Instrument Set-Up (Staff Only)
The alignment of MOS or SIS involves three separate adjustments:
A properly aligned MOS or SIS means that:
- The mask slide, and the LAMA machine.
- The grisms in the grism wheel.
- The CCD.
We can note that since MOS and SIS share the same set of grism wheels,
once the grisms are aligned for SIS, the grism wheel can be installed
on MOS. The MOS alignment will only consist of a CCD rotation until
spectra run along the CCD columns. There is then no need to align
grisms on MOS, if they were properly aligned on SIS.
- The LAMA machine XY coordinate frame is aligned with the
CCD XY coordinate frame. This ensures that slits run parallel
to CCD rows.
- The spectra dispersion for each grism is running along CCD
columns, the blue part of the spectra being usually at small CCD
Y, the red part at large CCD Y.
Of course, this is not valid for grisms installed in the MOS/ARGUS
wheel, since the orientation here is at 90° to that expected in
MOS or SIS.
To properly align MOS, proceed as follows:
Once MOS is aligned, and if you need to align SIS:
- Insert a mask with the reference cross in the entrance focal
- Select the proper raster to isolate only one hole of the cross,
with a full size raster in Y.
- Adjust the MOS camera focus to get an approximate focus (accurate
focusing is done later, see "Spectrograph Focusing" below).
- Select a raster (vertical strip) to isolate one arm of the
- Take an image of the cross (V filter + Ar lamp = 1s). Use
"align" or "graph" to cut the image on the two extreme hole
images. Measure the angle of rotation that needs to be applied to the
- The rotation of the CCD is done in two steps: coarse rotation by
unlocking the "dogs" if a rotation of more than ±2.5° is
needed, or fine rotation by using the black rotation knobs on the side
of the CCD mounting plate after releasing the locking screws around
the plate. One turn of the black knobs will rotate the CCD by
~0.4°. If the CCD is rotated clockwise, the image of the cross
rotates clockwise on SAOIMAGE with Loral3. The alignment should be
done to better than 1 pixel over 2000.
- Insert a grism.
- Select a raster (vertical strip) on one of the holes of the cross
(off the vertical arm).
- Take an image (comparison) with the halogen image lamp (approx
- If the grism needs to be aligned, two setting screws are available
on each grism cell for fine rotation. To access these screws, the
grism wheel needs to be driven to the position exactly opposite to the
grism being adjusted. One turn of the set screws will rotate a grism
by ~0.78°. An accuracy of alignment better than 1 pixel over 2000
is required. Spectra rotate couterclockwise on SAOIMAGE (with Loral3)
if the grism is rotated clockwise as seen from the CCD toward the
- Align all grisms.
- Remove the grism wheel from MOS and install on SIS.
- Drive a grism into position. Since the alignment was done on MOS,
all grisms are aligned with respect to each other. Any grism will
therefore provide a reference to the dispersion direction which can be
used to accurately rotate the CCD for SIS alignment.
- Take a spectrum and use "align" to cut the spectrum at two extreme
positions. Compute the rotation angle that needs to be applied to the
- Rotate the CCD as described for MOS.
The internal focusing of the spectrograph is done by moving the MOS
camera focus. Each filter in the filter wheel will have a specific
camera focus; this is due to the fact that the filters are not
To focus the camera, proceed as follows for each filter:
Note that the values are significantly different for different CCDs
(for instance, close to 1000 for Loral3 and close to 3000 for
- Insert a mask with a reference pinhole in the entrance focal
- Select a 200x200 subraster centered on the pinhole at 1024x1024
and take a sequence of images by moving the camera focus in the MOS
form (starting with 50 unit steps, then finalize with smaller
steps). Evaluate the image quality with IQE. The best focus image
should be round and with X and Y image quality close
- For reference, the spectrograph intrinsic image quality is around
- When the internal focus has been determined for all filters, save
the values entered in the MOS form.
Once satisfied with the internal focusing, the zero point for the
Computer Aided Focus is ready to be computed. CAF is normally set with
a zero point which indicates the difference delta Y between the two
images produced by the Hartmann focusing device to be taken out of the
CAF computation when the telescope is exactly in focus. Inversely,
when you know that the spectrograph is in focus, (such as after
focusing the camera) setting zero point = 0 will force CAF to
compute delta Y, hence giving the zero point. This comes from the
linear formula which transforms delta Y into telescope encoder motion:
Proceed as follows:
- Select the grid or pinhole as an entrance mask.
- Select CAF, enter the proper exposure time (1s with one Argon lamp
on) and enter "0" for the CAF zero point. Select "accept" to start a
CAF sequence. The CAF program outputs a telescope focus motion T in
telescope encoder units.
- Enter the output value from the CAF progam into CAF as , this value being the separation delta
Y (in microns) between images when the telescope is exactly in
focus. Select "save values" to retain the zero point value for later
use of CAF.
- Verification: running CAF again with this setup should give a
suggested value for telescope encoder motion very close to 0 (no more
than ±2-3 units).
The fine offset calibration is done on the sky with the "autocal"
tool. Over several runs we found that a new calibration was not needed
for each new run, provided that the alignment procedure as defined
above is thoroughly executed. However, if you need to calibrate the
fine offset, proceed as follows:
- Select a field with a bright enough star for short (10s)
exposures. Ask the TO to find a guide star and guide with the star
close to center of the guide TV.
- Define the exposure time in "expose".
- Start the "autocal" sequence, with deplacement steps of 3mm. You
need to take 3 exposures according to the following pattern:
- The first exposure is taken and the image displayed. Select the
chosen star with the cursor in SAOIMAGE and record its position.
- The bonnette is moved one step in X. You then have to
drive the guide star back to the guide box by moving the telescope (it
is imperative that during the whole sequence the guide box is left at
the same XY TV coordinates).
- A second exposure is taken. Select the same star as previously
with the cursor.
- The bonnette is moved one step in Y. You then have to drive the
guide star back to the guide box by moving the telescope.
- A last exposure is taken. You have to select the same star as
previously on SAOIMAGE.
The calibration values are then entered into the respective "offset"
fields. The "autocal" program maps telescope coordinates versus CCD
coordinates and allows very accurate offsetting.
Checking the LAser MAchine:
The set-up procedures should include the following:
- cutting of a test mask.
- measurement of cutting accuracy:
- relative positions should be accurate to better than 1-2µm
over a full MOS mask.
- slit edge roughness should be on the order of 2-3µm
Send comments to: firstname.lastname@example.org
Copyright © 1997, CFHT Corporation. All rights reserved.