(Kiwamu, Alexa, Matt, Stefan, from yesterday)

As part of the ALS noise hunting, we revisited the PLL and PDH loops at EY.

- First we realigned the PDH diode that was intentionally misaligned the night before. We verified the the DC oscillations observed on that diode's read-back are not present in analog, so there must be some issue with the acquisition of that signal.
- Next, we looked at the PLL loop. With a UGF around maybe 20kHz it produced a lot of gain peaking around 30kHz. We turned the BOOST stage 1 on the PLL Sigg board off. That helped, but the loop was happiest at another 10dB lower gain.
- This suggested the idea of an "additive offset" like path, bypassing the PLL loop all together. We rigged this control scheme by using one additional BNC cable:
  - We moved the cable that runs to the VCO tune input from the PDH Sigg board fast output to the PDH Sigg board EXC B monitoring point (either 1 or 2 - they have opposite sign). We would like to use the slow path for this option, but it had additional undesired low-pass filtering.
  - We added a cable from the PDH Sigg board fast output to the PLL Sigg board EXC A input. This keeps a high-frequency roll-off as well as some PZT notches in the path.
  - The PDH Sigg board fast gain was lowered from 0dB to -22dB to set the AO gain. The PDH Sigg board input 1 gain was set to -4dB.
  - With this setting we got a nice, roughly 1/f total loop shape, and a UGF of 20kHz.


Our hope was that this configuration will change the noise seen at the end corner, but we had no luck - the corner noise was unchanged.

PS: Attached is a plot of the calibrated frequency noise in Hz/rtHz seen by 3 sensors:
 - Blue: PLL control signal: dominated by free-running laser noise
 - Purple: PDH control signal: above 2kHz this shows coherence with blue (not in figure), indication laser noise. But below 2kHz we seem to be limited by some other PLL sensing noise (electronics dark noise ?) Note that this noise was taken in the old control loop topology, because the control signal calibration was simpler.
 - Red: ALS in-loop common mode error signal from the corner beat node. Note that it is NOT coherent with any of the noise we see at the end.


PPS: We also noted that we are able to grab an arm lock with the PLL input disabled (i.e. without relying on the fiber for per-stabilization). However we would need some more low frequency gain to keep the lock in this configuration.