stefan.ballmer@LIGO.ORG - posted 03:10, Friday 07 August 2015 - last comment - 09:22, Friday 07 August 2015(20307)
Victory over 508.2896Hz in 2nd round
Evan Stefan, Sudarshan
Previous alogs 20298, 17610, 19720
- With the configuration from alog 20298
- MODE5: FM1, FM2, FM4, G=-100
- MODE8: FM1, FM4, FM5, G==500
we noticed that the 508.2896Hz rang down but then settled at a fixed and still large value of about 9e-15mRMS (2e-14m RMS/rtHz with BW=0.187Hz)
- This had two causes:
- The MODE5 filters (broadband feed-back for all ETMY modes) were interfering. Thus we designed a notch for the 508.2896Hz mode in MODE5.
This made the phase slightly worse for the 508.22Hz mode - but it still damps.
- While MODE8 it driving pitch, it has has some coupling to length, leading to some gain peaking at the violin mode frequency.
This effect limits the gain we can have.
Next we fine-tweaked the damping phase with the following trick:
- Right after turning on the MODE8 damping we see the immediate effect of the pitch to length coupling change the violin mode height.
- by fine-tweaking the phase, we can find the phases for which this immediate effect results in...
- ... the biggest step-down (violin mode and direct drive are exactly out of phase), and ...
- ... the biggest step-up (violin mode and direct drive are exactly in phase)
- We then added roughly 90deg to the step-down version of the filter, such that there is no immediate step when tuning on the damping.
This implies that the direct drive is lagging about 90deg behind the violin mode, which is what we want for damping.
- The setting in this configuration was (see snapshot)
- MODE5: FM1, FM2, FM4, FM5, G=-100
- MODE8: FM1, FM3, FM4, FM5, G==800
- In this configuration we saw a slope in the 508.2896Hz mode of about 1 decade reduction per 3.5h. (slow, but consistent!)
- We tried increasing the gain, but ran into a growing beat with about 2min duration. I am guessing that this is related to the P2L coupling induced gain peaking.
- After about 3h of this, we were again able to turn on the OMC DCPD whitening.
- MODE8 is now turned on in the guardian.
Images attached to this report
Comments related to this report
Some numbers on the feed-back phases and gains The filters currently used have: MODE5: Mode | 507.992 508.146 508.010 508.206 508.220 508.585 508.661 phase | -173d 177d -173d 170d 167d 167d 160d gain | 260dB for all MODE8: Mode | 508.289 phase | 72d gain | 277dB A transfer function measurement of the MODEx_OUT/DARM_IN1 gives roughly the same numbers, but lagging about 12deg, corresponding to 1 sample delay. So, roughly speaking, the 508.289 mode likes to be driven with an additional 90deg phase lag, and a 17dB higher gain.
After about 4h of damping, the mode is now about 10x smaller.