UIST – Notes on Thermal Imaging
Overheads for broadband thermal imaging can be significant. Either the 0.06 or 0.12 “/pixel cameras can be used; the smaller pixel scale is more efficient but of course gives a smaller field. The counts from the night sky with the minimum exposure times with the full 1024 x 1024 array using the 0.12″/pix camera are given in the table below. Both 512×512 and 256×256 subarrays are available, although it is expected that the smallest subarray will not be frequently used. Exposure times and overheads for L’ and M’ for each camera are given in the table below. These are preliminary values, which will be updated after more measurements.
a The minimum exposure with the 1024×1024 thermal readout is 0.20 seconds. A 0.20 second readout gives ~8000 ADUS in the L band and ~17000 ADUS in the M band in dry weather. Make sure to observe your standards with the same offset patterns and in the same region of the array, especially in the M band to take care of any non-linearity.
b We have found that the 512×512 readout and the 0.06 camera at L’ is not as stable as the 1024×1024 readout – the background is not always cleanly subtracted. We strongly recommend using the 1024×1024 readout with the 0.06 camera at L’ – efficiency is still quite high.
Photometric Behaviour and Telescope Offsets
We have found that the low-QE lower right quadrant (upper left now) and the central 100-pixel wide area of the array are not always well behaved, in that subtracted pairs of images can show some variation in these regions. Also photometry in these regions can give fainter values compared to positions elsewhere in the field. This is not fully understood, but the problem in the centre may be related to the fact that the array is read from the outer corners into the middle, so that the central region is exposed to sky for longer in a reset-read-read NDSTARE observation which may lead to non-linearity effects. The problem is especially bad at L’ with the 0.06 camera and the 512×512 readout – we recommend not observing in this configuration but instead use the full 1024×1024 readout if using the 0.06 camera at L’. The 512×512 performs reasonably for M’.
To deal with the problem of the lower QE, the template and standard star libraries were set up to use offset patterns that avoid the lower right and centre as far as possible, where the low-QE region was located before flipping. For the nod-8 pattern and the 512×512 subarray this means that we use the left (North) side of the array only, as the telescope aperture places the target slightly left of centre. The array orientation was flipped (up-down) in the cryostat in March 07. To take care of this change in orientation, the offsets in the libraries and the templates will be modified soon.