[evlatests] VLA follow-up of WIDAR oddities

[Michael Rupen]

I made two observations with the old VLA correlator yesterday, aimed at
understanding the current WIDAR-0 problems.

1- 30min C band continuum observation of 3C84 with 0.488 second dumps
   The main goal was to look for amplitude wobbles.

   * There is no evidence of the amplitude wobbles seen with WIDAR-0.
     In particular, the noise on baselines 18-19 and 19-28 looks similar
     to that on other baselines, in contrast to WIDAR-0, which sees
     a factor 2-3 larger noise on 19-28 always, and a somewhat smaller
     but significant and obvious increase for 18-19.

   * There are no phase jumps during this run.  Note however that I
     used quite long scans (10-ish minutes), so this is not conclusive.

2- 1.5hrs C band spectral line (25 MHz per IF, mode 4 --> 7 channels per IF)
   observation of 0217+738 with 3.3 second dumps; again long (~10 min) scans.

   The main goal here was to confirm that 0217+738 is a point source at 5 GHz
   in the current configuration.

   * 0217+738 is 4.2 Jy (by comparison with 3C48 = 0137+331)

   * The 0217 images show substantial crud associated with the source,
     reminiscent of but not identical to what is seen with WIDAR-0.
     Quantitatively the rms for various options is as follows:

                                                 rms
        Nant        IFs     Stokes   Weighting  [mJy/b]  4.22 Jy/rms
        27        AC + BD     I         N0        0.37   11350:1
        27        AC + BD     V         N0        0.31  [11460:1]
        27           AC       V         N0        0.37  [11350:1]
        27           A        RR        N0        0.55    7650:1

        27        AC + BD     I         R0        0.41   10120:1

        10        AC + BD     I         N0        0.76    5580:1
        10        AC + BD     V         N0        0.53   [7930:1]
        10           AC       V         N0        0.79   [5310:1]
        10           A        RR        N0        1.45    2910:1

      For comparison, WIDAR-0 on 8jun09 using my reduction gave:

        W0-10        A        RR        N0        0.56    7450:1    [8jun09]
        W0-10        A        RR        R0        0.24   17860:1    [8jun09]

     after adjusting the flux density scale to match the 11jun09 VLA data.


     In these tables:
       Nant=  27 --> all antennas, calibrated and imaged together
              10 --> WIDAR-0 antennas (1 2 3 9 18 19 23 24 25 28),
                     calibrated and imaged as a group (all other
                     antennas flagged)
           W0-10 --> WIDAR-0 antennas as above, but actually using
                     WIDAR-0 rather than the old correlator

       Weighting= N0 --> natural weighting, setting all vis.wts. to 1.0
                  R0 --> robust weighting, robust= 0.0

   * There are no obvious confusing sources, out to 20 arcminutes from
     the phase center.  The errors seem directly associated with
     the source itself (i.e., the errors are much larger near the source,
     and are centered on the source), in agreement with previous findings
     that blank fields with WIDAR-0 look noise-like.

   * I tried looking at closure errors directly in AIPS, and believe
     there are more non-zero closures than I would expect.

   The Calibration Manual claims 0217+738 is a perfectly good point source
   for all arrays and at all frequencies.

3- Upcoming VLA work
   Ken worries that 25 MHz/mode 4 may have troubles; I'll try taking some
   standard continuum data to eliminate that particular worry.

   I will also start offsetting the calibrator from the phase center,
   unless anyone objects.  This seems basic good practice, and as Bob
   points out, may eliminate certain possible origins for the problems.

   I'd like to try X rather than C band, to be sure there's nothing
   magical about 5 GHz.

   Finally, I should try looking for phase jumps using very short scans
   with the VLA correlator, to be sure this is purely a WIDAR problem.


Should anyone else wish to look at these data, they may be found at:
   /home/imager-a/mrupen/vla11jun09_0319cont.cband
   /home/imager-a/mrupen/vla11jun09_0217line.cband
These FITS files include various calibration and flagging tables; I suggest
you retain the flags and re-do the calibration so you're sure you know
what's going on.  Note that FG/2 is the same as FG/1 except for additionally
flagging all non-WIDAR-0 antennas.

   Cheers,

                            Michael