Ed called about difficulty locking, since I don't see anything obvious I looked at BRS channels I foudn trended here: 51338. It looks like the low frequency BLRMS of BRS X has been growing exponentially.
Here's th BRS HEALTH ndscope image during ENGAGE_SOFT_LOOPS
I don't see anything problematic really in the time series. But I did notice from the Detchar summary pages that there was some weird behavior with the inside temperature for a few hours from 16 UTC yesterday (02/09/2019). It takes hours until the BRS itself responds to temperature changes so I'm just guessing but you might be seeing the effects of that. You might consider turning the sensor correction to NON BRS state and lock since the wind is low now anyways. Here is the plot of the temperature
?
I don't think there is anything wrong with BRSX. First attached trend are 7 weeks of the both BRS driftmons, which are kind of a measure of the DC position. BRSX is close to the level where I would want to recenter, but I hope to push that off until we try to install Eyal and Arnaud's heating pad. I don't think the "strangeness" Eyal notes is anything to be worried about. It's a tenth of a degree, from 8am local to about 4pm, probably the aircon.
Second plot are 4 days of trends for ISC_LOCK (top left), the sensor correction control signal generated from the tilt subtracted STS (bottom left) and the drift (top right) and velocity of BRSX (bottom right). I don't see any signals in the right 2 plots that would cause the spikes in the sensor correction (so I'm assuming they are earthquakes, but I'll keep looking), or would be any cause for concern.
I don't know why the lowest frequency BLRMS of the BRS seem to accumulate these large numbers but I don't think it's a problem. Or if it is, both BRS are busted. Third image is the BLRMS for both BRS for the last 2 weeks. Maybe this is some sort of computational weirdness? Some integrator in the BLRMS filter? Maybe we should periodically clear the histories of some of the filters.
I've looked at few non-brs blrms channels and all of the dc-30mhz blrms channels show this same behavior. The 30mhz low pass is just integrating non-sense. There is no easy way to clear the history of the filter that does the calculation, without restarting the model. Dave and I are talking about adding an epic variable to zero out the filter. We will try prototyping on the test stand.
can you inject in parallel a negative large DC to try and zero it out?
The 30mhz BLRMS filter shouldn't be integrating like this. Filed FRS https://services.ligo-la.caltech.edu/FRS/show_bug.cgi?id=13602