[evlatests] Antenna/ACU/servo behavior in OTF mosaic tests

[Vivek Dhawan]

An update on ACU behavior during fast tracking (OTF-mode). Issues found
in previous tests are listed, with their current status. Plots exist to
support these assertions, but are not attached here.

1. Interpolation of commands was incorrect (to all ACUs) near azimuth due
    South, when tracking in the counter-sidereal direction.

    This was fixed in the executor.


2. Tracking error (deviation from commanded position) had a significant DC
    offset plus oscillations, in azimuth, especially on the NEW ACUs  ea01,
    14, 17, 21. For these units, Doug W. has tuned the bias currents (in AZ
    only, I believe) and eliminated most of the oscillation.

    The elevation errors remain, ~1’ deviation, and seem to be random, i.e.
    they are not antenna or servo resonances. There are differences between
    antennas, its likely this too be tuned away: ea01 and 17 have large
    EL rms, ea14 and 21 are very much better.

    All 4 new ACUs have a DC tracking offset when we calculate the error
    = (Position - Command) from the logged pointing data, but do not show
    the offset on the error signal that drives the servo. This still needs
    to be clarified, it may be another timing issue (distinct from OLD ACUs
    in item 3).

    The settling time for new ACUs is ~3sec, a factor of  ~2 better than
    the old ACUs, and unchanged from before the tuning.


3. OLD ACUs had small tracking errors and no oscillations, but showed jumps
    in the calculated error, ascribed to timing slop in labeling the current
    command, future command, and measured position (in the MIB).

    Hichem has updated the MIB code (on old ACUs) to fix the data handling,
    and these timing errors are now eliminated. This code has been tested
    on the sky, on a subset of antennas, but is not yet deployed anywhere.


4. The command sequence for OTF rasters was examined in detail, refined,
    and tested again on May 19, near transit, 77deg EL, and at ~40deg El.
    Bryan is looking at this, I believe it looked good.


5. MCAF (in test version) is writing the pointing table in binary form,
    and as of Friday, Rich had successfully read these into CASA.
    Promotion of test MCAF to production is expected soon.


In summary,

OLD ACUs : Look good to go, after the MIB update is deployed.

New ACUs : Look good in AZ, need some tuning in EL. DC offset between
            external and internal error signals needs to be understood.

--------------------------------------------------------------------------