IMC TRANS refinement broken

Some recent events clearly indicate that the IMC TRANS based time refinement is not working correctly:

• https://ldas-jobs.ligo-wa.caltech.edu/~lockloss/index.cgi?event=1239933083
• https://ldas-jobs.ligo-wa.caltech.edu/~lockloss/index.cgi?event=1239933509
• https://ldas-jobs.ligo-la.caltech.edu/~lockloss/index.cgi?event=1239819268

added Analysis Bug labels

changed the description

@jameson.rollins I will talk to commissioners and find a better channel to refine the lockloss time with

@yannick.lecoeuche The issue is not the channel, but the logic. I think the channel is a fine indicator, but we clearly don't have the logic correct, when it's easy to see by eye when the lock failed but it's still not able to refine.

@jameson.rollins I think this relates directly to #99 (closed), since the logic prohibited the lockloss time from being after the ISC_LOCK-determined lockloss. Regardless, I would imagine that the ISC_LOCK lockloss time should always be later than our indicator channels.

This is due to the GRD lock channel looking at multiple channels to check we are locked. As explained in #156 (closed). Because we are only looking at one channel we cannot see the real refinement. e.g The guardian ISC_LOCK bases a state 15 lockloss on the the following being locked: @ISC_library.assert_dof_locked_gen(['IMC', 'XARM_GREEN', 'YARM_GREEN']) we just look at IMC.

Merge request #105 (closed) will make sure that the refined time isn't any worse. Closing this issue.

closed

mentioned in issue #156 (closed)