I've been looking to see if LHO needs to pursue better L2A de-coupling in the corner station suspensions to improve our wind and earthquake robustness. The good news is I had to look for a while to find a candidate, but I know better what to look for now, so I'll see what else I can find. Looking at a couple of recent earthquakes, I noticed that we seemed to lose lock when the IM4 TRANS qpd pitch hit a threshold of -.6. After talking to Jenne about it, we looked at other QPDs close by and it was immediately obvious that MC2 trans qpd pitch was being driven by MC2 M1 length drive. The attached plot shows the story.
Both plots are time series for and earthquake on March 27 of this year, where we lost lock at around 1174648460 UTC. The top plot shows the MC2_TRANS_PIT_INMON, MC2_M1_DRIVEALIGN_L_OUTMON and MC2_TRANS_SUM_OUT16. The bottom plot is the ITMY STS in the Y direction. The first 600 seconds are before the earthquake arrives and is quiet. The spike in the STS at about 700 seconds is the arrival of the P waves. This causes MC2 sus to move more, but the MC2 trans sum isn't affected much. At about 900 seconds the R waves arrive and MC2 starts moving more and more, moving the spot on the qpd more and driving down the qpd sum. I've looked at the other pds used for asc and only IM4 trans and MC2 trans seem to move this much during an earthquake.
[Vaishali, JimW, Jenne]
We started looking at transfer functions yesterday to do the length-to-angle decoupling, but I mis-read Jim's plot, and focused on the lowest M3 stage, rather than the low frequency top stage.
Anyhow, hopefully we can take some passive TFs over the next few days (especially now, with the >90%ile useism and >98%ile wind), and have a decoupling filter ready for the next commissioning window.