HIRES Cross-Disperser Position Drift and Oscillation

The HIRES cross-disperser stage has two shortcomings that should be recognized by HIRES users and maintainers. First, the position of the stage can drift during observing; and second, the stage can go into an uncontrollable oscillation following a reboot or home operation. This page contains information on monitoring the CD position , and the oscillation recovery procedure.

  1. Position Drift

    2. The rotational position of the cross-disperser in use, be it Red or UV, is measured by use of a high-resolution Heidenhein encoder mounted directly on the rotation turntable shaft. The positional resolution is 1.8 arcsec, or 0.0005 deg per encoder step. The HIRES user requests a destination position in degrees through the XHIRES GUI, and the control software iteratively slews the stage to a setting accuracy of +/-1 encoder count or +/-0.0005 deg. One encoder count in rotation projects to 0.54 pixel on the HIRES CCD.

    The rotational stage is mounted on a compound angle and is air-cylinder preloaded, but it is implicitly out of balance except at one position. Due to vibration in the telescope structure, the largest source of which is the azimuth drive roller-ring frictional coupling, the cross-disperser position sometimes drifts several encoder steps during telescope slews. Drift can also occur on longer timescales if the imbalance of the stage against its preload is large enough.

    Users need to be aware of the potential for drift in the cross-disperser position, and monitor its position in XHIRES (which displays only to 0.001 deg), or via an xshow of the keyword "xdangl" which is displayed automatically adjacent to the XHIRES display; the encoder position is available through the keyword "xdraw". In addition, there is a monitor available that will poll the position every minute, and broadcast a voice message if the position changes from the previous sample. This monitor is started from the menu entry HIRES Control Menu -> Cross-disperser Monitor. If the monitor becomes bothersome while making intentional position changes, then kill the process 'xdwatch' on makua (found with ps -auwx | grep xdwatch; kill -9 PID#).

    If the cross-disperser drifts during science object exposures, it is permissable to reset the position while exposing. Note that a small amount of image blur may occur if the drift is more than 2 encoder counts (1.08 pixel), so resetting the position should be pondered with this in mind.

  2. Position Large-scale Oscillation

    3. The servo-control for this stage on occasion gets into an uncontrollable state, whereby the position drives from limit-to-limit at relatively slow velocity (about 5 deg/s) without halting. This fault occurs most commonly after a motor crate reboot, which puts all HIRES stages in a reset (not homed) state, followed by a homing operation which occurs automatically with the first requested move to a new position after reset. The reason for this oscillation is unknown, but can be recovered from in the following manner:

    (The following has been incorporated into the script "stop_drift", so you can try typing that command rather than the series of commands below).

    Halt the stage by inhibiting its motion: 

    makua{hires}69: m xdinh=1
    The stage now needs to be homed; do this manually, and monitor its position while the homing operation proceeds: 
    makua{hires}70: m xdcal=homed
    Next, set the position to a new value to test its control, either through XHIRES or by keyword: 
    makua{hires}71: m xdangl=0.5		for example.
    If this last command completes OK, then the stage control is recovered. If the stage begins to oscillate again when homed or set to a new position, halt it again as above, and repeat the homing and new position command.

    Note that rebooting the motor control crate VME crate will effectively halt the stage, but is unnecessary.