[evlatests] Some Good News about Switched Power (Really!)

[Rick Perley]

    In order to measure the variation of antenna gain with elevation, 
data were taken last evening of 3C345, from its transit (at elevation 
82) to setting (at elevation 8).  I used the 'old-OSRO' setup (2 
subbands with widely different frequencies) and all eight bands, running 
in a continuous loop.  Referenced pointing was was determined for each 
cycle -- about every 10 minutes -- and applied to all bands.  The 
weather conditions were exceptionally good. 

    I have reduced only the X-band data so far -- I used 8300 and 11300 
MHz.  The results are very impressive. 

    1) Because the bands were changed continuously (in fact, 8 bands * 
37 observations = 296 times), 'tuning failures' and fringe amplitude 
irregularities are expected.  And they were found.  At X-band, 10 of the 
37 observations had one or two (usually just one) antennas without 
fringes.  The failures were most common with antenna 8 (4 times), then 
antennas 24 and 13 (twice each), then antennas 25 and 20 (once each).  
All others fringed for every scan.   The failure rate is  10/(37*27) = 
1.0%. 

    2) But the failure rate -- as defined above -- is only half the 
story.  We have the attenuation change issue as well - the switches 
apparently have a slightly variable attenuation, which causes both a 
change in amplitude and a change in bandpass shape.  This problem was 
certainly evident in the X-band data, with antennas 4A, 13C, 14C, 18C 
and 25C being notably poor.  (4A and 14C are especially bad).  Note 
however that the majority of the antennas do not show this problem 
beyond, say, a single occurrence. 

    3) Now for the  Good News!  Application of the PDif values 
beautifully corrects nearly all of the gain variations noted!  Following 
an elevation fit (via ELINT), the post-correction residuals vary for 
nearly all antennas by less than 1% over the length of the observation.