We can send infrared light from input-coupler side. After that, we take reflection and transmission power. By doing ring-down measurement, we can extrapolate well information of P0, P1, r1, t1.

P0 is power coupled to OPO cavity, P1 is power not-coupled to OPO cavity, r1 is amplitude reflectivity of input coupler, t1 is amplitude transmissivity of input copuler.

I did a calculation with the ideal parameters of OPO, which tells us decay time of ~4us.

To measure this decay, we should be able to lock OPO and switch off incident laser faster than ~100ns. To do this we need to use signal generator to send a square wave to an AOM which is before OPO. According to the spec of MT110-A1.5-1064, the rise time can be smaller than 100ns if the beam size is smaller than 0.6mm (diameter). We will design a small enough beam to make this rise time small enough to measure ring-down.

The expected ring-down for reflection and transmission is attached as figure 1.

Using measurements of entry 2391 I could compare the IR and red beams propagation directions as in the attached figure (axis unit in mm):

• the angle the 2 beam : ~0.07rad
• the height difference between the 2 beams : ~ 100 um (might explain the difficulties to align)

I'll try to use the pinhole to shift the IR beam height to the red beam one.

Earth quake happened on 20210320 18:09, which was quite strong and long.

We checked the oplev signal of all suspended mirrors (attached figure 2). The watchdog of BS and END mirrors were switched off. However, since we didn't give large offset, we can just use the oplev signal to check how much mirror drifted. We could see these change

Pierre and Yuhang

To investigate if SNR can be improved by an improved photo detector design, we conducted TAMA resonant PD simulation by using two simulation tools. One is a python code called 'zeros', the other is a Texas Instrument software 'TINA'.

The configuration is based on the scheme in this link. In the real case, we have modified this scheme, which results in two cases. One case is using the same op-amp with R1 10Ohm and R2 100Ohm. The other case is using op-amp LMH6624 with R1 1kOhm and R2 10kOhm.

To measure PD noise, we used a 32dB amplifier to make sure the PD noise is well above the instrument noise. The use of this amplifier has been considered also in the simulation software. In the first attached figure, there are electronic noise measurement and simulation results. We could see that python code simulates well the floor noise. However, the noise around 14MHz is better simulated by TINA.

We anticipated a signal of 0.11uA from CCFC field. To compare signal and noise, we divided the voltage noise from figure one by the PD gain. Then we get the current input noise as the attached figure 2. This DC value of signal is much higher than the noise level at 14MHz. From the python simulation, LMH6624 has better performance. However, simulation of TINA tells us that AD8057 has better performance.

To compare, we put here also a measurement of PD noise budget (figure 3). From this noise budget, TINA simulation is closer with the real measurement.

Figure 4 and 5 show the SNR of different PD configurations with TINA simulation

Marc, Yuhang

Following the End mirror OpLev check (entry 2412) and the measurement with the sensing matrix of entry 2411 we decided to use a 4x4 matrix to try to remove the coupling of unwanted degrees of freedom on the reconstructed signals from the 2 QPDs.

• We locked the FC and only close End mirror length control.
• Tune the phases : We found out that all the phases changed by ~40 degrees. So we used time series of all segments and maximized the I signals
• We sent a line at 2 Hz on Input and End pitch and yaw. Using the calibration of entry 1877 we injected a 3.78 mrad line for every dof. Namely

  	Pitch	Yaw
  Input	600	172
  End	727	141

• We measured the amplitude of the injected line (removing the background level noise) on each QPDs.
• We computed the driving matrix and had to modify the medm configuration (outmatrix.adl) to allow to use the 4x4 matrix

• 	QPD1 pitch	QPD2 pitch	QPD1 yaw	QPD2 yaw
  Input pitch	6	-5	-0.5	-1
  End pitch	-5	10	0.5	1
  Input yaw	1	-2	4	-3
  End yaw	2	-4	-4	12

• We checked each QPDs signals as in entry 2412 but coupling was still visible.
• We checked the demodulation phases and they seem to have moved by ~-40 degrees back to the previous phases...
• We checked possible reasons for this phase changes :

1. We aligned the FC to another position : no phase changes
2. We tried to put the optical bench to the ground (before putting the electronics inside of a clean room the optical table was connected to the nim racks ground) : no changes

• We putted back the 2x2 matrix of entry 2412 and tried to tuned the coupled coefficients by hand to remove coupling. It could approximatively work but during this tuning we could also see that the level of coupling increases or decreases sometimes without noticeable reasons...

Marc, Yuhang

We went to check to End mirror OpLev : All optics seemed well fixed.

We tuned the beam height to be identical on the injection and detection windows.

We moved vertically the lens on the detection window to have the beam centered.

Unfortunately it didn't change the coupling of pitch to yaw visible in the figure 1 of the parent entry.

We can see that pitch couples to yaw but not yaw to pitch.

This seems to indicate that the sensing matrix of OpLev is not the culprit.

Furthermore, when we inject a line on the End mirror pitch, we can indeed see the beam on the FC transmission camera moving in pitch but not in yaw. Even if the OpLev yaw senses this line...

On input mirror we can see both pitch and yaw lines on pitch and yaw signals from the oplev so it could be solved by finer tuning of the oplev rotation matrix.

Marc and Yuhang

We sent several sine wave to different DOF for filter cavity as following:

(At the beginning, 2Hz/3Hz/5Hz/7Hz were used. We will use 7Hz instead of 9Hz in the future.)

By sending these sine waves to different DOF, the check of coupling in reconstruction becomes easier. As shown in the attached figure, we checked many signals before and after excitation. They are:

Oplev signals: Red line: Oplev signal before excitation Green line: Oplev signal after excitation

AA signals: Blue line: AA noise when light goes to QPDs are blocked (actually this is DGS ADC noise) Brown line: AA reconstructed signal before excitation Pink line: AA reconstructed signal after excitation

(AA signal is not calibrated, so it is much larger than oplev signal)

Oplev issues:

Since a large 9Hz peak appears in the oplev signal of End mirror yaw, either End mirror pitch driving or End mirror yaw sensing is strange.

AA issues:

End mirror pitch signal couples to End and Input Yaw.

Input mirror pitch signal couples to End Yaw.

Plan:

1. We are going to check end mirror oplev and coils. 2. Consider to realize a matrix including pitch/yaw coupling

Marc and Yuhang

We have checked the important parameters and green power as following:

SHG temperature: 3.09kOhm

AOM setting: 109.036MHz, 5.5dBm

green power before AOM: 50.1mW

green power before FC injection: 24.9mW

Yuhang and Michael

Unfortunately in 2398 the measurement was performced incorrectly. Random noise from the spectrum analyser was injected into the "Perturb IN" port of the IRMC servo controller rather than the IN port of the IRMC HVD. This mistake was corrected, and we measured the following

• Power spectrum of dark noise, LO jitter and phase shifter excitation noise from the X, Y and T (Total) channels of the 500 Ohm position sensitive detector (PSD) - figures 1, 2 and 3 respectively
• Transfer functions and coherence of source noise to the X, Y and T channels of the PSD - figure 4
• Transfer functions of source noise to the IRMC error signal (at EPS1), correction signal (at Servo Out) and IRMC reflection - figure 5

The high frequency numbers for pitch are now much more in line with 1904. Notably, the mid frequency discrepancy between 2398 and 1904 still remains, where the mid frequency noise magnitude in this rearranged configuration is much higher than for 1904 despite having similar levels at the extremes of the measurement window. The PSD of yaw noise from the excited phase shifter is now signifcantly reduced compared to 1904, although the LO jitter is slightly above the PSD dark noise this time. We also measured the T-channel of the PSD to obtain the amplitude noise of the LO beam.

There is quite a lot of coherence from the source noise to both T-channel noise and pitch noise, especially amplitude noise above 30 Hz. The T-channel noise transfer function and coherence is fairly similar to the IR reflection, which is to be expected since they essentially just amplitude noise transmitted and reflected from the IRMC.

Source noise coupling to the error signal increases above 100 Hz, while the correction signal transfer function does not follow said increase. Howver, they have similar coherence across all frequencies.

Just as a note, when I checked and realigned the IRMC at the start of the measurements for the day, it seemed to have drifted out of alignment with a pitch HOM prominent in the reflection spectrum.

Marc and Yuhang

Marc provided us a code written by Julia Casanueva Diaz, which was used by Marc to do FIS auto alignment for Virgo.

After reading this code together, I found out what can be improved for the construction of driving matrix. Especially, I finally understand what is the meaning of matrix of inversion. The idea is the change of basis after inversion. With this in mind, we checked again the sensing matrix and inverted it. We got following driving matrix

With this driving matrix, we did preliminary check of Input and End mirror pitch/yaw reconstructed spectrum. Compared with the old measurement (elog2263, elog2245), it seems the reconstruction is more reasonable (attached figure).

More investigation and optimization will be done soon.

Marc and Yuhang

After moving picomotors, filter cavity alignment was recovered.

Spectrum of all mirrors were checked as the attached figure. No touching induced peaks were found.

For the Labview control for INPUT mirror picomotor, I have some comments.

***notice: check point of reflected beam is inside PR chamber

1. Pitch positive is to move the reflected beam to the left.

2. Yaw positive is to move the relfected beam to the up.

(currently, the input mirror pitch and yaw picomotors are swapped)

As reported in entry 2399, there was an earthquake in March 16th around 5 am.

I went to check the setup and it seems that at least the imaging unit is misaligned (moved by few 10 um in distance to the translation stage and the red beam was misaligned by ~ 1 cm to the left on the photodiode). I'll try to characterize the beam again after maximizing the red and IR beams on the photodiode/powermeter.

Also, I suspect that there was some troubles with the computer as :

- several errors messages on the LabView program referring to troubles with the translation stage

- waveplate moved around the earthquake time but no errors

- the chopper frequency changed to 1kHz.

Michael and Yuhang

We completed the setup as per 2393. The spectrum of aligment noise from excitation of the IRPS was made and compared to 1904.

The measurement was taken using a 500 Ohm PSD placed after the IRMC and homodyne flipping mirror. We measure the dark noise, followed by the jitter noise when transmitting the local oscillator, and then the jitter noise when the phase shifter PZT is excited by random noise from the spectrum analyser, applied via the high voltage driver. In the original measurement of 1904, the PZT was excited using 200 mVpk random noise, where the beam had a horizontal angle of incidence of ~ 45 degrees. Here, the beam has perpendicular incidence, so we only use 200/sqrt(2) ~ 140 mVpk excitement for the yaw measurement, but keep 200 mVpk for the pitch measurement.

For yaw, we seen that the noise floor seems to be the same or perhaps slightly better than 1904, it is hard to tell just from the picture. However, the new measurement reaches the noise floor at approximately 30 Hz, compared to ~ 10 Hz in the old measurement. Also, both measurements have similar values at either end of the measurement frequency window, but the new measurement looks to be a lot higher in the mid frequencies than that of the old measurement. The PD dark noise is consistent in both cases.

For pitch, the measurement is very strange. Unlike yaw in this case and both degrees in 1904, the LO jitter noise does not converge to the PSD dark noise at high frequency. Both the LO jitter and excitation jitter seem to be about 15-20 dB higher than their yaw counterparts.

Afterwards, we measured the transfer function of input excitation (from the spectrum analyser) to measured PSD noise. We see that the pitch transfer function has significantly more coherence from the excitation to the PSD. The yaw coupling at 140 mVpk excitation is also much lower than for pitch excitations.

A few other notes:

- The IRMC alignment seems to drift quite a lot. Last week it had to be relocked about 4 or 5 times during the afternoon. It also had dropped to about 1.6 mW transmitted power before starting the measurement described in this elog. I realigned it for the optimal transmission each time (1.8 mW).

- The phase shifter is not quite flat, it has some pitch tilt. As a result, the incoming alignment is not perfectly horizontal, but rather is adjusted so that the outgoing beam maintains a constant height of 76 mm.

- PSD output voltage was zeroed to within 0.5 mV for both X and Y

Aso-san, Marc

The air leakage of the gate valve was indeed due to a loose screw.

Aso-san tighten it more and no more audible sound nor air coming out.

On Japanese standard time 04:56 JST 16 Mar. 2021, an earthquake happened close to Tokyo area. (attached figure one) 

I checked the oplev signal during this time in the saved data of DGS. PR, input and end mirrors reached the maximum of oplev sensing range. Besides, this oscillation lasted for almost 20 seconds and the exponential decay lasted for almost 40 seconds. (attached figure two)

After the earthquake, all mirrors drifted away from their original position. (attached figure three)

According to the above table, we are going to move picomotor to recover mirror position.

Before moving picomotor, I also checked the spectrum of all mirrors pitch/yaw. No touching induced peaks were found. (attached figure four)

Marc, Yuhang

Yesterday we wanted to check again QPD demodulations pahses before implementing the new driving matrix.

However, the BS pitch seemed to be totally out (180 offset required to get the beam on the end camera).

We had a look on one week data trend of every mirror oplev signal (attached pdf).

This large BS offset has 3 origins as indicated by the red arrows :

1. An earthquake not too far from Tokyo in March 3d caused a -500 pitch offset of BS (also visible on IN and slightly on END but not so evident effects on PR )

2. Switching the air conditioning to 'cold mode' cause a 1500 pitch offset of PR. Indeed, PR is located directly under the air conditioning.

3. Strong rain/storm on March 13th cause a further 500 pitch offset of PR

Therefore we decided to use PR pitch picomotors to move PR back to its previous good position (~300 steps).

But this made the required BS offset -180... This is quite close to coils saturation (it corresponds roughly to 20 000 counts) so we might have to move again BS picomotors...

After the movement of PR picomotor we took the oplev signals (fig 1) where green/brown are references and blue/red yesterday data.

We can see that PR seems fine. However, comparing the low frequency spectrum of BS and PR between reference and yesterday datas, PR low frequency spectrum changed quite more than BS one. It seems that PR is far more sensitive to seismic activities than BS?

We also found out that there is an air leakage on the BS gate valve. It might be due to a loose washer pointed by Yuhang in Fig 2. It might be another after effect of the large earthquake as the air leakage sound was not so easy to hear.

Marc, Yuhang

Yesterday we spent some time to tweak a program made by Julia Casanueva for Virgo automatic alignment and adapt it to our case.

Then, we tried to measure the sensing matrix.

However, the green beam was not anymore centered on the QPDs..

We found out that the last steering mirror before the QPDs board was not well fixed. We fixed it back but we could not recover the alignment by acting only on this mirror.

It could mean that another mirror on the green reflection of the FC has been misaligned but we couldn't find which one so we acted on the both galvos to recenter the beam on the QPDs.

We measured the sensing matrix by injecting a 2Hz line on Input and End mirror pitch (300 amplitude) and yaw (200 amplitude) and extracted a driving matrix and phases to get all signal on I quadrature.

Today, we saw that the phases are tuned at the level of each segment so we will have to tweak a bit more the program as it gives the optimal phases at the level of pitch and yaw (not individual segment).

Anyway, today we tuned each segment phase 'by hand' to get all signal on I quadrature by either looking at the 11 Hz resonance of pitch or injecting a 2 Hz line. During this measurement the misalignment of the FC affected the accuracy of the measurement so we had to realign it, but also there were moment with large changes of the optimal phases...

The new phases are :

We completed the arrangement of the optics described in 2393.

A bit of misalignment from an ND filter in front of the 250mm lens was corrected, and some slight account had to be made for the phase shifter not being perfectly vertical (i.e. introducing pitch misalignment from horizontal incidence). The IRMC mode matching was optimised through alignment and fine tuning of the 75mm and 250mm lens positions, and we recovered the transmission of 1.79 mW of TEM00 from the IRMC (power meter measurement).

We will compare the new result to that of 1904

Michael and Yuhang

We worked on replacing the phase shifter for the beam going into the IRMC, highlighted in figure 1. Afterwards we will also look at the one going into GRMC. The angle of incidence causes beam jitter noise when the phase shifter acts on the beam. We decided to replace this with a perpendicular incidence setup as per the sketch in figure 2, using a PBS and waveplates.

We have reference for the values of the IRMC (390 µm beam waist, 1.8 mW transmitted power). Using the reference waist and the distances of the holes, the beam should be collimated before the 250 mm lens (fig 3). So there is not much need for complicated rearrangement of the lens positioning. We just have to move the 250 mm forward by however much the path length is from the PBS to the phase shifter and back. Fine tuning of the lenses can be done via the mount. A rough indication of the distances is shown in fig 4.

We took the following items:

PBS: CVI PBS-1064-100

HWP: CVI QWPO-1064-08-2-R10

QWP: CVI QWPO-1064-09-4-AIR-R10

During the aligment we made sure to recheck the beam propagating along the west end of the table to the 250 mm lens, to make sure it aligned with the holes and was consistent with the more recently aligned beam height from the SHG (76 mm).

Using the power meter, the power before the PBS was 4.14 mW. After double passing through QWP and PBS, it was ~ 3.8 mW.