According to the actuation calibration in elog1877, the driving of 5urad can be represented in the unit of counts.

The excitation is therefore decided to be the following table (at 2Hz).

According to this excitation, the response (sensing matrix) is as the following table

The offset of these signals is listed as the following table

Then subtracting the offset from the sensing matrix. The sensing matrix (after substracting) is used to decide the driving matrix index magnitude. The principle is that if the response is larger, the driving needs to be also larger. Without considering the coupling between pitch and yaw, the driving matrix is decided to be

The sign of the above driving matrix is decided by the time-series measurement yesterday.

According to the WFS error signal in time, it can be inferred that situation of QPDs location is shown in the attached figure.

In this case, the error signal from input mirror is shown with opposite sign on QPD1/2. While the error signal from end mirror has the same sign.

[Aritomi, Yuhang]

We checked nonlinear gain with 56mW green and found that the nonlinear gain is below 20 while the theoretical nonlinear gain is 37.5. So we optimized the nonlinear gain. We measured OPO transmission of BAB with power meter (power meter range: 8.8 mW) while OPO and green phase is scanned.

Then we measured FDS, but we couldn't find any squeezing with 56mW green. We decided to reduce the green power to 41.5mW (MZ offset 4.5). We optimized OPO temperature and p pol PLL for 41.5mW green. The nonlinear gain is 13.4 and consistent with theoretical value which is 12.8.

The error signal in time tells phase information(error signal in/out of phase with driving) and locking information(if error signal is around zero, the locking is good). Therefore, I checked WFS error signal in time on QPD1/2 while driving/nodriving INPUT/END mirror Pitch/Yaw separately. This time, the driving is at 2Hz.

Figure 1: no driving. Even when there is no driving, the pitch error signal oscillates around zero, which indicates mirrors have very large oscillation in pitch direction (known before in other ways).

Figure 2: Drive INPUT pitch. We could see WFS1 reconstructed pitch error signal is in phase with the local control pitch motion (driving). But WFS2 reconstructed pitch error signal is out-of phase with the local control pitch motion (driving). We could see that there is some signal in Yaw error signal, but if you compare it with the no driving case, this Yaw error signal fluctuation seems not to be from Pitch driving (or coupling is covered by yaw original fluctuation). (The coupling to Q_phase is also very small)

Figure 3: Drive INPUT yaw. We could see WFS1 reconstructed pitch error signal is out-of phase with the local control yaw motion (driving). But WFS2 reconstructed pitch error signal is in phase with the local control yaw motion (driving). The pitch/yaw is also not visible. The coupling to Q-phase is also quite small.

Figure 4: Drive END pitch. Both WFS1/2 are giving out-of phase error singal. The coupling to yaw could be seen in WFS1. There is also lots of coupling to WFS2 Q phase (will not be used to feedback, so it's fine).

Figure 5: Drive END yaw. Both WFS1/2 are giving in-phase error signal. The coupling to pitch is not obvious. The coupling to Q phase is also not obvious.

The phase information could be used to decide the sign for feeding back.

I'm monitoring the finesse of the cavity and it is slightly decreasing.
I'm planing to try to desorb the molecular layer by illuminating the main laser.

Firstly, I checked that PDH signal coming from Q-phase gives basically 0 when crossing carrier frequency (DDS2 channel3 phase 150deg). Then to check the RF gain/demodulation gain, a check of PDH singal in time was performed by looking at I-phase channel.

To have values for each segment, the alignment was checked to be as good as possible. (green tra dc is around 5200~5300) (a moment of flash is shown in attached figure 1) For each measurement, the light beam is centered on separate segment of QPDs by adjusting its closest steering mirror. (as shown in attached figure 2)

After checking demodulation phase, alignment and beam centering, the PDH singal was measured (several measurement is attached) and summarized as the following table.

Note that: the value showed in above table comes from a single measurement. The PDH signal actually have small variation around 10mV.

Although this measurement is not very much precise, it gives information about the signal coming from each QPD's segment.

[Yuhang, Aritomi]

[Aritomi, Yuhang]

First we fixed DDS AA phase to 150 deg and optimized I/Q demodulation phase again. As long as FC alignment is good, changes of the optimal demodulation phase are within a few degrees. During this measurement, we checked that WFS1 I3/Q3 coupling is less than 3%.

Then we injected a 12Hz line to INPUT PIT and measured sensing matrix, but there is still large PY coupling. We found that 12Hz peak heights on each QPD1 segment are quite different (following table).

This gain unbalance may cause the coupling problem. So we compensated this gain unbalance in matrix in DGS (attached picture). Each number in the matrix is decided by 10/(12Hz peak height on the segment). In this case, there is no coupling in WFS1 I YAW for INPUT PIT driving, but there is still 16% coupling in WFS1 I PIT for INPUT YAW driving.

After that we aligned FC well and measured 12Hz peak height again. This time the gain unbalance is different from previous measurement.

To decide the gain unbalance precisely, we will check PDH signal on each segment and calibrate it by sending a 12 kHz line to PZT as we did for FC PDH signal. 12kHz is within DGS bandwidth and it is around UGF of FC loop. Since what only matters is ratio of gain of each segment, it is not a problem even if the injected line is suppressed by FC control loop.

There was strange coupling observed in elog2206, but I think it may due to the not proper excitation singal sent to END mirror.

Excitation: the excitation is a sine wave, with amplitude from 700 to 11000, sent to END mirror pitch. (An example shotscreen is shown in attached figure 2)

Measurement: the spectrum of end mirror optical lever pitch/yaw were observed. (An example shotscreen is shown in attached figure 3)

The response information is extracted by using the cursor at 6Hz on each spectrum, and read the value of cursor. The coupling is the ratio of yaw/pitch peaks.

The coupling was always around 5.2%, which is the minimum could be found now.

The method used to find minimum was by adjusting the coefficient for H1 and H3 coils (giving them same/slightly different values from 0.02 to 0.06). The minimum is around -0.05 for H1 and 0.05 for H3. 

What I did

I replaced the PD at transmitted port in order for precise measurement of decay time.
Current PD has the bandwidth of 150 MHz.

Then I measured the ringdown of transmitted beam around 8 K.

Results

The PD has fast response such that the measurement can be done well.

The finesse was obtained by fitting the measured data, and it was about 1.65*10^4  though 1.67*10^4 at room temperature.

Next step

As the UGF of PDH servo is about 3 kHz and cannot achieve stable lock, I have to modify the servo to set higher UGF.
In addition, I will monitor the finesse behavior for a while.

Aritomi and Yuhang

As reported in elog2173, the driving for end mirror has some coupling between pitch and yaw. To decouple them, we decide to modify the driving matrix.

However, we found out that the coupling between pitch and yaw is different for different excitation strength. The coupling situation is shown in the attached figures and sumarized in the following table.

This measurement is done after optimizing the coupling with excitation of 1000. The pitch driving matrix is as following:

I compared green, BAB, CCFC locking accuracy. 

As pointed out by Aritomi-san, the formula used to calibrate the measurement had some problem (check entry642). After correcting that, the measurement result becomes reasonable.

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.

As pointed out in the last FC meeting, the error signal for green and infrared around 10kHz is similar. This is actually strange for me. Due to the cavity pole for infrared and green has a factor around 25 difference. Above their pole frequency, the green error signal should be around 25 times larger than infrared.

However, I checked several times this entry and compared with elog642, I couldn't find what is wrong. I will try to measure it again.

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.

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.

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.