[evlatests] Referenced pointing troubles ...

[Rick Perley]

    Now that the weather has stabilized, Ken and I felt that a careful 
check to see if the subreflector rotation trick is actually working as 
expected. 

    To do this, we constructed a test last night, observing a northern 
calibrator (1800+784), with the following sequence:

    1) X-Band observation
    2) X-Band referenced pointing determination
    3) X-Band, ref pointing applied
    4) K-Band obs, ref pnt off
    5) K-Band obs, ref pnt on
    6) Q-Band obs, ref pnt off
    7) Q-Band obs, ref pnt on
    8) K-Band obs, ref pnt off
    9) K-Band referenced pointing determination
    10) K-Band obs, ref pnt on
    11) Q-Band obs, ref pnt off
    12) Q-Band referenced pinting determination
    13) Q-Band, ref pnt on. 

    This sequence was run twice. 
    All source observations requested 1.67 seconds integration.  All 
referenced pointing observations requested 10 seconds integration.
    The data were filled as correlation coefficients. 


     Some remarkable, and unexpected results were found:

     1) The first source observation (#1 above) was observed with 10 
seconds integration -- the executor clearly used the integration time 
for the *following* observation. 

    2) The first referenced pointing determination (#2 above) failed, 
despite having more than 3 minutes duration, and having the antennas on 
source.  I suspect something to do with the integration time.  On the 
second round of observing, the referenced pointing worked (and the 
preceding scan was observed with the correct integration). 

    3) At X-band, the amplitudes were stable when referenced pointing 
was not applied.  But when the referenced pointing results were applied 
(whether there were any valid solutions or not), the amplitudes *rose* 
uniformly over the 1.5 minutes of observation, with no sign of a 
flattening.  In both cases (the first one with *no* referenced pointing 
solution, and the second with one), the amplitudes began at the same 
level as the preceding stable observation. 

    All antennas behaved identically.  The rise in amplitude is quite 
large -- 15% in power!!! 

    4) At K-band, a similar effect was seen, but it is not so obvious.   
Some antennas decline, but others show no change in amplitude with the 
referenced pointing applied. 

    5) At Q-band, the effect is nearly invisible -- some antennas drop, 
some antennas don't change, and some rise.  In all cases, the effect is 
very small, a few percent at most.

    We have no idea whatever is going on here.  Most mysterious is why 
the effect is so large at X-band. 

    Another test will be run this afternoon, with a much longer 
duration, to see how long it takes for the gains to stabilize when the 
referenced pointing is applied.