During this afternoon's arm locking to tune the ALS / TRX handover, we measured the out of loop performance using the AS beam PDH detector.

The CARM feedback loop used for this arm locking has a UGF of ~100-200 Hz, so the noise measured in ASAIR_45_I is the sensing noise of the other detector impressed upon the arm length (assuming the sensing noise of the AS 45 detector is negligible at this level).

In the attached plot, you can see that the out of loop noise shows a bunch of acoustic noise when the ALS (BLACK) is used as the sensor. In the RED trace, the low frequency noise is ~10x lower and the high frequency noise is higher, but smooth. Around 20 Hz-rms of the noise comes from the 60 Hz power line in the Thorlabs detector used for TRX and ~15 Hz-rms comes from the sub-Hz sensing noise which is most likely due to angular motion of the suspensions producing RIN in the cavity transmission. Presumably, the 60 Hz peak will go down as we start recycling and increase the power (and switch to the in-vac QPDs), but the low frequency stuff will not get better. To get the TR error signals to be better than 15 Hz-rms, we should figure out which suspension is moving at 0.67 Hz by so much.

Calibration: to turn the ASAIR45 signal into Hz, we looked at the peak-peak signal when scanning through the resonance using ALS. That was 400 counts. With a cavity pole of 42 Hz, this means that the peak-peak discriminant corresponds to 84 Hz. Since we are making this measurement not at the zero crossing, but ~20 Hz detuned from resonance, I assume that the calibration is close to (2*84 Hz)/(400 counts), with a low Q zero pair at ~60 Hz.