Nutsinee Daniel
We identified some of the laser power drift that we observed after the (red) EOM to be due to a small polarization error and the birefringence of the EOM. Due to a manufacturing error in the Faraday isolator preceding the EOM we need to rotate the polarization prior to the EOM. We are using a lambda-half plate but it is not easy to adjust, since we don't have a convenient polarizer in the path. This introduces a small polarization error incident to the EOM. The birefringence of the EOM will then introduced a drift in the polarization due to temperature variations, it will introduce polarization noise due to the TTFSS feedback path to the EOM, and it will introduce RFAM for the SHG modulation frequency. Variations in the polarization angle are then converted into intensity variations by downstream polarizers—compounded by the waveplates which are used for power adjustments.
A PBS cube has been installed after the 35MHz EOM in order to 1) minimized the wrong polarization coming out of the EOM and 2) Clean up the polarization before it hits other half-waveplate and cubes downstream. See attached photo for the setup.
Both CLF launch and SHG GR PD spectrum got better at >1kHz as they should be. All the funny features are gone. Sadly there were some other noise today that can't be explained by the EOM (the EOM/PZT crossover is at 38kHz). Could it be temperature drifts due to table open/close? There were also some locking issue today after Beckhoff restart (wrong error signal fed to laser PZT while OPO is locked, fixed now). I'll take another set of data tomorrow morning (this should also let temperature related things stabilize a bit). Note that IR DCPD looks at Mephisto power (the diode was right after the newly installed cube). That one hasn't changed from before (see alog45774). Also attached TTFSS fast path transfer function -- looks normal again.