[Sheila, Evan, Gabriele]

The goal is to cancel the PRCL length chane induced by driving MICH using only the BS. For this reason we plan to implement a path from MICH control signal to PR2.

1. we injected a MICH motion using a differential drive on the two ITMs, and a PRCL motion using a common drive on the two ITMs. The PRCL error signal (POP_9I) sees the common mode about 20 times more than the differential mode. So we conclude that POP-9I sees MICH about a factoir 20 less than PRCL. It's therefore a good reference to disentangle MICH and PRCL in the BS drive
2. we injected broad band noise in tBS and measured the transfer function to POP_9I. Let's call this transfer function TF_POP_BS. This measures how much PRCL we do when we drive BS
3. we injected broad band noise in PR2 and measured the transfer function to POP_9I. Let's call this transfer function TF_POP_PR2.
4. we fit the ratio TF_POP_BS / TF_POP_PR2 with a 6th order filter (see figure) and implemented this filter in the PR2 ISC input filter banks. We set the driving matrix to send MICH to PR2, and activated the filter

We checked the result with a MICH line at 132 Hz: when the compensation is on, the line amplitude in the PRCL error signal is reduced by at least a factor 10, as expected from the fit residual (figure 2).

We're not leaving this in the guardian.