I recovered squeezing and saw that the noise floor was completely stable with no glitches. The noise floor level is not so good, only -5 dB, and there is some excess noise in the spectrum from probably mode cleaner motion. Also the green path is still behaving strangely. Still, I saw ~ 5dB squeezing for about 10 minutes continuously with only stationary noise, which is a small relief to close out the year at least. The two main problems now are green lock stability and IRMC power level noise.

Piece by piece details

Second harmonic generator

SHG electronics have some weird behaviour. For a while the appropriate lock threshold has been about 10x what the transmission spectrum on the oscilloscope would suggest. Actually right now it's at the maximum of the range. I guess there is a gain in there somewhere that needs to be turned down. Also the SHG servo gain has some very annoying behaviour - it seems the ideal gain is about 1.5 to 2, however, sometimes on lock it will have a weird 250 Hz oscillation. This is a separate issue from the usual high gain oscillation of a generic control loop, which happens at gain ~ 2.5-3. The 250 Hz oscillation can sometimes be made to disappear by turning the gain a little bit up or down, but sometimes it also just loses lock. This also sometimes causes unlock of the SHG.

Green mode cleaner and Mach Zender

As mentioned a few times before, the green mode cleaner will arbitarily switch demodulation phase from I to Q and unlock the green path. This is now the main issue preventing long term measurement of squeezing. Specifically, I optimized the demod phase to I-phase and then proceeded with squeezing. The GRMC outputs TEM00 at 25 mW as usual, and from the CC error signal it produces the nominal level of nonlinear gain. But then after some time it unlocks, and looking at the error signal immediately after shows that it has gone completely to Q-phase. After some time it drifts a bit and 10 minutes later it's at I guess halfway between I and Q phase. I don't really know whether it's optical or electronic. I tried turning the green phase shifter PZT offset all the way up to 150 V (usually sits at 75 V), which is about enough to switch it from "halfway between" to "almost I-phase" . But in retrospect I don't think it has anything to do with the CC loop actuation of the GRPS since this happens even without the CC engaged. I think I really need to clean up my understanding of the green path and electronics.

Phase locked loops

No major issues. Seems completely fixed.

Infrared mode cleaner

As I noticed in my last post, the IRMC error signal had a large DC offset. I don't know where it came from, but it seems to be a recent issue. I used the "offset" knob on the servo to center the error signal at the appropriate level. The IRMC remained locked for > 1 hour using ND1 on the reflection PD with 424 mV error signal. However, the error signal peak size seems to be fluctuating quite a lot, and the IRMC reflected power also is not very stable. Qualitatively I didn't the see the same behaviour as much when looking at the transmitted power with the power meter, though I just lazily held the power meter and didn't mount it (the ring sensor doesn't have a clamping base and there weren't any left in FC cleanroom). I suspect that the sideband power is fluctuating a lot for some reason, a bit more compared to the carrier. However, the SHG error signal, which uses the same sidebands, is quite stable.

Local oscillator alignment

The LO was a bit misaligned to the homodyne so I rebalanced it. It was also slightly misaligned to the alignment mode cleaner but not so much. I recovered the mode matching to quite a good level. Usually this alignment ends up being the best of the squeezer table (it was 99.99% last time I measured it). The LO seems to be unbalancing quite a lot lately. It might be necessary to carefully re-check the alignment to homodyne.

Squeezing to homodyne alignment

I locked OPO for BAB (ppol 250 MHz) and saw that it was very misaligned at the AMC - TEM00, yaw and pitch HOMs were all about the same level. I realigned to ~95% mode match. That said, the BAB is combining with LO by reflecting at the balancing BS which is used to adjust 50/50 balance of LO to the homodyne, so the misalignment is was probably due to adjusting that BS too much.

Optical parametric oscillator and coherent control

I didn't optimize the green alignment to OPO, temperature or ppol frequency, but by scanning the CC1 (green phase shifter) phase I could see the level of amplification/deamplification in the OPO on the CC1 error signal, and it was the same size as last time I checked so I just moved on. ppol and BAB alignment are fine and lock with no problems. CC1 locked with no issues. CC2 missed a bit at first but that was how I came across the aformentioned sqz/LO misalignment. After sqz to AMC was fixed CC2 locked with no issues. With no CC1 scan, the coherent control error signals are nice and flat compared to the mess they were this time last year.

Squeezing

I observed squeezing to about -137 dB (-5 dB) at high frequency > 30 kHz. The noise floor seemed also ok at about 100 Hz but had a lot of 50 Hz harmonics (spectrum of LO only didn't have much). There is a large bump that looks like a mode cleaner resonance in the range of 5-20 kHz. This might be related to the IRMC power fluctuation mentioned above.

The spectrum noise was completely stationary for about 10 minutes, which is great. Eventually the squeezed state was lost by green unlock. I relocked green again so see how long the squeezing could be maintained, but green lost lock maybe 30 seconds after squeezing was reacquired, and the GRMC error signal had gone back to Q-phase. Therefore, understanding the renewed green lock regime and figuring out the IRMC carrier/sideband noises are the most pressing issues to recover the appropriate squeezing level for the Taiwan tomography project.