Attached below are the results (top to top TF measurements and raw OSEM spectra - both damped and undamped) taken after power cycling h1susex and replacing the faulty 18bit DAC card. The attached results are for Main chain (M0) and reaction chain(R0).
Attachment 01: At first lets look at the plots which needs urgent attention. This morning I took damped and undamped spectra of all the raw OSEM signals as a check to rule out any electrical/mechanical problem. The ASD results shows that for the main chain (M0) and especially when damping is ON, we see a spike at around 11Hz. Also when the damping if OFF, it is still very noisy below 10Hz. It looks like this problem is mostly affecting the main chain (however for the reaction chain for R, V and T signals there is a small peak visible as well). The attached results are not complete, there are in total 70plots, but only 33 attached here due to pdfmerge command issue in Matlab (I will look into this issue as well) . These results are consistent with what I saw in the dtt spectra - posted this morning alog 63775.
Attachment02 and attachment03: shows the top top top TF measurements results for main chain and reaction chain. This looks all good (when compared against the model and 2018 data).
Attachment04 and attachment05: shows the individual osem performance for main chain when offsets where On (time 1800) and OFF (time 1900).
Attachment06: shows the individual osem performance for the reaction chain - looks all good.If you look at page no 16 and compare it against 2018 data (LHO alog 63496).
Re-engaging the offsets in the T and V degrees of freedom seem to have removed this excess peak at 11 Hz (and maybe also some of that noise). So, perhaps the problem was never the DAC (although obviously if it failed its autocal, it did need replacing), and we do have some kind of mechanical rubbing somewhere (or some electronics problem other than the DAC).
Anyhow I've turned those offsets back on and we're hopeful that this will allow us to lock.
Short version:
Could this 11Hz thing be a loop oscillation of the OSEM damping and ISI combined?
Longer version:
I chose time before and after offsets were put back in today (2022/Jun/30 19:30:41 UTC and 22:16:43 UTC).
In the 1st attachment, left column shows M0 OSEM sensors (references=no offset) while middle/right column are for ISI ST2/ST1 outputs. All ISI ST2 outputs clearly show ~11.375Hz and its 3rd (~34.13Hz), 5th (~56.75Hz) and 7th (~79.88Hz) harmonics, some of them also visible in ST1 outputs.
2nd attachment shows coherence between M0 SD sensor signal (which shows about an order of magnitude larger 11Hz peak than other M0 OSEM sensors) and ISI ST1/ST2 output.
I don't know why changing mechanical position of M0 will suddenly create mechanical resonances of relatively good Q like these. Though it doesn't sound super-likely, loop oscillation is somewhat more believable than mechanical resonances.
Sensing gain of each OSEM against M0 motion DOFs could be dependent on M0 position. Or maybe there's a subtle rubbing that changes the damping loop phase depending on M0 position. If we're close to loop oscillation to start with, these changes might be enough to make the loop stable or unstable.
Georgia added some offsets to ETMX M0 to remove the 11.35 Hz peak. We think that it is a sign of more rubbing and likely what was preventing us from locking.
Transverse = 29000
Vertical = 140000
Roll = 6000
We don't know why these offsets must be so large, but they improve the spectrum. We also checked that the low frequency noise in the ETMX oplev spectrum has quieted down.
Photos added of the offset changes, and the spectra comparing the old rubbing peak, and new rubbing peak.
