[evlatests] 16D -- More Curious than you Think!!!

[Rick Perley]

    The 2-hour 'switch stress test' gives some *very* interesting 
characteristics about the 16-D amplitude problem. 
    Remember that in this test, I cycle around the bands, changing every 
minute, for 2 hours. 
    The behavior is quite different from band to band.  Here goes:

    1) L-band:  Only the first scan is affected, high by ~2 dB.

    2) S-band:  Only the first scan affected, high+ by ~2 dB.

    3) C-band:  The first 3 scans are affected -- and not randomly!!!  
The gains decrease stepwise until the correct, and stable, gain is 
reached on the fourth scan.  After this, all is stable.  The gains are 
high by:  7, 6, and 4 dB for those first three scans. 

    4) X-band:  Only the first scan is affected, high by ~0.9 dB (only!)

    5) Ku-band:  The first 3 scans are affected -- the first two are 
high, the third is low.  However, it is known that this third scan 
amplitude has been affected by a pointing error.   Gain offsets from the 
stable mean (following the third scan) are:  4, and 2 dB.   

    6) K-band:   The first 3 scans are affected, high by:  7, 6, and 4 
dB (i.e., same as C-band). 

    7) Ka-band:  The effect is entirely absent here!  All scans in 16D 
are stable and equal.  (This has gotta be a big clue).

    8) Q-band:  The first two (and probably three) scans are affected 
... high by 3, 2, and ? dB.  (The third scan is not measureable, as we 
had a major error in referenced pointing on that scan).