Abstract:

We should check the notch filters in the type-A suspension local controls because several resonant peaks around 40 Hz have drifted over the past month, likely due to temperature changes at EX and EY.
Additionally, further analysis of these peaks suggests that neither thermal noise nor coil driver noise in the IMs can fully explain the measured peak amplitudes.
Therefore, reducing actuator efficiency could be a worthwhile approach to suppressing these peaks.

Details:

Recent oscillations of the DARM loop around 40 Hz may originate from a mismatch between the actual pitch or vertical resonant frequencies and the settings of notch filters in the local control system.
To investigate this, I compared the DARM spectrum at moments when the DC readout remained stable for a relatively long period last month and this month.

Figure 1 shows a comparison of spectra recorded on Feb. 21st at 15:00 UTC and Mar. 18th at 16:00 UTC, both with a time span of 512 s and an averaging time of 15.
Several resonant frequencies have indeed shifted, likely due to temperature variations at EX and EY over the past month.
The IM temperatures changed from Feb. 21st to Mar. 18th as follows:

EX: 87 K -> 43 K
EY: 87 K -> 79 K
IX: 88 K -> 93 K
IY: 88 K -> 93 K

Given these changes, we should check the notch filter frequencies in the type-A local control loops, particularly for EX and EY.

Next, as suggested in klog:33131, the observed 40-Hz peak amplitudes are unlikely to be fully explained by thermal noise or coil driver noise.
However, we should reduce these peak heights to minimize nonlinear contamination in the most critical frequency range, at 60-90 Hz.

To quantitatively evaluate the peak amplitudes, I selected one pitch and one vertical resonant peak and performed fitting to the spectrum from Feb. 21st, assuming realistic thermal noise and coil driver noise (Fig.2 and Fig.3).

The assumed parameters are:

• Q-value: 7 × 10³

• Force noise from the coil driver: 6 × 10⁻¹² N/√Hz

  • Computed as 0.055 [N/A] / 180 [Ω] × 1 × 10⁻⁸ [V/√Hz] × sqrt(4)

The fitting parameters are:

• Vertical-horizontal coupling (VHC) for the vertical resonance

• Miscentering for the pitch resonance

If the peak amplitudes were purely due to vertical and pitch thermal noise, then:

• VHC would need to be 0.22, and

• Miscentering would need to be 3.5 cm.

If the peak amplitudes were due to coil driver noise, then:

• VHC would need to be 2.2, and

• Miscentering would need to be 15 cm.

Both scenarios are highly unlikely, indicating that these peaks are likely excited by additional noise sources.
To mitigate these peaks, we should consider reducing the actuator efficiency of the IM, or possibly MN.
This could potentially suppress the peak amplitudes and significantly improve the BNS range.