[evlatests] Switched Power Compression?

[Rick Perley]

    We have long noted that the values of 'PDif' (Cal On - Cal Off) 
changed when the antennas observed a strong source.  This, in spite of 
holding the antenna electronic gains constant.  In some cases, the 
change has been noted to be very large -- exceeding 25%, and generally 
in the 'compressive' sense -- the values of 'PDif' decreased when 
observing a strong source such as Cygnus A. 

    There was reported (by me), a few months ago, weak evidence that 
this 'compression' did not occur to the visibilities -- so that 
application of the switched power values would degrade the visibility 
data.    These data were taken on 3C273, which provides only ~2K of 
antenna temperature. 

    As we are now in D configuration, a much better test of whether the 
visibilities are not affected by the PDif compression issue can now be 
done using Cygnus A.  This source has a big effect on Tsys (up to 120K 
at L-band), and certainly causes the PDif values to change. 

    A short observation was taken last evening of 3C286 (calibrator) and 
Cygnus A.  Unfortunately, the observations of the former were taken at a 
low elevation, and four antennas were shadowed when on 3C286 and could 
not be used for calibration.  This is a problem only in that these 
represent many short spacings, which are the useful ones for this 
experiment.  Nevertheless, the major result seems to be clear:

    *** The Visibility data *are* compressed in a manner similar to that 
of the PDif Values *** 

    Application of the PDif values provides  flux values for Cyg A much 
closer to the known total flux.  Calibration without applying the PDif 
values results in Cygnus A flux density significantly too low -- 
especially at L and S bands, where the antenna temperatures due to Cyg A 
are high. 

    Almost every antenna shows a strong reduction in the 'PDif' values 
when observing Cygnus A at L-band.  In a following report, I will 
provide the compression factors.  In general, the amount of compression 
declines with increasing frequency. 
Curiously, the strong compression formerly seen on ea24 in RCP at C-band 
is now completely absent!!!

    L-band also provides another problem -- some baselines, notably 
those connecting antennas 8, 20, 23, and 28 -- are giving *enormous* 
closure errors when on Cygnus A, but not on 3C286.  'Enormous' means 
tens to hundreds of percent!  Something is seriously amiss here. 

    There are other (unhappy) issues with the PDif system which I shall 
report on separately.