[evlatests] 3bit/8bit PDif Compression Test

[Rick Perley]

    Ken ran a short test this morning, with the goal of better 
characterizing our 'PDif Compression' problem. 

    The idea is to drive the 3-bit-equipped antennas from zenith to 
horizon, then back again, while monitoring the change in PSum, PDif, and 
Tsys.  With the attenuators fixed, proper and expected behavior would be 
for the PSum to rise, and PDif to remain unchanged.  Attenuator levels 
are set twice -- at the zenith for the downward slew, then again at 
elevation=8 for the upward slew.  To ensure a good range of input power, 
the test was run at 42 GHz.  He used the A1C1, A2C2, and B0D0 channels.  
The last one of these is the 8-bit path. 

    For this trial, only the upward slew produced data to the archive.  
(I have a query out to the right people to find out what happened to the 
downward slew data).  The results are very interesting ...

    1) The 8-bit path data show the system temperature doubles (nearly 
exactly) over the range of this slew.  The 3-bit paths (for reasons to 
be explained below) claim a much larger range -- more than triple for a 
few IFs. 

    2) All PSum data decline, as expected, as the antennas go from 8 
degrees to 90 degrees elevation.   However, it is quite obvious the the 
3-bit power change is always less than the 3-bit difference, 
spectacularly so in some cases.  See below. 

    3) All 3-bit PDif values smoothly increase as the power level 
decreases.  The factor by which some of these rise is very high!  (See 
table below)

    4) Some of the 8-bit path data also show a rise as the power level 
decreases, but in all cases, the factor is small. 

Below is a table, showing the ratio of PDif(90)/PDif(8) -- ratio of the 
PDif values between the zenith and the lowest elevation. Remember that 
the power ratio between these is nearly exactly a factor of 2.0

    Antenna      A1      A2      B0    C1     C2      D0
----------------------------------------------------------------------------------------
       7             1.32   1.27   0.00    1.50   1.31   1.06
    10              1.54   1.20   1.02   2.78   1.40   1.04
    14              1.56   1.21   1.02   1.27   1.31   1.01
    18              1.67   2.26   1.04   1.18   1.76   0.00
    23              1.22   1.25   0.00   1.26   1.36   1.02
    26              1.27   1.57   1.03   1.38   1.33   1.04
----------------------------------------------------------------------------------

    Perusal of the PSum values clearly shows that some of the 3-bit 
PSums are providing a poor estimate of the total power (presuming the 
8-bit PSum represents 'truth').  Below is another table, showing the 
ratio of PSum(90)/PSum(8) -- the ratio of powers between zenith and 
horizon.  The correct value is about 0.53.  Note that the same 
antenna-IF-paths that give severely depressed PDif also give the most 
incorrect power ratios. 

Antenna          A1     A2     B0     C1     C2     B0
-----------------------------------------------------------------------------------------
    7                0.53   0.53   0.52   0.62   0.49   0.50
    10              0.63   0.56   0.50   0.86   0.54   0.51
    14              0.61   0.52   0.52   0.59   0.59   0.54
    18              0.54   0.79   0.54   0.56   0.79   0.54
    23              0.54   0.54   0.52   0.56   0.56   0.53
    26              0.57   0.51   0.52   0.55   0.56   0.54
--------------------------------------------------------------------------------