Sheila, Dave, Ryan Crouch, Tony

This afternoon after the maintence window when we first started using guardian, once the ASC safe.snap was loaded by SDF revert we started sending large signals to the quads.  We found that this was due to the camera servos having their gains set to large numbers.  This was set this was in the safe.snap file. 

After I set these two zero in safe.snap (which is really down.snap), Ryan again went through the guardian down, and this time we started to saturate the quads because of the arm asc loops (which we probably didn't notice the first time because we tried running down when we saw that there was a problem, and down would turn these off but not the camera servos).

Dave looked in the svn for this file, which he had committed this morning with this set of: diffs from this mornings svn commit  .  Looking through these, it kind of seems like somehow the safe.snap may have been overwritten with the observe.snap file. 

Dave reverted that to the file from 7 days ago, which has Elenna's changes to the POP QPD offsets. Then I reverted all the diffs, so that we set all settings back to 7 days ago except those that are not monitored.

After this, Mayank and I were using various initial alignment states to make some clipping checks, which Mayank will alog.  We noticed that the INP1Y loop (to IM4) was oscillating, so we reduced the gain in that from 10 to 40, on line 917 of ALIGN_IFO.py  We also saw that there is an oscillation in the PRC ASC if we sit in PRX, but we haven't fixed that.  These should not be due to whatever our safe.snap problem is, we hope.

Edit to add: We looked at the last lockloss, when the guardian went through SVN revert at 7 am yesterday Feb 3rd.  It looks like the camera gains were 0 in the safe.snap at that time, but it was 100 by the time we did SDF revert at 20:51 UTC (1pacific time) today.