[evlatests] Some K-Band WIDAR results from Thursday

[Rick Perley]

    Michael ran the K-band 'WIDAR' stress test Thursday afternoon, about 
3:30 local time.  The test consists of two northern objects, located 
about 12 degrees apart.  At the time of observation, their elevation 
difference was only 3 degrees. 
    The data were taken in dual polarization mode, 8 subband pairs, with 
250 kHz resolution and 512 channels/spectrum.  The test duration was 30 
minutes. 

    Basic Results:

    1) Amplitude stability is outstanding.   There are no 'zero' 
records, and no evidence for significant low records.    However, 
antenna 27, in all IFs and both polarizations, is very weak -- 
amplitudes low by about a factor of ten.  This is a real SNR loss, not 
just a low gain. 

    2) Phase stability, for a single source, looks o.k.  ***However***, 
we have a large and very peculiar difference in phases between the two 
sources, that is clearly dependent on particular antennas.  For example, 
using antenna 4 (at W01) as reference, antenna 8 (at N01, only ~30 
meters away) shows a clear phase difference between the two sources of 
15 degrees.  The phases for each of the two sources individually are 
excellent, but the plot shows the two are simply displaced.  Each 
antenna has a different offset -- there is no spatial relationship 
whatever.  Some antennas show no differential:  25 and 27 look fine.  
All others have some offset.  If this is an elevation effect, it is 
gross indeed, since the two sources are separated in elevation by only 3 
degrees. 

    3) Sub-band continuity:  The continuity in the bandpasses, in both 
amplitude and phase, is as perfect as can be discerned in the data (to 
~1% or so).  There are no sub-band to sub-band steps in amplitude or 
phase. 

    4) Delay.  A single delay was run (some delays are very large -- 
hundreds of nsec).  Everything looks stable in delay. 

    5) Bandpasses:  All look normal, except subband 1, which has the 
usual huge 'upturn' at the low frequency edge. 

    In summary -- the only worrisome feature is the antenna-dependent 
phase offset between sources.  This feature will make calibration very 
difficult indeed -- particularly so if it is dependent on distance 
between sources. 
    It might we worth repeating this experiment, using the VLA 
correlator and all the antennas, over the same time range utilized for 
this test.  We need to know if this problem is peculiar to the EVLA. 

    Rick