[evlatests] Effects of T304 attenuator settings on bandpass shapes

[Rick Perley]

     Earlier WIDAR tests have shown that changing the T304's attenuator 
levels causes changes in the bandpass amplitude shape by ~1%.  To 
determine which attenuator is responsible, whether particular settings 
have a more dramatic effect, and to otherwise illuminate the 
characteristics of the problem, Ken and Michael ran a specific 
experiment.   In this, they stepped through five different input and 
five output attenuator settings, spending one minute on each pair.  This 
was done on three antennas (1, 18, 25), the others were fixed.  The 
observations were of 3C84 in one continuous observation.  The attenuator 
pairs which were sampled were (in order):  (0,0), (-2,0), (-1,0), (0,0), 
(1,0), (2,0), (0,0), (0,-2), (0,-1), (0,0), (0,1), (0,2), (0,0).  All 
values are in dB. 

    The data were calibrated and the bandpasses determined for each 
antenna, for each of the paired settings listed above.  The AIPS program 
BPLOT was used to show the *differential* results.  (This means that the 
average of all bandpasses is made and subtracted.  The results displayed 
are the fractional deviations from the average). 

    Results:  Not as simple as had been hoped. 

    1)  The three antennas which had the attenuators changed clearly 
show larger changes in their bandpass shapes.  The maximum deviations 
are ~2.5%.   No commonality nor trends are seen in the bandpass changes. 
    2) The seven antennas with fixed attenuators showed much smaller 
deviations, much less than 1% for nearly all.   Changes that are seen 
are gradual, indicating some physical change over time.  Antenna 19 is 
by far the worst amongst these, with a 1.5% change over 13 minutes.  The 
amplitude pattern has a ~100 MHz periodicity, indicating a standing wave 
set up over ~1.5 meter length. 
    3).  Two of the changing antennas -- 18 and 25 -- showed by far the 
biggest deviations (>2%) with the setting (0,1) and (0,2) -- the output 
attenuator seems to have the biggest effect.   This is also true for 
antenna 1, but the amplitude of the bandpass change for this antenna is 
very much less (~1%). 
    4) For the three antennas with attenuators that changed, the four 
(0,0) settings all showed different bandpasses.  These changes are 
typical for the temporal variation of the seven antennas with unchanging 
attenuators. 

    It thus appears likely that we have two simultaneous effects 
confusing the issue:  A temporal change, operating over timescales of 
tens of minutes, and an attenuator-depending effect.  Both are of 
similar magnitudes -- up to 2% in bandpass amplitude. 

    All the above applies to the amplitudes only.  I'll determine the 
phase effects after running FRING to remove residual delay effects (this 
should keep my computer busy over the lunch hour ...)