[daip] DBCON

[Eric Greisen]

Ian Pattison writes:
 > I have a concatenated uv file (B and C VLA data) which I wish to convert
 > from 1950 to 2000 coordinates. UVFIX couldn't seem to handle this, so I
 > SPLIT the data into the individual files which were originally
 > concatenated. UVFIX was then run on each of these to covert from
 > 1950-2000. The files were then joined back together with DBCON.
 > 
 > Looking at a clean map of the new data, point sources become smeared and
 > their peak position shifted slightly (~0.5 to 1 asec), and the effect
 > seems to be greater as a function of distance from the centre. I have put
 > a couple of maps of the same source in the original (1950) data and the
 > new DBCON data (2000). The files are named 12_3GOOD.PS and 12_3BAD.PS and
 > can be found at http://ast.leeds.ac.uk/~ip.

     I could not find these at least with mozilla - it said the two
files did not exist.

 > 
 > I would appreciate any feedback on this, particularly if this effect has 
 > been noticed before. 

I have no bright ideas.  UVFIX depends on the coordinates and the AN
files being correct - it is the only aips task that uses the AN info
in any detail.  Any error there could cause major troubles in the
output of UVFIX that would not be noticed previously.  t is easy to
play with aips headers to suppress a minor rotation and the like -
this would cause trouble as well, trouble that would get larger the
further one was from the phase center.

Precesion over a VLA-size field is simply a shift of central
coordinate and a rotation.  The verb EPOSW should work on the 1950
image, although you will then need to rotate it (with e.g. OGEOM) to
compare with the unrotated 2000 images.

I did fix UVFIX in December 2003 to correct errors affecting the 2nd,
3rd, etc subarray in DBCON'd files.  But it is always possible that
there are still errors in UVFIX - it is not up to e.g. CALC standards.
If you find any, please let us know.

Eric Greisen