NAOJ GW Elog Logbook 3.2

Simon, Pengbo
Yesterday we maesured the beam profile again. We moved the first lens for 2.5 cm, make it closer to the FI. As can be see from the attachment, the result we get is 0.042 mm, which is better than before.

As I replaced some mirror mounts yesterday, the beam alignment needs to be adjusted.
In addition, the beam profile was also changed and the beam size at AOMs are relatively larger, about 800 um.
Therefore, I tried to minimize the beam size at AOMs with changing the lenses and their positions.
I could somehow adjust the beam size about 500 um at AOMs.
The next step is adjust the alignment in order to inject beams to AOMs.
Then play with STMs to maximize the diffracted beam power.
I adjusted the beam size at AOMs by changing the positions of lenses.
I aligned one of the AOMs and I could get about 70 % diffaction efficiency with double-pass.
Then I put a BS for ISS.
The next step is align another AOM and install a BS.

Eleonora and Yuhang
We tried to tilt lens by more than 10 degrees to avoid back scattering (see attached figure 1). Beside, we tried to put beam dump (see attached figure 2) or iris(SM1D12 with thread and can be mounted to the lenses before homodyne) to block the scattered light.
The comparison of shot noise with homodyne open to filter cavity(but only vacuum come from FC) is shown in the attached figure 3. Although we could see there are no obvious difference between use beam dump (iris) or not, the level of back scattering around 10Hz is less than the shot_FC measurement we did one month before. This means anyway, the tilt of lens helps to reduce back scattering.
However, the coating of lens is not optimal for non-normal incidence. So we tried to measure squeezing again. And the result is attahed as figure 4. We could see we have only 5.63dB squeezing. This means we have 3% more loss caused by the tilt of lens.
Anyway, we measured also the FIS from FC, the result is attached in figure 5. Although we have only 2.17dB squeezing from FC, but the low frequency squeezing is much better now.

Because of my stupid work, the optics located very upper stream part of the system were not tightly clamped.
So I decided to modify them and did re-alignment work.
It is still ongoing and the next step is the alignment of double-pass AOMs in HOMs.

I replaced 2 mirrors to BSs which are used for pick off the beam power in order to intensity stabilization.
In addition, I installed a HWP for adjust the polarization of the beam because the polarization is changed by double-pass AOM configuration.
I checked the transmitted beam power with 3.2 mW input.
The result was 0.5 mW and this value was consistent with the spec.
At that moment the angle of HWP was 300 deg.

[Shoda, Eleonora]
Yesterday we found that dataviewer and diaggui were not working properly. In particular, diaggui was showing the error: "Unable to obtain measurement data"
With the help of Shoda-san, we ssh into the standalone, check the disk status and discovered that the disk for data storage was full.
We deleted some of the old data stretch and the system started to work again.
We should delete the data stored in /frames/trend/second
Here some useful commands:
df # to check the disk status
ls -l # to check the data folder with information on dates when they were acquired
du -h --max-depth=1 # to check files size
rm -r (number of the folder) # to remove the data stretch in the selected folder
Command example:
ssh standalone
cd /frames/trend/second
ls -l (or ls -la to show hidden files)
rm -r (number of the folder)

Simon, Pengbo
Last friday and today, we measured the beam profile after install the FI.
First time we didn't change the distance between the telosocpe, the result is: w = 0.0582+/- 0.0009mm, z_position = 56.54+/- 0.34 mm. ( Figure 1.)
Then we adjust the teloscope distance from 22cm to 23 cm by moving the first lens, the result is : 0.0512 +/-0.0012 mm, z_position = 30.45 +/- 0.34 mm.
the waist position is closer to the optical bench than before but the waist radius is just a little bit smaller than before.(Figure 2)

I successfully glued another fused silica mirror.
Now that I have a set of fused silica mirrors and I will start installation of folded cavity.

Yesterday I tried to glue a fused silica mirror to a mirror holder using a jig as show in attached picture.
I used DP-190 for gluing which was borrowed from Tomaru-san.
This morning I checked the situation of the mirror and it seemed glued to some extent which can be used for foled optical cavity.
The AR side was enough clean.
I will glue another mirror.
[note]
The procedures are as follows:
- Apply First Contact on HR surface for protection.
- Set a mirror holder and a jig.
- Apply glue on the mirror holder.
- Attach the mirror to the mirror holder.

Eleonora and Yuhang
As we tested on Thursday, CVI mirror is ideal for our requirement. So we decide to use CVI BS. Although it takes us a lot of time to find a good position for it. We succed in balancing homodyne in the end and to find a good compromise between the position of a lot of optical components. The change of optical components is summarized in the last attached picture(the latest change is updated to naoj_wiki).
We verified that no backreflected beams could be found with the new BS.
After balancing homodyne, we measurde back scattering and squeezing again. The comparison is between Friday and Thursday result (attached figure 1, 2 and 3).
Note that the back-scattering is measured by placing a sensor card on the squeezing path. Since the peaks observed at low frequency are at the bench resonant frequencies, we assume that the sensor card (which is vibrating with the bench), is reflecting back the scattered light towords homodyne. If we put a beam dump instead of the sensor card the peak at low frequency disapperad.
In attached figure 1, we can see the back scattering is even worse. Based on the fact that this back scattering is not stable, we think that the change of BS doesn't improve the situation.
In attached figure 2, we can see the squeezing level is improved from 5.90dB to 6.06dB. This means CVI BS really has less loss. From the loss and phase noise information measured in entry 1916, we can infer we reduce loss by ~1% by replacing BS.
In the attached figure 3, we can see there are some new peaks appear around 3kHz to 10kHz. These peaks was there several months ago. They were removed after the use of Faraday. We will investigate why they show up again.
Conclusion: the back scattering peaks at low frequency observed when we place a sensor card on the sqz path are still present even after replacing BS with a better one, with no detectable back reflection, and even after replacing homodyne lenses with superpolished ones. The source of back scattering was not the cube BS. So, what can it be? Homodyne?
Next steps: Put beam dumps around Homodyne. Check the situation with the filter cavity aligned.

After 2.5 days cooling, the cryostat reached the temperature ~60 K.
This value is my guess because the sensor had some problems and it showed higher temperature than actual.
Anyway, the refrigerator worked well though longer cooling time than before.
This is due to some insulators were not installed inside the chamber...
But it can be cooled down at some extent and I gonna do with this configuration.
I turned off the refrigerator around 20:20.

[Yuhang, Eleonora]
Before Christmas we decide to replace the cube BS before homodyme (where SQZ and LO are recombined) with a plate BS in order to reduce back-reflection as much as possible.
The BS ratio must be as much as possible close to 50/50 to get optimal homodyne balance. We purchased 5 plated BS and characterized them with Simon's set up (entry #1931). We use s-pol.
Here the results:
code | ratio (T/R) at 45 deg | 50/50 ratio angle | note | |
Thorlabs1 | BSW41-1064 | 1.2 | 55 deg | bright ghost beam |
Thorlabs2 | BSW41-1064 | 1.17 | 37.7deg | bright ghost beam |
Newport1 | 10Q20HBS.33S | 1.006 | 45 deg | weaker ghost beam, low dependance from angle |
Newport2 | 10Q20HBS.33S | 1.006 | 45 deg | weaker ghost beam, low dependance from angle |
CVI | BSNP-1064-50-1025 | 0.89 | 41.7 deg | no ghost beam |
We note that the power mesured by the power meter is quite sensitive to the incidence angle. So, we adjusted power meter orientation every time to optimize it. Newport BS has the most balanced splitting ratio at 45 deg. and it is very much insensitive to the angle. Anyway CVI has no ghost beam, and the mirror quality should be better so we decided to go on with CVI BS and adjust the angle to get optimal balance.

Since the pressure reached 6.5*10-4 Pa, I turned on the refrigerator.
It seemed working and I left it running.
After 2.5 days cooling, the cryostat reached the temperature ~60 K.
This value is my guess because the sensor had some problems and it showed higher temperature than actual.
Anyway, the refrigerator worked well though longer cooling time than before.
This is due to some insulators were not installed inside the chamber...
But it can be cooled down at some extent and I gonna do with this configuration.
I turned off the refrigerator around 20:20.

The measurment in the first picure is done when the back reflected light towords the squeezing path is blocked with a sensor card.

We received the super-polished lens from CVI with 30mm focal length. With this super-polished lens we should have less parasitic interference.
So we compared the back scattering and squeezing level.
As shown in the attached figure, there are some difference at low frequency which is more back scattering(worse) and more squeezing(better).
Actually the low frequency part, we have already ovserved some uncertainty during different measurment. So it seems the replacement of lens doesn't improve system obviously.
The measurment in the first picure is done when the back reflected light towords the squeezing path is blocked with a sensor card.

Since the pressure was 1.5*10-5 mbar, I decided to turn on the cryostat on tomorrow morning.
I will leave the scroll and turbo pump running overnight.

I found that the DP (DSP250) back of the TMP was failed with "ALM06" in the south end. Though I tried to restart it some times, it was not rcovered.
I replaced the DP to the other DP (DSP500) from arm front where the TMP is not working now.

[Yuhang, Eleonora]
Filter cavity lock was recovered easily.
Then we checked green alignment into OPO: we sent BAB and maximized its alignement into OPO by moving the lens in pic 1. We scanned OPO and measure the BAB transmitted TEM00 with and without green pump. The green phase was modulated at 10 Hz. We optimized the alignment of the green beam into OPO by maximizing the BAB transmission. The ratio between BAB transmission with green (pic2) and without green (pic 3) gives the parametric gain which is ~ 25.4 for 50mW of pump. It seems fine. Note that recently we observed a reduced parametric gain but this measurement is in agreemen with the early value (see entry #1131).
We changed OPO temperature from 7.185 to 7.215 kOhm to maximize the non-linear gain, (with the PLL offset without green set to 180 MHz.)
We check the p-pol PLL offset when green is injected into OPO and optimized it by maximizing CC1 error signal. Now the good values for PLL offset are:
with green | without green | |
PLL offset | 180 MHz | 325 MHz |
We found that p-pol was not well aligned into OPO, and the alignment was very unsatable and difficult to optimize. Eventually we discovered the p-pol laser temperature was 2 degree away from the optimal temperature. So the issue was likely due to mode-hop. We put it back at the nominal value and we confirmed that all the lasers are at the optimal temperature, namely:
MAIN | P-pol | CC | |
temperature | 23.105 °C | 32.49 °C | 38.15 °C |
Then we aligned p-pol and CC into OPO. Spectrum is shown in Pic 4 (blue is p-pol, yellow is CC).
Finally, we measured again squeezing with 50mW green pump. The measured squeezing level is 5.9dB which is consistent with the measurement done before. Besides, the spectrum is as flat as before.

I installed the inner shields, forgot to take pictures though...
When the vacuum level gets enough low, I gonna turn on the cryostat.
Since the pressure was 1.5*10-5 mbar, I decided to turn on the cryostat on tomorrow morning.
I will leave the scroll and turbo pump running overnight.

Today, I did vacuum test with the repaired scroll pump and the turbo pump.
Since the chamber had been opend for a long time, the vacuum level was not good though there were no huge leaks.
In addition, one of the vacuum gauge did not work.
I will do another vacuum test, and then do cryogenic test.