[evlatests] More on PDif Compression -- (more confusion)

[Rick Perley]

    Ken modified the PDif compression test script to lower the power 
levels within the T304 by 10 dB from the nominal values.  It was hoped 
that in doing so, the compression seen in PDif when we observe Cygnus A 
(typically 15% for a power increase of ~4 dB) would disappear.  The test 
ran last night. 

    It didn't. 

    Some test details:  Ken first set the input attenuators 10 dB higher 
than is normal.   The output attenuators were then set to provide the 
correct power to the digitizers.  These values were 'remembered'.  The 
script then adjusted the output attenuators, in 3 dBsteps, from a level 
12 dB higher than the nominal to 12 dB lower.  For each of the nine 
settings, we observed cold sky, a local calibrator, and Cygnus A.  The 
requantizers were not adjusted (and the visibility data clearly show the 
overflow expected at the high power end). 

    The middle observation is at the nominal power level, except that 
the power *within* the T304 will be 10dB below that of the previous 
test.  So it is useful to compare the compression for the middle setting 
from last week (power within the T304 at the nominal level) to that from 
last night (power within the T304 10 dB lower). 

    The same characteristics noted before are seen again -- antennas 
which had very high compression before still do with the reduced power 
level.  Those with the compression low then are low now.  There is one 
small, but significant difference overall -- the compression levels are 
a little less with the lower power.   However, it is not useful to add 
another 10 dB attenuation  to the input -- we'd have little left to look 
at. 
    Last night's test also revealed another curiousity -- in the maximum 
attenuation state (output attenuators at the highest level) a number of 
antenna/IFs showed a small 'negative' PDif.  This went to the normal 
state (small positive PDif) as the output attenuation was decreased. 

    The only clear result I have is that the two antennas with the 
highest PDif (5 and 25) are just as compressed in last night's test as 
before.  For these two, it seems clear that the compression arises 
elsewhere, presumably 'upstream'.