[evlatests] Results from Switched Power at C-band

[Rick Perley]

    The advent of our switched power data has enabled many more system 
checks and measurements.  To explore these capabilities, a short (1 
hour) OSRO-style observation was taken 2 days ago, involving observing 
Cygnus A and a nearby calibrator.  Bands observed were L, S, C, and X.  
Ken arranged things so that the 'set and remember' function was turned 
off -- each observation 'auto-levelled' its power going into the 
sampler.  The reason for this is that at L-band, Cygnus A approximately 
quadruples the system power.  This is easily accommodated within the 
8-bit sampler, but might cause undesirable effects at the 4-bit 
requantizer following the formation of the digital subbands. 

    The rescaling has the beneficial effect (from a tester's point of 
view) of changing the T304 attenuation -- allowing checking of the 
ability of the system to monitor and correct for these changes via the 
switched power. 

    Here I report on C-band only -- I pick this one first as the 
attenuation effects are not extreme, and system behavior in general more 
reliable. 

    1) Antenna sensitivity:  The ratio of the derived system temperature 
when on Cygnus to on a calibrator gives a measure of the 'effective 
system temperature', Tsys/effic, independent of the noise diode power.  
With the known flux density of Cyg A, the measured ratio can be related 
to Teffec by:

    Teffec =  (Tcyg-R.Tcal)/(R-1)

    where R is the observed ratio, Tcyg is the antenna temperature due 
to Cyg A with 100% efficiency, and Tcal is the same for the calibrator 
source.    At 4.5 GHz, Tcyg = 76K, at 6.5 GHz, Tcyg = 48.  For both 
frequencies, Tcal is about 0.5.   For the known system temperatures and 
antenna efficiencies, we expect R ~ 2.5 at 4.5 GHz, and 2.1 at 6.5 GHz. 

    The observed values are close to those expected:  the median at 4.5 
GHz is 2.35, and 2.05 at 6.5 GHz.  There is a modest spread:  Antennas 
much better than the median are 1, 3, 8 and 28.  Antennas much worse are 
4 and 23.  (In both cases, the deviation is by about 15%).   The 
'formal' Teffec is 55K at 4.5 GHz, and 44 at 6.5 GHz. 

    2) The 'auto-levelling' caused all antennas to change attenuation 
when on Cyg A -- curiously, about half of them did *not* return to the 
same attenuation when we returned to the calibrator.  These antennas 
then give us an indication of how well the switched power accomodated 
the change in power level.  Of the 29 antenna-IFs which had a different 
attenuation (1 dB, or 12% in voltage) between the two calibrator 
observations, 24 had this change corrected to much better than 1%.  But 
5 had after-calibration voltage levels at or slightly more than 1%.  
(None were more than 1.5%).  More review will be needed to understand 
these 'failures'...