Koji is elogging separately of his exploration of different configurations. The lock stretch that I'm looking at here uses the same parameters as Koji had for PRMI sb lock, using AS55Q for MICH and REFL33I for PRCL, with MICH gain of -0.8 and PRCL gain of 0.05 .
All of these plots are the same few second lock stretch, with different zooming. Jamie's super-sweet getdata python script only accepts integers for the start time and duration parameters, so lots of this zooming happened by hand, but I tried to always keep the time axis aligned within each screenshot. Sometimes the plot axis labels say differently, but they're lying to you.
Plot 1: gps start time is 1050915916, duration = 6 seconds. Overall view of the lock stretch.
Plot 2: gps start time is 1050915921, duration = 1 second. We're looking at the lockloss that happens at the left side of the plots.
Plot 3: zoomed in (along the time-axis) version of plot 2, so much shorter time duration. Some zooming on y-axes.
Plot 4: zoomed in (along y-axes) version of plot 2.
It seems to me from these plots that maybe MICH CTRL is drifting away? It seems like we lose the MICH lock, and that destroys the whole thing.
Koji made some comments to me earlier, regarding his work this evening, that the MICH signal quality is poor in general, and that we should calculate/think about changing our schnupp asymmetry.