NAOJ GW Elog Logbook 3.2
[Takahashi, Aritomi, Yuefan, Matteo,Yuhang, Eleonora]
Today with the help of Takahashi-san we opened the BS and PR chambers.
First goal was to check a possible touching of the intermediate mass.
PR chamber
We found the intermediate mass was almost touching the magnet cage. Takahashi-san adjusted the cage position to leave more space between the cage and the intermediate mass. From the spectra taken before and after there is no evidence of improvement (pic1), but the spectra after the change have been taken with the chamber still open. So we will repet the measument in vacum to confirm the results.
BS chamber
BS intermediate mass seems fine.
The second activity was the assesment of losses inside PR chamber, from viewport, Faraday, dichroic mirror etc. Aritomi and Yuhang will report about that.
We closed the BS chamber and left the PR one open with the cleanboth. We wll close it tomorrow moring.
Yestarday afternoon at ~17.15 we had a blackout in TAMA due to a severe lightning storm. The power came back after ~10 minutes.
We spent the rest of the afternoon recovering the system. For the filter cavity everything seems fine. Takahashi-san came to help with vacuum system recovery.
I re-tried lock, but could not.
At first, I adjusted the offset of servo to set the monitored error signal to zero.
Also I minimize the gain of servo.
I tuned the laser frequency around the resonance, and turned on the integrator.
I tried several times with different conditions (inverted or not, and so on..), however, could not lock...
There may be some problems such as high finess, large gain, etc...
Actually, the efficiency of laser PZT is 50MHz/V, and the slope of error signal is 4.2e-7V/Hz.
Therefore, it needs to optimize the gain of the servo.
I found that green into OPO is completely misaligned. I recovered the alignment by measuring parametric amplification. Here is the information of parametric amplification of BAB.
| green power (mW) | OPO temperature (kOhm) | p pol PLL (MHz) | Vmax (V) | Vmin (V) | parametric gain |
| 0 | 7.2 | 305 | 0.112 | 0.112 | 1 |
| 40 | 7.185 | 165 | 1.82 | 0.04 | 16.3 |
I installed the lens (f=50mm), QWP, and mirror for another double-pass AOM.
They are roughly located to compose a cat's eye retroreflector, but not tuned yet.
I tried PDH lock with a servo borrowed from TAMA, but could not lock.
At that time I found the laser power was larger or smaller than usual.
I monitored output voltage from the servo, and it was about 14V which saturated PZT frequency tunig range.
I played with gain, but it did not affect the output.
Actually, I did not adjust the offset before turnig on the integrators.
So I will re-try the lock as Matteo suggested; tune the offset first, then turn on the integrators with smaller gain around the resonance.
I re-tried lock, but could not.
At first, I adjusted the offset of servo to set the monitored error signal to zero.
Also I minimize the gain of servo.
I tuned the laser frequency around the resonance, and turned on the integrator.
I tried several times with different conditions (inverted or not, and so on..), however, could not lock...
There may be some problems such as high finess, large gain, etc...
Actually, the efficiency of laser PZT is 50MHz/V, and the slope of error signal is 4.2e-7V/Hz.
Therefore, it needs to optimize the gain of the servo.
Simulink model and MEDM screen for AA is now completed and ready to be tested with quadrants.
Note: while restarting the model last sunday I was not able to restore the screeshot file taken just before the restart. I didn't find any problem with the command but some of the epics channels remained blank. I check the file and see that some of that channels ware actually missing. I don't know why it happend. Anyway I restored a previous snapshot file. I restarted the model and restored snapshot several times after that and this problem never showed up again.
As we decided to try the dithering technique to keep long term alignment of the cavity, in the past day I realized the software necessary for the job. Both simulink model and MEDM interface. I tried to reuse code from KAGRA as much as possible.
We have how the possibility to send 4 lines for pitch and 4 lines for yaw to each of the 4 suspended mirror. and we can demodulate a cavity signal (to be chosen between GREEN TRA, IR TRA and FC CORR) at the frequency of these lines. Then we can decide to which mirrors to send the correction after properly filtering it.
In order to implement the dithering we should switch from DC local controls for mirrors to AC local control. I quickly realized a preliminary, non optimized version of these AC loops (except PR), which I hope to improve soon. Pic 3.
Yuhang and Yuefan
According to the experience of Eleonora and Matteo, the yaw of PR has a very narrow range. Also, we moved -2000 counts of PR and we observed a stronger vibration of green beam. So we guess maybe PR is touching the intermedium mass.
So we checked how much we moved and them tried to move it little by little and then have a look the spectrum to see if there is any difference of peak values. The measurement was down from -1000 to 2000 counts while the original position is 750 counts. The spectrum is attached in the sequence from -1000 to 2000.
We could see that after 1300, the high frequency noise spectrum is a bit higher than the reference. And we could see at -200 and -1000, the spectrums have parts lower than the reference.
Yuefan and Yuhang
Today we finished all the cabling works for the AA telescope.
On the top left of the first picture shows the DGS board, 16 channels on the bottom are connected to the I and Q DC output of the demodulation board for both quadrants. On the top, the middle 8 channels are the DC output from the QPDio-base for two quadrants.
In the right middle of this picture, there are the two boards to control the galvo, four inputs on both boards are also from the QPDio-base DC output. and the x and y output will be connected to the galvo.
Second picture shows the demodulation board for two quadrants, one the left is the 16 demodulation signal mentioned before, and on the right, there are 8 channels which connect to the RF output of the quadrants. We didn't connect the LO yet.
Third picture shows the QPDio-base, the left four channels are the DC output, and the right two cables will be connected to the quadrant.
In the fourth picture, these are the cables that will connect the galvo motors and the RF output of the quadrants.
We put lables for all the cables we connected today to avoid any confusion in the future. We still need to organize a bit the cables, and we will also check if all of them are working tomorrow.
We also prepared the rest lens and mirrors for the telescope, if the green is availabe tomorrow we can continue the installation.
I swapped the lens to compose a cat's eye retroreflector for AOM path.
Next step is adjust the configuration and alignment by maximizing the efficiency with AOM modulation.
Activity of 09/06:
We removed spare ETMY substrate and applied first contact on the surface we didn't clean before.
We checked the calibration (figure 1 and 2) before installing TAMA size sapphire and measure the absorption.
R_surf = AC_surfref/(DC_surfref*P_in*abs_surfref) = 16.9 [1/W]
where AC_surfref = 0.40V, DC_surfref = 3.85V, P_in = 0.028W and abs_surfref = 0.22
R_bulk = AC_bulkref/(DC_bulkref*sqrt(T_bulkref)*P_in*abs_bulkref) = 0.797 [cm/W]
where AC_bulkref = 0.08V, DC_bulkref = 4.65V, T_bulkref = 0.55, P_in = 0.028W and abs_bulkref = 1.04/cm
We then moved the translation stage of the imaging unit to take into account the thickness of the TAMA size sample (60mm) from 70mm to 44.3mm.
Reminder: dL_IU = (n-1)*L_sample / n
After calibration and installation we checked the XY limits of the translation stage and modified the X one to take into account the size difference with respect to KAGRA size mirrors (IMPORTANT: change them to go back to KAGRA size).
The coordinates of the center of the sample are: [X_c, Y_c] = [327.75, 123.25]
Then we did a series of z scan to determine the z position of input and output surface. Fig 3 shows the last one performed with 10.1W of input power.
According to the phase profile the input and output surface are at Z = [44, 79]
Yuefan and Yuhang
According to Yuhang, the beam splitter after the Faraday, which is also at the bottom of the periscope was changed later last week after all the mirrors were installed. The beam doesn't seem so well aligned on the mirrors anymore.
So we tried to align again everything today, and at the same time we increase the beam height to around 224.5mm, to match the height of the quadrant on the breadboard.
During the alignment, we lost the green reflection very frequently, and at the last try when we want to install the second lens on the breadboard, the reflection beam became really low. Then we checked the end camera, half of the beam are out of the screen, and we could not use only the BS pitch local control to recover the beam height, but the PR pitch is saturating with zero corrections.
We checked the cables that connect the quadrant and board, if they are long enough for us to put the board above and below the bench. We tried to put the quadrant at the edge of the breadboard which is also the furthest position from the edge of the bench. In the attached pictures, we could see it is possible to do in both way. Anyway, I also contacted Matteo, he will go to ask about the cables next week when he will be in Nikhef.
Now the Faraday is located 10~14 holes after PBS, according to the measurement of beam parameter we could predict the beam size at both ends of this Faraday isolator. The result is shown in the attached figure 1. I also checked the aperture of FI(IO-3-1064-VHP), which is 2.7mm in diameter.
In this case, the FI aperture and beam size ratio is minimum as 2.7/0.945/2 = 1.4286. We could calculate the Gaussian beam power through an aperture is P/P0 = 1-e^(-2*(1.4286)^2) = 0.98. So in our current case, even the best-aligned beam will loss 2% of the incident power. Also, I remember that we achieved almost 97% of power transmission of FI even in the case of this beam clipping issue.
Usually, we make the beam five times smaller than the aperture. But it is really difficult to have space and meet this usual requirement. However, if we move this beam one and a half hole backward, the power cut by aperture will be 0.5%. This can be realized by moving the fork of green injection mirror (the last green mirror before going inside the chamber). Because this is blocking the way of moving backward.
The reflectivity of PBS is optimized with the same method of the optimization of the dichroic mirror(HBSY11).
However, there is a strange thing, which is the reflected field is even stronger than the incident field. See attached figure 1 and 2 (incidence and reflection). So the reflection is 109/106 = 102.83%. But anyway, the reflection is maximized.
[Aritomi, Yuhang]
We measured free running and closed loop CC2 phase noise from filter cavity (attached picture). CC2 error signal is 356 mVpp. Phase noise from filter cavity is large at low frequency. Although phase noise at low frequency is suppressed by CC2, bump at low frequency in squeezing spectrum might be phase noise of CC2. Odd number harmonics of 50Hz peaks appear when CC2 is closed. These peaks are related to electronics of CC2.
[Aritomi, Yuhang]
From previous measurement, dichroic mirror we are using (HBSY11) had only 96% reflectivity while the spec reflectivity is 99.3%. We tweaked angle of the dichroic mirror and measured BAB peak height when OPO is scanned. We improved the reflection by 4% from 95.2mV to 99.2mV. However, the optimal angle is not 45deg and we lose 18% of BAB at PBS after OPO with this angle of dichroic mirror. We tried to change the angle of the PBS by hand, but it didn't improve. We need to align the PBS.
[Aritomi, Yuhang]
We measured loss from PBS after OPO to filter cavity reflection just after PR chamber. We haven't measured loss from filter cavity reflection to homodyne yet.
| position | BAB power (uW) |
| after PBS | 273 |
| after faraday on the bench | 254 |
| after HWP | 248 |
| before PR chamber | 242 |
| after PR chamber | 211.5 |
From this measurement, loss from PBS after OPO to filter cavity reflection is 22.5%. Since we know that loss when squeezed light is directly injected to homodyne is 21%, total loss should be at least 40%.
We found that HWP we were using before had 5% loss and seems dirty. This explains 4% additional loss when I did additional loss measurement using this HWP. After replacing with new one, HWP loss becomes 2.4%.
Faraday on the bench has 7% loss. Alignment of the faraday should be optimized.
Current squeezing level is 2.4dB. Phase noise from laser around 10kHz becomes smaller. Spectrum at low frequency is bad due to CC2 lock loss during the measurement.
