Sheila pointed out that we're having a lot of scattered light for the microseism level we've got, and at the Commish meeting asked about our sensor correction / seismic configuration.
Since the CPS differential is designed to help at the wind and earthquake band, it makes some sacrifices of isolation performance in the microseism band. As a test, I've turned off the CPS differential system by ramping down the final output gain (H1:ISI-DIFF_CONTROL_BIT to zero, ramping down over 10 seconds), and we'll see if we see a bit less scattering.
I took the IFO out of observing for a few minutes to make the transition, accepted the SDF diff, and went back to observing. Note that I left the SEI_DIFF guardian in its nominal state which might make one think that it's on, but since the Run state does nothing but return True, it is not writing to that EPICS channel. I didn't want to edit the guardian's nominal state, which is why I've left it alone.
It's been a little over 2 hours, and the IFO seems to like having the CPS diff off, so I'll leave it off for now.
In the attached screenshot, the top row is the BNS range, 2nd row is the DIFF control bit showing when I turned off CPS diff. While we're still seeing glitches, they (with this limited amount of time) seem to be somewhat less frequent, and the range seems a little bit better. The third row is microseism BLRMS for the z-axis STS at each building, showing that the microseism has not decreased since turning off the CPS diff. The 4th row is X, Y, Z axis BLRMS in both the EX and EY stations for the wind and earthquake band, showing that we haven't had too much wind or very low frequency seismicity, which is where the CPS diff helps most anyway. Note for the seismic plots that these units are nm/sec, whereas they are displayed in um/sec on the control room wall.
Looking at the ETM ISIs, the current cps diff filter makes the microseismic table motion and above worse on Stage 1, this is fine most of the time, but not a good thing for us at the moment. The attached plot shows the ground and St1 motion before Jenne turned off the cps diff (cyan and brown) and immediately after (red and green). The effect on the BSCs is pretty dramatic One thing we could try is just making a state that doesn't engage the ETM cps diff. It doesn't seem to be a problem to impress this extra motion on the BS, it would be good to get some time to try engaging different parts and seeing what hurts us. It would also be good to see what the effect is when SEI_CONF is in the useism state (which I switched to a few hours ago, as the microseism has gone over 1 micron/s rms).
Turning off the CPS diff had a clear impact on the size of glitches apparent in omicron on the summary pages.
The attached plots is a stitch-together of 2 days of omicron plots from the summary pages, along with the same time period of data in ndscope for the BNS range, whether the CPS diff was on (high) or off (low), what state the SEI_CONF guardian was in (high=nominal WINDY, med-low = microseism state, low-off-plot = sensor correction off), and the ground STS BLRMS in the microseism band. I've included vertical bars at the locations where the CPS diff is turned off yesterday, where Jim changed the SEI_CONF state to microseism, and where Corey changed the SEI_CONF state to have no sensor correction, in preparation for maintenance day.
The glitches from about 20 Hz - 40 Hz seem to immediately drop away when I turned off CPS diff yesterday afternoon, but then they start to come back as the ground motion continues to get even worse. The microseism has gone down a bit, but it looks like the CPS diff got turned back on (not yet sure if that was a guardian thing, or an SDF accept thing), and JeffK pointed out that we were again seeing big glitches. I've again turned off CPS diff (all of it, not just ETMs, although that is an interesting suggestion by Jim that perhaps we should try sometime).
One might think at first glance that Jim's having put the SEI_CONF guardian into the microseism state increased glitches between 10-20 Hz, but I suspect that those are actually just coming from the increased microseism during that time. There's some subtle threshold, but when the microseism BLRMS fall below that level, those glitches fade away, even though the seismic configuration has not changed.