Summary: Tonight we successfully implemented active violin mode damping. We demonstrated damping of modes on all test masses by length actuation on the marionettas. Modes that are isolated in frequency are easily damped into the noise floor. We had trouble with modes close together in frequency and on the same mirror, as expected; specifically around 449.3 Hz.
- - - - - - - -
Damping scheme
The damping scheme and associated channels are shown visually here. It goes as follows:
-
take in DARM as error signal
-
a BP filter is applied around violin mode: bandpass is 4th order buttorworth, ~200 mHz wide. There are two BP filters to chose from. Filter 1 is centered on measured violin frequency, Filter 2 is shifted by ~60 degrees
-
apply gain to mode path
-
all mode paths belonging to one mirror are summed together to create a mirror-specific correction signal
-
correction signal is sent to either pitch or length of the marionetta
Damping notes
Gains and filters and drives that effectively damped are here. Times that damping started to occur are also there, in case anyone wants to look at data. All modes that are frequency-separated from other modes on same mirror were easily damped into the noise floor.
Ringdown activity here. Full before/after spectrum here. Bandpass coverage here. Note: we don't know what that broad bump in the spectrum is, but it's been there all night and we don't think its associated with our activity.
Only trouble modes are around 449.3 Hz. NE has modes at 449.30 and 449.33 Hz, making the sensing and independent damping difficult. We rang them up quite a bit before getting a setting that somewhat turned them around; they will need more work. Zoom in of spectrum around the modes here. Zoom of bandpass coverage mess here.
RMS channels are best indicators of damping/growing, but other nearby high modes can bleed in.
GUI
Yuta made an awesome GUI for violin mode damping, shown here!
For everything you can:
-
turn on/off a master switch. This is turned off by Metatron in the DOWN state, requires manual turn on (has 10 second ramp). In this way you can keep known good filters and gains set and eventually this can be turned on via Metatron.
For each mirror you can:
-
turn off the output
-
toggle between length/pitch marionetta drive
-
open relevant dataDisplay screens, which will start automatically.
For each mode you can:
-
get the current gain (wait time >5 sec)
-
set the gain
-
toggle between filter 1 & 2
-
emergency stop sending any signal along that mode path
GUI and associated config files are found in /virgoDev/Automation/scripts/LSC/ViolinDamping/violingui.py. It can be opened for all mirrors ( >>python violingui.py ) and also for one mirror, input as an argument ( >>python violingui.py NE )
Other stuff
Violin damping infrustructure is in LSC_Acl.cfg, filters in LSC_Filters.cfg. Note: we hijacked ASC_{mirror}_TX/TY_NOISE_OUT channels to send to DSP for now. This will be cleaned up some, matrix options for drive, etc.
Average elapsed time of LSC_Acl increased form 12.8 to 16.6 us with all added filters, etc. We plan to next optimize some filters, etc., to bring this down.