EleonoraP and Yuhang

Today we checked the PBS and the two following mirrors for OPO transmission. PBS has a transmission of 0.2% while we measured reflection more than incidence.  And the two mirrors are fine as well. We didn't use wrong coating components or they have some issues.

Chien-Ming and Yuhang

After the last very good measurement of squeezing, we start to think about improving coherent control 1 PD. Then we measured squeezing again. The result is attached in figure 1 and 2. As you can see the low frequency is covered by classical noise while high frequency seems fine. However, the squeezing level was only 4.9dB at that moment.

Then Chien-ming suggested cleaning optics. And then we checked optics together. We found optics are tremendously dirty. Especially one of the lenses is very dirty on one side of the surface. And for others, there are lots of dust on the surface of the mirror. However, after the cleaning of optics, the balance of BS is destroyed. After the alignment of this BS, we measured squeezing again. However, we didn't improve the squeezing. We observed squeezing level almost the same with last time. See attached figure 3 and 4.

There is some consideration after this measurement:

1. The fluctuation of measurement is about 4dB pk-pk. Is this coming from the measurement device?  

2. The low-frequency noise of the second measurement of squeezing is much higher than the first one. We also know that for the second measurement, the balance of the SQZ part is very bad. The unbalance signal on PD is about 50mV. Although we know this balance is not as important as the balance of local oscillator, we still need to make it stay at a decent level.

3. To keep a clean clean room is really important for squeezing to avoid some potential losses.

4. We also tried to change coherent control power, but we noticed that after we decrease coherent control power we may also increase phase noise. This measurement is better done when we have a better coherent loop. Also with some characterization of the coherent control loop.

[Yuhang, Eleonora]

Today we temporarily swiched off both the rotary and turbo pump in the central area (close to BS).

The attached plot show the comparison of the spectra of the suspensions in the central area, with pumps on and off.

We see that BS is strongly affected by the pumps vibration while PR and INPUT seem not.

[Note that at the time when I developed local control for the first time, the pump where swiched off].

The pumps have been switched off for about 3 hours ( from 5PM to 8PM).

Yuhang, Aritomi, EleonoraP, and EleonoraC

On 5th of June 2019, we locked the filter cavity again after almost half a year we haven't done this. We checked several references including irises on the bench, target on the film of PR chamber, target on the film of BS chamber, the first and second target inside the filter cavity vacuum tube. To recover these checking points, we used pico-motors with almost no failures. But it seems there are still some problems with end mirror pico-motor. Also, we observed some large mirror motion after we moved pico-motor of BS. But it went back to static after several minutes.

The alignment is also done with the standard procedure. It is first to misalign input mirror and find filter cavity transmission. Then center it on the screen. Then bring back input mirror and make injection and reflection overlap. Finally, align end mirror to have TEM00 dominant. After the lock, change the control point of each mirror to see if the transmission goes up to maximize.

Power:

Signal:

The demodulation phase for filter cavity locking is 65deg.

The setting of FC locking servo:

Today suddenly DAC died. Since I was working around the DGS I supposed I accidentaly disconnect some cables, but I check carefully and everything seemed fine.

We rebooted the standalone computer and the problem was solved. Note that in order to switch it on we had to disconnect and reconnect the power cables and to disconnect the timing singnal (pic 1). Timing cable was reconnected after the restart.

We observed that connecting end room picomotor driver to the power line in the same mupliple socket of OPLEV laser and PSD brings a HUGE 50 Hz in the signals.

Since picomotors are only used temporarily we have now disconneted them. We may try to connect them to another socket next time that we need them.

Local control on END mirror were implemented.

Pic 1 Pitch TF

Pic 2 YawTF

Pic 3 Comparison between open and closed loop spectra.

Pic 4, 5 photon model and damp for pitch  

Pic 6, 7 photon model and damp for yaw

I found a minimum of the sensing coupling when input signals (pitch and yaw) are rotated of -0.06 rad. So I updated accordingly the rotation matrix.

I used the following driving matrix  (pitch, yaw -> coils)   [0, 1, 0, 1; 1 0 -1, 1].  Note that the coil disposition is not the usual one!

For a comparison with the old control check entry #238.

Today I did alignment work to see the flash of transmitted beam using PD.
I though that PD detected transmitted beam, but it turned out that it was scattered beam...

Then I re-aligned so that the reflected beam reach to FI, and scanned laser frequency with monitoring reflected beam.
At that time, the reflcted beam power was fluctuated.
Actually, it was caused by vibration of cryostat chamber.
So far I have used long pedestal to hold silicon mirrors, and this may affect the vibration of cavity.
Using fixed spacer may reduce the vibration effect, but it might be better to consider about vibration isolation system.

I installed a PD for monitoring transimtted beam of optical cavity with one STM. (forgot to take a photo...)
Though the transmitted power is very weak, the PD detects the transmitted beam.

Then I did alignment and scanned the laser frequency.
However, I have not seen any fringe so far.

I will continue to align the beam with scanning laser frequency and try to find good alignment.

Taking advatage of the the DAQ I playbacked the data and check the oplev signal during a 4.9 Earthquak occured in Chiba prefecture, last saturday.

https://earthquaketrack.com/quakes/2019-05-31-22-58-08-utc-4-9-20

The EQ is cleary seen by all the sensors. Some dof ( in paricular  PR and END pitch) didn't go back to the original position.

The data on DAQ are referred to UTC but the time of standalone PC is not well sincronized (about 3 min ahead the correct one).

Chien-Ming, Yu-Hang,

We want to reduce the coherent noise coupled into the homodyne by reducing the intensity of CC beam entering the OPO. However, this action will cause the demodulation 14 MHz error signal (for coherent control of the pump beam phase) to become smaller. So, we tried to increase the gain of the photodetector at 14MHz by changing the capacitor CX and the amplifier resistors R1, R2, see Fig.1.

We let the capacitor CX become a plug-in base and test some combinations of capacitors. Currently we find the biggest signal at 14 MHz when using CX as 150pF+33pF. What is puzzling is that when we replace this combination with a single 180 pF capacitor, the signal is reduced by 8 dB on the spectrum analyzer. So, we are now using 150pF+33pF, and R1=500 ohms R2=5k ohms. The S/N now is improved 5 times more as shown in Fig.2.

I installed another silicon mirror inside the cryostat.
Then I did rough alignment of input beam with 2 STMs and the reflected beam reached to FI.

Tomorrow, I will scan the laser frequency and try to find transmitted HOMs.

Yuhang and Chien-Ming

As we said before, we need a better PD for the coherent control1 lock. (PD set at OPO reflection) So we talked and tested several things.

1. Change the amplification factor of the op-amp. We tried two combinations of (500Om, 50kOm) and (500Om, 7.5kOm).  For the first combination, we tested the signal, but the amplitude and SNR are the same as before. While for the second combination, the amplitude is smaller while the SNR is a bit better.

We checked the spec of AD8057. The maximum gain we should give for this op-amp is 10. The further increase of the gain will degrade the bandwidth. And this is exactly what we see for increasing further the amplification factor.

It seems that we are also reaching the saturation current of the op-amp.

2. We also tried to use the RF amplifier, because maybe there is an optimal input level for the mixer to make it work better. But it seems RF amplifier increases both signal and noise.

3. We also tried to change the capacitor of the PD circuit. But we are still trying.

[Eleonora P, Yuhang]

We replaced the FI's stage along the green path, after a hole has been drilled on it as explained in entry #1367. 

In fig.1 the picture of the stage on the bench.

I installed one silicon mirror, which was labeled "No.1", inside the cryostat and did rough alignment.
Then I measured beam profile for mode matching.

Today, I re-installed some optics for higher-order-modes.
I wanted to put a iris for the mark of alignment, but I could not find it...
From tomorrow, I'm going to install silicon cavity inside the cryostat.

I changed optical layout to achieve better alignment.
I will update the schematic figure of optical layout later...
Anyway, the alignment seems ok so far.
 

As shown in entry #1363, the error on the beam waist position is large ( around 8%).

I checked if the solution #3 of entry #1366 would be good also considering the border positions of beam waist in the error interval, running again the simulations (fig 1, 2, 3) and analyzing the robustness on Python.

Both focal lengths are 750 mm.

I summarize the results in the following table:

where the positions of the lenses is referred to the PBS after the OPO.

I computed on Python the position of the two lenses corresponding to the condition of minimum total mismatch.

In fig. 4, 5 and 6 are shown the robustness of the telescope with z=0.081 (min position), 0.089 (real position), 0.097 (max position), that are still pretty good (less than 10% mismatch).

[Yuhang, Chien-Ming]

Today we tried to decrease CC laser power. But when we just decreased it to half, we couldn't lock the CC1 loop. 

The cc error signals that we can use to lock and cannot be used to lock are attached in Fig1.

The RMS dark noise for CC1 and CC2 PD is measured and attached in Fig2. We can see that the CC1 PD has RMS dark noise almost 5 times higher than CC2 PD. 

We also suspected that this noisy PD for CC1 locking brings also phase noise.