[evlatests] More amazing phenomenology ...

[Rick Perley]

    Closer inspection of our test data taken Sunday reveals more 
coherent patterns, which are easy to see if the data for each of the 
offset positions are phase shifted to the phase tracking origin (this 
takes out the spatial component of the phase, leaving the time-variable 
effects bare for all to see). 

    I find:

    1) The residual phases of 25 and 28 are both rotating at a constant 
rate over the 1 hour duration of the file.  The two rates are exactly 
equal and opposite!  (That is, while 25's phase is increasing, 28's is 
decreasing at exactly the same rate).  It is clear that this is a 
continuous change (I can resolve it within the scans), and is not 
connected with scan boundaries. 

    2) The phase rates are also proportional to source offset distance.  
The slope is about +/- 0.6 degrees/(minute.arcminute), with + for 
antenna 25, and - for antenna 28. 

    3) There is a clear time at which all phases for baselines connected 
to 25 and 28 are correct (i.e. = zero):  This time is the beginning of 
the observation. 

    It looks like a slow slippage in phase tracking, which starts at 
zero at the observation's beginning.  Antenna 25 is at N2, while antenna 
28 is at N8. 

        Ken has a theory ...