Komori, Fujimoto, Tanaka

We tried to improve the stability of DRMI 3F lock. Current DRMI 3f signals (maybe MICH?) seems to has a very narrow linear range. Due to this, if build up powers fluctuate by fluctuating length or alignment, it causes the lock loss. And also the good offset value seems to be changed by the alignment for some reasons. Therefore, obtaining and maintaining good alignment is necessary to keep the 3f lock for a long time.      

## What we did

Currently, DRMI 3f lock duraion seems to become short, at most several ten seconds. We observed a specific phenomenon before the lock loss that the AS34 and POP90 dropped suddenly and error signals moved largely. Fig 4 is one of examples on lossing the lock. We confirmed that we can hand over the 3f signal from the 1f signal for 1 DoF. Also we could handed over the 3f signals of PRCL and MICH at the same time. However, if we handed over the 3f signals to the 1f signals for SRCL and either one of other two, the sudden drop happens. So we assumed the coupling in 3f sensors between SRCL and others seems to be large. we moved the decoupling.

### Decoupling DRMI 3f sensors

First, we tried to subtract the PRM motion from SRCL and MICH 3f sensors. We measured the 3f sensors' responses when PRM was excited with the same manner in klog37312. We applied for LSC-{CARM,XARM,YARM}_IN1 as 3f sensors of PRCL, SRCL, and MICH, respectively. Fig.1 shows the ratio between PRCL1 and 3f sensors (CARM, XARM, YARM). the ratio between PRCL 1f and SRCL and MICH 3fs seems to become smaller than the one between PRCL 1f to PRCL 3f.  From the results, we obtained the matrix like Fig.2. And Fig.3 show the spectra of each sensor and coherences among them. There seems to be lower coherences of 3f sensors from 10 Hz to 100 Hz. 

### lock loss investigation

We used these decoupled sensors to lock DRMI 3f. The situation became slightly better. We can hand over the 3f signal from the 1f signals of PRCL and MICH /SRCL maybe thanks to this. However, we could not handed over the 3f signals of MICH and SRCL at the same time. According to this results, we assumed that some or all error signals has narrow linear range.

We swept the offset of the 3f signals one by one when we handed over the only one 3f signal to estimate the range roughly. We obtained the center value of the range and applied the value for the setpoint. Then, we tried to 3f locked. but the situation was not changed.

At last, Fujimoto-kun found that the MICH offset, which makes AS DC power minimum seems to be better. In this offset, 3F lock kept ~30 mins.  

### reproducibility of the offset

Next day, we tried to lock DRMI with 3f with the same offsets which were found by Fujimoto-kun. However, we failed the lock. And we found that MICH offset seems to be changed by the alignment, especially SRM.  Therefore, we need the method to obtain and maintain good alignments.