There has been a long running issue with IRMC lock ever since the main laser EOM was switched from 15 MHz to 88.3 MHz. Before, the error signal was acquired with a TAMA 15 MHz resonant PD but we since switched to a Thorlabs PDA05CF2 broadband RFPD with 75 MHz linewidth. From old logs, the error signal was about 1Vpp before the change and decreased to 100 mV after changing to new modulation scheme. However, the SHG underwent the same change and only saw a factor of 4 decrease in error signal. Initially I though there was some issue with weak IRMC modulation or demodulation signal due to weakening local oscillator synthesis from DDS, but the SHG uses the same mod/demod electronics and works mostly fine so that isn't the main problem.

We currently inject about 2.5 - 3 mW IR to IRMC with the goal of about 2 mW transmission on resonance. So off resonance about 3 mW makes it to IRMC REFL, which is a bit much for the PD. For whatever reason, at some point there was a weak ND filter Thorlabs ND03A, which resulted in a base level of > 5 V on the reflection spectrum (fig 1). With the PD close to saturation, the side spikes of the PDH signal become very small (fig 2). I changed the ND filter to Thorlabs ND1 which is what it was a while ago, which gives a base level of about 2.0 V and dip of 1.3 V (fig 3). This makes the error signal have the usual PDH shape but it also became quite small, ~ 70 mVpp.

I checked the PDH sidband amplitude by manually putting the IRMC near resonance (servo scan amplitude zero and PZT HVD offset adjusted to make the reflection signal drop by 1.3 V). I could see the 88 MHz sideband from IRMC REFL RF was about -27 dBm. I sent it to a spare RF amplifier port on the rack +18 dB and could see that the sideband was amplified to maximum -8 dBm. The error signal becomes 400 mVpp (fig 4). I adjusted the cable length a bit to improve the demodulation phase and now the IRMC can automatically lock quite stably. IRMC transmission is about 2.16 mW.

For now this is reasonable to operate. There is some issue with the servo electronics that is reducing the signal. One fix we discussed a while ago with Yuhang and Pierre Prat is to replace resistor R33 in the servo schematic (feedback resistor of an inverting op amp) with a ten times higher resistance to increase the op amp gain 10x. Also I think we could use a *slightly* weaker ND filter on IRMC REFL RF. Could also make an 88 MHz resonant PD but maybe that's too much effort. For now it's fine, we still have spare RF amplifier ports so we don't need to mess with it until we need those for other things.

Next:

• measure IRMC tansfer function and optimise UGF
• recover BAB alignment to OPO and homodyne
• optimise green alignment to OPO via BAB nonlinear gain