Update on demod. for OPO cavity lock. Last related elog entry described prevalence of <= -77 dBm of odd line noise harmonics (60, 180...) Hz, along with poor SNR PDH error signal. First attachment is a drawing of the current RF connections. Upon completing list of suggested actions from this post, the difference was mostly made by looking at RFPD RF out power before mixer < -40 dBm. This was no good, so after realizing that the OD = 3 nd filter before RFPD was only allowing 80 uW of a nominally reflected ~25 mW, swapped the ND filter with HWP + PBS for adjustable power splitting. Then, a healthier -10dBm made it into the mixer and SNR improved considerably (see second attachment). Upon closer examination of err signal, low freq. sinusoidal modulation sat on top of it suggesting slightly off-resonant demodulation so finely adjusted the (Marconi) LO frequency from 36.000 MHz --> 35.999828 MHz until the error signal had a good enough shape (see third attachment below).
First attempt at cavity lock was done with ~46% mode matching efficiency and max. modulation depth (estimated ~0.21) on the EOM. The loop is achieved using UPDH box (v3) which I stole from CTN lab. Upon connecting all the inputs, scanning the phase shifter without making much of a difference, and enabling the lock, saw a stabler higher order mode on the cavity transmission which is nice. The natural follow up of scanning the PZT driver (i.e. as an offset) and re-engaging the lock resulted in what I can only describe as a "visit to the dentist", where the cavity PZT (on the output coupler) was resonating quite loudly (!!). After looking at the output monitor of UPDH box with engaged lock on SR785 an ~ 8 kHz peak explains the noise as an audible mechanical resonance. Adjusting the servo gain finely tunes it out a bit, and adding an SR560 in line before the PZT driver unit greatly helps, but changes the overall loop gain and the lock becomes unstable... Current efforts are therefore geared towards improving the pdh loop, for which an option is to bypass the thorlabs MDT694 HV piezo driver and directly connect the UPDH output to PZT (which it may be meant to directly drive) and use slow temp. control on pump laser to approach the lock point. Another option, involving way more time, would be to *not* use UPDH box at all and implement a digital feedback loop + filter with the Red Pitaya. Perhaps the pragmatic action is to get the analog solution working and develop digital solution on the side.