[evlatests] L-band gain 'bump'

[Rick Perley]

    More on the gain bump seen at L-band.

    First off -- my statement in the L-band report that there was no 
Tsys recorded
was a 'pilot error'.  I filled the front-end values, which don't exist.  
After retreiving
the back-end values ...

       Antenna 13 gives 1K temperatures, and antenna 16 hundreds of 
thousands of K.

    Antenna 14 gives 'reasonable' values.  Zooming in on the interesting 
event, we
find that the *reported* Tsys rises by about 6% at precisely the time of 
the apparent
gain drop, and drops again in concert with the gain recovery.  
Possibilities:

    a) We can reject the hypothesis that the source flux really did drop 
-- no such
effect was seen on any other antenna.  Even if such an event did occur, 
the magnitude
seen would not cause any effect to Tsys.

    b)  If the system temperature really did change, the correlation 
coefficient would
drop, and the application of the 'peculiar gains' would offset this, 
resulting in
constant visibility amplitude.  Since this didn't happen, then either 
the peculiar
gain was not applied, or the measured (and utilized) Tsys is wrong. 

    c) Ken peered into the system, and declares that Tsys (and the 
peculiar gains)
are being applied.  (Too bad -- if Tsys really did rise, and the gains 
were not
applied, the reported flux density would drop, more or less as seen). 

    d) We might then conclude that the reported Tsys is in error.  However,
a plot of the reported (and presumably utilized) Tsys, although showing an
event coincident in time, also displays quite a different shape, presumably
due to the averaging.  It would seem that an event occurred changing both
the correlation coefficient and the reported Tsys.  The former -- not being
subject to smoothing, is rapid, and caused a ~6% drop in correlation for
about 5 seconds.  The same 'event' disrupted the Tsys measurements, which
were applied long after the `event' ended -- causing the slower decline in
apparent gain reported earlier.