[evlatests] 3C273 images

[rperley at nrao.edu]

Michael pointed me to three observations made of 3C273, a simple double
source.  I analyzed the C-band one -- 40 minutes long.  Data quality very
good, except for antenna 3, which gave no fringes.

In this configuration, we get about 3 arcsecond resolution at X-band.  The
source is an 18" double,  so even 40 minutes observing with 10 antennas
should easily see the basic structure.  The jet of the double is heavily
resolved, so calibration consists of using the long baselines only to
obtain phases and gains.

After doing this, an image was made -- a terrible image.  The unresolved
nuclear source is present, but there is no hint at all of the secondary --
just masses of lumps extending in all directions.

Plotting the amplitude and phases against uv-distance shows the
visibilities oscillating in the expected manner.  However, the expected
patterns -- the largest oscillations must be on the shortest spacings, the
least on the longest -- is entirely absent.  There is no overall pattern
in the variations as there should be.

Could it be that the (correct) visibilities are being assigned to the
wrong baselines?  Such a scrambling would explain why no image, other than
of a pure point source, gives a sensible result.  Something like this must
be going on ...