NAOJ GW Elog Logbook 3.2
[Tomaru-san, Tanioka]
Today, we investigated the problem of the vacuum gauge.
Tomaru-san cleaned the inside of it, then I installed it to the chamber.
We checked the behavior by pumping down, but the situation was not improved.
Around 1.2*10-2 Pa, the vacuum gauge showed an error related to the sensor.
We will send it to the company and ask for repairing or buy a new one.
I found that the TMP in south end was failed. The TMP controller showed red indication but the LCD was disappeared.
Similar to the case in BS, oil vapor went from the RP to the TMP. Therefore the RP was replaced to the dry pump from the MC.
The TMP controller was replaced to the other one from the BS. The TMP is under baking to remove the contamination due to the oil vapor.
Baking time today (11/25): 14:00-17:00 (3 hours)
Baking time today (11/26): 9:30-10:30 (1 hour)
After the temparature settled, I opened the gate valves for the TMP. The pressure was decreased from 1x10-6 mbar to 2x10-7 mbar.
I tried to compute a rough calibration for the cavity mirrors actuation in DC.
I injected an offest of 10.000 counts in pitch and yaw, in INPUT and END mirrors, one at a time, and record the change seen by OPLEV.
Then I used the calibration reported in entry #1874 to convert it in urad:
| 10.000 counts injected |
INPUT |
END |
|
PITCH |
63 urad |
52 urad |
| YAW | 220 urad | 268 urad |
Pitch is stiffer than yaw. There is a difference of 15/20% beetween INPUT and END that can come from both errors in OPLEV calibration and actuation unbalance. Also It is a bit strange to me that for the same amout of injected counts pitch motion is larger for INPUT and yaw motion is larger for END.
[Federico, Matteo, Eleonora]
In the past few days we had many troubles with our DACs. Several channels were not working and moreover the issue seemed somehow "non stationary": some channels stopped working, than worked again, etc..
After we veryfied that the connections and the hardware have no evident problems, we decided to change the old DAC board with a new one. This seems to have fixed the problem. We also installed the cable to connect the second DAC board to the timing box and test the signal chain. We had to deal with several issues:
1) Dsub to BNC board installed in the cleanroom had a strange behaviour. We found some jumpers were missing inside and we soldered them (we could't find proper jumpers in the elec-shop). Now it is OK.
2) The AI board has only 8 channels (instead on 16) but only channels 2-3-4 seem to work fine. Channel 1 has a large offset and the signal is very much attenuated, channels 5-6-7-8 show crazy things.
3) In general both the dsub to BNC boards show an oscillation at about 1 MHz, with amplitude of volts, when the output is probed with a long cable. This issue was also observed in KAGRA. (Entry #6266 of Klog). According to our expert electronician this is due to the cable impedence which induces an auto-oscillation of the last op-amp. This could have been solved by inserting a resistence before the output. Note that we are currently sending such dirty signals to our coil-drivers. Hopefully the oscillation frequency is so high that they don't really care? Note that most but not all the channels show this behaviour.
Federico, Yaochin, and Yuhang
1. Test of small size beam with different power (attached figure 1 and 2)
power varies from 5mW to 12mW.
2. Test of 12mW light with a different beam size (attached figure 3 and 4)
3. The test of the large beam with different power (attached figure 5 and 6)
varies from 0.5mW to 12mW.
Note: According to the AA telescope design, the beam size is ~260um. The available power we can send to QPD is 2mW, the power distributed to rach quarter is 0.5mW.
I summarize the calibration for the OPLEVs.
For the conversion from counts to Volts, I used the following value: 1 V = 1638 counts -> cal = 6.105e-4 [V/count].
Note that for BS and PR we use TAMA PSD and no lens for shift/tilt decoupling.
PR (entry #276) --- SUM: 11 V
Yaw: 0.24+/-0.03 [mrad/V] = 0.146+/-0.02 [urad/count]
Pitch: 0.33+/-0.03 [mrad/V] = 0.21+/-0.02 [urad/count]
NB: for geometrical reasons pitch is multiplied by a factor sqrt(2) (see attachment in entry 276)
BS (entry #337) --- SUM: 13.4 V
Yaw: 0.37+/-0.03 [mrad/V] = 0.25+/-0.02 [urad/count]
Pitch: 0.37+/-0.03 [mrad/V] = 0.25+/-0.02 [urad/count]
NB: Since in this case the incident angle is almost normal the factor sqrt(2) can be neglected.
INPUT (entry #278 and pag 168 of my Phd thesis) --- SUM: 3.2 V
Yaw: 0.038 [mrad/V] = 0.027 [urad/count]
Pitch: 0.062 [mrad/V] = 0.038 [urad/count]
NB: We considered a factor 100 coming from SR560 amplifiers
END (entry #278 and pag 168 of my Phd thesis) --- SUM: 4.45 V
Yaw: 0.030 [mrad/V] = 0.018 [urad/count]
Pitch: 0.043 [mrad/V] = 0.026 [urad/count]
NB: We considered a factor 100 coming from SR560 amplifiers
It would be good to implement real time calibration for OPLEV channels.
Today I checked the beam jitter induced by the AOM frequency shift.
I measured beam shift by a beam profiler at ~2.5m distance from the AOM.
The beam shift was about 0.16mm by ~15MHz frequency shift and this is small enough for this experiment.
[note]
Tweaking the alignment of double-pass AOMs is needed when the beams are locked to the folded cavity in order to maximize the diffracted beam power.
The TMP in BS was replaced to STP-1003 on the 21st. After the one-day monitoring, I opened the gate valve on the pump. The pressure in the arm input was improved from 1.7x10-7 mbar to 8.5x10-8 mbar. Unfortunately the vacuum gauge on the BS is not working due to the over life-time.
We performed some tapping test in order to identify the resonances of the bench. On the attached plots x stands for the NS direction, y for the EW and z for the vertical.
- when tapping along NS we excited the 10,4 Hz peak (plot 3 shows the spectra during the tapping along the 3 directions)
- when tapping along EW we excited the 14,2 Hz peak (plot 4)
- when tapping along the vertical direction we excited a broad peak at 41.8 Hz and another broad bump around 84Hz, but not very clear (plot 5)
We also performed a switch off test of the fans over the SQZ bench. As you can see from plot 6, it doesn't seem to make any difference along the z direction
The last plot shows the FFTs of the accelerometers on the SQZ bench while excited and the ones of the seismometer on the floor. All 3 directions are shown together. It would be useful to compute the transfer functions between them.
In conclusion: 10,4 and 14,2 Hz are resonances of the SQZ bench, and the fan over it doesn't seem to create any excitation.
I checked the output voltage from the vacuum gauge and it was 0.2V.
This indicates that there is an error in the B-A gauge or the sensor.
I did degas, but this not change the situation.
One possible solution is to clean up the gauge.
Yaochin and Yuhang
The spectrum of BAB into AMC is as following. TEM00: 5.56V, Higher-order 1: 6.8mV, Higher order 2: 10.6mV.
So the matching is 5.56/(5.56+6.8e-3+10.6e-3) = 99.69%
Did you tune CC2 demodulation phase for each green power?
Yao-Chin and Yuhang
For the optimization of homodyne, we basically align again the homodyne. Including the matching of IRMC/AMC and OPO/AMC, the balance of homodyne, the beam height regulation.
The situation of IRMC/AMC matching is 99.95% while OPO/AMC is 99.7%.
The balance of homodyne now is done without aligning its BS but by aligning the lens before homodyne.
Benefit from not aligning homodyne BS, we could make squeezing go to homodyne flatly. Also, in this case, the balance of homodyne is easier.
After this work, we characterize squeezing again. We measured squeezing and anti-squeezing for power from 20mW to 60mW with an interval of 5mW. The result is shown in the attached first figure.
In the attached figure 2, we see there was a peak appearing. I need to mention that this peak appears after we did several measurements.
By fitting the squeezing and anti-squeezing, we could find out the loss and phase noise information. However, this time, the fitting and raw data has a larger discrepancy.
#############
I am sorry that I forgot to put the information about the Green power and demodulation phase.
| green pump power | MZ offset | OPO temperature | p-pol locking frequency | CC2 demodulation phase(sqz) | CC2 demodulation phase(asqz) |
| 20 | 4.1 | 7.166 | 175 | 75 | 160 |
| 25 | 4.2 | 7.167 | 175 | 90 | 170 |
| 30 | 4.3 | 7.18 | 175 | 100 | 175 |
| 35 | 4.4 | 7.18 | 175 | 120 | 170 |
| 40 | 4.5 | 7.19 | 180 | 120 | 160 |
| 45 | 4.6 | 7.19 | 170 | 125 | 155 |
| 50 | 4.7 | 7.19 | 150 | 125 | 160 |
| 55 | 4.8 | 7.195 | 150 | 125 | 155 |
| 60 | 4.9 | 7.2 | 150 | 135 | 155 |
#############
In the last attached figure, we can see the fit without considering the sqz-asqz for 55, 60mW.
Note: the anti-squeezing level is lower than the measurement did last week. This may come from the alignment issue of green pump.
Did you tune CC2 demodulation phase for each green power?
Yaochin and Yuhang
We found the two lenses before homodyne both have a scratch. One has scratch just in the center area. Another has a larger area scratched but not in the center.
The photo of the lens is shown in the attachment.
We will order new lenses.
I found that the vacuum gauge seemed to be working at relatively higher pressure level.
I vent a littele bit and started pumping down to see the behavior of vacuum gauge.
At some extent pressure, the gauge worked and the output voltage was not so strange.
The attached figure shows the pressure level (behavior of the gauge).
However, the voltage went crazy around 2*10-3 Pa and the display showed an error.
I checked the output voltage from the vacuum gauge and it was 0.2V.
This indicates that there is an error in the B-A gauge or the sensor.
I did degas, but this not change the situation.
One possible solution is to clean up the gauge.
Yaochin and Yuhang
In our lab, we have mainly three kinds of mirrors. One is from ALTECHNA, another is from LAYERTECH and the last is from Newport.
1. ALTECHNA mirror is the mirror we are basically using for squeezing injection to filter cavity. The company claim that it has a reflectivity of >99.5% for the HR side. And the surface quality is 20-10 SD. And wavefront distortion is lambda/8.
2. New port 5104 mirrors. From the website, reflectivity is >99.7%, wavefront distortion is lambda/10 and surface quality is 15-5 SD.
3. LAYERTECH. reflectivity is larger than 99.9%. We have only a few of them.
Yaochin and Yuhang
(Actually, we started to do this since last Friday)
After realizing the importance of this work, we started to move the two lenses between OPO and AMC. After reaching a matching level of 98.5%, we found the beam on the second lens is quite off-center. We know that this will cause the problem of astigmatism(see attached figure). So we should try to avoid this. However, actually, this is caused by the alignment of homodyne BS and dichroic mirror just after OPO.
For the homodyne BS, we adjust its pitch and yaw to balance homodyne. The consequence of this adjustment is that we need to tilt BAB in order to overlap it with LO.
For the dichroic after OPO, we forgot to change it back to flat after the characterization work of its loss. Then it may tilt the beam as well.
To correct these two things, we did again the check of beam flatness from OPO to homodyne. And at the same time, we changed the PBS after OPO to a normal mirror.
(This work is still going on)
Yaochin and Yuhang
We tried to match BAB to the filter cavity last Friday. However, we found the beam is quite misaligned. And we have a bad mode mismatch.
Later, we figured out that there are two different reasons causing this problem.
1. Change of Faraday isolator. The length of the old FI(IO-3-1064-VHP) is 114.3mm. The length of the new FI(FI-1060-5SC HP) is 58mm. This length difference is not negligible and will change the mode matching condition.
2. Change of lens1 position. This lens1 is the lens just after OPO transmission. It is the common lens for FIS and FDS path. After the installation of the new Faraday, the matching condition for FIS is changed a lot. To characterize FIS, we must move the common lens.
Anyway, we should first match BAB to AMC.
We will measure again the beam parameter and check the position of lenses after matching BAB to AMC.
The monitor indicates something wrong with the vacuum gauge...
Yao-Chin and Yuhang
We improved the matching between IRMC and AMC from 99.5% to 99.95%. As shown in the attached figure, you can find the TEM00 peak height(11V) and higher-order modes peak height(5.28mV).
I removed the gauge from the chamber.