Abstract:

We may be able to significantly improve the DARM sensitivity by optimizing the pitch control of PR3.

Details:

As noted by on-site commissioners, the PR3 pitch signal shows strong coherence with DARM around 20 Hz.
Addressing this issue should be a high-priority next step, so I investigated the current status of the PR3 feedback as an initial approach.

Figure 1 compares the spectra of PR3_{IM, TM}_SUMOUT_{L, P, Y}_OUT, {PRM, PR2}_TM_SUMOUT_P_OUT, and the coherence between DARM and {PRM, PR2, PR3}_TM_SUMOUT_P_OUT.
Among these, PR3 pitch is the noisiest degree of freedom and exhibits very high coherence with DARM, particularly at the resonant peaks (17 Hz, 23 Hz, and 59 Hz), as well as more broadly across the 10–40 Hz band.

In Figure 2, the current open-loop transfer function of the PR3 pitch control is shown.
One way to reduce the sensing noise feedback into PR3 pitch is to modify this open-loop transfer function—either by lowering the unity gain frequency (UGF), adding notch filters at the resonant peaks, or implementing sharper roll-off filters.
Another solution, suggested by Ushiba-san, involves changing the current feedback hierarchy—switching from feedback to the PR3 oplev error point to a more absolute control scheme using the IFO WFS signal.

It is also worth noting that the peaks at 17 Hz and 59 Hz appear in both PR2 and PR3 pitch signals, and the 59 Hz peak is common to all pitch signals from the power recycling mirrors.
This suggests that the type-Bp suspension may have mechanical resonances at these frequencies, although they are not currently listed in the documentation identifying type-Bp suspension resonances.