Instructions for Collecting Pointing Data and Building a Pointing Model
The old instructions for making a pointing model are found here
. These are being updated and moved here, but should problems be encountered please refer to the old instructions.
- make POINTINGSTARS callable from scripts (non-interactive) - may be done already
- automate data collection on each instrument if possible.
- automate TPOINT first-pass data analysis if possible.
- Make a set of pointing+collimation stars from UCAC3
- Use the new set of fainter stars (can then combine pointing and collimation)
Understanding what is needed:
Any set of pointing data needs a few basic things to be successful:
- Telescope logfile (snapshot of telescope pointing)
- Image with known orientation, traceable WCS, accurate timestamp
- Valid measurement of where the rotator center lies on the image
If you are missing either of the first two, then you cannot make a reasonable pointing model. For the second point, we use the centroid of the pointing star to define the origin of the WCS (CRPIX/CRVAL) and the orientation to define the CD matrix. From this WCS you can then easily find the RADEC coordinates on sky of the rotator center even if the absolute pointing is off. The timestamp is used to associate the RADEC of the rotator center from the image with a corresponding snapshot from the telescope (ALTAZ of the mount plus some other information). This is the input data needed to generate a pointing model.
- TELESCOPE IN THERMAL EQUILIBRIUM!
- Be sure that azimuth and elevation encoders have been properly reset.
- Re-measure mechanical rotator center coords on the instrument if it or its AGW has been dismounted/remounted.
Rotator Center Measurement:
- Take a pair of TRACK mode images rotated 180 degrees with respect to each other.
- The new rotator center is the averaged mean coordinates of matched pairs of source coordinates. That is, the rotator center is at the midpoint between the two images of the same source, averaged over all available sources.
- Reasonably dense fields work best, so that there are many matched source pairs.
- Turn on pointing logging in PCSGUI and record the filename
- For future reference: PCS is being configured to generate two pointing logs.
- They will be stored in /home/telescope/PointingLogs
- The files will be automatically named _SX and _DX.
- Multiple re-starts of the pointing logging will append to these files
- Use Pointingstars5 routine to determine star list and sequence:
- In your working directory on an observer workstation:
cp ~LBTO/POINTING/* . to copy program files
- Be sure to run prepdir in the working directory for the LBTtools.
- PointingStars5 is called by gopoint.cl script
- If Pointingstars5 returns a 999, skip this star and try again later
- In iraf, run the script gopoint (JohnSpecial)
- lpar gpoint to set agwNum, exptime, Nimages, active, gdelay, ploop.
- If you specified ploop+, this script runs until you ^C it, otherwise it does one star.
- Specifying a logfile makes an easy way to record star numbers and times.
- You can repeat or return to a specific point in the sequence by specifying the new gridID.
- Continue with gopoint until you have enough stars.
- If obtaining data for the LBC pointing models, gopoint is not the task. Instead:
- select and run the OB to slew to the pointing star indicated by the Pointingstars5 routine.
- input this pair of images to lbcrangebal (in LBTtools.Observe). This will adjust the pointing model and mirror positions to correct pointing and co-pointing.
- once mirrors and mount have stopped moving, run the OB again.
- move on to the next star and repeat steps.
- Stop pointing log in PCSGUI
Data Analysis (in progress)
- Measure pointing star centroids if the data were not collected in GUIDE or ACTIVE mode.
- Generate new pointing model (D.Thompson)
- Copy new model to ~telescope/TCS/Configuration/PCS/
- Update LUCIFER entry for LEFTPOINTINGMODEL in PCSInstrument.conf
- Re-authorize LUCIFER on IIF GUI to read in new pointing model
- Test new pointing model by pointing to a few stars or taking an additional set of pointing data.
Pointing Model Tips from Dave
Make sure to start the pointing log on PCS when taking pointing model data!
It is OK to correct the pointing with IE and CA during the run as the
pointing log uses the mount AZEL directly. If you need to tweak the
collimation, it is probably OK to do it all with M1, but I typically only
need to manually tweak the focus a little if things are stable.
A "minimum" dataset consists of the first 12 cells, from which I can make
a decent pointing model (<10 arcsec rms). The next 16 fill in
between the first 12, then the next 24 fill in more between
the first 28, and the final 36 complete the pattern. The full cell list has 88 positions,
so if you take pointing data, take sets of 12, 28, 52, or 88 points. If you are diligent, you can
take ~25 data points per hour with LUCIFER.
More details on Marco's pointing star program
Marco has set up a script for selecting optimally distributed pointing
stars. It lives in /home/LBTO/POINTING
You run it by copying it to a local scratch directory and saying: ./Pointingstars5
The program waits for you to input the "cell" number, just start with 1 and increment each step.
For example, I just ran it and it returns:
66 WT10_355 03 45 27.557 +47 39 37.25 0.00210 -.02700
The "1" is what you type (cell number), the next line is the star info
(cell number, ID, RADEC, pmRADEC). The program has the full WT10 and ACT
catalogs in it, it will give you the star closest to the middle of the
requested cell. You then do a "pointto WT10_355" and get the star on the
detector (acquisition camera is OK) with a spiral search if necessary.
Once there, take a short exposure and note the image number so I can pull
the UT out of the image header to match to the pointing log file. Take all of
the data with the detector at a PA of zero.
If there is no star, the program returns cell number 99 (or 999) and the
coordinates for the center of the cell. Just skip it and try again later.
Check the AZ wrap on the first star, the cells are distributed such that if
you are at the end of one wrap at the start you will not have to unwrap
during the observations, although it is not a problem for the pointing
model if you unwrap in the middle.
- 07 Apr 2010