November 20, 2012

The MMT has always been famous for exceptionally good pointing. The current pointing model (from mount/src/telrel.c) yields 1.15 arc-second RMS residuals, and this figure is often quoted as "the MMT points to within 1 arc-second".

See some notes on MMT pointing from 2001.

The pointing model is used by the mount software to make an adjustment to alt-az coordinates after all other corrections have been applied. (A reasonable question would be how well the MMT would point if the pointing model was turned off). Our present model was determined in May of 2007 and consists of 8 terms as follows:

           coeff        change   value     sigma

      1     IA         -0.026  +1191.16    1.289
      2     IE         -0.385     -1.81    0.304
      3     NPAE       -0.060     -4.23    1.533
      4     CA         +0.049    +16.71    1.878
      5     AN         +0.019     -1.32    0.103
      6     AW         +0.002    -13.25    0.103
      7     TF         +3.111     +7.65    0.889
      8     TX         -0.829     -2.80    0.291
Units are in arc-seconds.

This model was determined from a 146 star pointing run done the night of 5-31-2007. Every indication is that our pointing is very good at this time and that there is no urgent need to do another pointing run and update the model. The terms are:

Tpoint software

We reduce pointing data and fit a pointing model using the "Tpoint" software. (See the TPOINT software home page). Tpoint is a program written by Patrick Wallace at the Anglo American Telescope and has become somewhat of a standard tool to analyze telescope pointing. The MMT has owned several copies of the software. An ancient copy predates my time and was used at the original MMT. Tom Trebisky purchased a copy sometime in 1999. Tim Pickering obtained a more up to date copy in October of 2008. The copy we currently have is version 12.3, supplied and licensed to the mmt in October of 2008 and runs on linux systems.

There is a 131 page PDF manual for Tpoint, but you will need to purchase a copy of tpoint if you want it.

We keep all of the pointing related files on hacksaw:/mmt/pointing. The tpoint software itself, manuals, and every pointing run that we have ever done should be found here.


The "elcoll" scheme stores a table of corrections (as a function of elevation) that are used to adjust the secondary using the hexapod. There is an analgous "tempfoc" scheme that stores a table that is indexed by temperature and used to adjust the secondary.

Pointing and "elcoll" interact with one another. Originally we ignored elcoll (actually once upon a time there was no elcoll), and let the tpoint model account for every aspect of telescope pointing. Later we introduced "elcoll" (and "tempfoc") and correct for some aspects of telescope pointing (and elevation dependent collimation issues) using the secondary. Tpoint then takes care of any "left-overs" after elcoll has done what it does (these comments really only pertain to the TF and TX terms in the tpoint model).

A pointing run

The following instructions have been extracted from a comment in the scope GUI - Note that pointing data is now obtained via socket from the mount crate via the network "remote" protocol, then written to disk by the "scope" script. This means that scope must be run on a computer with /mmt mounted to give access to /mmt/pointing. It also means that the old business of defining CATALOGPATH in the mount startup script is obsolete and bogus.
 How to do a pointing run:
 1 - the following directory must be writeable on the machine whereon
     this script is run, by whoever is running this script.  The usual
     "deal" is to just make the directory world writeable.
	(presently the directory is /mmt/pointing/newdata)
 2 - use the Pointing menu bar thingie and click start.
     Three new buttons appear below the sky view map labelled
     Skip, Save, and Next
 3 - The next button runs a raster pattern on the sky and picks
     stars from the catalog that are near raster locations.
     Since the earth turns and all of this takes time, it is
     possible for the same star to be used more than once.
     The next start does it all, selects the star and commands
     the telescope to go there.
 4 - Once the telescope has slewed to a star, you are expected
     to paddle the star onto the rotator center, then hit Save.
 5 - If some star is hopless (let's say you are doing this with
     clouds or some such), hit Skip and go on to the next raster point.
 6 - The current raster is defined in this script by pnt_alt_min and
     so on, at this time from 25 to 65 degrees elevation in 20 degree steps,
     Azimuth steps in 30 degree increments from 0 to 350.
 7 - You could kind of man-handle this and pick your own stars by
     doing the "start" thing to give you the pointing buttons and then
     whatever method you like to pick stars, use the Save button to add
     entries to the file.  Just kill the script when you are done.
     Save always opens the file and appends information to it, and
     closes it for every star.
A pointing run uses a catalog of well known stars, and selects stars in all parts of the sky to calibrate telescope pointing. Any existing pointing model can be active during a pointing run (and will be a big help, unless it is totally screwy). The operator commands the telescope to move to a new star, paddles the star to the rotator center (which they will locate on screen prior to starting the run), and then saves the data to a file. The data saved will have the coordinates prior to any pointing correction and paddling, as well as the coordinates after paddling. This allows a model to be calculated which will replace the one in force, rather than a model which increments the one being used.