We saw a big 3.1 Hz instability in DHARD P (and other loops) a few minutes after getting to low noise in our second lock today.  We were constantly saturating ETMX PUM, and the rage dropped.  This seems similar to a problem seen at LLO (maybe 42442, I'm not sure if  43785 is related as well).  

We stepped down the output matrix elements that send DHARD P to the ITMs, and that made things better although the oscillation and saturations continued.  Once we engaged the ADS loops and the converged things got better.  For now we still have DHARD P sent to the ETMs only, because Anamaria told us that is what LLO is doing for this problem.  

We are hoping that this was a cross coupling due to our spot positions being way off for ADS, so we are hoping that for the next few locks we can try to wait and make sure that ADS is converged a bit before doing low noise ASC.