[evlatests] Test data from Tuesday

[Rick Perley]

    Michael took one hour of L-band data on 3C286, and a 'blank' field 5 
degrees south, on Tuesday afternoon.  

    Most of the problems noted earlier remain with us:

    1) sub-band shapes for 8 LCP, all subands remain crazy -- no longer 
a single sinusoid.  Looks like about 12 cycles. 
    2) 3LCP, subbands 3 and 4 remain sloped, with near zero sensitivity 
at the boundary. 
    3) A huge number of identical zero visibilities noted -- the usual 
0.1% randomly placed, plus a large fraction of antenna-based zeros.  
However, all of these are associated with antennas 3, 4, 8 and 23, and 
only for the first ~7 minutes of so.  After that, all is normal -- 
*except* for 23 LCP, which is identically zero for all subbands 
throughout the entire run. 
    4) We are getting steps in gain, always associated with scan 
boundaries, and (with one exception) only on the LCP side.  All these 
gain changes are of the 1-db variety.  Antennas affected are 4, 24 and 
25.  The single exception is 25RCP -- this has 4 gain changes over the 
run, none of them of the 1 dB type.  I recall having noted this before, 
last month. 
    5) Delays are the same as at C-band (plus/minus a couple of nsec). 

    Despite these issues, the data calibrate beautifully after 
flagging.   Most bandpasses are stable (there are exceptions that I'll 
note, this afternoon).  The famous VLA suckout near 1465 MHz is 
completely absent on antenna 24 -- as it should be. 

    Decent images have been made -- I'll return to this in the afternoon. 

    Michael's attempt to observe a noise field almost succeeded, but he 
failed to note that the calibrator 3C287 is located 5 degrees south of 
3C286!    But this little accident is actually good -- a full-field 
image of the 'blank' field shows a nice response of 3C287, 30 arcminutes 
from the randomly-chosen phase center -- i.e. near the first null.