For future reference, the peak in the ISS_rpn spectrum at ~120Hz appears to be caused by the NPRO's noise eater. It also explains why I had to lower the reference signal to -2.5 from -1.9 to get the diffracted power to a reasonable level (33% down to 2%). This was fixed by turning the noise eater off and on. The attached plots show the noise eater noise (iss_rpn-009) and after it was reset (iss_rpn-001).

This was caused by swapping the FAST cable earlier last week. I was able to set the reference signal back to -1.9 afterwards.