[Hang, Gabriele, Sheila, Jenne, Georgia, PatrickG, Marc, Alex-the-job-shadow-student]
Summary: As a result of being concerned that the CARM transition to TR signals doesn't work when the green WFS have high bandwidth (we failed twice this morning, in addition to the night crew's several), we looked a little more closely at the green WFS. For most of the morning we thought that the Yarm WFS A had a problem, but in the end it looks like there was a filter setting that should be the same for all 4 WFS (X A&B, Y A&B), but this lowpass filter was on for 3 of them, and not on for one. So, when I looked at the WFS outputs after this filter bank, we were seeing differences. As part of our diagnostics, we removed the HWPs that had been temporarily put in on the Yarm when trying to figure out the DIFF glitches (see alog 42978).
The filters that are different are 0.3Hz lowpasses that are limiting the bandwidth of the WFS loops. They should be off for all the WFS, but the were only off for Yarm WFS A. This made the PIT and YAW outputs of WFS A look more noisy in comparison to the others. These lowpass filters were on throughout O1 and O2, but Hang's new WFS designs should have them off.
Tomorrow morning, in parallel with TCS work, we'll likely work more on increasing the bandwidth of the green WFS loops. Hang has been doing a bunch of good design work today, and we should know the settings that we need to try out tomorrow. This will hopefully make the first part of initial alignment much faster than the ~45 minutes it sometimes takes for the loops to converge right now.
For tonight, we're sticking with the low bandwidth version that we've been using so that we can work on locking.
As a follow-up investigation, Gabriele is looking at the digital loops that we use to center the input green beam.
Measured the open loop transfer functions of the PZT centering loops for the Y arm ALS.
PZT1 pitch and yaw have a bandwidth of about 2 Hz and the transfer function does not show any high frequency feature.
PZT2 yaw has a bandwidth of about 2 Hz, but PZT2 pitch has a much lower bandwidth (about 0.5 Hz). But the important thing is that both PZT2 pitch and yaw show a resonant peak at about 375 Hz, which is in the same place of the large peak visible in the QPD signal (see plot attached to main entry). In both loop the gain margin at the peak is very large (at least 80 dB), so it's not that our centering loops are unstable there.
However, this might be an indication that the PZT closed loop controller is oscillating at that frequency. Looking into the PZT and driver manuals (LIGO-T1000646) this frequency seems compatible with the yaw resonant frequency of the loaded PZT. Also, Keita pointed out a long time ago that using the controller in the "high bandwidth" configuration caused oscillations (8535). Tomorrow morning we should try to open the controller and switch the jumper to use the "half bandwidth" configuration.
