With takaaki.yokozawa and takafumi.ushiba
Follow-up: ITMY MN 33 Hz oscillation after warm-up

• In DAMPED, a ∼30–33 Hz oscillation was observed, causing MN actuator saturation as reported.
• Noise investigation showed:
  • In ISOLATE, the 33 Hz feature disappeared in all sensors.
  • Enabling controls step-by-step revealed that the oscillation is introduced by MN DAMP T:
    • T ON → 33 Hz appears
    • T OFF → disappears
  • See IY_OSEM_noise_003.png, where
    • Blue: original DAMPED state
    • Green: ISOLATE state
    • Brown: MN DAMP ON except T (L, R, P, Y ON)
    • Pink / Red: T only ON
• OLG check:
  • The MN DAMP T open-loop gain shows a clear gain peaking around 30 Hz, consistent with the observed oscillation.
  • See IY_MN_DAMP_T_OLG.png.

Mitigation

• Since the instability originates from the MN DAMP T loop, the control filter was modified.
• The elliptic low-pass filter ELP30 in the MN DAMP T path was changed as follows:
  • Before: ellip(LP, 4,1,100,30) gain(1.12202)
  • After: ellip(LP, 2,1,100,13) gain(1.12202)
• The filter parameters were adjusted so that the phase at the UGF remains unchanged, while suppressing the gain peaking around 30 Hz.
• The before/after transfer function of the filter is shown in before_after_ITMY_MN_DAMP_L_2.png.
• As a result, the DAMPED state became stable.

• After stabilization:
  • TM pitch was offloaded using the moving mass.
  • Y offloading was performed by adjusting the BF Y offset in ALIGNED, reducing the MN Y actuator load.
• Current status: ITMY now proceeds to LOCK_ACQUISITION.

After reaching LOCK_ACQUISITION, ITMY starts oscillation at 2.2Hz today (fig1).
It takes about 45 minutes to see the oscillation in time series (fig2).
According to the oscillation frequency, NBDAMP_{L1,T1,Y4} would need tuning.

I checked NBDAMP filter R1 and V1 for ALIGNED state.
It seems the FM1(300K) looks better than FM3(117K) for R1 filter, so that I changed it.
V1 looks fine.

Figure 1: R1

Figure 2: V1 (brown lines are the current situation)

Figure 3: Changed R1 filter