Planning and preparing RSS MOS observations is somewhat more involved than other modes. On the other hand, once the principles have been understood, it can be very quick to define your observations, so first-timers should not be put-off by it.

Observers should endeavor to prepare high-quality masks and submit their phase II proposal early. A typical turnaround time from submitting a mask and getting it cut on a carbon-fibre bland in Cape Town, then shipped to and installed at the telescope is two to three weeks.


Please note the following when preparing and submitting MOS masks and blocks.

  1. Planning MOS:  Be sure to read the MOS section of  the call for proposals page. For your science requirements notice in particular the difference compared to Long-Slit mode in that most of the slits are not at the optical axis – this results in both wavelength range and S/R differences (see also FAQ 3 further down this page.)
  2. Preparing masks: Please use the RSMT program for defining your masks. Slits and reference stars maybe selected interactively by centroiding on targets, or you may upload target lists. You may alternatively want to try the new tool that has improved capabilities to automatically optimise fields from larger input target catalogues. Please contact if you’d like to try this new tool.
  3. Position Angle and coordinates of the Mask: It is absolutely crucial you remember to the check PA constraints at your target coordinates  (see this document and a figure reproduced below, where you should attempt to use the dark blue regions).
  4. Inside the RSMT the coordinate and PA information is given in the tab called Slitmask Header; position angle was labelled “rotation angle” in older versions of the RSMT (always use the newest version). Note also that whatever you define as the PA, RA and DEC inside the RSMT will be used for the telescope pointing, the PIPT will not allow you to submit if there is a inconsistency. Buttons in the MOS mode RSS config panel of PIPT will allow you to easily just transfer mask coordinates and PA to target pointing information. Finally, do not confuse slit tilts  with the PA of the whole mask! The first are defined per slit in the Slit Mask Content tab of the RSMT and will allow you to produce arbitrarily tilted slits (making your reductions more difficult).
  5. Reference stars: We are currently using 5″x5″ boxes for the reference star holes in the masks, the PI cannot define or modify this on the RSMT. The stars are currently just shown as small circles there. However, you should be aware of the size of the hole that is actually cut so the spectra don’t overlap with neighbouring slits. Experience has shown that the best results come from using 4 to 8 reference stars with magnitudes in the range 15 – 18 mag. Additional guidelines regarding reference stars may be found further down this page.
  6. Finder charts: Please use the online tool for generating your FC. You will need the mask XML file, which you can obtain as follows.
    1. Change the file extension of your RSMT file from .rsmt to .zip.
    2. Unzip the file.
    3. The file called Slitmask.xml is the file you need.

    These FCs are the most useful for field and reference stars identification at the telescope and will help speed up the acquisition process. You may submit additional FCs if you so wish. Secondary FCs are recommended in cases where the field is very crowded, for example a globular cluster field. If there is any ambiguity of what the reference star is in the default DSS view, do supply a higher spatial resolution secondary FC. But otherwise, just pointing out your science targets in the slits is really irrelevant because we concentrate on the reference stars during the acquisition stage. If you do make secondary FCs, make them in the conventional N-up, E-left orientation, indicate the scale, and mark clearly the locations of the reference stars, and the slits.

  7. Astrometry: It is the PIs responsibility to make sure the astrometry of the field used to define the masks is correct and of high quality. If your astrometry is good to 0.3″ we should be able to align the field with that level. It is crucial that your targets and reference stars come from the same WCS solution.
  8. Slit Lengths: We recommend using slits with lengths of at least 10″, even longer if you have space, for a decent sky removal.
  9. Slit Widths: We do not recommend using slit widths smaller than 1.5″. This is mostly due to non-negligible rotation drift, which we cannot model yet. Assuming the telescope guide star is at the edge of the field-of-view, at the other edge one can expect drifts up to ~0.8″ during a 1h track, which should be added to the ~0.3″ rms uncertainty of alignment. Southern azimuths, DEC < -55 or so, are much more stable than others, 45 mins without significant rotation has been seen. For other Azimuths, the above values still apply.
  10. Mask realignment: Re-alignment during a track will take ~5-7 minutes, in case the rotation drift mentioned above is a concern. The best way to define this in the current phase-2 is by defining 2 (or 3) separate configs under the same Block. The PIPT should charge you 5 mins for it. However, in addition, you must tell the observer in the Block comments and the notes for SA section that you wish to do a re-alignment. A new version of PIPT might have tick-box for this?
  11. Observing very faint targets with MOS:  General SALT and RSS status throughput issues apply (check the PDF document available from the call for proposals page), as well the fact that we still do not have an edge-sensing system for keeping the primary mirror alignment stable. The PSFs will hence vary from observation to observation. Adding to these the above mentioned rotation drift, and small ~0.3″ level alignment inaccuracies, it should be clear that stacking of several MOS observations taken at different times and conditions will be extremely difficult.Do NOT simply use the RSS simulator for an observation of 12000s and assume that you will get the same SNR with 6 x 2000s separate blocks, you must be much more conservative. How much more? We do not know exactly. We would be delighted if some PIs gave feedback how the stacking of multiple observations of same masks has worked for them, there are data around already. Having said that, if you do plan to stack several observations, it is imperative to have some brighter targets well spread out over the FOV to attempt some de-shifting of the frames in case there are drifts and misalignments.


I want to read in slit data into the RSMT. What file format is required?

Please see this example file. The file must be called Slitmask.xml. Zip this file and a  JPEG image of the mask (called Slitmask.jpg).

I created an RSMT file with the beta version of the new Slit Mask Tool, but I cannot import it into the PIPT. There even is no error message. What can I do?

This is caused by some incompatibilities between the Slit Mask Tool and the PIPT, soon to be fixed. But the following steps should resolve your problem:

  1. Change the suffix of your RSMT file from .rsmt to .zip.
  2. Unzip the file.
  3. If unzipping created a new folder, go into that folder.
  4. Convert the PNG image (Slitmask.png) into a JPEG image, and make sure that the JPEG file is called SlitmaskImage.jpg (not the “Image” in the name).
  5. Delete the PNG image.
  6. Zip the files. (The files must be at the top level in the zip file, so make sure you don’t zip their parent folder.)
  7. Change the suffix of the zipped file from .zip to .rsmt.

You should now be able to import the new file into the PIPT. If that didn’t work, please contact, attaching the RSMT file to the email.

I am currently planning a Phase I observing proposal for SALT for a MOS observation. As part of this I need to work out the amount of observing time that will be required and am using the RSS simulator tool. However, this does not allow the calculation of S/N for a slitmask. Would I be correct in assuming that if I set the simulator to longslit mode, then this would tell me the signal-to-noise I would expect from the MOS mode using the same sized slit?

The S/N will actually be a function of position over the field of view. It will be a maximum at the centre and drop (non-symmetrically) at off axis angles along the dispersion direction (left/right). So you can use the results from the simulator tool for long slit as a maximum S/N value. The RSS observer’s guide discusses this on page 10. There is also an RSS simulator tool (though unfortunately it does not include the pg0300 grating) which can be found here. This produces an efficiency plot for off axis sources which will help you estimate how your S/N will drop for your off axis sources


Some of these notes were kindly provided by Eric Hooper, University of Wisconsin

Three setup stars is the absolute minimum, but picking between 5 and 7 is strongly recommended for the following reasons:

Some setup stars can prove to be unusable:

  • They fall too close to, or inside, the chip gaps. The RSMT shows the chip gaps, but currently the fiducial size is too narrow, triple the width at least.
  • They are too close to other stars in crowded fields, or they turn out to be binary stars.
  • The astrometry of some just turn out not to good enough, with more options it will be easier to spot the clear outliers and ignore them at the fine-alignment step

A very large number of setup stars will likely be counterproductive just because real estate on the slit mask will be lost to science targets, and too many stars (>10) will slow down the identification and centroiding procedure performed by the SA.

We have found the optimal magnitude range for reference stars to be in the magnitude range of SDSS r’ ~ 14 – 18 mag. Stars fainter than this can be centroided, but it just takes more time. We typically use 3 to 5 second exposures during alignment, and the said mag range works well. To see fainter stars we need to rack up the exposure time, but then other stars might start saturating, or at least filling the 5″x5″ box and get harder to centroid. Don’t use setup stars that are saturated in the image used to design the slit masks in the first place, as the coordinates for these stars will be less accurate. Some observers have had trouble finding a sufficient number of sufficiently bright, yet unsaturated, stars when using extremely deep images from other large telescopes to design their slit masks.

Pick stars distributed widely about the field, but either avoid the outer edge of the field, or pick brighter stars there. Avoid the gaps between the CCDs. Avoid picking setup stars that have other objects, such as stars or galaxies, near (<5-10″) them, especially if these other objects have comparable brightness to the setup star. Finally, do remember that placing slits or reference holes in the same horizontal region will cause their spectra to overlap!