[evlatests] Continuous phase

[Rick Perley]

    Barry pointed out in the WIDAR test meeting that a single delay 
solution, when applied to data whose frequency is continuous from 
subband to subband, should preserve the phase continuity. 

    It's clear that the calibration methodologies that some of us 
(including me) have used in the past treat each subband as separate, 
with the result that although the delays are removed, phase offsets are 
introduced so the result of applying a single delay destroys the phase 
continuity.  (The phase slopes within any one subband are flat, but each 
subband is offset from all the others).  This is no real problem in the 
grand scheme of things (these offsets would be removed in subsequent 
calibration), but is a nuisance as we try debug the correlator. 

    So, with help from Craig and Eric, I've figured out a set of magic 
AIPS parameters, which will preserve phase continuity. 

    1) In FRING, solve for the delay and phase, using a short segment of 
the data.  Use a 1-minute long TIMERANG, set SOLINT = 1, APARM(5)=1, and 
DPARM(8)=1.   This enforces a single solution for each IF, over all 
subbands, for both delay and phase. 

    2) In CLCAL, put SOURCES = '',  SAMPTY 'BOX', and BPARM = 24 .  
(This will apply a single averaged solution for the delay and phase to 
all sources).