We no longer see excess noise when controlling DARM with ETMX bias -125V,  72111, and one potential change that might explain this would be the recent reduction in DARM RMS (this wouldn't be a normal upconversion as it is seen at -125V and not at +125V).  To test the idea that the improvement in DARM RMS is the reason for the change, we went back to the -125V configuration and added an injection into the DARM error point at 2.1Hz.  

We don't see any change in DARM for increases of a factor of 2-3 increase in the DARM error signal RMS, but we also don't see the injection in the control signal at this level because the drive to the ETM is dominated by SRCL FF at this frequency 71994. When we increased the error point injection enough that we start to see it in the drive to the ESD, we are increasing the error signal RMS by 2 orders of magnitude, and start to see it upconverted into the GW band, unsuprisingly.  If we want to pursue this further we may need to think more carefully about where to inject. 

While transitioning DARM control back to the usual configuration, we lost lock.  I think that the configuration I was setting things back to looks OK (ie, there was no typo or anything), so I'm not sure what happened.