The DEIMOS internal focus can change significantly on a daily basis. Hence, it is advisable to refocus the spectrograph each afternoon. Since the offsets between various filters appear to be stable over time, the standard procedure is to focus the instrument in direct imaging mode using the R filter and dome illumination. The focus sequence can also be run using the internal lamps, when the telescope cannot be pointed at the dome flat position, e.g. during adverse weather at the summit. Once the focus in this fiducial filter is measured and set, the instrument's automated filter focus discipline software will adjust the internal focus as needed to account for the changes in the filter.
The focus sequence consists of the following two steps:
Prior to January 2020, the focus analysis was done using an IDL-based script running on an old Solaris machine. On January 2020, we developed a new Python-based focus analysis script on a Linux system. The new script is ~15 times faster than the old one. This helps optimizing the DEIMOS afternoon checkout procedure.
This page shows the results of a comparison between the focus maps obtained with the new Python-based and the old IDL-based scripts.
The following figure (click on the figure to expand it) shows the pinhole centoids for the GOH_X mask in detector pixel coordinates. The focus images were taken on January 21, 2020. The centroids calculated using the Python-based script are represented by red plus symbols and the centroids from the IDL-based script are represented by blue circles.
The following plot shows the best-focus stage value for all GOH_X mask pinholes as a function of detector row position. This graph was created using the focus sequence obtained on January 21, 2020. The plus symbols represent the best focus from the Python-based script and the blue circles represent the values from the IDL-based script. The best-focus stage value is, on average, slightly smaller when obtained with the new Python script than with the old IDL script.
The next figure shows the best-focus stage value as a function of X and Y position on the detector. Red plus symbols correspond to the Python script and blue circles represent the values obtained with the IDL script.
The following plot shows a comparison between the measured focus curve (FWHM in pixels vs. focusing stage value) on one of the pinholes and the corresponding parabolic fit. Note that a change in +/-100 counts on the focusing stage about the minimum of the parabola changes the FWHM of the PSF by only 0.003 pixels, i.e 0.064%.
The following plot shows the focusing stage best-focus value using the old IDL script (circles) and the new Python script (triangles) in the period between June 2019 and January 2020. Different colors represent different CCDs on the detector mosaic. The dashed black line represents the average of all four CCDs using the IDL method. The solid black line represents the average of all four CCDs using the Python method. The dataset is a mixture of focus values obtained with the dome flat lamps and the internal lamps.
The following plot shows the difference on the best-focus value using the old IDL and the new Python script in the period between June 2019 and January 2020. Different colors represent different CCDs on the detector mosaic. The dashed black line represents the average difference including all four CCDs.
One of the advantages of the new Python-based script, in addition to being ~15 times faster than the IDL-based script, is that it is more robust. The following figure shows a focus map obtained on November 22, 2019 using the IDL-based script.
In contrast, the same data set as on the previous figure analyzed using the new Python script produces a nominal focus map (see next figure). We found several instances between June 2019 and January 2020 when the old IDL script failed to generate a correct focus map, while the new Python script generated a nominal focus map.
In this section, we analyze the difference between using the internal calibration lamps and dome flat lamps to determine the best DEIMOS instrument focus. In order to do the analysis, we run focus sequences with both, dome flat lamps and internal lamps, on 9 different nights in June 2020.
The following plot show the comparison between best-focus value obtained with the internal lamps and with the dome lamps as a function of date. Different colors represent different CCDs on the detector mosaic. The black solid line represents the average best focus for data taken with the internal calibration lamps. The dashed solid line represents the best focus for data taken with the dome flat lamps.
The following plot shows the difference between the best-focus value obtained with the internal lamps and with the dome lamps as a function of date. Different colors represent different CCDs on the detector mosaic. The dashed black line represents the average difference including all four CCDs.
The main take aways from the previous two plots are: