Vicky, Daniel, Dhruva
Today we tried to make a number of changes to the CLF. Seeing as LLO could operate their squeezer at much higher CLF power levels, we decided to turn our CLF up and try and investigate what might be preventing us from getting squeezing at high CLF levels.
Changes made :
1. Replaced 90/10 BS in front of CLF fiber with a mirror
2. Removed amplifier from the CLF PD's RF path
We could seem to get any squeezing with CLF AOM drives ranging between 20dBm and 30dBm, no matter what we tried to do with the CLF, LO and ISS loops. We did however, manage to make some measurements of seeding at different CLF AOM drive offsets. We also offset the SQZ laser VCO by 500Hz, and saw some seeding peaks at 2kHz.,
My theory for how seeding prevents us from getting more squeezing : The reinjected (flat) sensing noise from the LO loop creates flat phase sidebands in the LO loop bandwidth which beat with the seed to produce this 'flat' noise that looks like squeezing loss. This would also explain why we ended up getting more squeezing after reducing the LO loop gain (bandwidth) at 2W.
For more context to this line of investigations, from Dhruva's alog exploring PSAMS for SQZ 66877, his trace "LO Loop Adjusted" meant that he turned up the LO CMB's common gain, and turned off both LO boosts. This is the difference here between the lime green and purple traces, both taken at the same PSAMS voltages.. but turning off LO boosts ncreased high-freq squeezing by over 0.5 dB -- very strange. This weirdness launched us into phase noise investigations the following day 66884, to figure out why we ended up having no boosts in the LO or CLF servo's, which does not make sense from the perspective of in-loop phase noise measured in 66884, where it is clear that we want both LO+CLF boosts to suppress phase noise. In fact, when locking FC-IR on the IFO, it is even clear in the FC_WFS_I phase signal that engaging both LO boosts reduces laser phase noise, as compared to the quieter filter cavity. So why do we end up turning off boosts and seeing improved squeezing?
Then today Dhruva is noticing that LLO can get good squeezing with very high CLF levels (eg LLO:63184), which are about 10x what we are operating at here. Last December 2022 LHO:66473, we even saw a broadband squeezing reduction with marginally higher CLF levels. Then we also looked for seeding at these low CLF powers, 66533 but nothing clear. So how come LLO can operate with >10x higher CLF power, especially when it seems both sites had CLF issues in O3?
So, that is why today we super cranked up the CLF power to study things (swapping back to a mirror on the CLF path, and removing the ZHL-* mini-circuits amp from the CLF RF6 PD). CLF optical power increased like ~20dB, mini-circuits amp removal decreases like -20dB from the RF PD readout, so CLF loop gain kinda stays in the same ballpark, while LO loop gain was decreased significantly to lock. More to figure out here, but with the huge increase of CLF power, we can finally see the seeding peaks like the O3 alogs e.g. LHO:52762.
Here are some measurements of seeding. I tried to change the AOM drive while adjsuting the waveplate to keep the optical power (and RF6) constant at around -17dBm.
The two measurements are on the homodyne with
2. LO loop locked at 1kHz offset) which produces a seeding peak at the same frequency. This is done by changing the LO loop demod frequency to 3.124MHz
Changing the RF drive seems to have a pretty clear effect on the the spectrum, although the trend is a bit hard to decipher. A low RF drive (22dBm) seems to produce a smaller peak but more broadband noise, which seems unrelated to any of the loops (more noise on the AOMs??).
