Reports 1-1 of 1 Clear search Modify search
VIS (IY)
Carl Blair - 15:00 Thursday 30 January 2025 (32506) Print this report
Violin mode damping

[Carl, Haoyu]

We propose that the violin modes could be actively damped.  Also that notching violin mode freqencies in control paths could reduce mode excitations.
This could potentionally give a factor of 10 improved sensitivity between 176Hz and 186Hz during some recent lock stretches.  First attached figure top panel shows violin mode spectrum during some recent lock stretches (Dec and Jan).  The blue trace shows a trace where a factor of 10 to 100 reduction in mode amplitude could gain significant bandwidth sensitivity in DARM. The most recent trace from 22 Jan is similar to the red trace with low violin mode amplitude.
The violin modes are often small in amplitude in the figure, however we identify several times when the amplitude of individual modes grows dramatically during lock. One such transient is shown as a sequence of PSD (fig 1 bottom panel).  It grows in amplitude in a few seconds then rings down over a minute of so (03:13:00 31/12/2024 UTC).
We guess this must be the result of some control loop driving the modes (or intentional driving), notching violin mode frequencies in control loops that actuate on IM (probably pitch vertical and roll) may be a way to avoid these transients in violin mode amplitude and active damping may be able to reduce amplitudes faster and to a lower level after excitation.

In klog31963 excitation and ring down measurements of violin modes is described.   We understand the drive signal to excite the modes is IM pitch.

We therefore propose that the DARM signal could be fed back to IM pitch to actively damp violin modes.
A model change would be required to send the DARM signal to a VIS model to damp one violin mode.  Cartoon of where this change would be required in the RCG is shown in the second attached figure. The VIS models would also require DARM from IPCx_PCIE if this is available on VIS computers.  This type of damping has been used at LIGO Livingston for violin mode and bounce and roll mode damping.   Also similar notching of control signals (in addition to existing notches in damping filters) may decrease excitation of lower frequency suspension modes from glitches.

Images attached to this report
Comments to this report:
takafumi.ushiba - 14:46 Thursday 30 January 2025 (32507) Print this report

Thank you for your suggestions.

In fact, we already have feedback paths from LSC/ASC signals to any stages/DoFs of Type-A payload.
So, please use NBDAMP blocks if you would like to try to test violin mode damping.

Note that NBDAMP filters are implemented into the different model, feedback signals will be delayed by 2 sampling for sending the signals between models (but hopefully not so problematic because the sampling rate is high enough compared with the violin mode frequency).

Carl Blair - 10:58 Friday 31 January 2025 (32524) Print this report

[Carl, Haoyu]
EDIT
We have set up violin mode damping for one ETMX mode at 184Hz as identified in klog31977 as a test.  The damping uses the NB filter bank P4.  The input matrix receives the ISC lock signal going to that suspension.  Currently DARM goes to ETMX and ETMY so damping is possible on these two test masses using this method.  The filtered signal will then be sent to IM P for a test.  The P4 filter bank (first image) has a BP filter for the mode with a gain of 0.001 and 3 phase adjustment filters +60, -60 adn +30 degrees to give 12 possible damping phases.  The gain with the band pass filter should result in a drive signal to ETMX IM P that is about 1/1000 of the current p-p amplitude of the signal to the suspension.

We engaged the input to the P4 filter in the previous lock a few seconds before lock loss.  There were no outputs.  It does not look like the lock loss was accociated with this activity.  The attached plot show the DARM (the signal being used for violin mode damping) and the filter input. 

Images attached to this comment
Search Help
×

Warning

×