[evlatests] X-band WIDAR report

[Rick Perley]

    Michael took 30 minutes of data on our perfect point source 0217+738 
at Xband.  The results are simply superb:

    After dead-easy calibration (flagging the first 10 or so seconds, 
and flicking away two or three bad records), and inserting the usual 
large delay via CLCOR, a single bandpass solution was derived.  The data 
were then integrated over frequency (channels 50 through 1000), and the 
results self-calibrated, with 1-second solutions.  A few minor issues 
are reported below.

    I then made an image, using all 6 IFs.  The peak/rms ratio is 
38000:1.   In all my years in this business, I've never seen that kind 
of dynamic range utilizing only four antennas, and 30 minutes of data. 

    The amplitude and phase deviations (on a single 100 MHz-wide 
sub-band) from a perfect point are less than 50 milli-degrees, and 
0.1%.  (These are, in essence, the 'closure' errors, which I think are 
entirely due to noise). 

    ISSUES:

    Not all is perfect.  The following were noted:

    1) FRING (used to find the large delay) aborted.  There is no 
combination of adverbs which lets it run.  Since the data are organized 
'backwards' in frequency, I have to suspect this reversal of the 
frequency axis is the problem.  Fortunately, it was clear that the 
delays are in the same (to within 1 nsec) of those determined at C-band, 
so I used these (but negated, to offset the reversal of the frequency 
axis). 

    2) Antenna 4 had two small (10 degrees) phase jumps -- at 'scan 
boundaries'. 

    3) Antennas 3 and 4, in sub-band 6 only, showed a remarkable 
degradation in SNR in the last 10 minutes (SNR was cut in half, 
roughly).  No idea why. 

    4) I attempted to use BLCAL to measure and remove baseline 
('closure') errors.  Although it worked, it's clear that it has no idea 
about the 6 IFs (sub-bands).  It seems the program is hardwired for two 
IFs (i.e., it assumes a VLA-like pair of frequencies). 

    On the basis of these observations, my opinion is that the 'closure 
test' is satisfactorily completed.