We further investigated the MICH signal observed in REFL51.

Estimation of Coupling Contributions Including Phase Information

First, we estimated the coupling contributions from PRCL and SRCL, taking into account the phases of their TFs.
For REFL51I, the phases of the coupling TFs through the PRCL and SRCL paths are 127° and −37°, respectively, which are nearly opposite in phase.
As a result, the two coupling contributions largely cancel each other, and the estimated total coupling gain for MICH-to-REFL51I is −31.2 dB, whereas the measured gain is −15.4 dB.
This suggests that the observed signal in REFL51I is dominated by the genuine MICH response rather than the actuator coupling contributions via PRCL and SRCL.

Dependence on the Demodulation Phase

Next, we estimated the coupling contributions at various REFL51 demodulation phases.
As described in klog37830, the PRCL and SRCL signals in REFL51Q are minimized at demodulation phases of 130° and 145°, respectively.
Using this information, we calculated the PRCL/SRCL coupling contributions to REFL51I/Q as functions of the demodulation phase.
We also measured the MICH-to-REFL51I/Q TFs by exciting MICH(Figure.1) and compared the measured gains with the estimated coupling contributions (Figure.2).
The direct coupling contributions from PRCL and SRCL are also shown.
The results indicate that the estimated actuator coupling contributions are negligible compared with the measured MICH-to-REFL51 signals, except where the measured gains fall below approximately −40 dB.
However, the response to MICH is much weaker than that to PRCL and is comparable in magnitude to the SRCL response.

According to the optical design, the MICH and SRCL signals are expected to be well separated in the I/Q plane.
However, the measured responses are nearly aligned, with a phase difference of only approximately 5°.
The reason for this discrepancy is not yet understood.

Note: We cannot rule out the possibility that the observed "MICH signal" contains additional coupling contributions that have not been considered in our estimation.
If such contributions are significant, the genuine MICH response could be much smaller than the measured value.

[Sugioka, Hirose, Komori (online), Ushiba, Fujimoto]

Abstract

We investigated whether the MICH signals observed in REFL51-I and REFL135-Q were actual MICH signals or originated from coupling due to imperfect decoupling of the MICH actuator.
As a result, we found that the MICH signal observed in REFL51-I was likely due to coupling.
On the other hand, the estimated coupling contributions in REFL135-Q were more than one order of magnitude smaller than the observed signal, suggesting that REFL135-Q was likely sensing the actual MICH signal.
In addition, the signal-to-noise ratio relative to the noise floor was about six times better in REFL135-Q than in REFL51-Q, suggesting that REFL135-Q could be used for MICH control.
However, as reported in ⁠klog 37835, REFL135-Q also has coupling from PRCL and SRCL. In particular, the PRCL coupling is larger than the MICH response. Therefore, decoupling by subtracting the coupled signals is essential.

Details

Previous investigations of signal coupling in the DRMI suggested that MICH signals might appear in REFL51-I and REFL135-Q.
However, when measuring the MICH response of each sensor, we used the MICH actuator, which combines the BS, PRM, and SRM actuators. If the MICH actuator was not sufficiently decoupled, it could also excite PRCL and SRCL. Therefore, the observed signals might have originated from coupling through PRCL and SRCL rather than from MICH itself.

To investigate this possibility, we estimated the excitation amplitudes of PRCL and SRCL when injecting a signal into the MICH actuator, using the 1f sensors.
We then multiplied these amplitudes by the transfer functions from the 1f sensors to the 3f sensors to estimate the peak amplitudes appearing in REFL51-I and REFL135-Q through PRCL and SRCL coupling.
By comparing these estimated amplitudes with the measured amplitudes, we evaluated whether the observed signals originated from MICH itself or from coupling.

Figs. 1, 2, and 3 show the spectra of the signals when excitations were injected into MICH, PRCL, and SRCL, respectively, while the DRMI was locked with the 1f sensors.
We used the peak amplitudes and transfer functions obtained from these measurements to estimate the coupling contributions.

MICH to REFL51-I

The measured peak amplitude in REFL51-I when exciting MICH and the estimated peak amplitudes due to coupling through PRCL and SRCL are as follows:

• Measured peak amplitude in REFL51-I: 0.074
• Estimated peak amplitude through PRCL: 0.022
• Estimated peak amplitude through SRCL: 0.031

Since the estimated coupling contributions are of the same order of magnitude as the measured peak amplitude, the signal observed in REFL51-I when exciting MICH is likely dominated by coupling.
However, if the relative phase between the PRCL and SRCL coupling contributions is close to 180 degrees, these contributions could cancel each other out.
Therefore, we cannot rule out the possibility that most of the observed signal is actually a genuine MICH signal.

(The following values were used to calculate the estimated peak amplitudes:

• Peak amplitude in PRCL_IN1: 0.015
• Transfer function from PRCL_IN1 to REFL51-I when exciting PRCL: 3.16 dB
• Peak amplitude in SRCL_IN1: 0.11
• Transfer function from SRCL_IN1 to REFL51-I when exciting SRCL: -11 dB)

MICH to REFL135-Q

The measured peak amplitude in REFL135-Q when exciting MICH and the estimated peak amplitudes due to coupling through PRCL and SRCL are as follows:

• Measured peak amplitude in REFL135-Q: 0.023
• Estimated peak amplitude through PRCL: 0.0017
• Estimated peak amplitude through SRCL: 0.0019

Since the estimated coupling contributions are more than one order of magnitude smaller than the measured peak amplitude, the observed signal is likely a genuine MICH signal rather than a signal originating from coupling.
Therefore, REFL135-Q could potentially be used for MICH control.
However, as reported in ⁠klog 37835, REFL135-Q also has coupling from PRCL and SRCL. In particular, the PRCL coupling is larger than the MICH response. Therefore, decoupling by subtracting the coupled signals is essential.

(The following values were used to calculate the estimated peak amplitudes:

• Peak amplitude in PRCL_IN1: 0.015
• Transfer function from PRCL_IN1 to REFL135-Q when exciting PRCL: -19.1 dB
• Peak amplitude in SRCL_IN1: 0.11
• Transfer function from SRCL_IN1 to REFL135-Q when exciting SRCL: -35.3 dB)

Abstract:

MICH signals on REFL 51Q seems reducing by a factor of 10 compared to the values before MC length tuning.

Detail:

I compared the ratio between POP 17Q and REFL51Q/REFL135Q while exciting MICH DoFs (fig1).
Following table shows the signals (preak height at excitation frequency) of PDs during the measurement.

From these results, I calculated the ratio between POP 17Q (we thought MICH signals are dominant on this signal) and the others.
The result is as follows:

So, the ratio between REFL51Q and POP17Q was reduced by a factor of 10 after MC length adjustment.
On the other hands, ratio between REFL135Q and POP17Q was slightly improved.

We are not so sure why this signal reduction happens, so further investigation is necessary.

I ran a finesse script and confirmed that REFL135 (and POP45) is much more sensitive to the BS motion than REFL51 (and POP17). Probably better decoupling of the BS motion or actuating MICH in different ways (eg. both ITMs oppositely) is necessary. You'll see the finesse code here.

Fig. 1 shows the frequency response of the sensors (1f/3f, f1/f2, POP/REFL) when we move BS for MICH control; blue-orange lines are for 1f, green-red lines are for 3f. The demodulation phases of POP45 and REFL135 are optimised for PRCL, and those of POP17 and REFL51 are optimised for SRCL. The most sensitive signal to BS motion is REFL135I, and it's ~ 400 times more sensitive than REFL51Q.

Fig. 2 shows the frequency response when we move the ITMs in the opposite direction, rather than BS, to keep PRCL and SRCL unchanged. In this case, the signals are well decoupled, of course.

I'm not sure how precisely the actuators are decoupled in terms of DoFs for DRMI, but much harder decoupling would be necessary to resolve the coupling between REFL51-REFL135 and PRCL-MICH.

[Fujimoto, Komori(online), Sugioka]

Abstract:

According to klog37833, MICH signal appears in REFL135Q.
We therefore investigated the potential of taking MICH3f signal from REFL135Q.

Details:

We measured the response of each PD signal to MICH/PRCL/SRCL in the same manner as klog37832 while DRMI1f locked(Figure.1,2,3).
Although we minimized the PRCL signal in REFL135Q by phasing(klog37829), REFL135Q is approximately 4 times more sensitive to PRCL.
Simply using REFL135Q for MICH signal thus seems difficult.

Furthermore, since an inexplicably large MICH signal is present  in REFL51I and REFL135I, we suspected that these signals are originate from MICH actuator coupling.
Although we previously performed MICH actuator compensation to avoid changing PRCL and SRCL(klog37812, klog37816), imperfections in MICH actuator decoupling might still remain.
Then we estimated how much this actuator coupling causes the 'false MICH signal' in each PD siganl by  measuring the MICH IN1-to-PRCL/SRCL IN1 TFs and PRCL/SRCL-to-REFL135/51 I/Q TFs.

Since there should be two coupling paths for each PD signal- one is via PRCL and the other is via SRCL- we estimated the coupling factor of all of these coupling paths.
For example, in the case of REFL135Q, the TF gain for the first path is the product of MICH-to-PRCL gain(~-26dB) and PRCL-to-REFL135Q gain(~-15dB), while for the second path, it is the product of MICH-to-SRCL gain(-6dB) and SRCL-to-REFL135Q gain(~-40dB).
The total coupling TF gain is thus estimated to be approximately -37dB.
Likewise, MICH-to-REFL135I, -REFL51I and -REFL51Q are estimated to be approximately -5dB, -13dB and -34dB respectively.

These results roughly agree with the measured MICH-to-REFL135/51 I/Q values.
In particular, MICH-to-REFL135I and -51I exhibit high coherence (>0.9), and the estimated values match the measurements well (within ±2 dB).
If the unmeasured REFL135 demodulation phase response of SRCL peaks around -90°, this coupling picture becomes consistent with the REFL135 demodulation phase response of MICH (klog37833).
In this case, the contribution from the SRCL path would be maximized at 135I with a demodulation phase of -119°, while that of the PRCL path would peak at a demodulation phase of around -90°, resulting in the largest overall response at a demodulation phase of around -100°.

A REFL135 demodulation phase sweep should be performed to test this hypothesis. (We actually carried out this measurement using Hirose-san's scripts, but somehow we could not find the saved data file.)
If this hypothesis proves correct, it implies that the genuine MICH signal in REFL is too faint to be used for the MICH sensor, meaning new approaches to extracting the MICH signal from other ports may be necessary.

[Sugioka, Komori (online), Fujimoto]

Abstract

We tried to decouple the DRMI 1f signals again.
Although some couplings were slightly reduced, the PRCL-to-MICH coupling, which we particularly wanted to suppress, became worse.
Therefore, we did not implement the newly obtained input matrix in the Guardian.
The most significant couplings before the decoupling were changed as follows:

• PRCL-to-MICH coupling: −1.9 dB → 7.5 dB
• SRCL-to-MICH coupling: −7.4 dB → −11 dB
• MICH-to-SRCL coupling: −5.2 dB → −5.1 dB

Coupling Measurements Before Re-Decoupling

Before the decoupling work, we evaluated the coupling between the DRMI 1f sensors by exciting each DOF while the DRMI was locked with 1f signals and measuring the transfer functions between the error signals calibrated in nm.
Fig. 1 shows the measured couplings.
The following couplings were particularly large:

• PRCL-to-MICH coupling: −1.9 dB
• SRCL-to-MICH coupling: −7.4 dB
• MICH-to-SRCL coupling: −5.2 dB

Re-Decoupling of the 1f Signals

POP45 Phasing

We adjusted the demodulation phase of POP45 to minimize the peak in POP45-Q when PRCL was excited.
Fig. 2 shows the spectra before and after the adjustment.
The demodulation phase was changed as follows:

• K1:LSC-POP_PDA1_RF45_PHASE_R: 73.870 → 73.300

POP17 Phasing

We adjusted the demodulation phase of POP17 to minimize the peak in POP17-Q when SRCL was excited.
Fig. 3 shows the spectra before and after the adjustment.
The demodulation phase was changed as follows:

• K1:LSC-POP_PDA1_RF17_PHASE_R: −139.600 → −141.300

Reduction of PRCL-to-SRCL Coupling

We attempted to remove the PRCL-to-SRCL coupling by subtracting POP45-I from POP17-I.
Specifically, we adjusted the subtraction factor to minimize the peak in POP17-I when PRCL was excited.
The resulting signal was sent to CARM_IN1 to check the reduction in the spectrum.
Fig. 4 shows the reduction in coupling achieved by this adjustment. The coupling was reduced to approximately 1/2 of its original value.
The adjusted input matrix elements were:

• POP17-I to CARM: 1.00
• POP45-I to CARM: −2.50

Reduction of SRCL-to-PRCL Coupling

We attempted to remove the SRCL-to-PRCL coupling by subtracting POP17-I from POP45-I.
Specifically, we adjusted the subtraction factor to minimize the peak in POP45-I when SRCL was excited.
The resulting signal was sent to CARM_IN1 to check the reduction in the spectrum.
Fig. 5 shows the reduction in coupling achieved by this adjustment. The coupling was reduced to approximately 1/5 of its original value.
The adjusted input matrix elements were:

• POP45-I to CARM: 1.00
• POP17-I to CARM: −0.032

Reduction of PRCL-to-MICH Coupling

We attempted to remove the PRCL-to-MICH coupling by subtracting the PRCL signal from POP17-Q.
Specifically, we adjusted the subtraction factor to minimize the peak in POP17-Q when PRCL was excited, using the PRCL signal obtained in the previous step.
The PRCL signal was sent to DARM_IN1, while the resulting signal after subtraction was sent to CARM_IN1 to check the reduction in the spectrum.
Fig. 6 shows the reduction in coupling achieved by this adjustment. The coupling was reduced to approximately 1/4 of its original value.
The adjusted input matrix elements were:

• POP17-Q to CARM: 1.00
• POP17-I to CARM: 0.0203
• POP45-I to CARM: −0.636

OLTF Measurements

We adjusted the overall gains of the input matrix elements so that the OLTFs obtained using the newly decoupled PRCL, SRCL, and MICH signals matched those measured before the decoupling.
Figs. 7, 8, and 9 show the resulting OLTFs of PRCL, SRCL, and MICH, respectively.
The final input matrix elements obtained from this adjustment were:

PRCL

• POP17-I to PRCL: −0.16
• POP45-I to PRCL: 5.0

SRCL

• POP17-I to SRCL: 3.55
• POP45-I to SRCL: −8.88

MICH

• POP17-I to MICH: −0.203
• POP17-Q to MICH: −10.0
• POP45-I to MICH: 6.36

Coupling Measurements After Decoupling

Fig. 10 summarizes the couplings measured after the decoupling adjustment.
Comparing these results with those measured before the adjustment, the most significant couplings changed as follows:

• PRCL-to-MICH coupling: −1.9 dB → 7.5 dB
• SRCL-to-MICH coupling: −7.4 dB → −11 dB
• MICH-to-SRCL coupling: −5.2 dB → −5.1 dB

Although the SRCL-to-MICH coupling was slightly improved, the PRCL-to-MICH coupling, which we particularly wanted to suppress, became worse by approximately 9.4 dB.
The reason for this degradation is unclear, since we adjusted the subtraction factors at each step to reduce the coupling compared with the original signals.
However, we observed that the coupling fluctuated over time during the decoupling work. Therefore, the non-stationarity of the coupling may have prevented us from subtracting the coupled signals properly.

Note: Since the newly obtained input matrix resulted in worse coupling, we restored all input matrix elements to their original values.

[Kataoka, Sugioka, Fujimoto, Ushiba, Hirose]

Overview

Since the MICH and SRCL signals at RF51 are likely to have similar demodulation phases, we wrote a Python script to measure the RF signals while sweeping the demodulation phase. The script was developed for DRMI commissioning based on the script of klog36343. (/users/cHirose/script/schnupp_asymmetry_measurement_DRMI.py)

1. RF51 demodulation phase sweep with MICH excitation 

We first measured the RF51 signal while exciting MICH.

• Excitation frequency: 210 Hz

• OSC CLK gain: 30000

• S/N ratio: approximately 3（When RF51 demodulation phase: −145 deg）

The measured signal did not show a clear sinusoidal dependence on the demodulation phase, possibly due to the low S/N ratio.
In addition, the excitation amplitude of 50000 was directly applied to the OUTPUT matrix, which may have been too large. The DRMI 1f lock was lost during the measurement.
Therefore, the phase sweep could only be performed from −180 deg to −23 deg (FIg1, FIG2).

2. RF51 demodulation phase sweep with SRCL excitation

Next, we measured the RF51 signal while exciting SRCL.

• OSC CLK gain: 5000

• S/N ratio: greater than 10（When RF51 demodulation phase: −180 deg）

The measured signal showed a clear sinusoidal dependence on the demodulation phase.
At the current RF51 demodulation phase setting for 3f (−145 deg), the Q signal was close to its minimum, as expected.(FIg3, FIG4)

3. RF135 demodulation phase sweep with MICH excitation 

Finally, we measured the RF135 signal while exciting MICH.(FIg5, FIG6)

• Excitation frequency: 210 Hz

• OSC CLK gain: 50000

• S/N ratio: approximately 10（When RF135 demodulation phase: −180 deg）

The S/N ratio of RF135 was better than that of RF51 under the same MICH excitation conditions. 
Currently, the RF135 demodulation phase is set to -119.8 deg, where the RF135 I signal is maximized for PRCL excitation. 
It seems difficult to achieve coupling between MICH and PRCL due to the phasing of RF135.

[Ushiba, Hirose, Aritomi, Fujimoto, Kataoka, Sugioka]

Abstract:

To construct input matrix for DRMI3F and decouple each signals, we measured coupling levels.
We then assessed the 3f signals and attempted the 1f-to-3f handover. 
The handover failed, and it is likely that the MICH signal is still not sufficiently decoupled.

Details:

We measured the MICH-, PRCL-, and SRCL-to-3f PD signal transfer functions (TFs) while DRMI 1f was locked. 
We injected a 150.125 Hz excitation into each DoF 1f control loop and measured the TF from the 1f IN1 signal to the 3f PD signal. 
Since the IN1 signal is calibrated in nanometers, the gain is calibrated in dBcnt/nm.

The measurements were performed with the REFL51 demodulation phase set to  145(Figure.1) and 130(Figure.2)(as detailed in klog37830).
The left/middle/right pads shows the measurements under PRCL/SRCL/MICH excitation, respectively.
In both demodulation phase cases, the MICH-to-REFL51Q gain was remarkably small even compared to MICH-to-REFL51I.
This might imply that MICH signal is either in phase or in antiphase with PRCL and SRCL.

Based on these measurements, we constructed the 3f control signal and evaluated it. This part was performed exclusively with a demodulation phase of 145°.
We constructed a 3*3 matrix with nine each measured TF gain, and used the inverse of this matrix for Input Matrix.
We utilized the CARM, DARM, and XARM IN1 channels—which are not currently used for control—as the 3f MICH,PRCL, and SRCL signals.
Then we injected 150.125Hz excitation to each DoF 1f contorl loop again, and measured TF from 1f IN1signal to 3f IN1 signal(Figure.3).
Figure.3 shows that 3f PRCL and SRCL signal is well decoupled.
Although the 3f MICH signal exhibits some coupling from PRCL and SRCL, its coherence is low.
We also measured the OLTF of these 3f signals while DRMI1f locked(Figure.4 for MICH, Figure.5 for PRCL, Figure.6 for SRCL) which generally matched to those of 1f.

Subsequently, we attempted the 1f-to-3f handover. 
First, we tried the MICH handover, but it was unsuccessful. 
Next, we attempted to hand over PRCL and SRCL first, followed by MICH control. 
The PRCL handover—which had already been achieved yesterday—was successful, but PRCL 3f was unstable, and the SRCL handover attempt was skipped.

Plans:

There are several potential approaches for DRMI 3f locking:

1. The surprisingly weak MICH signal in REFL51Q is unusual; therefore, further investigation was conducted, which will be posted by Hirose-san later.

2. Taking the SRCL signal from another port and using REFL51I for MICH should also be feasible.

3. There is a possibility that imperfect phasing and decoupling in 1f compromised the matrix adjustment. Consequently, further fine-tuning of the 1f control could serve as a solution.

[Kataoka, Sugioka, Hirose, Aritomi, Tanaka (online), Komori (online), Ushiba, Fujimoto]

Abstract

We performed decoupling and calibration of the DRMI 3f sensors.
We also attempted to lock the DRMI using the calibrated 3f signals. We successfully performed the 1f-to-3f transition for PRCL and SRCL individually.
However, the transition for MICH was unsuccessful. Further adjustments, such as optimizing the error signal offset and improving the sensor decoupling, are likely necessary.

Details: DRMI 3f sensor decoupling

Since we successfully locked the DRMI with 1f signals yesterday, we used the 1f-locked state to perform decoupling of the 3f sensors.

REFL135-I demodulation phase adjustment

While the DRMI was locked with 1f signals, we excited the PRM at a frequency outside the PRCL control bandwidth and adjusted the REFL135 demodulation phase to minimize the peak in the REFL135-Q spectrum.
The demodulation phase was changed as follows, and the resulting spectra are shown in Fig. 1.

• K1:LSC-REFL_PDA2_RF135_PHASE_R: 60.2 ⇒ -119.8

REFL51-I/Q demodulation phase adjustment

While the DRMI was locked with 1f signals, we excited the SRM at a frequency outside the SRCL control bandwidth and adjusted the REFL51 demodulation phase to minimize the peak in the REFL51-Q spectrum (i.e., to minimize the SRCL coupling).
The demodulation phase was changed as follows, and the resulting spectra are shown in Fig. 2.

• K1:LSC-REFL_PDA2_RF51_PHASE_R: -35 ⇒ 145

Reduction of PRCL-to-SRCL coupling

Since REFL51-I contains PRCL coupling, we decoupled the SRCL signal by subtracting REFL135-I from REFL51-I.

First, while the DRMI was locked with 1f signals, we excited the PRM at a frequency outside the control bandwidth and measured the transfer function from REFL135-I to REFL51-I at the excitation frequency.
We then entered the measured value into the input matrix to subtract the PRCL contribution from REFL51-I.
To check the spectrum of the decoupled REFL51-I signal, we routed it through the unused CARM filter bank and monitored its IN1 signal.

Fig. 3 shows the spectra of the raw REFL51-I signal and the decoupled signal (K1:LSC-CARM_IN1). The PRCL coupling was reduced by at least a factor of approximately 7.
The input matrix values were:

• REFL51-I to CARM: 1.0
• REFL135-I to CARM: -0.218

SRCL 3f calibration

We calibrated the decoupled SRCL 3f signal into nm, as for the 1f signal, by applying a constant scaling factor.

Specifically, while the DRMI was locked with 1f signals, we measured the transfer function from SRCL1_IN2 to CARM_IN1 to obtain the 3f OLTF.
We then adjusted the scaling factors in the input matrix so that the 3f OLTF matched the 1f OLTF.

Fig. 4 shows the 1f OLTF and the calibrated 3f OLTF for SRCL.
The input matrix values were:

• REFL51-I to CARM: 4.45
• REFL135-I to CARM: -0.969

Reduction of SRCL-to-PRCL coupling

Since REFL135-I may also contain SRCL coupling, we decoupled the PRCL signal by subtracting REFL51-I from REFL135-I.

First, while the DRMI was locked with 1f signals, we excited the SRM at a frequency outside the control bandwidth and measured the transfer function from REFL51-I to REFL135-I at the excitation frequency.
We then entered the measured value into the input matrix to subtract the SRCL contribution from REFL135-I.
To check the spectrum of the decoupled REFL135-I signal, we routed it through the unused CARM filter bank and monitored its IN1 signal.

Fig. 5 shows the spectra of the raw REFL135-I signal and the decoupled signal (K1:LSC-CARM_IN1). The SRCL coupling was reduced to the noise floor, corresponding to a reduction of at least approximately one order of magnitude.
The input matrix values were:

• REFL135-I to CARM: 1.0
• REFL51-I to CARM: -1.8

*Note: While writing this report, I noticed that the noise floor increased by approximately a factor of 2 after the subtraction. This is because the relatively noisy REFL51-I signal was used for the subtraction, adding noise to the decoupled signal.
However, as described below, we were able to lock PRCL with the 3f signal alone, so this increase in the noise level may not be a serious problem.

PRCL 3f calibration

We calibrated the decoupled PRCL 3f signal into nm, as for the 1f signal, by applying a constant scaling factor.

Specifically, while the DRMI was locked with 1f signals, we measured the transfer function from PRCL1_IN2 to CARM_IN1 to obtain the 3f OLTF.
We then adjusted the scaling factors in the input matrix so that the 3f OLTF matched the 1f OLTF.
Fig. 6 shows the 1f OLTF and the calibrated 3f OLTF for PRCL.

The input matrix values were:

• REFL135-I to CARM: 0.208
• REFL51-I to CARM: -0.375

Reduction of PRCL-to-MICH coupling

We decoupled the MICH signal by subtracting REFL135-I from REFL51-Q.

First, while the DRMI was locked with 1f signals, we excited the PRM at a frequency outside the control bandwidth and measured the transfer function from REFL135-I to REFL51-Q at the excitation frequency.
We then entered the measured value into the input matrix to subtract the PRCL contribution from REFL51-Q.
To check the spectrum of the decoupled REFL51-Q signal, we routed it through the unused CARM filter bank and monitored its IN1 signal.

Fig. 7 shows the spectra of the raw REFL51-Q signal and the decoupled signal (K1:LSC-CARM_IN1). However, the PRCL coupling was reduced by only approximately a factor of 2.
The input matrix values were:

• REFL51-Q to CARM: 1.0
• REFL135-I to CARM: -0.05

In this measurement, the peak in the REFL51-Q spectrum was only about 6 times higher than the noise floor. A larger excitation may allow more accurate adjustment in the next measurement.

MICH 3f calibration

We calibrated the decoupled MICH 3f signal into nm, as for the 1f signal, by applying a constant scaling factor.

Specifically, while the DRMI was locked with 1f signals, we measured the transfer function from MICH1_IN2 to CARM_IN1 to obtain the 3f OLTF.
We then adjusted the scaling factors in the input matrix so that the 3f OLTF matched the 1f OLTF.

Fig. 8 shows the 1f OLTF and the calibrated 3f OLTF for MICH.
The input matrix values were:

• REFL51-Q to CARM: 36.7
• REFL135-I to CARM: -1.84

Details: DRMI 3f lock trial

We attempted to lock the DRMI using the calibrated 3f signals described above, together with the filters, gains, and offsets previously used for the 3f lock. However, we could not acquire the lock at all.
Therefore, we investigated the problem by handing over one DOF at a time from 1f to 3f while keeping the other DOFs locked with 1f signals.

PRCL 3f handover (successful)

We successfully handed over PRCL from 1f to 3f using the same filters and gain settings as for the 1f lock.

SRCL 3f handover (successful)

When we attempted to hand over SRCL from 1f to 3f using the same filters and gain settings, the feedback signal saturated and the lock was lost.
We then reduced the SRCL1 gain from 1.4 to 0.7, which allowed us to complete the handover without any problems.

MICH 3f handover (unsuccessful)

For MICH, the feedback signal initially saturated, as in the case of SRCL.
We therefore reduced the ELP from 500 to 100 and the gain from 9.0 to 2.0.
With these changes, the feedback signal no longer saturated during the transition. However, AS DC gradually drifted, and the lock was eventually lost.
This suggests that we need to optimize the MICH offset for the 3f signal.
In addition, when we measured the PRCL-to-MICH coupling again after some time, we found that the coupling had become worse.
Furthermore, the gain and phase of the 3f OLTF had changed significantly compared with those measured during the calibration described above.
Further investigation is necessary for MICH.

Plans for tomorrow

When we first succeeded in locking the DRMI with 3f signals (⁠klog 37297), the sensor decoupling was not as well optimized as it is now.
Therefore, we would like to spend a little more time trying to achieve the 3f lock using the parameters adjusted yesterday.

• First, try handing over both PRCL and SRCL to 3f simultaneously.
• Find an appropriate MICH offset for the 3f signal, since an incorrect offset may cause a large MICH detuning during the handover.
• If we want to further improve the MICH decoupling, one possible approach is to choose a different REFL51-I/Q demodulation phase, as discussed in ⁠klog 37830.
• It may also be useful to diagonalize the MICH actuators using the DRMI sensors, as we did previously.

Concerns

Unexplained peak around 14 Hz in the OLTFs

All the OLTFs measured this time, for both 1f and 3f, show an unexplained peak around 14 Hz.
This peak was not present in the OLTFs measured during the recovery work in October (⁠klog 37681).
Although this peak is probably not causing control instability, its origin is still unclear and worth investigating.

Significant change in the MICH optical gain?

In this measurement, the input matrix coefficient required to calibrate REFL51-Q into the MICH error signal in nm was 36.7.
However, during the recovery work in October (⁠klog 37681), this coefficient was 4.8.
Thus, the required coefficient is now almost 7 times larger than before.
This suggests that the MICH optical gain (or possibly the actuator gain?) has decreased to approximately 1/7 of its previous value.
Such a large reduction is not expected from the IMC length adjustment alone, and the cause is currently unknown.

[Sugioka]

Abstract:

I measured the coupling of PRCL and SRCL into the MICH signal in DRMI 3f (REFL51Q) with various demodulation phases for REFL51.

Background:

As will be reported by Fujimoto-san later, we performed REFL51 phasing by minimizing the peak in REFL51Q while DRMI1f locked and SRCL was excited.
We also set the input matrix to minimize the peak in the MICH signal during PRCL excitation.
We then attempted the DRMI control handover from 1f to 3f; however, the MICH handover failed. 
This indicates that the MICH signal in 3f is not sufficiently decoupled.

Details:

According to Kawakami-kun's simulation (the figure on the right side of the 11th slide in the latter file of JGW-G2617586-v2), there should be an optimal REFL51 demodulation phase that minimizes the PRCL signal. Therefore,minimizing PRCL-to-MICH coupling via phasing and subtracting SRCL using INMTRX is a potential solution for achieving DRMI 3f lock. I measured the coupling of PRCL and SRCL into REFL51Q across various REFL51 demodulation phases.
As Fujimoto-san will post later, SRCL signal was minimized with the demodulation phase of 145 while exciting SRCL.
The three pads in the right side of Figure.1 comfirm this result.
Meanwhile, PRCL signnal was minimized with the demodulation phase of 130 while exciting PRCL, as the three pads in the left side of Figure.1 show.

Since the SRCL/PRCL IN1 is calibrated to nm, the SRCL/PRCL2POP51Q TF gain is also calibrated to dB cnt/nm.

The approximately 15 degree difference between the phases that minimize PRCL and SRCL coupling is consistent with Kawakami-kun's simulation.
However, the measured gain of the PRCL/SRCL2REFL51Q TF appears unreasonable.
PRCL2REFL51Q gain is approximately 4 times larger than that of SRCL2REFL51Q, whereas almost the same value in Kawakami-kun's simulation.

Plans:

There are 3 options for constructiong the MICH signal for 3f control.

1. Minimize the SRCL coupling by phasing and subtract PRCL coupling by INMTRX
With the demodulation phase of 145.
The handover failed using this method previously, but further matrix adjustment may help achieve 3f locking.

2. Minimize the PRCL coupling by phasing and subtract SRCL coupling by INMTRX
With the demodulation phase of 130.

3.Mitigate PRCL and SRCL both by phasing and also adjust INMTRX
With the demodulation phase of 130-140.
 

> Tomorrow, we plan to install and connect patch panels to the racks, as well as test the cables.

We have completed the installation and connection of the patch panels in the rack, as well as cable testing.
Connections:
From 2F EY1 (Fig. 1) to 1F EY0: 12-fiber MTP cable with 3 wires (Fig. 2); to EYA: 2 wires (Fig. 3)

Since the delivery was delayed, only the two wires for EYA have female plugs on both ends.
Ultimately, all connections will have male plugs (with guide pins) on both ends.

I changed the setpoint of the F0 GAS from 4300 to 4250.

This log summarizes the Guardian modifications made during the commissioning work reported in klog 37825.

1. DRMI Guardian updates

[DOWN]

• Set the demodulation phase of POP RF45 to 73.870 deg. We plan to move the DRMI-related settings to the DOWN state, but only this setting has been implemented so far.

[PREP_FM_FOR_DRMI]

• Set the demodulation phases for DRMI 1f locking as follows:

  • POP RF17: -139.60 deg

  • POP RF45: 73.870 deg

• During this commissioning, the demodulation phases were adjusted to maximize POP45I for PRCL and minimize POP17Q for SRCL.

[LOCKING_DRMI_1F]

The following settings were updated.

• Input matrix (INMTRX). To reduce the coupling between the longitudinal DOFs, the following elements were set:

  • POPA45I (MICH) → SRCL: -9.5

  • POPA17I (SRCL) → PRCL: -0.4

  • POPA45I (PRCL) → MICH: 5.3

• MICH offset. Changed the MICH filter OFFSET from -25.8 to -35.0. After this change, the AS port camera image became slightly darker.
• Output matrix (OUTMTRX). These values were taken from the PRMI and SRMI settings, respectively.

  • MICH → PRM: -0.0404

  • MICH → SRM: 0.23479

• PRCL gain. Set PRCL1_GAIN to 0.85 based on the OLTF measurement.

2. SRY initial alignment

[LOCKING_SRC_3F]

• Changed LSC_INMTRX['SRCL', 'POPA17I'] from 80 to -80. The original value of 80 was the same as that used for SRX, but SRY could not be locked with this setting. After reversing the sign, SRY was successfully locked. The reason why the opposite sign was required is still unclear.

[Ushiba, Hirose, Aritomi, Fujimoto, Kataoka, Sugioka]

Abstract:

We performed adjustments for DRMI 1f.

Details:

 We achieved DRMI 1f locking,and completed the phasing, decoupling, calibration, and UGF adjustments for DRMI 1f.

First, we performed phasing for POP90I/AS34I because they are used for thresholds in DRMI locking state.
Since there seemed low coherencein excitation measurement, we tuned the demodulation phase to minimize the offset in timeseries of POP90Q/AS34Q While PRMI3f/SRMI1f locked(Figure.1).
We also updated LSC_CUST_OUT_MTRX in DRMI to reflect the values from PRMI and SRMI.
Then, we achieved DRMI 1f locking.

After DRMI locked, we performed the phasing for POP17 and decoupled SRCL from PRCL IN1.
We injected 150.125Hz excitation to SRCL. Then we changed the demodulation phase from -149.6 to -139.6 to minimize the peak in POP17Q.
We also set the K1:LSC-PD_DOF_MTRX_4_9  -0.4 to minimize the peak in PRCL IN1(Figure.2).

Next, we performed the phasing for POP45 and decoupled PRCL from MICH IN1 and SRCL IN1.
We nnoticed that exciting IN1 causes the contamination by the different G in each control loop, so here we injected 150.125Hz excitation to K1:VIS-PRM_TM_TEST_L_EXC.
Then, the PD signal is just the product of excitation and A(Actuator efficiency) which is independent from control loop.
We changed the demodulation phase from 57.87 to 78.87 to minimize the peak in POP45Q.
We also set the K1:LSC-PD_DOF_MTRX_3_11 5.3 and K1:LSC-PD_DOF_MTRX_4_11 -9.5 to minimize the peak in MICH IN1 and SRCL IN1(Figure.3).

Then, we adjusted control gain.
We measured OLTF for MICH, PRCL and SRCL(Figure.4,5,6).
The UGF for each control was approximately 50Hz, 30Hz and 100Hz.
Out-to-In measurement was consistent with the suspension model, which verifies that the IN1 signal is calibrated to nm.

All of these parameter changes were updated in Guardian,and the specific Guardian modifications will be detailed in Hirose-san’s upcoming post.

YamaT-san, DanChen-san, Nakgaki-san, Infofarm(4 people), Ikeda

We installed fiber-optic cables from the 2nd floor of the Y End to the mini racks near DGS and EYA on the 1st floor.

Tomorrow, we plan to install and connect patch panels to the racks, as well as test the cables.

With Nakagaki-san

We carried out electrical safety improvements in both the EXA and EYA areas.

Work performed

• Replaced or corrected power cables that had been identified as issues in the fire safety assessment.

• Rearranged the cables and equipment so that the power distribution boxes could be properly closed. (pictures)

• Installed dust-proof safety plug covers wherever possible.

Impact on equipment

• The power cables for the network switches were also included in the replacement work. As a result, network communication with devices connected to the Pcal network switches was temporarily interrupted.

• The affected devices included Tcam and  a vacuum monitor in EYA

I offloaded the F0, F3, and BF GAS filters with the FRs.

I changed the setpoint of the F0 GAS from 4200 to 4300.

I measured the IP tilt, which depends on the IP position. The IP position was changed with the setpoint for the LVDT L. The IP tilt was measured using the DC feedback signal in the FLDACC H1 (aligned to L). The plot shows that the DC acceleration in the FLDACC is proportional to the IP position. The coefficient was 2.8μg/mm which corresponds to 2.8μrad/mm in tilt.

This log summarizes the Guardian updates described in klog37816. (SRY locking and the SRC initial alignment have not been tested yet.)

1. Updates to lsclib.py

• The lock threshold for SRMI 3f (is_srmi3f_locked) was added: ezca['LSC-AS_PDA2_RF34_I_NORM_MON'] > 0.05
  This is the same threshold as that used for SRMI 1f. Unlike PRMI, both SRMI 1f and 3f have the 1f sidebands resonating in the interferometer, while the carrier is dark at the AS port due to the Schnupp asymmetry. Therefore, the same 2f1 RFPD signal is used to determine the lock status for both SRMI 1f and 3f.
• The @check_srmi3f decorator was also added.

2. SRX 1f/3f commissioning

[PREP_FM_FOR_SRC]

This state is shared by SRX and SRY, although SRY has not been tested.

• The SRCL1 gain is set to 0.02 for 1f and 0.015 for 3f.

• The HWP angle is set to 146 deg.

[LOCKING_SRC_3F]

In the DOWN state, LSC_INMTRX['SRCL','REFL51I'] is set to 1 by default.

For 1f locking:

• The default SRCL -> REFL51I input matrix element is cleared, and SRCL -> POP17I is set to 80.

• ELP100 is enabled.

• The POP17 demodulation phase is set to 153.00 deg.

For 3f locking:

• SRCL -> REFL51I is set to 200.

• ELP500 is enabled.

• The REFL51 demodulation phase is set to 130.100 deg.

Previously, this state only checked is_src1f_locked. Since the same state is now used for 3f locking, the is_src3f_locked condition was also added.

3. SRMI 1f/3f commissioning

[PREP_FM_FOR_SRMI]

• The HWP angle is set to 140 deg.
• The demodulation phase is set to:  1f: -149.60 deg, 3f: 70.100 deg
  For 1f locking, MICH and SRCL are sensed using POP17Q and POP17I, respectively. Considering that the SRCL signal may be larger than the MICH signal, the demodulation phase was adjusted to minimize the SRCL contribution in POP17Q.
• For both 1f and 3f locking, the output matrix elements from MICH to BS and SRM are configured. MICH -> BS is already set to 1 by default. In this state, MICH -> SRM is additionally set to 0.23479. This value was estimated from the theoretical ratio and measured responses of BS and SRM during SRX commissioning, in order to predict the coupling between MICH and SRCL in SRMI. With this setting, the SRMI control loops were stable after lock acquisition. This is the same approach as that used when transitioning from PRY to PRMI in klog37812.

[LOCKING_SRMI]

• The default input matrix elements configured in the DOWN state are cleared before setting the SRMI input matrix.
  For SRMI 1f, the following elements are set: LSC_INMTRX['MICH','POPA17Q'] = 57.5, LSC_INMTRX['SRCL','POPA17I'] = 13.335
  For SRMI 3f, the following elements are set: LSC_INMTRX['MICH','REFL51Q'] = -176.4, LSC_INMTRX['SRCL','REFL51I'] = 66.976
• For both 1f and 3f locking, the following settings are applied:

  • MICH2: ELP100 ON and ELP500 OFF.

  • SRCL2: ELP500 ON.

  • MICH1 gain: 0.081.

  • SRCL1 gain: 0.297.

  • When the interferometer approaches resonance, the SRCL2 and MICH2 gains are set to -0.1 and 60, respectively.

• Previously, only is_srmi1f_locked was used to check the lock status. The is_srmi3f_locked condition was added to support SRMI 3f locking.

[SRMI_3F_LOCKED]

• The @check_srmi3f decorator was added to monitor the SRMI 3f lock status.