Michelle Performance and Use

The estimated image quality delivered to the instrument is given as part of the observing condition constraints.

Mid-IR observing overheads can be significant and must be taken into account when proposing for time on Michelle.

Experience indicates that each new imaging target will incur a configuration overhead of 10 minutes (e.g. slewing, centering, setting the peripheral wavefront sensor for tip-tilt AO correction on a guide star). For setting up a spectroscopic measurement an additional 5 minutes should be allowed for accurate centering on the slit and for switching between imaging and spectroscopic modes.

Once a typical observation with chopping and nodding (i.e, imaging and low-resolution spectroscopy) begins, the time spent actually integrating on the source - the value entered in the observing tool as the "total on-source time" - will be about 25% of the elapsed time. Thus, you specify the actual on-source time for Nod A Chop A + Nod B Chop B (i.e., one beam for each nod position). If chopping on-chip you in fact get double the on-source time that you specified but, because guiding is only possible in one of the chop beams (see here), the quality of one of the two images will be somewhat degraded and unlikely to be useful for science. It is not necessary (or, indeed, possible) to specify the time for a single exposure of the array, or the time (and number of individual exposures) between nods - this is done internally and depends on the instrumental configuration. You only need be concerned with the total on-source time and total elapsed time. The factor of two on-source overheads are due to the chop duty cycle and the time required for the telescope to nod and then stabilise. We hope that this eventually can be reduced to a factor of ~1.5.

In early 2007A a new observing mode (longer frame time and fewer read-resets) for the Qa filter was tested which increased the observing efficiency in this filter by about 40%.

When observing with the echelle and medium resolution gratings, which are performed in stare/nod (as opposed to chop/nod) mode, the observing overhead, after the initial 15 minutes of setup time, is a factor of about 1.5.

In the case of imaging polarimetry the time requested in the observing tool (and the time that should be specified in the integration time calculator) is the total on-source time (this is change from 2007A and 2006B where the time per wave-plate position was used). The exposure time in any individual wave-plate position is 1/4 the total on-source time. The overhead involved in moving the half-wave plate during the observations means that the overall efficiency is generally of order 7.5%, rather than 25% as in regular imaging mode. Thus for a 2 minute on-source time specified in the OT, the actual time spent per wave-plate position will be 0.5 minutes; coupled with the 7.5% efficiency, the total time needed to take the observation will be about 2 minutes/0.075 = 27 minutes for the filters in the N-band window (after acquisition). In 2007A we hope to test a new sequence of waveplate moves which will significantly increase the observing efficiency in this mode.

Currently the maximum chopper throw allowed on Gemini is 15 arcsec, which is considerably smaller than the Michelle field of view. Also, guiding with a peripheral wavefront sensor is presently performed on one side of the chop only, so the quality of the two chopped images will differ significantly and the value of the two unguided images will be marginal except under exceptional circumstances. Two basic chop/nod methods are available:

1. Beam-switching, in which the nod is parallel to and the same amplitude as the chop. If the chop is sufficiently small and the target is also small this will result in a final frame containing three separate images of the target, the central one being guided and the other two unguided, with the central image having been obtained in half of the integration time.
2. Perpendicular nodding, in which the nod is perpendicular to and a similar amplitude as the chop. For a small chop and nod this would result in four images of the target on the frame, two guided and two unguided, with each image corresponding to one-fourth of the integration time.
   

Mid-IR calibrations are discussed in detail on the general mid-IR web pages. A baseline calibration set is taken for each observation. Wavelength calibration is derived from the telluric lines in the science target and/or calibration spectrum.