Marc, Yuhang

The past trials to measure the AA sensing matrix taking into account possible coupling between pitch and yaw of both input and end mirrors were all based on injecting 2 Hz lines on each degree of freedom (dof).

This gave us strange results, especially visible in the TF phase between each excitation and sensing QPDs (ie far from +/-180 or 0 degrees).

As suggested by Raffaele in last FC meeting, this could arise because 2Hz is too close to the mechanical resonance of the mirror.

Therefore, we decided to use 15 Hz lines to measure the TFs.

We also tried to measure the demodulation phase between I and Q of each QPD segment using a 15 Hz line but this was not so much conclusive ( the line amplitude might have been too low and also hard to distinguish in the time series from the natural 11 Hz pitch resonance). So we used 2 Hz line to tune this demodulation phases.

They are now :

Then we injected a line at 15 Hz on each mirror dof with amplitude 1000 except for end pitch which was 1500 (without particular reason...)

Figure 1 to 4 present the TF magnitude (top left) phase (bottom left), coherence (top right) and time serie (bottom right) for respectively input pitch excitation, input yaw, end pitch and end yaw.

The TFs magnitudes give the sensing matrix absolute value while the phase the sign of each element. We used only TF when the coherence was above ~0.4. From this all phases was +/-180 or 0 deg within less than 10 degrees.

This gave the following driving matrix :

Note that the actual sign of the computed matrix is the opposite but we used a negative gain.

Using gain of -0.05 for in and end pitch and -0.05 for in and end yaw we could close the AA loop.

The comparison between the QPD and OpLev signals are presented in figure 5 (pink is live QPD, green is live OpLev).

We can see that there is almost no coupling between the various dof !

However, the AA loop gain needs to be reduced a bit more as we can see that it introduces noise (eg comparing the green with the brown := OpLev reference)

Abe, Aritomi, Marc

Previously we were using the pinhole to characterize the beams.

We found out that this technique is heavily dependent on the alignment on the pinhole and that the power fluctuations affect strongly the quality of the measurement.

Also, as the translationStage.v3.vi that allows to automatically perform many scans along the Z direction is now available, we switched to using the razorblade with this vi.

The 2 lenses have been moved so that L1 (closest to the periscope) and L2 (closest to the imaging unit) are now respectively :

L1->L2 = 21.2 cm

L2-> z=0 of translation stage = 23.4 cm

With this configuration, we characterized the red and ir beams when cutting the beam vertically. For this, we placed the razor blade as close as possible to the 2" holder at the edge towards the lenses.

The measurement and fit are presented figure 1 (red := red beam, black := ir beam) where the z position corresponds to the real one (ie the half width of the 2" holder (9 mm) has been substracted)

It gives :

ir : waist size = 37.5 um at z=38mm

red : waist size = 76.6 um at z=42.8 mm

During the week-end, we started several measurement of the red and ir beams with the blade at the edge of the 2" holder towards the imaging unit (ie 1.8cm closer to the imaging unit than the measurement presented in this entry). This will allow to check the crossing point of these 2 beams.

We should get all the required measurements on Monday.

Note that the height of the ir beam has been corrected following this measurement as shown in elog2444.

Marc and Yuhang

I measured frequency independent squeezing on Thursday (20210408), the difference between this measurement and last measurement is the improvement of IRPS. The result is shown in the attached figure 1.

After this, I also flipped lens inside homodyne. The minimum ROC of gaussian beam with w=390mm is (x_R+x_R^2) = 650mm. This fits better the flat surface of lens, which makes it reasonable to make light hits on lens flat surface. After this work and tilting lens a bit (few degrees), I got squeezing spectrum as the attached figure 2.

Figure 2 represents the cleanest squeezing spectrum we have ever got.

[Aritomi, Abe, Marc]

Since IR beam height was higher than red beam by ~3 mm as shown in elog2446, we adjusted the height of two lenses for IR beam. We made the first lens lower by 3 turn of screw to make the beam lower after the first lens. By doing this, the IR beam height just after the second lens changed from 55 mm to 52 mm. Then we made the second lens lower by 6 turn of screw to make the beam flat.

We did knife edge measurement in X direction at Z = 25 mm and 70 mm to check if the IR beam is flat. Attached figures show knife edge measurement at Z = 25 mm and 70 mm. The beam height are X = 316.96 mm and 316.85 mm, respectively.

Now IR beam is flat and the IR beam height is almost the same as the red beam height.

Axis of figure 2 should be "beam radius"

I checked some optics in the ATC cleanroom to prepare for the construction of the new OPO.

We can use the Faraday isolator Thorlabs IO-5-1064-VLP - aperture 5mm diameter, 9cm length, maximum power 1.7 W, 25 W/cm^2 (blocking), 100 W/cm^2 (transmission).

Using an f = 100 mm lens placed as shown in figure 1, we can maintain the beam waist below 1mm diameter over a distance of [distance] from the 100mm lens, so it is sufficient to use for the FI.

I did a rough estimate (figure 2) and measurement (figure 3) of the beam size to make sure. We can maintain < 1mm beam diameter to about 35cm from where I placed the lens.

First picture: X axis

Second picture: Y axis

Today, I tried to lock OPO but not successful. While checking OPO error signal, I found the error signal didn't seem to be reasonable. By adjusting only the phase of channel 3 (LO for OPO PDH demodulation), I found the optimal phase changed from 135deg to 205deg. This change of 70deg probably comes from the 90 phase change of DDS, combined with the original phase was not optimized.

Attached figure shows the PDH signal when new optimal phase is set.

Marc and Yuhang

Yesterday, we found p-pol PLL couldn't be locked. While checking it, we realized that it will be better to put p-pol PLL on the second layer breadboard.

The new location of p-pol PLL fibers are shown in the attached figure.

Yuhang and Michael

The IR phase shifter position was shifted to the beam waist as per 2435. The arrangement is shown in figure 1, 2. The half wave plate is contained inside the lens mount. We then measured again the noise spectra using the PSD after transmission through the IRMC.

The results show that there is now considerably less change in the noise spectrum when excitation is applied to the phase shifter, versus when there is no excitation. In figures 3 and 4 we show the normalised spectra for pitch and yaw. The X/T and Y/T noise look to have a noise floor of about -30 dB, versus -20 to -25 dB as shown in 2422. The broad peak at ~2.3 kHz also looks like it is suppressed. Figures 5 and 6 show the absolute measurements from the PSD compared to the dark noise. The result is improved versus that in 2407, we don't seem to have mysteriously high pitch noise anymore.

Based on the design from Pierre and me, I modified the TAMA PD today. The modification is as following:

1. Change op-amp to LMH6624

2. Set R1 to be 33Ohm

3. Set R2 to be 330Ohm

4. Put 3pF capacitor in parallel with R2

5. Change photodiode parallel capacitor to be 150pF

Figure 1 is after I remove op-amp. Figure 2 is after I put the new op-amp. Figure 3 is after I put R1, R2, 3pF and 150pF. Figure 4 shows the label I put on this PD.

I also measured the TAMA PD noise level with bandwith of 2GHz and 10MHz. The 10MHz measurement is around 14MHz. I found:

1. The offset is about 0.1V

2. Figure 4 shows the main oscillation is around 268MHz with amplitude of 5uV_rms/sqrt(Hz). 

3. Figure 5 shows the noise peak is a bit higher than the designed value. This is due to the photodiode parallel capacitor was chosen to be 150pF, but the designed value is 160pF. We can choose to put a 10pF in parallel with the 150pF one.

Michael and Yuhang

We moved IRPS to be very close to the waist position. According to a simulation I have done, this will reduce quite a lot the amplitude noise.

To test if this really solves problem. Firstly, after realigning everything, we got the IRMC spectrum as shown in attached figure 1. Then we gave a high voltage difference of 70V, which is almost the maximum voltage we give for IRPS while CC2 loop is locked. In this case, we got IRMC scanning spectrum as attached figure 2. We computed misalignment to be 4.5%.

Compared with the result we had from elog2424, the LO amplitude noise should be reduced by a factor of ~5. From this result, combined with input mirror feedback, we can completely avoid amplitude noise coupling at low frequency when CC2 loop is closed.

Michael and Yuhang

We recovered filter cavity alignment. While doing the recoverment, we found some issues of picomotors. Especially, the BS yaw had problem. The problem is when we move yaw, the actual motion was pitch. The solution is to disconnect the pitch control wire. Then when we move yaw, we can really move it.

After recovering mirrors positions, we got cavity flash as attached figure 1. Then we centered all oplev signals on each PSD. After that, the medm interface is shown in the attached figure 2. In the end, we also measured the spectrum of each oplev signal. FIgure 3 shows the comparison with old measurement. No issues were found.

Manuel (remote), Marc

Recently Manuel is helping me to try to restore the translation-stage v3 and v4 VI.

Yesterday, Manuel found out that the problem of v3 is that the left panel to choose all z values for the scans are overwriting the z value in the usual panel (same one as v2).

So it is needed to put first the z values on the left panel, then start the vi and and then everything works properly.

Aritomi, Marc, Matteo, Simon

Yesterday, we started by aligning back the ir beam on the 2 lenses [ before I wasn't thinking about the fact that adjusting the beam position by acting on these lenses will introduce astigmatism...]

Then we used the pinhole to characterize the ir beam.

The beam waist was too far (~60mm) so we moved the first lens after the periscope closer to the periscope.

I discovered that previous fits were wrong because i was trying to fit at the same time beam waist [um] and position [mm]. I corrected this and results are presented in figure 1 (red is red beam, blue is ir).

The situation is :

- crossing position is roughly matched to the 2 beam waists

- ir waist roughly 2 times too large -> need to move the second lens after the periscope closer to the periscope by ~1mm

- ir beam is not horizontal -> higher than red beam (larger z corresponds to smaller height)

Marc, Michael

On Wednesday we started the FC recovery.

First we checked the OpLev spectra and it seemed fine (no extra peaks indicating touching).

The PR references are good meaning that the squeezer alignment were good.

We moved PR picomotors to recover the reference on BS chamber.

We started to move BS picomotors to recover the beam on the 2 targets. But at some point the pitch picomotor stopped moving..

We did ~2000 steps for pitch.

We checked the OpLev spectra again and seemed fine so no touching.

I measured again the red beam when passing through the pinhole and extracted waist size and position (from power transmitted by a circular aperture) and the propagation directions (fig.1 in red).

The goal is to have the ir beam crossing the red beam at their waists with angle between their propagation directions of 0.1 rad (in fig 1 it corresponds to Y direction).

The corresponding ir propagation direction is shown in blue in figure 1.

I placed the pinhole at 2 positions (extracted from the figure) and acted on the 2 lenses after the periscope to maximize the power transmitted by the pinhole.

However, the ir beam was quickly clipped on the last lens before the translation stage.

Also, I wanted to use a small rail to more precisely control the first lens position and therefore the waist position.

However, the lens mount is too large and it is not possible to use the rail with this mount.

Marc, Matteo

Yesterday we tried to shift the ir beam using the pinhole at 2 positions.

However, after some back and forth between these positions, the power transmitted started to have a steep decrease...

Recentering the power-meter did not help.

We tried to remove the pbs and the waveplate between the 2 lenses as the beam started to get misaligned on these optics.

It did not change the power level.

Then I did several scan of the position of the imaging unit to check which is the good distance of the imaging unit with the new lens position. Before each measurement I tried to maximize both AC and DC.

I recorded both AC and DC and in figure 1 you can see the AC/DC in function of the imaging unit distance.

It seems that 74 mm corresponds to the maximum (previously it was 70 mm). Also, at this position there was a maximum of R~14 (goal is 18) maybe due to a better alignment compared to the other measurement?

I also preformed scan at each of these positions which still show umbalanced peaks.

Now I'll try to check again the beams propagation direction and the ir waist position as it might have changed again due to the removing of the pbs and waveplate...

If this does not work I'll try to remove the 2 lenses and start from scratch.

Today I tried to put the new Shinkosha sample (1'' thickness) into the new holder for this thickness.

No problems on this side.

Yuhang and Michael

The vertical alignment of the beam reflection from the pase shifter was improved. We can see that the incident and reflected beams stay at the same height for a long distance (figure 1).

Afterwards, the pitch efficiency of the IRMC was checked. In 2424, the phase shifter was given a DC voltage in order to cause mode mismatch in the IRMC. Much more light went into TEM01 than TEM10 at high offset - at 70V offset on the high voltage driver, the ratio of TEM01 to TEM00 was approximately 23%, as seen in figure 2 of 2424. This time, we did a quick measurement of the reflection spectrum at 70V offset and obtained 544 mV in TEM00, 132 mV in TEM01 and 32 in TEM10, corresponding to 18.6% power into TEM01. Which is improved but still not optimal.

From finesse simulation and experimental results, we theorize that the pitch misalignment is coming from bending of the piezo element controlling the phase shifter. In this case, there may be a static bending in the pitch direction causing misalignment when the piezo is excited. Ideally we only want longitudinal motion of the piezo. Yuhang also simulated the effect of phase shifter misalignment versus the phase shifter proximity to the beam waist, and found that the noise is reduced as the beam waist is brought closer to the phase shifter. In the previous arrangement, the beam waist after the 250 mm lens was located approximately two breadboard holes from the phase shifter (see 2393 for a screenshot of the relevant optical table layout). Now, the beam waist is located approximately one breadboard hole after reflection from the phase shifter, and we see a corresponding reduction in pitch noise from the measurements in 2407 compared to 1904.

Following this, we will move the phase shifter to the beam waist position. However, there isn't enough space on the optical table to move the phase shifter one hole backwards, so the IRMC mode matching lens setup may need to be reworked.