Last modified: 4 Dec 2014


Converting between CHIP and SKY Coordinates with dmcoords

CIAO 4.12 Science Threads




The information presented in this thread is no longer accurate. In CIAO 4.7, the SIM drift is now automatically accounted for as long as the observation has been processed through chandra_repro or has a version from the archive with ASCDSVER greater than 8.4.2.

This thread will be removed at a later time.

dmcoords does not automatically take into account drifts in the Chandra geometry when calculating sky coordinates. These drifts were negligible near the beginning of the mission, but have grown to the order of 10 arcseconds and so cannot be neglected.

Run this thread if:

you are running dmcoords to convert from sky to chip coordinates or from chip to sky coordinates.

Last Update: 4 Dec 2014 - This thread is no longer applicable as dmcoords now automatically accounts for the SIM drift.


Background Information

Download the sample data: 1838 (ACIS-S, G21.5-09)

unix% download_chandra_obsid 1838 evt2,asol

To connect sky and chip coordinates, we need to take into account not only the celestial pointing direction (RA and Dec) of the telescope optical axis, but also the position of the instrument bench in the frame of the telescope mirror.

  • the motion of the SIM which puts ACIS or HRC in the focus at a particular location
  • the time-variable bending of the telescope tube and other drifts in the Chandra geometry.

This latter correction is measured by the fiducial optical system and is contained in the aspect solution (asol1.fits) files as columns dy and dz, giving shifts in mm of the focal plane relative to the optical axis. These values change very slowly over the course of an observation (about one or two pixels); it is taken into account in the calculation of sky coordinates in the pipeline by acis_process_events.

Because the mean values of dy and dz were also small early in the mission, dmcoords does not by default take them into account. However, five years into the mission the mean value has now grown to the order of 10 arcseconds and so cannot be neglected.

Accounting for the Mean Offsets

To get an accurate chip to sky conversion with dmcoords, one must determine the mean offsets and supply them to the tool. You can do this very simply by providing the aspect solution files in the asolfile parameter.

In some rare cases, there will be more than one aspect solution file (pcad_asol1.fits) for an observation. All the files must be input, either as a list or as a stack. Here we use:

unix% cat pcad_asol1.lis

unix% pset dmcoords asolfile=@pcad_asol1.lis

This allows dmcoords to pick up the correct conversion from the aspect solution.

Verifying the Results

You can check the results by finding a bright source (for good statistics) and using dmstat on the events in the event file. Then confirm that the results agree with those given by dmcoords. Since acis_process_events has always applied the full aspect solution correction, including the dy/dz correction, the chip and sky coordinates for individual photons in the event files are fine - the problem only arises in dmcoords when you are trying to predict positions.

unix% punlearn dmstat
unix% dmstat "acisf01838N003_evt2.fits[sky=circle(4070,4249,14.024608)][cols chip]"
chip(chipx, chipy)[pixel]
    min:        ( 340 350 )           @:        ( 2722 375 )
    max:        ( 400 410 )           @:        ( 2360 2047 )
   mean:        ( 367.3435916 382.13359884 )
  sigma:        ( 12.667743873 12.784088222 )
    sum:        ( 2029206 2110906 )
   good:        ( 5524 5524 )
   null:        ( 0 0 )

unix% dmcoords acisf01838N003_evt2.fits asolfile=@pcad_asol1.lis x=4070 y=4249 option=sky mode=h
unix% pget dmcoords chipx chipy

The discrepancy is small - about 3 pixels in chipx and half a pixel in chipy. Compare this to the results obtained from dmcoords without the aspect solution:

unix% punlearn dmcoords
unix% dmcoords acisf01838N003_evt2.fits x=4070 y=4249 option=sky mode=h
unix% pget dmcoords chipx chipy

The difference is greater than 10 pixels without the aspect solution applied.

This correction is only for the mean SIM drift during an observation. If the observation is very long or if the SIM driftly rapidly, then the mean may not be a good estimate.


30 Nov 2004 updated for CIAO 3.2: new asolfile parameter in dmcoords
09 Dec 2005 reviewed for CIAO 3.3: no changes
01 Dec 2006 reviewed for CIAO 3.4: no changes
08 Jan 2008 updated for CIAO 4.0: filenames and screen output updated for reprocessed data (version N002 event file)
13 Jan 2009 reviewed for CIAO 4.1: no changes
05 Feb 2010 reviewed for CIAO 4.2: no changes
11 Jan 2011 reviewed for CIAO 4.3: no changes
10 Jan 2012 reviewed for CIAO 4.4: no changes
03 Dec 2012 Review for CIAO 4.5; changed statement about frequency of observations with multiple asol files; added note about limitations of mean approximiation.
25 Nov 2013 Review for CIAO 4.6.
04 Dec 2014 This thread is no longer applicable as dmcoords now automatically accounts for the SIM drift.