J. Kissel

Similar to the BSC ISI ST2 X/Y (longitudinal) and RX/RY (pitch) broad-band excitations taken and projections made (see LHO aLOGs 37752 and 38122), I've processed the RZ (yaw) broad-band injections I took a few days ago (see LHO aLOG 38031).

The ISI basis RZ couplings are significantly less than that from pitch or yaw, but ETMX remains the worst offender.

Recall that an ISI basis RZ drive does not just cause Yaw of the suspension point, it also creates Longitudinal and Translational for the QUAD as well (see T1100617). This is similar to an RX / RY drive causing some X/Y and Z motion. If we still had the CART2EUL drive matrices installed, we might be able to create a pure Yaw drive, but this is likely not worth it based on these results. My gut says that what coupling we do see from RZ drive, is just the small Longitudinal component of the RZ drive.

However, the L, P and Y couplings may be a little underestimated: the coherence between X/Y IN1s ,RX/RY IN1s, and RZ IN1s is non-negligible, but it's still small enough (0.1 to 0.2) that adding these noise predictions are incoherent (i.e. adding them in quadrature) may be close enough.
RZ of the platform is controlled, but bandwidths are limited to 30-40 Hz.

The remaining open questions regarding the L and P coupling / noise:
- Is this the cause of some transient features we've seen in the 70 and 90 Hz regions throughout aLIGO at H1?
    - The coupling is likely stationary, but we know the input is not ... 
- Should we check the HAMs or Beam Splitter for similar nastiness?
- What is the coupling mechanism?
    - Charge interactions between the cage / EQ stops / ring heaters and the test mass?
    - Scattered light?
    - Impressive that H1 was so much more unlucky than L1 in this...
- Is there any evidence of this happening before the July 6th 2017 EQ?
    - We'll be taking long-duration coherence measurements to check for coherence, but this may be thwarted by the non-stationary of the input noise (and jitter + new mystery noise masking the issue)
- What should we look for when we first go in chamber for vent in 1 month?
- We presume this is the pre-damped QUAD cage resonance. What can we do about it in the future?
    - Even the non-resonant, 1/f^2 slope is high enough that the noise is well above the designed sensitivity. We need to fix this if we're to get there!
    - Strive for increasing the bandwidth of the ISIs? We've run out of phase margin because of signal processing phase loss to go beyond where we are now, plus table motion is already below the GS13 sensor noise
    - Try to move the cage resonance by subtracting / adding weight, or make the vibration absorbers more lossy to reduce the Q further?