On Friday and today we ran the noise budget templates for laser frequency noise, intensity noise, and the input beam jitter. Last year I split the intensity noise coupling measurements into 3 frequency bands, I put these incorrectly in the noise budget and we were summing some intensity noise were the bands overlapped. Craig added some frequency-bound functionality to the noise budget and fixed this.
Initial measurements showed some nonlinearity, with the a >1kHz intensity noise injection showing up in DARM in the bucket. This turned out to be coupling through the OMC length dither line, which we'd seen in the past with frequency noise injections. I added a 4kHz - 4.2kHz band stop and the high frequency injection only shows up at high frequency in DARM now.
The frequency noise and input beam jitter look pretty close to DARM at 20 Hz, and broadly between 10 and 100 Hz. I'm not sure if the IMC WFS not being centered will affect the input beam jitter projection (which shakes the IMC PZT in pit and yaw and uses the WFS as a witness sensor). The noise looks pretty similar to measurements we made last year (alog 58925 and comments).
When I tried to re-run the ISS injections this afternoon it turned out the second loop DC servo had railed and was just integrating away. The diffracted power was sitting at around 4%. I followed Craig's instructions (alog 63431) and was able to open the loop, tune the PSL-ISS_REFSIGNAL and PSL-ISS_THIRDLOOP_OUTPUT_OFFSET, and re-engage the loop with no problem. We originally thought the loop was running away again and opened it and cleared the history of the REFERENCE_SERVO by hand. Not sure this this helped - it turns out the output of the slow offset servo is often on the order of 20k counts. I tried to accept the offsets in SDF but I don't think they're monitored.
To make our frequency noise projections we put CARM on LSC-REFL_A and excite the OG common mode servo. We use LSC-REFL_B as a witness sensor. Craig pointed out that using the REFL_B digital readbacks is a bad idea since the whitening is funky and the gain is weird. This evening we re-ran the frequency noise injection with CARM on REFL_B (it's ok in analog), and using REFL_A as the witness sensor. Attaching the laser noise sub-budget. This week we plan to center the IMC WFS which may help with our jitter coupling.
I'm also attaching the full noise budget since it's been a while.
Craig changed the REFL_B gains (LSC-REFL_B_RF9_I_GAIN and LSC-REFL_B_RF9_Q_GAIN) to match REFL_A, and accepted it in the SFD safe.snap. We ran an extra frequency noise injection and confirmed REFL_B and REFL_A now see the same injection amplitude.
I re-ran the IMC PZT injections after recentering the IMC WFS, 64442 Craig was able to run the noise budget code, although I am not able to run it in gitcommon at the moment.
The jitter peaks are still predicted to be about a factor of 2 below DARM between 100-200 Hz.
