[evlatests] IF1-IF2 Phase troubles -- the bad and the good

[Rick Perley]

    Barry and I ran a number of tests yesterday and today in an attempt 
to understand its cause a bit better. 

    Thursday Tests:

    A few hours spent, cycling amongst the three low frequency bands (L, 
C, X), one minute each.  During this time, a number of changes to the 
system were made:
    a) Tsys was turned off
    b) Tsys was turned on
    c) We went the Flukeset 2
    d) We returned to Flukeset 1
    e) The round-trip phase was turned off
    f) Round-trip phase was turned on
    g) System returned to normal.

    The 'IF1-IF2' problem was rampant throughout on some (but not all) 
VLA antennas.  Antennas that are particularly bad include 3, 6, 11, 12, 
20, and 27.   The others either did not show any problem, or very few 
problems.  Hence, none of the tests above appear to have eliminated the 
problem.  The cause remains mysterious. 

    Point of emphasis:  There is no differential effect between the EVLA 
antennas.  Using antenna 18 as a reference, the R1-R2, and L1-L2 phases 
are steady (or drift slowly -- see below). 

    For this run, a very strong differential phase between the VLA and 
the EVLA was seen.  (In this mode where we change frequency, the array 
phases reset to a new value after change of band). 

   Now, we recall that all, or nearly all, the science data taken in the 
past 2 weeks do not show the IF1-IF2 phase problem described above.  We 
note that nearly all of these observations are taken at a single 
frequency.  This encouraged us to try a test today:

   
    Friday Test:

    The same correlator mode as yesterday (line, mode '4'), but at a 
single frequency -- X-band.  We observed 3C286 for an hour, with 
one-minute durations. 

    The results show *no* significant IF1-IF2 phase changes amongst VLA 
antennas. 
    Referenced to a VLA antenna, the EVLA IF1-IF2 phases slowly wind -- 
40 degrees over the hour.  It seems to be the same for all EVLA antennas 
-- indicating some global array differential. 

    Conclusion/Recommendation

    We have not isolated the cause of the IF1-IF2 phase differential.  
It is certainly a VLA issue.  But it does appear to be triggered by 
changing band.  Hence, my recommendation is that we stop spending 
considerable resources in chasing this down, in favor of recommending to 
our users that they ensure that there is a calibration included with 
every change of band.  I believe this is the advice that we currently 
have given users for using the EVLA antennas.