[daip] DBCON

[Eric Greisen]

Amy has gotten your data and confirmed your difficulties with it.  We
did the test of, one subarray at a time, replacing your actual data
with model data, imaging that, then UVFIX and then image fixed data.
The models come out smeared with the smearing in a radial direction
increasing with radius.

I have then repeated the test on an old data set of mine (the DDT data
available from the web).  In that case, everything works as it should.
 The 1950 and 2000 images are virtually identical excpet rotated
slighty and (after an EPOSW and OHGEO) vary close to identical.

This makes me wonder about your data.  Have you put more than one
frequency's data together in the one file as if there was only 1
frequency.  This is an acceptable thing to do so long as you do not
run UVFIX.  Each data set keeps its u,v,w in wavelengths at the actual
observe freq and it is those u,v,w used by the sky (or UVMOD) to make
phases.  But UVFIX does not know about the different frequencies,
treats all points as the same frequency and so makes an error on some
but not all points - an error that scales with baseline length.  This
seems to fit what we see with your data.

Any thoughts?

Eric Greisen