[evlatests] Negative Switched Power!!!

[Rick Perley]

    Another curious problem arises with our values of PDif. 

    It was noted in the 'solar tests' that for some antennas and bands, 
the PDif values averaged slightly negative when we pointed at the Sun.  
This cannot be correct for a properly operating system -- individual 
values can be negative due to noise fluctuations when the power in the 
cal signal is completely overwhelmed by the system/source noise (as is 
the case for regular cals when looking at the sun), but the average 
cannot be. 

    In the antenna 24 tests done two weeks ago, we also turned the 
calibration off in order to test for this offset.   This should result 
in a 'PDif' value averaging zero, since the system should be 
differencing two equal powers.  Non-zero values should only reflect the 
noise. 

  But -- all four IFs showed a 'negative PDif' when the switched cal 
signal was turned off.  The magnitude is significant.  The normal PDif 
values should be around 0.5 'counts'.  We see:

    IF A:      offset = -.01 counts.
    IF B:      offset = - .04 counts
    IF C:      offset = -.05 counts
    IF D:      offset = -.01 counts. 

    I can imagine one mechanism which would lead to this:  If the on/off 
durations used to form the values which are then differenced are not of 
equal duration, then there will be an imbalance between the integrated 
'on' power and 'off power', leading to an offset.  
   
    How sure are we that the 'on' and 'off' durations are precisely equal?