I took some time to rerun the noise budget injections for all ASC arm loops, PRC2 and MICH. I have attached the resulting plot below. I used 10 minutes of quiet time from tonight, gps start: 1364268964, in case you want to use the time for a bruco.
The most exciting point is the significant reduction in CHARD Y noise contribution. For reference, I have attached the ASC noise budget from March 7 showing the CHARD Y contribution as the largest contributor to DARM. The DHARD loops and CHARD P are the next loops to attack for noise reduction from ASC. Gabriele and I have a new DHARD Y design that should reduce the DHARD Y contribution by 15 dB at 20 Hz. I will also be looking into improving the HAM1 FF to win a little more in CHARD P.
I don't know exactly when was the last time the PRC2 and MICH budgets were ran, so I am not surprised to see a change.
As I guessed from Gabriele's recent bruco, it looks like we are starting to get into the regime where the arm loops will have about the same noise contribution as the cavity loops. I didn't update the SRC2 injections, but it will probably be necessary to start including the recycling cavity loops and maybe even INP1 in the ASC sub budget moving forward.
I also reran the LSC injections, but I am not posting them as I am sure they will soon be subject to change. One note is that the PRCL loop issue that I had the last time I reran the noise budget seems to have fixed itself. Specifically, there was an error where the PRCL noise was way overestimated at 10 Hz and I couldn't debug why. Feel free to check out the plots here: /ligo/gitcommon/NoiseBudget/aligoNB/out/H1/lho_all_noisebudgets
This noise budget shows the total sum of all noises exceeding the measured DARM noise. This might be because we are overcounting. ASC and LSC signals are not independent and therefore summing them in quadrature result in an overestimate. The huge unreadable first plot shows that there is indeed quite a lot of coherence between different ASC and LSC signals, and between ASC and LSC.
To make the result more readable, I computed the average coherence between all pairs of signals, between 10 and 30 Hz, that is where the ASC and LSC noise contributes the msot to DARM. There are several cases where pairs of signals are very coherent, as shown by the coherence matrix attached.
BruCo results for this quiet time: https://ldas-jobs.ligo-wa.caltech.edu/~gabriele.vajente/bruco_CAL_1364268964/