[Yokozawa, Ushiba, Tanaka]

## Abstract

(this is the work on 2024.06.19)

We tried and succeeded in the PRX lock with 3f and 1f signals. However, PR2 or PR3 oplev glitches sometimes seem to lose the lock. Also, I implemented the ADS for PRX with PRM and IMMT2 to maximize POP90.

The gain of 3f lock is 6~7dB higher than the previous one maybe because the phasing. The gain of 1f lock is ~1dB lower than the previous one.  Anyway, we need to investigate difference between current and previous configration carefully. 

## What we did

### PRX 3f lock

• I aligned ITMX, PRM, and misaligned ITMY with each guardian. I tweaked the PRM alignment to maximize the POP 90 flash, then I increased the flach amplitude of POP90 (K1:LSC-POP_PDA2_RF90_I_ERR_DQ) to ~0.012. 
• I requested the VERTEX guardian to go to the "PRX_3F_LOCKED" state, but it failed because the POP90 seemed -0.012 when the PRCL loop closed and then the VERTEX guardian lost the lock judging from this value. So I closed the PRCL loop manually and succeeded in the PRX 3F lock.
• Yokozawa-san and I checked the phasing for POP90 by exciting PRX length with PRM at 40 Hz and found that the peak in the POP90_Q was larger than the one in the POP90_I signal. This indicated that the phase of POP90 rotated near 270 deg. (Maybe this phase rotation was caused by the re-cabling for POP90 in the WFSf3 work in klog29779?) But anyway, we derived the rotation angle from the amplitude ratio between the peaks of the I and Q signals to minimize the Q signal. And then we implemented the angle in K1:LSC-POP_PDA2_RF90_PHASE_R. Then the sign of POP90 was flipped.
• After that, I requested the VERTEX guardian to go to the "PRX_3F_LOCKED" state again and it succeeded at this time. On the other hand, the lock did not keep for a long time. I found when the PR2 or PR3 oplev glitches occurred, the POP90 power got lower than the threshold of the lock(0.01). (fig.1) So I decreased the threshold to 0.005 temporarily.
• Also, I implemented the ADS for PRX with PRM and IMMT2 to maximize POP90. Before the implementation, I performed the phasing for their ADS signals by following the procedure in klog25199.
  • I modified the "ENGAGE_ADS_FOR_PRX" state in the ASC_LOCK guardian to input each phase value when we request the ASC_LOCK to go to "PRX_ADS_ENGAGED" and also add the script to restore the IMMT2 phase value in the "DOWN" state.
• After that, I requested the ASC_LOCK guardian to go to "PRX_ADS_ENGAGED" and it seemed to work well. the POP90 increased to ~0.015 by ADS. (fig. 2)
• Moreover, I performed the phasing for REFL RF135 and POP RF45 by exciting the PRC length with PRM. Fig. 3 shows the result. We succeeded in minimizing both Q signals of REFL RF135 and POP RF45.
• At last, I measured the OLTF of PRX3f. At that time, I turned off the ADS. Fig. 4 shows the result. Current PRX 3F lock UGF was 35 Hz, it is higher than previous one(~20Hz). I'm not sure the reason for now. One possibility is that the phasing optimization in this time is effective. we need to investigate difference between current and previous configration carefully. 

### PRX 1F lock

• I requested the VERTEX guardian to go to the "PRX_1F_LOCKED" state, then the lock itself was succeeded but the sign of POP90 was flipped again. Also the power on a REFL PD decreased but the power on a POP PD increased (fig.5). This indicates the carrier is resonance unlike the 3f lock case. So I modified the VERTEX guardian to resonate not carrier but the sideband in this time.
• I confirmed the phase value with PRCL excitation. there is no difference between the 3f case and the 1f case.(Fig. 6)
• At last, I measured the OLTF of PRX1f. At that time, I turned off the ADS. Fig. 7 shows the result.  The gain was lower slightly.

[Kenta, Yuzu]

We investigated the cause of the 6dB difference in the Fig. The 2.9dB (x1.4) difference could be explained by the phasing work(klog#30006). Still the 3dB difference is unexplainable.

Details

• We checked the filter bank at the latest (2024/06/19 11:38:19 UTC) and the past measurement (2022/12/04 06:56:33 UTC). But, the value in the filter bank (FM8) was not changed, except for the 1e-9 correction. (Fig)
• The normalization factors for input powers from the IMC are consistent with the input power. The normalization was not problem at both measurements. (Fig)
• The demodulated amplitudes were 0.079041 (updated at 2022/12/04, Fig) and 0.111175 (updated at 2024/06/19, Fig). The ratio is factor 1.4, that is 2.9dB (Fig).

Though there is only 1dB difference in the case of 1f lock, note that it's 1dB decrease not 1dB increase. Checking excited amplitude before and after phasing with same manner as the 3f case in klog#30037, excited amplitude was increased from -66.9dB to -65.3dB (please compare the left top panels of Fig.1 and Fig.2). OLTF gain of 1f lock decreased as 1dB in spite of increasing gain as 1.6dB by phasing. In other words, OLTF of 1f lock has unknown difference as -2.6dB from the past measurement.

According to the investigation about gain difference in the 3f case (klog#30037), unknown difference on the OLTF of 3f lock between measurements in 2024 and in 2022 is +3dB. Because actuator (PRM_TM_L) is common in 1f lock and 3f lock, the fact that a sign of unknown difference is opposite in 1f lock (+3dB) and in 3f lock (-2.6dB) couldn't be explained by only the difference in the actuator efficiency.

So it seems that all optical gain of 1f, optical gain of 3f, and actuator efficiency of PRM_TM_L changed. Or only optical gains of 1f and 3f changed to the opposite direction each other. As for the latter, I cannot think of a reasonable cause, but is there a good idea by experts? (How about beam clip only at POP?)

In addition to the consideration about changes in optics, more careful checks about digital filters, gain, matrices seem to be required. In klog#30037, only about PRCgain in LSC-REFL_PDA2_RF135_I FilterBank was mentioned. But all other filters except LSC-PRCL1 and LSC-PRCL2, EPICS norm factors, and in/out matrices can affect this issue.

(miss)

Cintinuous work of klog#30037

I compared the power at REFL_PDA1* (Fig). The DC power was decreased from 11600 to 8400, when we compare the data on 2022 and 2024. This is 2.8 dB decrease (Fig). In the both measurements, the VERTEX guardian was in PRX_3F_LOCKED(250).
As a next step, I will check the changes digital filters, gain, matrices between the recent and past measurements, as YamaT-san suggested.

* Note that I checked the value of REFL_PDA1, because I heard the sensor of REFL_PDA2 to monitor the DC power is broken.