Although Jeff and I reduced the appearance of triple bounce and roll modes in DARM by adding notches to the top mass damping for the HSTS (30970), there are still a few modes that show up. 

The largest in our last lock was at 28.2 Hz, which belongs to PR3 according to the resonances wiki and Jeff's measurement in 30646.  We had no notches for the HSTSs so I have added notches 1 Hz wide centered around 28.2Hz and 44.2 Hz for PR3 and SR3.  

The other thing which is large in our spectrum from last night is the remaining 40.9375 Hz mode from PR2.  I am not sure why the PR2 mode shows up in DARM larger than the others.  I looked at spectra of the osems to see if there is more noise on the PR2 osems.  The attached screenshot shows that the top mass osems for PR2 aren't any more noisy than PRM, but there is extra noise in the lower stage osems for PR2, and a large comb.  This might be noise in the readback only.  

I made the notches in the damping loops 20dB deeper for PR2, we will see if this helps in the next lock.