[evlatests] Last night's wideband data

[Rick Perley]

    I was asked to review the L-band observation, taken yesterday 
(Friday) afternoon. 

    The setup was the standard 8 subbands spanning the L-band window.  
All subbands were adjacent, and 128 MHz wide, with the center (between 
subbands 4 and 5) at 1500.0 MHz.   The first six minutes were spent 
observing 3C84, the rest of the time (1 hour, 34 minutes) on 3C273. 
    I'll comment on these choices later. 

    Perhaps the most notable fact is the size of the database, and how 
long it took to get it into AIPS:  the UVFITS file is 48 GB.  The total 
time taken to get it from the archive was 9 hours (which includes all 
necessary transfers and conversions). 

    Specific notes:

    * FITLD reports 5.1% of the data were flagged, and a minuscule 0.15% 
of the data, post-flagging, that were identically zero. 
    * Antennas 6 and 10 were out of the array. 

    * Antennas 17, 20 and 26 gave no fringes, but were not flagged.  (No 
receivers in any of these, I believe).  I guess we're still not using 
the power levels to trigger flags. 

    * Antenna 3 gave no fringes in LCP.  RCP fringed well -- but see my 
note below about gain changes. 

    *  Delays were larger on some antenna-IFs:

    4A   -10 nsec
    4C     46 nsec
    2C     14 nsec
    3C     15 nsec
    24R   -17 nsec

    All others were less than 5 nsec (w.r.t. antenna 1). 

    * Some '1-dB' gain changes were seen -- is not the 'set and 
remember' mode turned on? 
        Antenna 3, in LCP, changed its gain (amplitude) by 1 dB every 20 
seconds throughout the entire observation of 3C84.  How can this 
happen?  What is especially interesting here is that the phase changed 
by 5 degrees in perfect synchronism with the amplitude changes. 
        Antenna 24, in RCP had a pair of 1 dB changes.
        Antenna 5, in LCP had a pair of 1 dB changes. 

    * All other amplitude gains were very, very stable, except 28 in 
LCP, which is 'ratty'.  (Hard to describe -- I have a plot for those 
interested). 

    *  There were all manner of startup issues.  The first unflagged 
data appeared at 0:10:32.  Many antenna-IFs showed remarkable changes 
over the next 30 seconds or so -- although most in fact were stable from 
the start.  Antenna 1, in LCP, had a number of gain changes (looking 
like 1dB steps) until 0:11:18 -- almost a minute after good data first 
appeared. 

    *  After these initial problems died away, the quality of the data 
appears extremely good (other than the gain changes). 

    The observations made above come from the initial observation of 
3C84.  The volume of data on 3C273 has (so far) prevented any careful 
inspections.