There is still another lesser glitch source coming from somewhere. It has some relation to the ppol to OPO path, since it is present when all green is blocked and LO shot noise is measured with OPO locked. LO shot noise with OPO unlocked is fine.
BAB nonlinear gain
I optimized the OPO temperature, PLL settings and green alignment to maximize the amplification and deamplification of BAB. Method is to lock OPO using ppol, while transmitting BAB to the stick power meter. Green is unlocked. The phase of green is constantly being modulated sinusoidally so the transmitted BAB level is amplified/deamplified according to the green phase. The optimum OPO temperature is 7.116 kOhm (down from 7.118 kOhm before) and the optimal ppol frequency is 195 MHz (same as before). The green alignment was not too bad, only about 10% off of maximum. This gives a maximum transmitted power measured on the power meter as 1.484 mW, about a factor of 4 increase compared to the transmitted power without green.
LO/sqz overlap
The LO/sqz overlap is optimized by sending both, one at a time, to the alignment mode cleaner (a reference target). LO to AMC mode matching was improved to 99.89% and BAB (sqz) to AMC mode matching was improved to 98.39%. I checked homodyne balance afterwards and it is still zeroed so ok.
CC loop lock
I fixed the stability of the CC PLL loop a while ago, so now it's time to check if the squeezing controls are stable. I locked everything and checked the CC error signals. when turning on the servo scan I can see a modulation that indicates that there is nonlinear gain. Then, I locked. It works! And I could see squeezing. The CC PLL loop remains stable and there is no crazy destabilization of squeezing from before! (characterized by a huge rise in the squeezing noise floor and wacky green phase shifter PZT oscillation to the maximum of 140V).
Squeezing spectrum
There is still some minor glitch noise to be fixed before we can get fast data for Hsien Yi and the Taiwan people. The squeezing spectrum is experiencing a roughly 2 Hz pulse that kind of interrupts the measurement and raises the noise floor by about 1 dB. However, it doesn't raise the noise floor by 10-20 dB so it's nowhere near as bad as before. I can still see the "squeezing level" (about -6 dB). During BAB and ppol alignment I noticed that the infrared transmission through the OPO is experiencing these glitches. Perhaps the ppol PLL could also be fixed a bit. But I'm not entirely sure if it's coming from ppol PLL. It could also be the ppol laser since I noticed weird flashing while aligning it.
I had thought to just go home and brainstorm it later, but I wondered if it was a ppol laser mode hop. I checked and for some reason the ppol laser current and temperature were set to the CC laser values. However, replacing the correct settings didn't fix the new glitch issue.
Misc
GRMZ control loop is quite temperamental. The optimal demodulation phase seems to go between whatever value it wants every now and then. I optimize and then suddenly the PDH signal goes to a bad shape and the optimal demod phase has gone to some other value. The loop sometimes has trouble stabilizing the GR to OPO power level at the proper amount (for reference, GRMC transmission monitor reads 312 mV for 25 mW green injection to OPO) but it seems to get better by switching INV to NON INV or vice versa. It tends to unlock after about 5-10 minutes. Anyway it seems to be a problem primarily with the GRMC error signal but maybe also the MZ loopology.
I left lasers open and both PLLs locked to see if they stay locked over the long weekend.