[evlatests] Some new L-band phenomena ...

[Rick Perley]

    Some recent L-band data taken over the past couple days has shown 
some new, possibly worrisome, effects.  The observations were made of 
the strong calibrator 3C273.  This source is partially resolved, but 
this cannot be used to explain the issues noted below, as the extension 
(14 arcseconds) is much less than the typical resolution we now have at 
L-band (arcminute). 

    1) Large, coherent visibility variations.  All antennas, and all 
subbands are showing factor-of-two variations is visibility amplitudes.  
All channels within all subbands, and all antennas are varying 
together.  The amplitude of the variation (when factored out by antenna) 
is typically a factor of two (varies from 3.1 for antenna 1 to 1.5 for 
some others).  The timescale for this variability is typically ~5 
seconds -- all antennas have the same variability timescale.  There is 
some evidence that these variations do not factor out cleanly (i.e., 
there are 'closure' issues here), but this possibility needs more 
careful judgement. 
    The phases are also varying in concert with these visibility 
changes, but the amplitudes are much less -- perhaps 10 degrees (1/5 
radian). 

    According to Michael, the attenuators should be fixed throughout 
these tests, so this effect acts like power level changes, affecting all 
frequencies and all antennas equally, and coherently -- as if the 
amplifier gains are changing.  It is most unlikely that the source 
itself is varying on these timescales ...
    It would be good to have the autocorrelation spectra for one of 
these runs ...

    2) Non-closing bandpass shapes.  BPASS was run, with a 1-minute 
average, giving good solutions.  But an SPFLG display, with the bandpass 
solutions applied, demonstrate very clearly that we have large, 
sinusoidal variations of large amplitude (factors of many!) on some 
baselines.  The baselines which show this effect strongly are *always* 
from adjacent antennas, and the frequency scale of the oscillations is 
clearly related to the antenna separation.  The oscillations slowly in 
frequency as time progresses...
       3C273 is rather close to the sun now, and I'm hoping that the 
quiet sun will be sufficient to explain this effect.  If not, we'll have 
to start worrying about signal coupling between antennas ...