[evlatests] Closure Mysteries

[Rick Perley]

    I had a few minutes to take a closer look at the X-band data. 

    In the continuum, we see a consistent ~8% excess flux density on the
baseline 14 x 16 for all IF parallel-hand combinations, after antenna-based
calibration is done.  This is visible on 6 sources.  Five of these are pure
point sources -- one (3C345) has resolution of about 5 to 10 % over the
range of spacings, so is not a reliable closure calibrator. 

    I got a bit of spectral line data -- mode 1A at Xband, on 3C345 only
with 50 MHz BW (16 channels). 
The hope is that the closure error is due to some uncorrected passband
shape factor. 

    Not obviously so.  The passband solutions are remarkably good! 
Indeed, if I covered up the antenna numbers on the plots, even an
expert would have trouble identifying which are the EVLA antennas. 

    Yet, when I apply the bandpass solutions, the 14 x 16 closure
report essentially vanishes.  Antenna 14 shows no closure at all,
while (oddly) 16 report many small (~3%) closure errors.  Remember,
however, that antenna 16 is rather distant from the others, and that
3C345 is notably resolved.  My hunch is that these closure errors
are structure induced, and that there is no real closure problem at all.

    But I still cannot explain the clear closure excess on 14 X 16 in
the continuum data, on the clean point sources.

    This issue needs a bit more attention.