As reported in the original post, we encountered a 1.7 Hz oscillation, even though a notch filter appeared to sufficiently suppress this frequency in the feedback signal. 
We found that this oscillation does not occur during DC readout, suggesting that it originates from a radiation pressure effect, and that cavity detuning may have introduced additional stability—possibly due to a small optical spring.

To mitigate this issue, we followed the procedure below and successfully reached the observation state with an IX drive alignment value of -1.0:

• Reached the “Engaging DARM offset” state

• Set the IX drive alignment to -1.0 and waited for BPC stabilization

• Performed initial OMC RF alignment

• Reached DC readout

The resulting DARM sensitivity is shown by the orange line in the attached figure.
Although the sensitivity improved at low frequencies, it degraded at mid and high frequencies.
We attempted to recover sensitivity by adjusting the INP2 and PRC2 offsets, but the improvement was limited.

We then restored the IX drive alignment to -2.5, and the sensitivity returned to the previous level (blue and green lines in the figure), without needing to redo the OMC RF alignment.

A possible cause of the degraded sensitivity with the -1.0 alignment is reduced common-mode rejection.
In fact, frequency noise increased, and the arm transmission at both ends became more asymmetric than usual.

The IX beam miscentering was previously measured to be approximately 1 cm at a drive alignment of -2.5.
Therefore, with a drive alignment of -1.0, the miscentering should be around 4 mm.
Although this represents a noticeable shift, it did not improve sensitivity—indicating that this direction of beam spot adjustment is likely unfavorable.

Next, we will evaluate the sensitivity after shifting the beam spot further toward the edge, using a drive alignment value of -3.0 or more.