[evlatests] Gain and switched power issues from yesterday's test

[Rick Perley]

    A 30 minute observation of 3C286 at all cassegrain bands was done 
last night to check the accuracy of our 'blind' calibration methods. 

    The major goal for this test will not be reached, as the high 
frequency bands were clearly affected by very poor pointing.  The test 
was run about 6PM, and the referenced pointing solutions show *median* 
offsets of one arcminute in elevation, and nearly 1/2 arcminute in 
azimuth.  These, combined with modest winds (20 mph, or so) seem to be 
sufficient to explain the rather wide variations in amplitude found at 
Ka and Q bands.  (The other bands seem to be fine, and I'll report the 
results later).
    Each band was observed twice, with two subbands (only) per band.  I 
went in simple band order:  L, S, C, ... Ka, Q. 

    Here I want to note some other oddities -- some new, some old. 

    1) ea25B had a major change in delay and amplitude part-way through 
the run.  The bandpass solution clearly shows that something happened to 
the DTS transmissions.  This was seen at all bands.

    2) We have some real problems in gain settings at L-band.  A number 
of antennas showed major changes in amplitudes.  For all three, the 
change affected both IFs of the same polarization. 
    a) ea12 in RCP:  Changed by 50% between the two observations.
    b) ea15 in LCP:  Changed by 25% "
    c) ea18 in RCP  Changed by 80% in IF A, and 10% in IF B. 
       For all of these, the bandpass changed in the way expected for a 
large change in the attenuator settings. 
   
    So, despite the 'set and remember' algorithm being run, it's clear 
that at L-band, for these antennas,  something is causing a major change 
in gain.  If it's not the T304 attenuators (and because both IFs of the 
same polarization were affected in each case, it seems unlikely), then 
something 'upstream' is causing a major gain change to occur between 
these two observations (which were separated by only 11 minutes). 

    3) Three more issues, unique to L-band were noted in the switched power:
    a) ea03 had zero (actually, slightly negative) PDif on all four 
IFs.  I conclude the switched power is not, in fact, switching.  PSum 
was fine.
    b) ea28, in LCP, gives values from 10 to 600 milli-counts -- all 
randomly distributed.   *** This problem has been in place every since 
switched power was turned on!  It is unique to this band, this antenna, 
and this polarization.  It has to be something in the receiver.  Can 
somebody please look at this? ***
    c) ea17, in RCP only, has a 'ratty' PDif and PSum.  Although the 
values are useable, they are widely scattered by factors of up to six, 
and cannot possibly represent real gain variations. 
   
    4)  At S-band, we got no fringes from ea09, although the 
coordination meeting summary says there is (or was) a receiver 
installed.  Has it been removed?

    5) At X-band, ea12 gives very weak fringes in the BD IFs (tuned to 
11.25 GHz), while giving great fringes in AC (tuned to 8.25 GHz).  It 
acts like there is an old-style receiver installed, but according to the 
coordination meeting summary, this antennas has a new-style wideband 
receiver. 

    6) On all four high frequency bands, a small number of antennas are 
showing slow and significant changes in gain:  'Slow' means on 
timescales of tens of seconds, 'significant' means changes of 20% or 
more.  These appear to be real -- both PSum and PDif change equally.   
The same antennas are doing this at all bands -- which implicates the UX 
converter:  ea01, 06, 16, and 19.  In all cases, it's a single IF which 
is varying.  I don't recall seeming this phenomenon before ...