NAOJ GW Elog Logbook 3.2

We have a problem of CC1 saturation and we found that the CC1 saturation is caused by FC lock/unlock. We should lock FC first and then lock CC1.

I checked the splitting ratio of pickoff BS (BSW11) for CCFC with BAB.
This BS is roughly 55:45 and gives 45% loss for squeezing.

Aritomi and Yuhang
Recently we found the lock of filter cavity always has problems. For example, after we try to remote lock, it takes a while to lock or it doesn't lock.
So we checked several setting for that. Firstly, the injected green power confirmed to be 24.2mW. According to elog1886, we checked the setting of remote lock.
Offset from DGS(unlock) | 3V |
Offset from DGS(lock) | 0V |
Threshold in rampeauto | -0.5V |
If I understand well entry1886, I think they are fine. (But it seems the offset from DGS(unlock) should be 1.5V. Although it is different from what we measured, it should not make difference for the remote lock performance)
Then we checked the Green_tra_DC with osilloscope, the transmission peak was only around 1V. So we decide to scan FC_green by sending a ramp signal to END mirror length. The measurement is shown in the attached figure.
It is clearly shown in the figure that there are visible sidebands around Green TEM00. Surprisingly, these sidebands are about half the magnitude of TEM00. Compared with elog1674, this is so much different.
We also checked the sideband of modulation from EOM by looking at the spectrum of GRMC reflection DC channel on oscilloscope. For GRMC ref_DC, the sideband is barely visible. So it should not be the problem of EOM modulation.
Today, we find that we were injecting 25kHz noise inside the PZT.
After removing the injected signals, the cavity scan was performed again. The diaggui file for cavity scan (green) is saved in Desktop/cavity as cavity_green_scan.xml.
This time, the spectrum is good.
Apart from this, FC green locking is normal again.

segment | WFS1 1 | WFS1 2 | WFS1 3 | WFS1 4 | WFS2 1 | WFS2 2 | WFS2 3 | WFS2 4 |
optimal DGS phase | 98 | 105 | 104 | 101 | 145 | 120 | 125 | 130 |


I started some analysis for finesse of the folded cavity.
The finesse was obtained by fitting the transmitted signal.
The finesse at room temp. was about 1.678*1e4(+/- 11) though that at 8 K was 1.579*1e4(+/- 26).
Further analysis is ongoing...

Aritomi and Yuhang
There was END mirror pitch/yaw coupling problem found in AA sensing matrix, as reported in elog2165. Besides, it was found the END mirror driving from coil/magnets has already been not clean, as reported in elog2173.
So we decide to find a good driving matrix for END mirror. Before doing that, the check about if the END mirror PSD is sensing well the pitch/yaw was done.
The method is to excite yaw of END mirror and check the response of signals in pitch/yaw channel relative to this excitation. If the sensing is not good, the resonant peak will go from yaw channel to pitch channel.
The test result is attached. By comparing the resonant peak at 1.58Hz, the sensing coupling was found to be around only 1%. This means that sensing coupling of oplev is not a problem.


What I did
- Ringdown and doppler measurements @room temp.
- Open loop TF measurement of PDH locking loop
Results
The obtained data are still in USB and floppy disk...
But the UGF of TF was about 3.1 kHz and phase margin was much larger than 30 deg.
Therefore, the lock seems to be stable one.
Next
- Cooling down to 10 K
I am planning to cool down the cavity to 10 K which is the designed temperature for the ET-LF.
And then monitor the finess in order to estimate the optical loss at 10 K.

[Aritomi, Yuhang]
We found that FC misalignment changes the shape of CCFC error signal and/or adds offset. This can cause the detuning fluctuation and we think that auto alignment is necessary to obtain the stable detuning fluctuation.
We measured CCFC error signal again (attached picture). CCFC error signal today is larger than last week at 100Hz-10kHz region. This may be related to the fact that we turned on the lasers today.

What I did
- Ringdown and doppler measurements
- Injected He gas to raise the temperature
Results
The attached file shows the very preliminary results of the ringdown measurements.
As shown in the figure, the finesse at 134 K, and 165 K is smaller than that of 170 K (judged by my eyeballs).
This might imply that the molecular layer i.e., amorphous ice can desorb around between 165 K and 170 K.
It should be noted that the fitting has not yet done...
Next
- TF measurement of PDH loop
Tomorrow, I will measure the TF of PDH locking loop.
Then I will re-cooling the cavity.
At this moment, the target temperature is 10 K and monitor the behavior of finesse in order to estimate the loss induced by cryogenic molecular layer.
I think this result would provide some implications to the Einstein Telescope.

I found one of the integrator was not working due to the insufficient soldering.
I remedied it, then the TF of the servo seems what I intented.
However, I found that the gain was too high for stable lock...
I will modify the servo or locking scheme.

On Friday I modified one of the two QPD demod box adding high pass filter (SHP-50+) and amplifier (ZKL-1R5+) for each QPD segment channel (picture1). See PDF pag.2 for current components of the demodbox. The second QPD demodbox is not yet cabled (picture2).

[Aritomi, Yuhang]
We replaced a flipping mirror for CCFC with flipping 50:50 BS. We originally had 36% loss and we have 50% additional loss from BS. The total loss will be 68%.
We measured FDS with CCFC lock (attached picture). We also tried to measured anti squeezing, but we cannot have anti squeezing larger than ~11dB so it is not shown in this plot. It seems the nonlinear gain is not optimized.
For HOM angle -1,-14,-28 deg, detuning is around 60Hz and more or less stable, but for HOM angle -38 deg, detuning is 44Hz. I'm sure that green alignment is fine for HOM angle -1,-14,-28 deg, but not sure for HOM angle -38 deg.
We will measure FDS again and check green alignment at the same time.
In this measurement, we assumed that green power is 56mW and generated squeezing is 21 dB, but nonlinear gain was not optimized in this measurement and actual nonlinear gain (or generated squeezing) should be lower. I assume that generated squeezing is 16dB in this measurement and fitted the FDS measurement again (attached picture). In this case, homodyne angle changes from 0 deg to 90 deg and detuning is between 89-98 Hz. The detuning fluctuation might be better with CCFC.

Aritomi, Matteo and Yuhang
Matteo implemented the RF amplifiers and filters.
The comparison of each segment is attached. Two orders of magnitude amplification is observed at low frequency.
This file is saved in DGS system in Desktop/AA with name WFS_amplify.

Just a reminder, sometimes we still suffer from the unlock of CC1 loop. This is mainly due to the saturation of CC1 correction signal. We need to investigate how to improve this.

I plotted the obtained ringdown data at room temp., 120 K, and 8 K, respectively.
The results shown in this entry are preliminary though, the decay time at 8 K is less than that at room temp.
It should be noted that the measured data at 120 K was 3days after the cavity reached 120 K.
I will analyze the optical loss of the folded cavity.
In addition, I am planning to compute the dynamical response of the cavity.

[Matteo, Eleonora]
In order to transform one of the broken TAMA demodulation board into a simpler phase shifter (to be used for CCFC lock) we removed the mixer and shortcut the output of the phase shifter to RF IN.
We measured the trasfer function between LO in and RF IN which is now an output. See attached picture. The TF is in agreement with what we expect from the phase shifter data sheet (SPH-16+).

[Aritomi, Yuhang]
Carrier and CC AOM frequency are as follows.
We changed a cable length for CCFC LO and checked the CCFC error signal. The result is as follows (Pic.1,2,3 are 0,2,4 m cable length).
CCFC LO cable length | phase |
0 m | 0 deg (I phase) |
2 m | 90 deg (Q phase) |
4 m | 180 deg |
Then we added the CCFC I phase error signal to perturb of green FC servo. We used SR560 with lowpass filter 0.1 Hz and gain 200 before injecting the perturb. Then CCFC stably locked!
When SR560 gain is 1000, it oscillates with 88Hz (Pic. 4).
We measured CCFC phase noise. For CCFC calibration, we used Pic. 5. Since AOM scan speed is 1600 Hz/s, the calibration factor (Hz/V) is 1600 Hz/s*70 ms/40 mV = 2800 Hz/V. Measured CCFC phase noise is shown in Pic. 6. Note that free run CCFC error signal is out of linear range and the spectrum could be wrong. I also compared with IR locking accuracy we usually use (Pic. 7).
We measured the ratio of CCFC OLTF and green OLTF with SR560 (Pic. 8). The crossover frequency is 40 Hz.
When the CCFC demodulation phase is optimal, CCSB frequency separation can be obtained from the distance of two dips in Q phase signal since the two dips correspond to CCSB resonance (Pic. 9). From Pic. 2, time difference of the two dips is about 120 ms and therefore frequency separation of CCSB is 120 ms*1600 Hz/s = 192 Hz. This is a bit larger than optimal value 108Hz. Note that when the CCFC demodulation phase is not optimal, it will also change the distance of the two dips.
Just a reminder, sometimes we still suffer from the unlock of CC1 loop. This is mainly due to the saturation of CC1 correction signal. We need to investigate how to improve this.
I attached OLTF of CCFC and green lock. Note that I flipped the sign of measured data to match the measurement and theory. The measured phase is not consistent with theory.

What I did
Today, I tried to measure the finesse of the cavity by locking the laser.
However, I could not lock and found the servo had some problems at low-pass parts.
I tried to remedy the circuit but could not solve the problem.
So I decided to make another circuit.
Actually this work has not finished yet...
Besides it, I did some measurements without locking the laser, i.e. doppler method which depends on the dynamical response of the cavity.
Also I measured the frequency split between transmitted p- or s- polarized beam.
As shown in the attached picture, there is a frequency shift between the s- and p- polarization.
The injected beam is pure s-polarized beam, but by playing with a HWP one can inject p-polarized beam.
This frequency spilt measurement corresponds to a cavity enhanced ellipsometry.
Results
To be honest, the analysis has not yet done...
The frequency spit was about 150 usec by 100 Hz 3 Vpp triangle wave.
Notes
The temperature of the cavity was about 120 K.
At first, the reflected (and transmitted) power was about 70 % less than I expected.
I doubted that the misalignment, but it seemed not.
Though I optimized the alignment, the reflected power did not improve.
After one or two minuites illumination, suddenly both of them recovered.
I could not figure out the reason, but the desorption of moleuclar layer is one of the possible reason.
Further investigations are needed.
I found one of the integrator was not working due to the insufficient soldering.
I remedied it, then the TF of the servo seems what I intented.
However, I found that the gain was too high for stable lock...
I will modify the servo or locking scheme.
Just to add a bit more information about my understanding. If CC1 loop is locked first, the lock of filter cavity causes the light from main laser having a large phase change. Then CC1 loop needs to give a large correction to keep the same phase. In the end, it causes the saturation of CC1 correction signal.
Matteo also suggested to feedback correction signal to the end mirror, which will offload the large correction sent to main laser.