[evlatests] More on switched power ...

[Rick Perley]

    Another test of switched power was run early Wednesday morning:  
3.25 hours of tracking 3C147, as it moved from elevation 49 to elevation 
19 degrees, cycling around L, S, C, and X-bands (in that order). 

    Many odd things have been (and are being) found.  Unfortunately, it 
again proves impossible to properly apply the switched power to the 
visibility data -- it appears that the SY table provided is only 
partially populated (is sort of resembled swiss cheese ...), so only a 
fraction of the visibilities are in fact corrected. 

    However, by inspection of the data we do have (at X-band), and 
comparing them to the visibility amplitudes, I think  it safe to state 
the following:

    1) Short-term (~hour or less) changes in visibility amplitudes are 
very well matched by corresponding changes in switched power level.  
There is thus high expectation that the application of the switched 
power will remove these system-induced gain changes to an accuracy of at 
least 1%, and probably quite a bit better. 

    2) There is a long-term decline in visibility amplitude, seen on all 
antennas, of about 5% that is not matched by changes in switched power.  
This must be an elevation -dependency of antenna gain/atmospheric 
absorption.  The standard gain corrections were apparently applied (as 
seen in the CL#1 table), but are clearly not sufficient.  I don't know 
if the improved absorption model has been applied by default, but doubt 
this can provide 5% attenuation at an elevation of 20 degrees. 

    3) The short-term amplitude variations noted in Item #1 above nearly 
always affect single scans.  They will be largely corrected by the 
switched power, but they shouldn't be there in the first place.  Only a 
few antennas show these -- and there is no unique pattern.  That means 
that on some occurances, all four IFs are affected.  On other, the two 
IFs of a single polarization are affected.  On others, only one IF is 
affected.  As I was switching around bands, suspicion centers on the 
band switches...

    4) As this experiment was run through the morning of 'maintenance 
day', 7 antennas were taken out at one time or another, then returned.  
Most returned in very peculiar states:  antennas 5, 8, and 10 were 
returned with their fringe and total (and switched) powers down by 
factors of nearly 10!  Antenna 6 returned with perfectly good switched 
power, but quite grotty visibilities.  Antennas 17 and 22 showed no 
significant changes.  Although we should expect no better for running 
during maintenance, it might be useful to know what was done, so as to 
understand the cause of these dramatic changes...