NIRSPEC Update

NIRSPEC Reliability Improvement
Progress report: 8 November 2004

Overview:

This report (the first one in several months) marks the end of project status for our efforts to increase NIRSPEC reliability. Several significant events have occurred within this reporting period. The project was on hiatus during much of June and July while the instrument was warmed up to work on a failed filter wheel. After being cooled down and returned to service, the new keyword and rotator server codes were released in late August. During this time period, only two tasks remained officially part of the project.

Release of the new code has encountered difficulties that have resulted in an increase in time lost on sky. It is hoped that these problems are now understood, solutions are in place and we will witness a return to improved reliability.

Upgrading the host.

The NIRSPEC host computer represents a critical vulnerability since no spares exist. During the period covered by this report, no progress was made on this task due to the diversion of resources to more urgent tasks. Now that the project has ended, it is unlikely this goal will be accomplished unless resources can be diverted from other areas.

Test and release new server codes.

On August 22, the new server codes were released. Testing to that point indicated that all the issues described in the last report had been resolved. The code was used successfully by us on prior Chaffee science nights.

Almost immediately problems were encounterd by the first science observer.

Nodding performance was unacceptably poor, with objects consistently landing off the slit with each throw.
Reverting to the old code fixed the problems.
Efforts to reproduce the problem during subsequent engineering time were unsuccessful.

During the first week of September, NIRSPEC was again on sky.

This four night run was a single observer who agreed to use the new code.
Problems with nodding performance were again encountered, but despite offers to revert the code, the observer chose to stay with the new code.
By the end of the four nights, anecdotal evidence implied that the nodding performance could be maintained by frequently re-initializing the rotator.

The third week of September saw NIRSPEC on sky again.

This four night run consisted of three observers, none of whom wanted to risk using the new code.

The first half of October 3rd was NIRSPEC engineering.

Significant improvements to the rotator server code were successfully tested.
The rotator correctly set slit PA on double stars.
Nodding performance was satisfactory.

The nights of October 4th, and 5th, observers used the new code.

Once again, problems with nodding performance and/or slit PA on sky were noticed.
The problems were at least partly ameliorated by frequent rotator inits.
Most importantly, what may have been a breakthrough in understanding of the problem was accomplished. It appeared that some or all of the degradation to rotator performance resulted from it losing steps.
Testing during the days after these nights indicated that increasing the size of the minimum allowed move, by a factor of four, dramatically improved rotator performance without loss of PA accuracy.

The second half of the night of October 21st offered a chance to test the new version of the server codes with minimum step size increased.

Thick cloud prevented rigourous testing at times but it appeared the rotator was working as it should. PA's were setting and holding correctly for up to 45 minutes with a variety of guiding methods. Nodding was satisfactory and rotator status was displayed correctly on FACSUM.

A four night run in early November was the first use by an observer of the new code with larger minimum step size.

The rotator was mainly used in PA mode for all four nights with both SCAM and optical guiding.
No keyword server crashes were experienced.
The rotator server experienced only minor problems that were quickly fixed by simple software restarts.

Proceeding From Here:

During the time covered by this report, time lost on sky increased. As complications associated with the new server codes are understood and fixed we hope and expect time lost to drop again. In this respect, we anticipate that this penultimate report will be followed by a final wrapup report some months in the future once sufficient statistics have accumulated to properly judge the success of this project.