The Spectral Energy Distribution (SED) mode applies only to the 32x32 Ge:Ga array, since it requires an offset of the scan mirror that deflects light away from the optical trains for the other arrays. This operating mode provides long-slit, low-resolution (R ~ 15-25) spectra in the far infrared (52 - 97 µm), with a dispersion of 1.70 µm /pix via a reflection grating. The SED optical train illuminates a slit approximately two detector pixels (~20´´) wide and 12 pixels (~2´) long where the full wavelength coverage is available. An inoperative detector module at the blue end of the dispersion of side A restricts the wavelength coverage to only 65 - 97 µm over the last 4 columns of the 16x32 detector array. Only side A data is calibrated for the SED mode.
A standard SED observation provides pairs of data frames between the target position (ON) and a nearby sky position (OFF). The scan mirror is used to chop between the ON and OFF, and an observer can choose a chop throw of +3´, +2´, +1´, or -1´. The observer can request either a pointed observation or an MxN raster map (with M, N between 1 and 100). For a pointed observation (or at each raster position in the case of a raster map), there is a basic set of 6 pairs of ON and OFF frames (plus bracketing frames on the internal stimulator (for tracking detector responsivity variations), of which the first 3 pairs are obtained with the target placed near the center of detector column 10 (hereafter referred to as dither position 1) and the next 3 pairs with the target near the center of column 5 (dither position 2). The observer repeats this basic observing set by a number of cycles (NC) in order to reach the desired S/N ratio. For a mapping observation, the value of NC remains the same for all the raster positions.
Using the same detector and internal stimulators, the SED mode shares some calibration characteristics (e.g., dark current) with the 70 µm imaging mode. As a result, most discussion on the data products and general detector calibration of the 70 µm imaging mode in this handbook applies directly to the SED mode.
SED observations provide pairs of data frames with the target in the slit and at a nearby off position. The scan mirror is used to chop between the target and the off position, and the observer can pick a chop throw of +3´, +2´, +1´, or -1´. An SED observation begins with an image at the off position followed by a stimulator flash frame. The scan mirror brings the target into the slit, and a series of 3 target - sky image pairs is taken with the scan mirror chopping between the two positions. The series ends with a stimulator flash frame following the final sky frame. The telescope then nods to move the target along the slit to a second position, and the above sequence of images, stimulator flashes, and scan mirror motions is repeated. If additional integration time is needed, the sequence is repeated except that the first sky - stim flash pair of frames is omitted. Consequently the AOT provides 6 target - sky image pairs per cycle; see Figure 3.19. The observer specified only the total number of basic observation cycles required; the number of pairs to be taken at one spacecraft setting before offsetting to the other has been optimized, as with the photometry mode.
Figure 3.19: Schematic representation of SED mode; see text. These are real data.
As a result of the on-orbit realities, we made changes to the original SED AOT, such that we avoid side B and the dead part of side A as much as possible. Note that full spectral coverage is limited to only 32x12 because of the dead readout on side A; the spectral coverage for 4 columns of the array is reduced to about 65-95 µm.
One can derive a simple median (or average) image from all the non-stim BCDs on the target, and a similar one from those off the target. The difference between the two coadded images is the sky-subtracted, 2-D image of the target. The PBCD products are created by properly combining the On-target and Off-target meaurements and producing a final mosaic from where a 5 pixel extraction (aperture correction of unity) is carried out and a ASCII table of the final spectrum is obtained.