[evlatests] Level setting problems

[Ken Sowinski]

Hichem, Keithm Jim and I were able to track down why the fast
downconverter ALC did not always work.  The problem is that
the input power to the downconverter is still changing long
after the ALC loop is started for many band changes.  The
cause of the power change is subreflector rotation which can
change noise power by 2 or 3 dB between the wrong and right
rotation setting.  This was very obvious when switching between
X and K bands which are on opposite sides of the feed ring.
How important it is depends on how long the move in rotation
is and, to some degree, the system temperature at the band
we are switching to.

The slow ALC loop was a little more acommodating to this change,
but it does not always work correctly either for the same reason.
If we want to do a better job, there are only two choices, to
pre-move the subreflector which breaks the old scan, or wait
a little longer before invoking the ALC.

In principle, this could contribute to phase changes between
successive scans at the same band if the first one mis-sets the
attenuators at a band change, and the second sets them correctly.