Elenna, Ryan S, Sheila
This morning Ryan and Elenna were having difficulty getting DRMI to lock, so we locked PRMI and checked the phase of RELFAIR with the same template/PRM excitation described in 84630.
45 MHz phase changed by 7 degrees, 9MHz phase changed by 5 degrees. The attached screenshot shows that the phasing of RFL45 did have an impact on the OLG, and improved the phase margin. In that measurement we also added 25% to the MICH gain.
We accepted the phases in SDF so they were in effect at our next (quick) DRMI lock, but not the gain change.
When we next locked DRMI, we measured the MICH OLG again, and here it seems that the gain is a bit high. We haven't changed the gain in the guardian since this seemed to work well, but the third attached screen shot shows the loop gain.
After this we went to DARM to RF, and manually increased the TR_CARM offset to -7. The REFL power diped to 97% of it's unlocked value while the arm cavity transmission was 24 times the single arm level, so following the plot from 62110 we'd estimate our power recycling gain was between 20 and 30.
We made two attempts to close the DHAD WFS. In the first, we stepped the TR_CARM offset to -7 and tried the guardian steps. Looking at the lockloss it looked like the sign of both DHARD loops was wrong and pulling us to a bad alignment.
In the second attempt, we stayed at the TR_CARM offset of -3 (from DARM_TO_RF), and tried manually to engage them. The yaw loop was fine and we were able to use the normal full gain. For the pitch loop, it did seem to have the wrong sign so we tried flipping the sign. The guardian would step this gain from -1 to -10 at the end of the DHAR_WFS state, we stepped it from 1 to 3, which seemed to be driving the error signal to zero, but we lost lock partway through this.
I have made three more attempts through DHARD_WFS, both unsuccessful. Each time, once in DARM_TO_RF, I've manually engaged DHARD_Y's initial gain, watched the error signal converge, then increased to the final gain without issue. I then would engage DHARD_P's initial gain with the opposite sign and watch the error signal slowly converge after many minutes. Both times, whenever I would increase the DHARD_P gain, soon after the control signal would start moving away from zero (away from the direction it came) and there would be a lockloss.
On the thrid attempt I did the same as before, but this time stepped the CARM offset from -3 to -5 before engaging DHARD_P, but ultimately this attempt didn't work either. I noticed that once the DHARD_P error signal crosses zero, DHARD_Y starts running away (?). If I'm fast enough, I can turn the gains to zero before a lockloss happens, then bring the buildups back up by adjusting PRM. This juggling act of turning the DHARD loops on and off while adjusting PRM went on for a while, but inevitably ended in a lockloss.
Elenna, Ibrahim, Ryan, Sheila
We had one more attempt at locking, we were able to close DHARD Y wfs with the guardian, and we stepped DHARD P as we stepped up the CARM offset. Elenna fixed up DRMI along the way as well.
We engaged the DHARD P loop with the usual sign and gain when the TR_CARM offset was -35. Then we let the guardian continue, and lost lock in the CARM_TO_ANALOG state.
Ibrahim has a few plans of what to try next.
I took histgrams of 8 days before the Sep13th changes and 8 days After the Sept 13th Changes.
