[evlatests] Switched Power successes and failures at X-band

[Rick Perley]

    The recent elevation test gives us a good idea of the effectiveness 
of applying the PDif values at each band.  Here I report on X-band.  At 
this band, we expect virtually no change in antenna gain with elevation. 

    1) Nearly 50% of the antennas had non-steady visibility amplitudes, 
due either to switch issues, or (probably) to temperature-related 
changes in gain.  Virtually *all* of these were corrected by application 
of PDif.  Only two clear exceptions to this are noted:
    a) ea26 had water in the feed, which caused a strong dependency of 
amplitude and system temperature on elevation.  Part of this was 
corrected by PDif (the part due to the water vapor emission), and part 
was not (that due to the absorption).  This is no failure of the PDif 
system!
    b) Application of PDif to ea15C introduced a scan-scan variation of 
~ 5% which was not present in the raw visibility amplitudes.  These 
variations are easily visible in the PDif values, but are not present in 
PSum (a measure of total power).   Hence, these changes are not a 
measure of the system gain.

    2) The PDif values for ea05, on all four IFs, have near zero mean, 
but the expected variance.  It is acting as if there is no switched 
power at all.  This problem is seen only at the band on this antenna -- 
all other bands give good PDifs.  Hence, this problem is something to do 
with the receiver. 

    3) The Pdif values for ea27A have very high variance, more than an 
order of magnitude higher than any other antenna-IF.  The mean value for 
PDif is within the expected bounds ( about 250 milli-counts).  The rms 
is many tens of millicounts, where the normal is about 5 millicounts.  
(This is not yet a wideband antenna, so I don't know if this problem is 
also on IF B). 

    4) The test observed 3C84 and nearby cold sky -- the goal being to 
utilize the Tsys value differences (Tant) as a means of roughly checking 
that the Tcal values are correct.  This is important as we want to use 
the derived Tsys measurements to derive system temperatures (for 
monitoring purposes) and to assign data weights to visibilities (for 
provide optimum sensitivity in imaging). 
    For a source of flux density S, and a 25-meter antenna of aperture 
efficiency e, the observed change in system temperature (Tant) should be:
                  Ta = 0.18.e.S (Kelvin). 
    At X-band, we expect an efficiency between 0.6 and 0.65.  The 
observations were calibrated against 3C147, from which we derive the 
flux density of 3C84 to be 23.7 and 27.9 at 8300 and 11300 MHz 
respectively, so a Tant of 2.8 and 3.3K are expected at these two bands. 

    Perusal of the results shows that most antenna-IFs are quite close 
to these expected values.  Below I give those wide-band antennas which 
varied by more than ~0.5K from the expected value.  (Note that a number 
of the older narrow-band antennas give deviant results -- I choose not 
to report these to save space.  Anybody who wants the list should email 
me directly). 

     ea10C has Tant = 2.4 and Tsys = 22.  Both are a bit low.  Tcal is 
listed as 1.08, which I think may be ~20% too low. 
     ea14B has Tant = 2.4 and Tsys = 24 -- again a little low.  Tcal = 
1.12 -- also suspiciously low. 
    ea21A and C both show Tant = 2.1 and Tsys  in the low 20s.  Tcal = 
1.0 and 1.3 on A and C. 

    Note that astronomical calibration against a source of known flux 
density allows correction of data weights, so these errors (if real) 
should not affect astronomical imaging.