Summary


Introduction

We have recently received some reports of large (> 10 Å) wavelength drifts in DEIMOS science spectra. Such drifts in the spectral direction cause the data reduction software to fail. We have performed several tests to better understand the problem and to try to find a possible explanation for this behavior. This page shows the results of our tests and some operational recommendations to prevent this type of problems.

Tests

Description

Tests consisted of taking sequences of DEIMOS arc spectra at different rotator angles. Arc spectra at each rotator angle were taken only once FCS was tracking. A reference FCS image was taken at the beginning of each test at a rotator angle of -30 degrees, following the standard observational procedure for DEIMOS. Tests were performed for different DEIMOS optical configurations.

The position of the arc lines was measured in each DEIMOS spectrum using the task IMEXAM (command "k") in Pyraf. In some specific cases, we also analyzed the positions of the spots in the FCS images as a function of the rotator angle.

Results

1200G at station 4, 6100 Å, GG445 (2016-01-30)

Figure 1 shows the drifts in the DEIMOS arc spectra (red curve) and the FCS correction (green curve) as a function of rotator angle (POS mode). Note the correction in the FCS detector X axis corresponds to the spectral direction in the DEIMOS detector (see FCS coordinate system). All points in the green curve correspond to FCS in tracking mode. It is important to emphasize that FCS claimed to be tracking before taking the arc spectra with DEIMOS.

Figure 1. Drifts in DEIMOS arc spectra (red curve) and FCS correction (green curve) as a function of rotator angle (POS mode). Drifts and corrections are in detector pixels with respect to the reference images taken at a rotator angle of -30°.

Figure 1 indicates there is a -75 pixels (25 Å with the 1200G grating) wavelength drift in the DEIMOS arcs when the instrument is rotated from the reference orientation (-30°) to 0°. This large drift is basically constant up to a rotator angle of 150°. Then, spectra start drifting back to the original postion. DEIMOS spectra become aligned with the reference spectra at a rotator angle of 180°. Once 180° is past, the wavelength drift starts increasing up to a maximun of +75 pixels at 210°. This offset remains constant up to 300°, when it starts decreasing againg down to zero at a rotator angle of 330° (-30°). It is important to note that the wavelength drift in the DEIMOS spectra was ocurring while FCS was tracking normally without showing any warning.

Figure 2 shows the FCS images corresponding to the same data set as in Fig 1. The same drift as the one detected in the DEIMOS spectra is observed in the FCS images, however, FCS never complained about detecting large offsets.

Figure 2. Offsets in the FCS CCD2 images for different rotator angles corresponding to Fig 1. The reference FCS image was taken at the usual angle of -30°. The green crosshair indicates the position of a nominal FCS spot for all images. It can be clearly seen how the spot bounces back and forth as the instrument rotates. In all these images FCS claimed to be tracking, which is totally puzzling, since there are clearly large offsets between most of the images and the reference image.

The working hypothesis to explain this problem is that the cross-correlation script in the FCS control software is detecting some local maximum instead of the real maximum, which makes it track on the wrong spot. This might be caused by the the FCS spot pattern for this particular optical configuration not having significantly distinct features.

The problem is therefore twofold in this case. On the one hand, the grating 1200G loaded in station #4 produces a very large offset in the spectral direction (amplitude about 50Å). On the other hand, FCS is not capable of detecting such drift. Hence, FCS does not issue any warning or error message.

This problem of FCS not being able to identify large offsets in the wavelength direction has only be observed in the configuration concerned here; 1200G (station #4) with 6100Å central wavelength and the GG445 filter. To prevent this problem from affecting science data, the following operational recommendations should be followed.

830G at station 4, 7750 Å, GG495 (2016-02-12)

Figure 3 shows the drifts in the DEIMOS arc spectra (red curve) and the FCS corrections (green curve) as a function of rotator angle (POS mode). When we compare this this figure with Figure 1, we realize that the situation in this case is completely different. There are still large drifts in the spectral direction, but FCS detects them and warns the observer. When the observer uses the script Fix FCS, the spectral drift is removed. This is why the red curve (DEIMOS arc lamp spectrum line drifts) shows spikes in the -30° to 0° and in the 150° to 210° rotator ranges.

Figure 3. Drifts in DEIMOS arc spectra (red curve) and FCS correction (green curve) as a function of rotator angle (POS mode). Drifts and corrections are in detector pixels with respect to the reference images taken at a rotator angle of -30°. When comparing this figure with Figure 1 note that this plot starts at -30°, i.e. at the rotator angle where the FCS reference was taken.

600ZD at station 4, 7600 Å, GG495 (2016-02-12)

Figure 4 shows the drifts in the DEIMOS arc spectra (red curve) and the FCS correction (green curve) as a function of rotator angle (POS mode). In this case the wavelength drifts are very small (note the vertical scale spans less than one pixel) and FCS can easily correct such drifts.

Figure 4. Drifts in DEIMOS arc spectra (red curve) and FCS correction (green curve) as a function of rotator angle (POS mode). Offsets and corrections are in detector pixels with respect to the reference images taken at a rotator angle of -30°. When comparing this figure with Figure 1 note that this plot starts at -30°, i.e. at the rotator angle where the FCS reference was taken.

Conclusions

Operational Recommendations

Our primary advise to minimize the flexure problems underlying the wavelength drifts reported in this page is to avoid loading the grating 1200G in station number 4. In general we recommend to use the slider station number 3 for the primary grating of any science program.

If for any reason observations need to be performed with the grating 1200G in station #4 and a central wavelength of 6100Å, it will be necessary to run the Fix FCS tool after every slew, before starting the field alignment. This is required to visually inspect if there is a large offset in the FCS images: