Abstract:

I estimated P2L coupling due to MN HPCD noise (1e-8 V/rtHz at coil driver output).
If we assumed 6% of P2L coupling, slope around 44Hz (ITMY pitch resonance) can be explained.

Detail:

To estimate the MN HPCD noise, I used TF from MNP to TMP, which was measured for peak identification.
Figure 1 shows the measured TF from MNP excitation to TMP.
Since excitation was had a band-pass filter from 10Hz to 60Hz, TF seems to be well measured only from 3Hz to 45Hz but it is enough for checking the coupling around 44Hz.

High power coil driver noise is measured as 1e-8 V/rtHz, which corresponding to the noise of 3.23e-5 cnt/rtHz at COILOFTF input (1e-8/310e-5).
Since EULER2ACT matrix elements from pitch to V1 and V3 coils are -2.9585 and 2.9585, respectively.
So, HPCD noise corresponds to 1.1e-5 cnt/rtHz at SUMOUT_P_OUT.
Since the HPCD noise are incoherent between V1 and V3 while pitch excitation is coherent for them, HPCD noise equivalent MNP excitation is 7.7e-6 cnt/rtHz.
Figure 2 shows the estimated pitch motion of TM stage due to MN HPCD noise.

By using this estimated TMP motion and asuuming P2L coupling rate, we can project the noise into DARM.
Figure 3 shows the projection by assuming 6% of P2L coupling.
Blue and green lines show the HPCD noise projection before and after inserting resistances reported in klog33426.
As you can see, 6% P2L coupling can explain the DARM noise around ITMY pitch resonances and it should be reduced by the inserted resistance effect.

Anyway, let's see the sensitivity curve when PRFPMI will be locked next time.

Note:

Peak height and Q-value in TF data are slightly different from the current peak in DARM, so the projection cannot explain the peak height on the resonance.
No reasonable explanation of 6% P2L coupling but I think it is not so unrealistic.