After fixing the good value of the SSFS filter gain during the last shift, we have implemented a quiet filter (flat until the cavity pole, 1/f until 5000 and then 1/f3) in order to void glitches engaging the lock.

The problem is that with the level of the 150Hz we expect to almost saturate in steady state, but the transient would be critical at this point. So we have triedlowering slightly the gain (not too much in order not to cross filters and become unstable), but we saw that in any case the transition to the MC was saturating and very critical.

At this point we have decided to charge the filter before, qhich is not a problem in this case because there is no integrator on the filter. We saw some improvements on the transient where we thought that we were limited by the MC loop hand-off. At this point we have implemented a hand-off for the MC where we pass with a ramp from the CARM error signal to a sum of the CARM signal and the SSFS_corr. This has been the more promising trial so far, and it is shown in Figure 1 with 400ms of duration. However, the ramp was not finished and so the loop was not fully engaged.

We wanted to keep on researching on this line, specially since we have seen that the loop that seems critical is the MC one. However, taking in account the state of the MC, the height of the 150Hz and the fact that we are trying to close a loop which is already difficult by itself, with a UGF which is 10 times lower than it should be, we have given up for today, waiting for more favorable conditions.