From SEI log 2252, 27 July, 2023 - Brian Lantz I updated the script I used to measure the excess tilt on HAM7 (SEI log 2251) to make it easy to collect and process data from all of the HAM chambers. I've run it for HAM2-HAM7 at LLO and LHO. Still need to do HAM8, For LHO, I used the time as the HAM7 work in SEI log 2251. For LLO, I used June 2, 2023, 2:00:00 UTC - the vertical microseism BLRMS is about 300 nm/sec (very similar to LHO vertical) - The horizontal microseismic BLRMS is 2-3 times larger, which is typical LLO vs. LHO. no EQ, IFO is locked. wind is ~ 0, no anthropogenic band spikes Central Time: Jun 01, 2023, 21:00:00 (Thursday evening, local time) GPS Time = 1369706418 I used ITMY as the floor sensor for all of these, except the original H1:HAM7, which used the HAM5 STS-2. HAM8 needs a different sensor. Monitor for excess motion at the microseism, likely from table tilt Here is a first look at the excess motion of HAM7 caused by tilt of the optical table. Do do this, I compare the horizontal GS-13 signal (stage 1 inertial motion) to a synthetic signal of CPS + ground. Since (nominally) CPS = stage 1 inertial motion - support table motion, CPS + ground is an interesting signal. I also compute the residual = GS-13 - (CPS + ground), the residual should be 0. I'm using HAM7 with recent data because I think it will be a decent test of the process so we can compare the before/ after of updating the GS-13 vertical calibrations (the likely source of excess tilt) I need to be able to add and subtract the signals, and I'd like to be able to do this in realtime at the models so we can monitor the residual for shenanigans. I decided to use a band-limited velocity signal as the 'common calibration'. This seems to work well in matlab. This means I will convert all of the signals to velocity. I decided to roll off below 50 mHz with 2 poles at 30 mHz (1+2i, 1-2i)/sqrt(5) to give decent phase matching at the microseism without blowing up the GS13 signal too much at lower frequencies. I roll off the high freq with 2 real poles at 1 Hz. I tried 3 Hz but the residual time series was dominated by CPS noise. Velocity is nice because: 1. It's a little more white, you can see the microseism easily with your eye in the residual, and you are not dominated by the lower frequency wander or the CPS noise about 1 Hz. 2. Usually differentiation is more stable than integration, so I expect we'll get better numerical behavior with differentiating the CPS than with integrating the STS-2. The GS-13 needs 2 integrations from 50 mHz to 1 Hz instead of 3, so I guess that's better. I have not tested this numerically, but as you see below the BL velocity seems fine. So I grab 2000 sec. of HAM7 data starting at GPS=1368165618 This is May 15, 2023, 06:00:00, UTC (local time is Sunday night May 14, 11 pm PDT) I tried to find a time where I could retrieve the data, the IFO is running, there are no earthquakes, the wind is low, and the microseism was big. There were no times which met this criteria - but the microseism BLRMS was relative large - between 200 and 300 nm/sec (right now it's about 60 nm/sec) f1_ideal_response.jpg - This is the target response for all the sensors f2_filters.jpg - These are the filters applied to each sensor f3_ground_motion.jpg - Ground motion during the test f4_BL_Z_3signals.jpg - Z motion version of the 3 signals as BL velocity - see that the CPS and the ground look very similar - this is a good test that the filtering is working. The GS-13 signal is much smaller - that is because the isolation is working. The lower trace is the residual. It should be 0, and it's pretty close. That's good. f5_BL_Z_2inertials.jpg - same data, but showing the GS-13 and the "virtual GS-13" = CPS+ground. the signals look similar. f6_ASD_Z_3signals.jpg - At the microseism - The table motion is ~10x below the ground, and the residual is 10x below that. There is some noise at the microseism - about 2e-8 m/rtHz. I don't know if this is real or numerical. I'm happy at the level, though. f7_ASD_Z_2inertials.jpg - At the microseism - The GS-13 and the "virtual GS-13" match very well. Conclusion: Tilt contamination is not a big deal for Z. Real Conclusion: The method is not crazy. on the to-do add the mccs2 residual to this plot f8_BL_X_3signals.jpg - X-DOF timeseries - there is a much bigger residual in X than Z, but not a lot of amplitude at the microseism, compare to Y in a moment f9_ASD_X_3signals.jpg - ASD of the X residual - it's much smaller than the X GS-13, which is good. X is only slightly contaminated, and the X GS-13 is mostly truthful at the microseism. Not so much at lower frequencies, though, but it never gets terrible. f10_coherence_X_residual.jpg - This is the coherence of the X residual with the Z ground motion and the ry CPS signals. There is some at the microseism, but not a huge amount. f11_BL_Y_3signals.jpg - YIKES - the Y residual shows clear excess motion at the microseism. The Y GS-13 and the Y CPS are NOT seeing the same signal. The Y GS-13 is contaminated by tilt. We'll see this in the coherence plots in a moment f12_BL_Y_2inertials.jpg - YIKES - compare the GS-13 with the "synthetic GS-13" and the difference is pretty clear. f13_ASD_Y_3signals.jpg - ASD of YIKE - the residual is 1e-7 m/rtHz at the microseism f14_coherence_Y_residual.jpg - coherence of the Y residual with the Z ground and the rx CPS shows very high coherence at the microseism f15_coherence_Y_GS13.jpg - The Y GS-13 signal also shows high coherence with Z ground and RX CPS. note that at 20 Hz, the GS13 has more coherence with the RX CPS than Y of the ground. This might be changed if I use the L-4C on stage 0. or if we balance the vertical GS-13s. From SEI log 2258, and comments: In this log, I've attached 2 pdf files, 1 for each site. There is 1 page per HAM chamber - showing the Z motion and the X&Y residuals (except I still need to do HAM8). H1:HAM7 has gotten lots of attention, but it is not the worst offender. I've tried to make a list of worst to best at each site. LHO: HAM6, HAM5, HAM4, HAM2, HAM3, HAM7 (this is totally by eye, your mileage may vary) LLO: HAM5, HAM4, HAM3, HAM2, HAM7, HAM6 This is roughly the order we should use to try and tweak up the performance. Except, I think we should actually start with LHO:HAM7, because we've spent so much time looking at it. Next step: repeat the commissioning drive, but with a vertical drive, not a single-corner drive - and see if the answer changes. We have a good idea what we should see with H1:HAM7 from SEI log 2239. All the resources for this analysis are in the seismic svn at Common/Documents/T2300286_tilt_induced_excess_motion/ The main analysis script is HAM_excess_motion_plot_script.m, to use it just add (or uncomment) the name of your data file and a text string which gets added to the plots. The data is collected with the script get_HAMdata_script.m. The analysis script assumes the same channel list and order, it's not very smart so be warned. The channel list is (with the correct ifo and chamber, of course): 'H1:ISI-GND_STS_ITMY_X_DQ' 'H1:ISI-GND_STS_ITMY_Y_DQ' 'H1:ISI-GND_STS_ITMY_Z_DQ' 'H1:ISI-HAM5_BLND_GS13X_IN1_DQ' 'H1:ISI-HAM5_BLND_GS13Y_IN1_DQ' 'H1:ISI-HAM5_BLND_GS13Z_IN1_DQ' 'H1:ISI-HAM5_BLND_GS13RX_IN1_DQ' 'H1:ISI-HAM5_BLND_GS13RY_IN1_DQ' 'H1:ISI-HAM5_BLND_GS13RZ_IN1_DQ' 'H1:ISI-HAM5_SCSUM_CPS_X_IN_DQ' 'H1:ISI-HAM5_SCSUM_CPS_Y_IN_DQ' 'H1:ISI-HAM5_SCSUM_CPS_Z_IN_DQ' 'H1:ISI-HAM5_BLND_CPSRX_IN1_DQ' 'H1:ISI-HAM5_BLND_CPSRY_IN1_DQ' 'H1:ISI-HAM5_BLND_CPSRZ_IN1_DQ' All the ground data is from STS_ITMY, except LHO:HAM7 which used STS_HAM5. HAM8 will use the one in the HAM shack. The plots are described in SEI log 2252. The plots here were combined with print_all_merge_fun.m (and a new version of ghostscript) here again, your mileage may vary.