[evlatests] ea25 pointing

[Ken Sowinski]

Since it underwent a bearing replacement in 2011 ea25 has exhibited
pointing errors which are both not modeled by the usual pointing model
and are hard to explain.  This note documents one example of what we
see in the hope of inspiring an explanation or repair.

This example is drawn from the pointing run of October 26, 2018;
other pointing runs could be examined to find similar examples if
that is useful.

  Az    El    Azerr Elerr (errors in arc-min)
248.4 109.8 -0.09  0.24
248.2 109.5 -0.26 -0.37
248.1 109.2 -0.47 -0.91

334.5  11.2 -0.13 -0.59
286.5  13.6 -0.05 -1.15
293.2  36.6  0.04  0.20
313.0  78.2 -0.03 -0.27

The first example is especially striking because the large change in
elevation error occurs when neither Az nor El change by even one
degree.  The second example covers a larger range of Az and El, but
still has a change in El error which is large and sudden.

An examination of the post-fit residuals to the pointing model fit,
shows that since the bearing replacement the pointing behavior has
deteriorated, and is limited to azimuths between 240 and 330 degrees.
This limited range makes it very difficult to explain the poor
pointing by software errors, FRM faults, loose mechanical structure,
or hysteresis.  There is nothing left to invoke except for a defective
bearing.  What ever the mechanism, if must be able to explain erratic
poinitng changes greater than an arc-minute occurring on azimuth
scales of a degree or less.

If a measurement of the bearing run out error confirms this suspicion,
then the only effective remedy is another bearing replacement.  None
of our usual solutions to this kind of problem: referenced pointing,
modeling the error profile, or measuring it with a tiltmeter, can
correct for errors that change so fast in both time and azimuth.