I was able to get some testing in on the SR3 P and Y estimators today with the most up to date estimator and blend filters. There were still some filter issues at first, but thanks to all the hard work from Brian, Edgard, and Ivey, we were able to get everything working!
Each test was done with the other estimator off, so as if the other damping degrees of freedom were just their normal -0.5 gain damping.
SR3 Y Estimator
We had tested the SR3 Y estimator twice before (85615, 86319), but this should be the best case of everything for the Y estimator. Each test was 30 minutes.
Estimator filters: Clean_fits_H1SR3_Y_2025-08-05.mat (86366)
Blend filters: Estimator_blend_skinnynotch_SR3yaw_20250723.m (86265)
Tests
DAMP Y -0.1
2025-08-19 18:22:06 - 18:52:06 UTC
DAMP Y -0.1 and OSEM Y -0.4
2025-08-19 17:22:05 - 17:52:17 UTC
DAMP Y -0.1 and EST Y -0.4
2025-08-19 17:52:23 - 18:22:00 UTC
Preliminary results: /ligo/svncommon/SusSVN/sus/trunk/HLTS/Common/FilterDesign/Estimator/Estimator/SR3_Y_EstTest_2025-08-19.xml r12603
The tests gave really nice results!
SR3 P Estimator
We have never tested the SR3 P estimator, but the Stanford peeps were able to get all the filters needed just in time for me to get a short (~7-10 mins each) set of estimator measurements.
Estimator filters: Clean_fits_H1SR3_P-08-05.mat (86430)
Blend filters: blend_SR3_pitchv1.m (86452)
Tests
DAMP P -0.1
2025-08-19 19:17:05 - 19:25:10 UTC
DAMP P -0.1 and OSEM P -0.4
2025-08-19 19:44:45 - 19:54:45 UTC
DAMP P -0.1 and EST P -0.4
2025-08-19 19:33:01 - 19:44:27 UTC
Preliminary results: /ligo/svncommon/SusSVN/sus/trunk/HLTS/Common/FilterDesign/Estimator/Estimator/SR3_P_EstTest_2025-08-19.xml r12603
These tests don't have as high of a bin width or number of averages as the Y estimator did, but the results still show a decent decrease in noise in ADD_P_TOTAL, besides a new peak that popped up at 3.5 Hz.
Next steps:
Both P and Y estimator tests are looking pretty good, so the next steps are to test them both out while we are locked. On Thursday I'll be doing that for at least SR3 Y and P if there is time. To make sure we can transition from the regular OSEM damping to the estimator damping, I have both SR3 P and Y set to their 'DAMP P/Y -0.1 and OSEM P/Y -0.4' settings. This means -0.1 of their DAMP gains are in the DAMP filter bank, and -0.4 of their gains are in the EST_{P,Y}_DAMP_OSEM filter bank, which give us the same amount and type of damping as -0.5 gain of normal DAMP damping would give us.
On Thursday I will be able to test out the estimator by swapping the -0.4 DAMP_OSEM gain to -0.4 DAMP_EST gain.
These damping settings have been accepted in sdf
Woot! This seems really good.
A few things about pitch
- The peak at 3.5 Hz is expected. This is a pitch mode, but I guess it's really hard to see with the OSEMs. It seems to be only barely damped. You can see it in the blend design plots of alog 86452, it's there in many plots eg. plot 3. For the "to-do" list, we could consider adding a bit more damping to this peak - if it is causing trouble for ASC.
- I would expect the low freq drive (.1 to 1 Hz) to be lower. It will be good to check the updated data.
- You can see the down-side to this particular blend - the 'bottom' setting of 0.1 means that the OSEM noise gets added at 0.1 for this path at high frequencies, so the total goes from 0.2 in light-damping to 0.2 + (0.8 * 0.1) = 0.28 with the light damping + estimator damping. It's fine, but could be played with more if useful. This is not really inherent to this design approach, just to the first one I built.