I ran a PRCL feedforward injection since the PRCL to DARM coherence has increased. Indeed, the injection showed the PRCL feedforward was doing worse around 30 Hz. I found that by increasing gain of the feedforward from 0.6 to 0.9 this improved the subtraction again.

You will notice that below about 15 Hz, PRCL coupling is still not great. That would probably require a more "invasive" adjustment of the feedforward fit, but maybe it's not worth it since it doesn't seem that it's causing much noise.

In the first attachment, blue trace shows and old coupling measurement with no feedforward, green was the "as found" previous feedforward fit measurement. Brown was the measurement today before I started adjusting the gain and red is what I got after increasing the feedforward gain. Second attachment shows the SDF accept. I also updated the gain in lscparams.