[evlatests] Bandpass stability test

[Rick Perley]

    The 2-hour long track on 3C147, taken last week to monitor the 
behavior of the switched power system, proves useful as a check on 
bandpass stability.  The observation was in OSRO1 mode -- two 
128-MHz-wide blocks, with 2 MHz frequency resolution, at 4500 and 6500 
MHz. 

    I made a normalized bandpass solution with a 5 minute average, 
without smoothing.  The AIPS program 'BPLOT' shows the residuals about 
the mean of the 24 individual solution.  These residuals are very low, 
and for nearly all antennas, appear limited by thermal noise.  The rms 
deviation about the mean is certainly less than 0.05% (in amplitude). 

    Some antennas/IFs don't behave as well as others.  The rogue's 
gallery includes:

    Antenna 4, IF B:  a narrow feature, near channel 22, which changes 
by ~ 0.5% over the 2 hours.  
    Antenna 6, in both IFs B and D (6500 MHz) -- is notably less stable 
-- perhaps 2X worse. 
    Antenna 10, in both IFs B and D -- much less stable (5 X worse), 
with oscillating variation of period ~70 MHz. 
    Antenna 11:  similar to antenna 10, but not as severe.
    Antenna 14:  Similar ~70 MHz changing frequency structure, slowly 
changing throughout, on all IFs. 
    Many (perhaps most) antennas show greater noise and/or changing 
frequency structure on scales of tens of Mhz in IF 2 (6500 MHz). 
    A special case:  Antenna 27, IF 'B' shows a spectacular (~1%) change 
in amplitudes, in three distinct stages.  The bandpass is stable within 
each 'stage'.  This is unquestionably due to a change in T304 
attenuation, as the switched power, total power, and fringe power all 
changed in two steps. 
   
    I have the plots for perusal.