[evlatests] Results from WIDAR K-band test

[Rick Perley]

    Michael took some K-band test data on the late afternoon of Dec 3.  
Twelve antennas were operating, observing two sources separated by about 
6 degrees.  Dual polarization (RR,LL) was utilized, four 128 MHz-wide 
subbands, with 512 channels for each polarization product -- 250 kHz 
resolution.   A total of 30 minutes observing, with 1 second 
averaging.   Scans were 90 seconds long, alternating between the two 
strong (2.3, and 1.6 Jy) sources. 

    In general, data quality was very good.  Exceptions are:

    1) Perfectly-zero records are back.  About 0.1% of the visibilities 
are exactly zeros. 

    2) About half the antennas had a notable phase jump between the 
first and second scans.  Those seemingly affected are 2, 3, 19, 25, and 
27.  Perhaps antenna 3 as well.  I flagged the first scan for all 
antennas, to simplify matters...

    3) Antenna 27 was very weak throughout -- amplitudes down by a 
factor of 3 (so it's like we had a constant offset, to the 10 dB power 
point).  I flagged this antenna out.

    4) Antenna 19 was bizarre throughout:  three subbands in LCP (2, 3, 
4) had ridiculous bandpass shapes -- I flagged all three.  All 
polarizations had phase jumps for the first 6 minutes.  Following this, 
the amplitudes all dropped by about 50%.  Amplitudes and phases were 
stable thereafter. I removed the first 6 minutes, and crossed by fingers ...

    5) Phase stability was very poor, with smoothly changing phases, by 
up to 100 degrees, throughout.  The amplitudes of these phase variations 
clearly increases with baseline length, and there is good evidence that 
the phases connect between sources, and the variations are identical 
between polarizations and subbands -- all are consistent with an 
atmospheric origin.  I hope the weather was truly bad on that evening ...

    I calibrated the data with a flat 6-hour average throughout.  After 
removing the issues noted above, the stability in the bandpasses, and 
between sources is very good.   There is nothing to indicate any 
problems (but -- the sources are not really strong, and the scans are 
short, so sensitivity is not great).  The differential bandpass 
solutions (solutions made for each observations for each source) is dead 
flat -- only noise is seen on all antennas. 

    I ran POSSM, to check the connectivity of the subbands, for each 
baseline, for each source, for each scan.  All amplitudes and phases 
connect smoothly -- there are no subband (aka spectral window) 
discontinuities at a level of ~1 percent.  The LCP window of subband 2 
is occasionally displaced by a couple of percent w.r.t. the adjoining 
windows (subbands), but this is seen equally for both sources, and I bet 
it's caused by something fishy with the bandpass solutions.