aladdin array manufactured in the original foundry run. While the array is cosmetically nearly perfect and has an average dark current of less than 1e-/s, the array has a large number of pixels with high dark current. Perhaps 5% of the pixels have dark current significantly higher than the mean. In long exposure images, high dark current pixels in the area of the array with the signal can significantly degrade the signal-to-noise of the extracted spectrum. We find that high dark current pixels dominate the noise for all but the brightest stars and significantly degrade the resulting spectra for objects fainter than 7-8th magnitude. The figure for limiting magnitude as a function of wavelength is based on count rates ignoring these bad pixels. In principle it should be possible to remove some of the bad pixels with software mapping. However, the bad pixels may be so numerous that this will not result in spectra of the quality that should be possible from the optical performance of the spectrograph.
Detailed plans for sharing Phoenix between Gemini, CTIO, and Kitt Peak are being discussed. As announced previously in this Newletter, Phoenix is scheduled to be at Kitt Peak only through 2000A.