Sheila, Evan H., Jenne, Elenna
Summary: Added new resg filters to PR3 pitch and yaw damping. Allows us to reduce the gain by a factor of 5 in both loops, no issues. Reduction in coherence of PR3 damping to DARM, PR3 damping to CHARD, noise reduction in DARM
Current state of the PR3 loops: implemented new resgains to PR3 pitch and locked successfully. During NLN we stepped the overall PR3 yaw damping gain down, eventually by a factor of 5. The IFO is stable. Our idea is at next lockloss to turn on the resgain filter we titled "1Boost", and reduce the gain in FM7 from 0.2 to 0.04. The we can return the overall filter bank gain to -1 (right now it is at -0.2).
Figure1 shows improvement in the coherences with PR3 to DARM, CHARD to DARM, and the PR3 oplevs to CHARD.
Figure4 shows the change in the pitch OLG and Figure5 shows the change in the yaw OLG, red shows the new loop, blue shows the old loop.
Other notes:
- There is coherence with SR2 and DARM above 10 Hz- figure2
- There is coherence with PR2 and DARM below 10 Hz - figure3
Attachment shows yesterday's DARM error signal spectrum compared to the new spectrum with less PR3 damping. The error signal rms is reduced to 0.01 mA.
A low-frequency bruco of this latest lock is available here: https://ldas-jobs.ligo.caltech.edu/~evan.hall/brucos/2022-07/lho_omc_dcpd_2022-07-21_133000/
Here are the step by step instructions to log into a machine where bruco seems to be working:
ssh sheila.dwyer@ssh.ligo.org
choose CIT
chose pcdev12
conda activate igwn-testing
python -m bruco --ifo=H1 --channel=OAF-CLEAN_DELTAL_EXTERNAL_DQ --gpsb=1342402218 --length=600 --outfs=4096 --naver=200 --dir=~/public_html/brucos/lho_cleaned_1342402218 --top=100 --webtop=20 --xlim=5:2048 --ylim=1e-24:1e-19 --plot=html --nproc=20 --excluded=/cvmfs/oasis.opensciencegrid.org/ligo/sw/conda/envs/igwn-testing/share/bruco/lho_excluded_channels_O3.txt
When I came in, Sheila had already turned on "1Boost". In FM7 there is a gain filter that is nominally 0.2. With our changes, we reduced the overall Yaw damping gain from -1 to -0.2. I put the gain back to -1, and changed FM7 to be gain 0.04. I have also SDFed the changes we made to the PR3 damping loops in pitch and yaw.
I tweaked the PR3 yaw loop even further to reduce gain peaking.
- The filter "1Boost" in FM2 was combined with the resonant gain in FM3. FM3 is now just called "resG". I also combined the bounce and roll stopband filters into a single module in FM9, now called "BounceRoll". Therefore to run in the configuration Elenna describes above, use FM3,4,5,7,8,9 with EPICS gain of −1.
- I used FM1, FM2, and FM10 to install Jeff's Level 2 HLTS damping design. The foton attachment compares this configuration to Elenna's configuration. The Level 2 configuration in total requires FM1,2,5,9,10 and also uses an EPICS gain of −1. A ringdown plot is attached at the end (30 ct into M1 yaw test). We may want to consider turning up the damping gain a bit, but nonetheless I accepted the changes in SDF. For this design I used Jeff's script in SusSVN under
HLTS/Common/FilterDesign/Scripts/design_damping_HLTS_20130710.m. See also G1400151.
The second attachment measures the PR3 M1 yaw loop gain of the old configuration (before Elenna and I changed anything), the new configuration that Elenna describes, and the newest configuration which is Jeff's Level 2 design.
