KAGRA Logbook

MIF (General)

kenta.tanaka - 15:09 Thursday 12 March 2026 (36560)

Investigation of mysterious GR-IR coupling: Day 4

Ushiba, Tanaka

## Abstract

Similarly with the PD case in klog36550, We succeeded in solving the couplilng issue when the beam was on REFL QPDs by replacing the power supply.

## What we did

At first, we recovered the shutter for REFL QPDs with the almost same procedure in klog36514. Then, When LSC_LOCK was in FIND_IR_RESONANCES, we opened the shutter so that the reflection beam hit on QPDs. After that, we aligned PRM manually. Figure 1 shows the time series of f DARM error/feedback and PRM guardian state from MISALIGNED_FOR_LOCK_ACQ(2300) to LOCK_ACQUISITION(2000). As PRM alignment got better, DARM signal got noisier. At last, ALS_DARM lock was lost due to the TM actuator saturation. This situation is the almost same as the one in klog31299. So we tried to solve issue by changing the power supply for QPD whitening filters.

Next, we measured DARM spectra in some different situations. Fig. 2 shows the DARM error spectra when IR beam was flushing in FIND_IR_RESONANCES. Blue lines shows the reference when IR flash amount was very low because PRM misalgined and Black lines shows the noisy reference when IR flash amount was very high because PRM aligned. Also, green lines shows the reference of yesterday's last result.

Brown shows the spectrum when the shutter for REFL QPDs was opend but all of QPD whitening filters were turned off. It seems to be the same level as green lines. Therefore, other circuits in the QPD path, ex. IQ demodulaters, did not contaminated the noise to DARM loops.

Magenta shows the spectrum when we turned on the only whitening filter for REFL QPD1 RF17. Then, DARM spectrum got noisy. This is the same situation as the PDA one. So, we changed the power supply to an independent one. Orange line in Fig.2 shows the result after changing the power supply. The noise level got quieter and got the same level as the brown one. This indicates that separating the power supply is valid for reducing the coupling.

Therefore, we need to change the power supplies for all of QPD whitening filters or to modify the current buffer in the circuits. .

This also indicates that we can use REFL QPDs during the lock acquisition term.

## Note

Above 20 Hz, the brown, green, and orange spectra is slightly larger than the most slient one (blue). This indicates the power supplies for other circuits, ex. PD interfaces, may also be changed.
Sorry, we completely forgot to turning on the other whitening filters (QPD1 RF45, QPD2, QPD3, QPD4). we need to turn on them before starting ASC commissioing.

Images attached to this report

36560_1773290353_Screenshot from 2026-03-12 10-49-57.png

36560_1773295785_Screenshot from 2026-03-12 15-09-40.png

Comments to this report:

takafumi.ushiba - 16:36 Thursday 12 March 2026 (36563)

We also examined the fluctuation of the current from the power supply to the circuit when the PRFPMI was flashing.
A rough diagram of the measurement setup is shown in fig. 1.

We connected three whitening filters (WF1, WF2, and WF3) to the external power supply (TEXIO DC power supply).
Between the +18 V from the power supply and the circuit (WF3), we inserted a 1-Ohm standard resistance to measure the current fluctuation to the circuit.
The data logger measured the voltage difference between the two edges of the resistance.

Figure 2 shows the time series data of the supply current to the circuit, and Figure 3 shows an enlarged view of Figure 2 from 190 to 200 seconds.
The red dashed line shows when we opened the laser shutter on the REFL WFS QPD path.
The green dashed line shows the time when we tuned the CARM/DARM offsets so that the PRFPMI flashes increased.
The current fluctuation becomes larger when the beam hits the QPDs and when the PRFPMI flashes increase.

Since the current spike from the power supply is less than 100 ms (which corresponds to several tens of Hz), the DC power supply must supply not only DC current, but also an AC current of ~1 A to the circuit in total because there are ten or more whitening filters for the REFL PDs/QPDs.
We are unsure whether these values can be supplied by the DC power supply in the computer room, which is far from the DGS rack. However, one possibility of the IR-GR coupling is due to the DC power supply's poor performance at high frequencies.

In any case, we need to discuss how to address the GR-IR coupling issue, as we would require an additional 10 A power supply near the ALS/IOO racks if we were to replace all the WFs for the REFL PDs/QPDs.

Images attached to this comment

36563_1773301486_power_supply_current_fluctuation_comment.png

36563_1773301489_power_supply_current_fluctuation_enlarge.png

takaaki.yokozawa - 11:53 Friday 13 March 2026 (36568)

I turned on the 5 whitening filter circuits.