[evlatests] Test for 10-second dropout problem

[Rick Perley]

     Vivek ran an hour-long script yesterday, which was designed to 
characterize the '10-second' problem.
     The idea was that if the effect is due to the change in 
tuning/configuration between X-band pointing and x-band observing which 
uses a different correlator/tuning setup, we should try multiple 
observations of this transition, compared to a transition using the same 
tuning/configuration between pointing and observing. Thus, we generated 
a script which did the following cycle eight times, observing the 
super-strong calibrator 3C273:

     X-band Ref pnting (two-SPW mode)
     X-band observing (wideband)
     X-band Ref Pnting (two-spw mode)
     X-band observing (two spw mode)

     The effect I saw in the October data from last year was *not* 
reproduced.  However, a similar, and much more comprehensive effect was 
seen:

     1) The transition from two-SPW pointing to two-SPW observing at the 
same frequency setup showed no amplitude dropouts of any kind.

     2) The transition from two-SPW pointing to multi-SPW observing 
showed complete absence of coherent visibilities at the beginning of 
each scan, with the following characteristics:

     a) SPWs 2 through 8, and 9 through 16 showed 10 seconds (exactly) 
of complete noise at the beginning of each scan -- every scan.   Both 
polarizations, and all antennas are affected equally. The return to full 
amplitude is immediate -- no transition.
     b) SPW 1 did not show this effect at all -- looks just like the 
two-SPW situation.
     c) SPW 9 showed a 'partial effect' -- the first ~ 4 seconds had 
reduced amplitude.

     In my October calibrator data, I did not check to see if the 
dropout behavior in SPWs 1 and 9 were different than the others. I'll do 
that later this morning.