Yuhang, Aritomi, Michael

From 2024-01-09

We optimized IR transmission from the OPO. The BAB transmission through the OPO when locked for IR is 320 uW. However, the ppol frequency went to 245 MHz - previously, the "IR only" ppol frequency for OPO was 220 MHz, so it has changed quite a lot in a short time, so this OPO/ppol frequency issue might be independent of green.

To align the filter cavity for IR, we do 1) align to reference targets on PR window, 2) align injection beam with vacuum tube targets, 3) check reflection on squeezer bench, 4) lock FC for green and adjust detuning of IR (green AOM) to find IR resonance, 5) align IR to maximize IR FC_TRA. This constitutes locking of the filter cavity ready to measure FDS. The infrared reflection signal is obtained through a 0.5% transmission polarizer, so quite difficult to see. Anyway after just a small adjustment of input steering yaw I could see IR reflection on a sensor card on the squeezer bench already, so we really didn't need to do much intermediate alignment. It was moving quite a lot (there was an M5 earthquake in Niigata at the time). The reflected IR power at first glance was about 240 um, 75% reflection. 

Eventually we could see flashing of the IR at about 109.037 135 615 000 MHz on the AOM detuning. On second inspection of the targets in the vacuum tube the GR/IR overlap seems quite bad though, so it will have to be fixed.

Yuhang, Michael

From 2024-01-08

OPO kept locking to higher order mode until I flipped INV switch on the servo. I should really figure out how the INV, +/- and threshold work together because it seems it isn't written down properly anywhere, but it's important for setting the proper behavior of the lock threshold to not catch HOMs, and depends on the sign of the error signal and use of transmission or reflection specturm.

To we keep BAB injected into OPO. The transmission is not so good, about 52 uW (should be about 400).

The green beam was aligned to the filter cavity using the targets inside the vacuum tube and the coils were offloaded to keep the best position around 0V actuation. The filter cavity was then internally aligned via reflection through the second target and to the squeezer bench. The END SR560s were turned to 1000 Hz LPF, Gain 100, low noise mode, 6dB/oct, DC coupling.

Filter cavity locks for green but is not super stable. If we put 1/f^4 filter it unlocks. We changed sideband frequency to 88 MHz so this might affect PDH.

We took the transfer function of the green FC lock (Source -> Signal:Perturb, Ch1 -> EPS2, Ch2 -> EPS1). The unity gain frequency is about 12.5 kHz, should be a bit higher - the gain was adjusted to give 13.256 kHz.

There is some weird behavior on the MEDM screen where the displayed FC_GR_TRA goes up when the filter cavity is unlocked. It seems to be an electronics issue regarding the displayed value.

Yuhang, Michael

From 2024-01-03 to 2024-01-05

We took the replacement OPO from the ATC cleanroom to TAMA and (finally) intended to replace it.

We decided to do a quick check for squeezing and afterwards just swap out the OPO. Since BAB, CC and ppol all end up being commonly aligned to the OPO via 2 steering mirrors after the recombination beam splitter, we decided we could just take it out, replace it and then recover the alignment with either of those beams. ppol is probably the most straightforward since it has its PD connected to one of the monitor channels on the OPO servo board. We should probably get a camera to check the shape of the SHG mode mismatch but for now we will just use transmission spectrum to align.

We drew lines around the old OPO to refer to exactly where to put it. After realigning the ppol reflection to itself we looked to scan the transmission spectrum, but the HVD would not turn on. After extracting the HVD -> OPO connection it turns out the HVD is fine, so the problem is somewhere in the electrical connection to the PZT. 

It seems the new OPO PZT is broken. So we had to put the old one back in. I accidentally burned a mark in the POM casing (the ppol laser is < 50 micron radius at the entry point of the OPO housing). After readjusting the alignment and mode matching of ppol to the old OPO, it was seen that the mode matching was improved by a small amount compared to the reference level shown on the wiki, so there is no issue with the crystal. There are two lenses on a ruler rail before the OPO - the positions before taking out the OPO were 170/264, and after replacing were 168/262, with 91.3% mode matching.

The old OPO PZT gives about 800 nF capacitance (about the same as the datasheet) while the new OPO doesn't give a reading.

I have taken two power meters from ATC clean room. They are now in Tama. 

They will be used to characterize our laser. 

Schematics can be found here. All filter shapes are written. Actuation gain of PZT and temperature actuator can be found in previous works, Niwa's master thesis for example.

I measured the open loop transfer function of the main-cavity system.

UGF is ~2 kHz and phase margin is 80 degree.

See the whole plot and local plot around UGF.

Let me leave a brief note about "frequency response analyser" of Moku:Lab, especially the definition of the vertical axis of the upper panel (see this screen shot).

Since it uses a unit dB, sometime I'm (and someone should have been) confused if this means in power or amplitude; as a conclusion, it means power.

To confirm this I used 1st-order low pass filter in SR560 with cutoff of 100 kHz as a reference. The screen shot is the measured transfer function of that. 

At 100 kHz, you see the decrease of gain by -3 dB. This means:-3 dB loss in power (=in Moku:Lab) corresponds to *1/sqrt(2) in amplitude.

More generally if the output gain of Moku is G dB, the amplitude gain g in log-scale should be calculated as:

g = 10**(G/20)

Ex.) -20 dB in Moku:Lab corresponds to 1/10 in amplitude.

I measured the finesse of cavity by scanning the laser frequency through the temperature actuator. The result shows that the finesse is twice lower than the designed value.

Designed power transmissivity and loss of mirrors from Layertec are:

T_in = 4000 ppm, T_end = 35 ppm, L_in=L_end=30 ppm,

therefore the cavity finesse should be ~1500. With cavity length l=15 cm, cavity pole should be:

T_in*c/4/l/2/pi = 320 kHz.

However, the measured cavity pole was 635 kHz i.e., finesse is =750. You can see the Lorentzian plot here. Calibration factor from scanning time to frequency is derived from the RF sideband frequency of 15.24 MHz.

This result means that cavity loss is ~4000 ppm, which makes this cavity totally useless for speed measurement.

Some items that were taken from TAMA to ATC clean room were brought back.

1. Beam Profiler and PC
2. Power Meter Display Monitor

See image here.

PS: Since the problem of not being able to upload images/files to elog persists, I decided to upload pics in my google drive, and put a link in the elogs. I would recommend others to do so too, because in near future everyone will face difficulty when they will wish to refer back to certain things.

Yuhang, Michael

From 2023-12-30

Don't accidentally send BAB and LO to homodyne at the same time otherwise you will get a large DC signal :O

Replace the power supply of the CC fiber PD

We realigned the LO and BAB to alignment mode cleaner and rebalanced the homodyne in order to reduce the optical loss for the mass data taking. We did some other checks. The CC PLL fiber PD for whatever reason is using a hard-to-find battery (A23 12V). We really should replace it with a power supply. While checking the green path we found that the SHG has some ridiculous 3rd order mode mismatch. It's about 30%. It cannot be reduced by beam steering, laser line filter or adjustment of mode matching lens position, and also doesn't appear to be induced by clipping anywhere. Still, it generates enough green power so it's workable for now. We saw that GRMC could not lock, and it turns out it wasn't even aligned. This is seemingly due to the reason mentioned previously, that the GRPS PZT had some hysterisis from sending the too large wrong signal. In fact, it could be realigned just by adjusting yaw from the steering mirror just past Mach-Zender. This mirror normally shouldn't be used for alignment to GRMC since it is common alignment for arms in the Mach-Zender, but in this case it worked because the PZT introduced a common misalignment. While the SHG has some weird issue, the GRMC still sends 25 mW to OPO at the appropriate MZ offset setting.

We checked CC alignment to OPO (send CC laser to OPO, block LO, check homodyne). It seems fine. So we decided to check again the ppol PLL frequency for 25 mW green and found that the CC1 error signal becomes quite large. The optimal ppol frequency, representing the OPO co-resonance condition between p-pol and s-pol, changed from 160 MHz to 250 MHz to 75 MHz in the space of about 10 minutes. This is some very strange behaviour, later we asked Matteo but he didn't have a suggestion other than multimode laser mode hop, which we are quite sure we do not have at the relevant laser current/temperature settings. Anyway we just decided to keep going on the basis that we had to take data before Hsien-Yi left. So the 25 mW green gives ppol frequency 75 MHz, for now. CC1 error (indicating squeezing level) was 384 mV, a bit larger than last time. 

Anyway, the appropriate signals were sent to DGS:

CC2 eps1 (representing local oscillator phase -> homodyne angle): K1:FDS-FC_GR_CORR
Ramp signal from function generator: K1:FDS-FC_GR_ERR
Homodyne (from SR560 with gain 1000): K1:FDS-ADCspare_1

We ended up taking data at MZ offset 4.1, 4.2, 4.3, 4.4, 4.45. For MZ offset 4.3 and 4.4 I had to reduce the SR560 gain from 1000 to 500 since it was overloading, and then again to 200 for the highest green injection power. This will also increase the proportion of dark and ADC noise. Also, when changing from 4.2 -> 4.3, the ppol frequency went from 75 MHz to 240 MHz - normally it is 160 MHz and doesn't change very much when the green injection power is changed slightly. But we kept on with taking data. During each green injection value I also made sure to check that the scan still covered squeezing and antisqueezing. The scan range also had to be reduced slightly for higher green injection due to the CC2 loop unlocking more easily.

After 30 minutes for each data point, I then took 5 minutes each for homodyne shot noise and homodyne dark noise at each SR560 gain level used in the measurement.

Yuhang, Michael

From 2023-12-29

We felt the squeezing (CC1 error signal when scanned) was too low, so we decided to realign alignment mode cleaner (i.e. realign infrared to homodyne). The alignment mode cleaner can have some backlash if adjusted over a large range too quickly, i.e. if the system is out of use for a long time. The LO was a bit misaligned, and BAB was very misaligned with about a 10% mode mismatch appearing since the last alignment a few days ago.

We then saw that CC2 has some issue with a quite fast oscillation in the error signal (when scanning, it should be going slowly up and down on a timescale of about 0.5s). It turns out I accidentally put the AMC scan signal (1V 10Hz) into the CC2 phase scan for data taking (should be 10 mV 0.01 Hz). Apparently this misaligned the green phase shifter (before GRMC) quite a lot, and the GRPS PZT had some hysterisis and didn't recover properly.

Otherwise, LO and BAB alignment were individually fine afterwards.

From 2023-12-28

Yuhang, Hsien-Yi, Michael

The goal of this day was to properly characterize and set up the squeezing/antisqueezing scan for taking the data for Taiwan's machine learning project. From the squeezing vs power data that was taken previously, it seems we can only go up to about 47.5 mW injection before the data becomes less meaningful.

We send a scan signal to CC2 Perturb in but the error signal becomes quite noisy. It was sent to an SR560 (although it's just in low noise mode, DC coupling, 1 gain - no filtering is selected). While inspecting the output of the homodyne time series in DGS, we suspected there was some DC power being amplified and deamplified inside the OPO. Perhaps it is backscattering of the LO from the homodyne detector into the OPO (a recurring issue). We wanted to look at bit further back but get 'synchronization error' in dataviewer. This must be improved with the installation of new DGS. Anyway, we saw that the possible scan range of CC2 phase without unlock is somewhat low, so we decided to check visually if it covers squeezing and antisqueezing. We sent a constant voltage at several points from -10 to 10 mV to CC2 Perturb In and checked the homodyne spectrum in diagguie, where the electronic and shot noise levels were saved last time. We managed to set the homodyne angle to a good point where we can start just before peak antisqueezing and end just after peak squeezing. Also in DDS 40 degrees homodyne corresponds to squeezing while 90 degrees corresponds to antisqueezing. A bit odd.

So we confirmed the feasibility of the squeeze anti-squeeze scan and were ready to leave it and take data. Considering that the fitting previously also showed large optical loss, we decided it would be good to optimize alignment of BAB (squeezing) and LO to alignment mode cleaner (i.e. homodyne). But we ended up taking data at 25 mW green injection for about half an hour.

From 23-12-27

Yuhang, Michael

To summarize, we verified the capability of the CC2 phase scan for Taiwan's Machine Learning project and also took squeezing measurements from 4.1 to 4.6 mV MZ offset (20 - 45 mW injection). We saw a total of 28% optical loss and 24 mrad phase noise, however, this measurement is dubious due to continued issues with CC2 glitch noise and rushed alignment of tabletop mode cleaners.

The PLL was left on from the previous day and remained locked when we started. To prepare for the data taking via the DGS system, we took measurements of the background noises. The baseline is Analogue to Digital Converter noise, which is at a set level, measured by putting a 50 Ohm terminator on the ADC port and taking a spectrum with diaggui. For the Taiwan machine leaning data, we send homodyne to SR560 and then to the spare ADC port K1:FDS-ADCspare_1_. The electronic noise of the SR560 is taken by terminating it with 50 Ohms at the input. The gain of the SR560 is set to put the SR560 electronic noise level just above the ADC noise level. This ensures that the optical signal from the homodyne is amplified enough to clear the ADC noise baseline. We take homodyne dark noise by measuring homodyne with no light input, which is only valid at one gain selection of SR560 (we end up having to change it later). Then we take homodyne shot noise by measuring the spectrum injecting just the local oscillator (after checking homodyne is balanced). Still pictures cannot be posted to elog, but the homodyne power spectrum shot noise is about 20x above homodyne dark, 40x above SR560 electronic noise and 50x above ADC noise (approximately).

For the Taiwan ML project, we want to somehow scan the squeezing and antisqueezing and take a long time record in the homodyne signal sent to DGS. The high voltage drivers in TAMA for whatever reason cannot scan over a range large than corresponding to 120 degrees phase shift of the local oscillator, and then the loop unlocks. Anyway, we sent a triangle wave to CC2 Perturb In to check. We want to take data with a scan of about 0.01 Hz, ~ 16 mVpk (any more than this and CC2 unlocks), for about 10 cycles. So we allocate about half an hour for each level of green power injection. We were a bit confused about how to exactly extract long term data from individual channels considering that we can only pull up quite recent data on dataviewer. In the end we just took the basic option to let Taiwan have the full sized GW frame files (.gwf) and they can extract the data using the usual python extraction tools. The size of the files are quite large so we should find a better option - upgrade DGS, set up a remote connection accessible to Taiwan, or both.

We proceeded to characterize losses using the method of measuring squeezing vs antisqueezing for different green pump power to OPO. After changing the pump power, we use CC1 error signal to characterize roughly the amount of squeezing obtained. Normally when the servo is set to "scan" mode, there is a ramp signal sent to some actuator that allows us to inspect the error signal over an FSR or some other selectable range, however, the CC1 scan mode amplitude should be set to zero for normal operation (I personally don't know the exact reason but I was told the response time of the CC1 servo to ramp signals affects lock acquisition). If we want to see the error signal we should turn the amplitude knob on the CC1 servo - after some value we can see a sinusoidal wave whose amplitude corresponds directly to the amount of squeezing applied (with the scan amplitude adjustment just turning the sine wave "on/off"), - CC1 actuates at the green phase shifter, so a scan of GRPS covers amplification and deamplification of infrared (i.e. CC) inside the OPO. Before we would use some more convoluted method like looking at BAB transmission amplification/deamplification caused by modulating the phase of the green input at the green phase shifter high voltage driver. But now we just check quickly without rearranging anything. We don't have any reference levels for the CC1 error signal characterization and it turned out to be quite erratic though. Still, we used the error signal to optimize temperature and ppol frequency for each pump power. 

At 50 mW injection the CC1 error signal is very unstable. In the past we have been able to reach 70 mW so this should be fixed. There is a lot of noise coming from the CC PLL control loop at this level of green injection, for whatever reason. 

Looking at the data afterwards, it seems that while we can perhaps fool ourselves into thinking that the squeeze level is about 7 dB on the spectrum analyzer, comparing the average levels of shot noise (-132.5 dBVrms/rtHz homodyne power spectrum) vs optimal squeezing (-137.5 dBm) gives only 5 dB squeezing at 35 mW injection. The squeezing versus antisqueezing follows:

Actually the 45 mW point was recorded wrong, and accidentally wrote the same spectrum as 40 mW squeezing. So it was removed to leave 6 points. The resultant curve fitting gives 28.6% optical loss and 24.1 mrad phase noise. Even the 40 mW injection point was noted to have quite a lot of CC2 glitch noise, so we tried removing it. The result is 29.1 % optical loss and 17.5 mrad phase noise. This is quite suspect, since phase noise is mostly determined from the high injection power part of the curve, but we still expect that the real phase noise level is lower than what the current CC PLL can provide.

Since the cryostat only has stripped wires (no connectors at the end), I prepared the connection inside the cryostat for the Nodal system.

The cables were prepared in KEK and brought to NAOJ along with the copper and kapton tubes.

I prepared connection for Electrostatic actuator (Brundy-♂), temperature sensor and heater (Samtec, 4-pin, ♀).

I will make two additional Smactec (4-pin, ♀) connections for future use.

I am not sure if it is better to introduce Samtec connectors to all the cables or not.

From 23-12-26

Yuhang, Michael

Apparently there was some issue with the connection of rampeauto to main laser (filter cavity servo). This was turned off and CC PLL improved, but the issue is still delayed into the future.

The current situation with the CC PLL is that the cable from the slow loop correction of the servo to the CC laser wasn't working, so we improvised and it's dangling over the optical table. This will need to be fixed at some point. The issue with the 2V required/200 mV provided correction signal was due to the sideband being on the wrong side of resonance. We should be such that decreaing the temperature of the laser decreases the frequency of the peak on the spectrum analyzer, and then position the peak about 20-30 MHz away from resonance before turning on the loop. This is a strange issue of the CC PLL system where somehow the correction signal has some offset and only properly corrects for sidebands initially far enough away from resonance.

We wanted to test taking the data for the Taiwan machine learning measurement. We injected to CC2 Ramp IN but it showed little if no effect on homodyne. In the electrical diagram EPS1 and Perturb IN add to give EPS2 so we just decided to add it at Perturb IN. Then we can see what looks like squeezing and antisqueezing on the homodyne. However, the maximum voltage sent to Perturb IN while CC2 is locked is quite low, only up to about 10 mV. Still, it seems to work, so we should verify squeezing and antisqueezing level over the scan range using the DGS system.

From 23-12-26

Yuhang, Hsien-Yi, Michael

We discussed taking data for the Taiwan group machine learning project. We would like to take a long scan of homodyne subtraction DC port while continuously scanning the homodyne angle, in order to have an appropriate time series for FIS training data. The proposition is to scan the phase of the CC2 servo, which controls the phase of the local oscillator via the infrared phase shifter, while CC2 is locked, and then leave it for some time with several periods of squeezing/antisqueezing traversal. Typically we set homodyne angle in DDS through DDS3 DAC3 CC2 Demodulation (though it's actually Channel 1 in the software). So we figured to inject a ramp signal to CC2 Perturb IN, which in principle should allow us to scan the homodyne angle. Unfortunately homodyne is demodulated using a single chanel mixer. We should buy an IQ mixer to allow us to tell if it is amplification or deamplification.

Yuhang, Michael

We investigated the cause of the CC PLL glitches and saw that it was not so stable as it seemed on Friday. I had to leave earlier but Yuhang continued to investigate afterward (next entry).

On the spectrum analyzer we normally look at the DC peak and visualize the CC PLL sidebands at -7 and 7 MHz. Whenever the system unlocks it seems to be precluded by small bumps about 1 MHz away from the sidebands.

Another thing to check was the magnitude of the correction signal. The CC laser has a piezo tuning coefficient of ~ 1 V/MHz. If we move the sidebands about 2 MHz away from lock, we see that the fast loop monitor output only shows 200 mV, which isn't enough to properly correct, though it moves in the correct direction. From the specifications of the ADF4001 board the fast loop should be able to provide 10 V. It seems that there is definitely some problem with the electronics. We see that ppol PLL gives roughly the correct magnitude of correction signal (2 V for 3 MHz detuning from 160 MHz).

We once again thought it was good enough, and decided to move on to characterizing optical loss and phase noise by measuring the sqz/asqz ratio for different green pump powers, since the purpose of switching to the "one EOM" layout was to reduce squeezing phase noise.

However, we then ended up seeing some CC2 phase jitter despite the board saying it was locked. We checked CC1 and CC2 error signals and saw that the glitches were correlated in each of these channels. At this point we were not sure whether or not the glitches were coming from the lasers or the electronics.

We sent the PLL CC monitor channel to a mixer and demodulated at 7 MHz. It should be flat but instead we saw spikes. We tried also for ppol, reducing the modulation to 80 MHz (DDS has a bandwidth of 110 MHz), and then demodulated at 80 MHz. We once again saw glitches in the ppol. Looking at the time domain SHG error signal on lock shows none of these glitches though. If the problem is in the main laser, then it should be in the SHG, but it isn't. The OPO error signal also has no glitches despite being locked with ppol. It should also be noted that we didn't see any of these spikes in the fast or slow output monitor channels of either PLL.

Freezing one of the glitch spikes on the oscilloscope shows it has a width of 100 us, corresponding to a frequency of 10 kHz. This seems like a mechanical frequency, but at this point I had to leave for the day.

Yuhang, Michael

From Friday last week.

We recovered frequency independent squeezing to about 7 dB (vs 5 dB in April). The CC PLL is still glitchy and occasionally bumps the noise of the squeezed spectrum but the interval is enough to take an averaged spectrum.

CC PLL 

Initially we attempted to check if the connection of the PLL actuation to the CC laser was broken. We saw that the slow signal monitor of the CC board has voltage, but the actuation cable from the slow servo did not.

After fixing this issue, we tried some other adjustments of the PLL parameters. Eventually we saw 10 MHz sideband detuning locks well. We went to lunch, came back and it was still locked.

We decided to check the sign behaviour of the PLL control loops. There are two control loops for each PLL, "fast" and "slow". The "slow" loop corrects long term drift. For each loop, we held the PLL sidebands at a location on the spectrum analyzer and plugged the output monitor to the oscilloscope. Then we saw the following:

Fast loop: When the fast loop is closed, the voltage of the correction signal goes up while the sideband frequency separation from the carrier is reduced. This implies a negative sign for the fast loop which is consistent with negative polarity set in the PLL software.
Slow loop: When the slow loop is closed, the correction signal goes down while the sideband frequency separation from the carrier is also reduced. So the slow loop has a positive sign. "Inv" should be off.

We were already using the smallest gain but the control loops still overshoot. We tried removing 12 dB attenuators, and this seemed to make the CC PLL lock good enough.

Frequency Independent Squeezing

We optimized ppol further by making the CC1 error signal large. We started from a ppol frequency of 160 MHz. But actually 160 MHz was already good enough and CC1 locked.

CC2 error signal was seen to go up and down on a timescale of 0.5s which is the correct behaviour.

Using the SR785, we saw -132.34 dBVrms/rtHz shot noise at high frequency (9 kHz) while blocking homodyne. CC2 fast loop seems to still be sending a lot of glitches to bump the squeezing spectrum every now and then, but the interval between glitches is still long enough to obtain 100 average traces.  The presence of glitches is indicated by the reading of the green phase shifter high voltage driver going to zero, which indicates that the problem is somewhere in the CC signal. To speed up the inspection we just looked at very high frequency of 80 kHz. The lowest level of noise we saw was -139.2 dBvrms/rtHz, or 6.96 dB of squeezing.

Yuhang, Michael

From Thursday last week. We made many fixes to just about all parts of the frequency independent squeezer.

SHG

The RFPD that detects the SHG error signal was moved to the reflection of the first dichroic in the green path, aided by a lens and a green block filter. The error signal becomes like the reference on the wiki. We took the open loop transfer function eps1/eps2 on the spectrum analyzer but it was strange compared to the reference version (Aritomi thesis). We tried on the old spectrum analyzer and it seemed fine, with the correct low frequency response and first mechanical mode at 21 kHz. So maybe we just missed some settings. We changed the servo gain to give 2.2 kHz unity gain frequency and 45 degree phase margin. Coherence was close to one apart from a spike at 500 Hz.

OPO optimisation

The OPO was optimized using a fast method: scan green phase (ramp signal sent to green phase shifter high voltage driver) without locking OPO and measure amplification/deamplification of BAB. We saw that for some reason the green beam was greatly misaligned in pitch on both the filter cavity and OPO paths, so we adjusted with the first steering mirror after the green FI. Then green to OPO was maximized using the second 45 degree incidence mirror after GRMC transmission to maximize nonlinear gain. We saw:

74 mV without green
268 mV with 25 mW green, before optimization
496 mV after optimization

giving a nonlinear gain of 6.7. This is really quite high compared with what we saw before. We did not see the need to adjust temperature (though maybe we should).

IRMC

We had issues with mutual lock of SHG and IRMC ever since switching to the one EOM scheme. Currently SHG and IRMC are both modulated with one DDS channel and demodulated with one DDS channel. We took some cables to add phase delay to the IRMC and got the good PDH signal. So now it locks properly, although the threshold needs to be optimized a bit.

Homodyne alignment

The homodyne was realigned and balanced. The coarse alignment steps are as usual:

1) Block all OPO side infrared signals from reaching the homodyne
2) Lock IRMC to send local oscillator to homodyne
3) Adjust pitch and yaw of lens closer to IRMC to bring the homodyne photocurrent close to zero (will go far down if misaligned)
4) Adjust pitch and yaw of lens closer to edge of table to bring homodyne photocurrent close to zero (will go far upif misaligned)
5) Adjust balancing beam splitter to finish bringing the HD signal to zero

Then local oscillator is aligned to alignment mode cleaner for fine alignment.

The spectrum of the homodyne output was seen to be flat to about 40 Hz, with 100 Hz peak from ceiling light. We turned off the ceiling light but then it somehow turned back on. We decided to investigate using the security camera system but could not login.

After LO is aligned, the squeezed path is aligned by sending BAB to the alignment mode cleaner, but first we have to lock OPO with ppol.

GRMC

For whatever reason GRMC misaligned a lot as well. Maybe had something to do with the green misalignment in pitch above. During this process we could see the filter cavity reflection moving a lot on the green injection mirror though (there was an earthquake), so the green beam shouldn't be too far misaligned to FC.

PPol

The ppol frequency was changed to optimize BAB transmission (without green). It goes to 220 MHz, down from 260 MHz some time ago.

CC

On this day we saw that the CC PLL had erratic locking. The sidebands would come close to lock and then eventually quickly be ejected away from the resonance point.

Nishino,

By scaaning frequency, I measured the mode-mtching ratio of the current setup.There were 4 HOMs (both HG and LG). The mode-matching ratio is:

4.90 / (4.90 + 0.337 + 0.097 + 0.097 + 0.097)*100 = 88.7 %

The injection current was 1.1 A.

Also I locked the cavity by Moku:lab. Environmental sound seems to be too big and the locking was not stable only with a feedback onto PZT. It would need something like a shield for a stable locking.

Yuhang, Marc, Michael

We once again tried to find transmission sidebands at 88 MHz for the SHG locking but could not see them, so we had to rethink the lock scheme.

We had some discussion of how to resolve the BAB direct reflection issue (in fact, the beam is forming an interferometer with OPO HR and SHG input as the two arms, and SHG refl as the output port).

Our first thought was to put a Faraday isolator to decouple the two beams.  For the SHG input path, there is not very much space for an FI, while on the BAB path the beam is quite large and would require a redesign of the mode matching telescope for BAB to OPO.

We had some thought about trying to decouple them using polarization. But given that SHG must be p-pol and BAB must be s-pol, we cannot decouple them using just HWPs and one PBS, and besides, we need those polarization optics for the Taiwan and Korea activities in ATC.

We then decided to try a pick-off from the SHG reflection before it reaches the SHG/BAB separation beam splitter and recombines with BAB. Since the current SHG refl PD has an ND2 attached and has about ~ 100 mW incident, we only really need about 1 mW of power to get a good signal. At this point we remembered that there are two dichroic mirrors on the green output of the SHG that reflect away residual IR into beam dumps. Actually, they reflect quite a lot, we measured 3.6 mW infrared reflected from the first dichroic after green generation.

And then we found that our attempts to see the transmission sidebands on the transmission PD were separately mistaken. On the first attempt, a low bandwidth PD was used in reflection, which is not fast enough to see RF sidebands. On the second attempt, an ND2 filter was left screwed on to the SHG refl RFPD when it was moved to the transmission side. So in both cases the sidebands were suppressed. Correctly implementing an RFPD in transmission of the SHG with sufficient incident power resulted in recovery of the error signal. So we didn't really have to change the locking scheme very much if at all.

We decided to put both SHG lock PDs in transmission. We selected a BSF10C beam sampler with an output wedge - this has ~ 1% reflection for p-pol at 45 degrees AOI, and maybe 5% at imperfect angle. To improve the visibility of the small transmitted power, the SHG correction signal SERVO OUT was disconnected, and an offset applied to the SHG PZT to see strong green generation. Then the IR transmission also becomes stronger, to about the level we would expect during SHG lock - about 3 mW transmission. In 99% transmission of the beam sampler I placed the RFPD Thorlabs PDA05CF2 which goes to the SHG error signal. In 1% reflection I placed the DCPD Thorlabs PDA36A-EC switchable gain, which goes to the servo threshold check and TRANSMIS OUT. I looked at TRANSMIS OUT on the oscilloscope -it was quite small so I increased the DCPD gain to 30 dB and it became about 550 mV. Then I checked the RFPD signal going to the mixer, which has 4.5 V max signal. I could see 95.7 % mode matching (4480 TEM00 + (96+56+48) HOMs). Then I placed a large beam dump at the former SHG refl PD point, which has quite large infrared power.

I set the threshold to slightly less than half of the DCPD signal (about 250 mV) and tried to lock, but the SHG would still only lock to a bad point (i.e. not to the level of the TEM00 peaks on TRANSMIS OUT), I looked at the error signal and tried to optimize the SHG demodulation phase but could not. From the reference signal on the wiki, it should have jump from a large negative point to a high positive point with a strong positive slope at the lock point (once again the NAOJ elog image upload issue is not very nice). However, instead it goes from large negative peak -> small positive peak -> small negative peak -> large positive peak, essentially making a small bump in the middle of the error signal with the opposite sign compared to what the polarity of the large peaks should give. I cannot get rid of this by changing the demodulation phase. Tomorrow we will try using the dichroic reflection for RFPD to see if that fixes the error signal.