-
CHARD P -I had previously made the CHARD cut off more aggresive, because in long locks where our spot positions move a little bit this noise will probably lead to the range fluctuating. I was concerned that this was too agresive, so I reverterd to the old cut off filter tonight and set the gain back up to -0.14, but we should try something intermediate once we have seen that the IFO is stable again.
-
Patrick directed me to his conlog test machine (http://conlog-test-replica) and I checked for differences in ASC settings between now and saturday's long lock. (Note that we have not has a stable long lock since Monday's accidental reboot of everything). The most suspicous wrong settings were in the refl DC centering loops, I've reverted to the settings in the screenshot attached to 31382.
-
Corey and I did add back in the RF9 modulation depth increase, and locked on the AS36 signal for SRM, and it seems OK so far. (before doing that we had a couple of locks were switching to AS36 clearly made the build ups worse) I suspect that the funky AS36 input matrix is sensitive to the RF9 modulation depth, so we either need to retune it if we are going to skip the reduction, or just put the reduction back in.
-
We were locked for about an hour like this, with PR3 controlled on POPX instead of REFL WFS. When we tried to switch over the IFO unlocked. For tonight I am just suggesting that Corey skip the switch and leave the beam diverters open. We will have to think about a real solution in the morning. (We have tried measuring the matrix, which seemed to work short term anyway). Other options would be to try to increase the recycling gain or to see if we can give the PRC2 loops more gain margin.
Operator information:
I've added the Reduce RF9 modulation depth state back into the normal guardian path, but if you are having locklosses durring this state or right after you can try skipping it. If you skip it, you might want to keep the beam diverters open so you can see that the SRC1 loops are OK (ie, the ratio of AS90/POP90 is staying reasonably high over the lock). You might need to open the SRC1 loops if you skip reducing RF9 modulation depth.
I've also moved PR3 REFL WFS out of the normal path, so that it will be skipped for now if you just request nominal low noise.
In the morning POPX DC PIT was routed to PRC2. Didn't make sense (POPX DC is controled by its own centering loop), so manually zero-ed it.
Should have had no impact on anything as POPX DC is at least 4 orders of magnitude smaller than POPX RF over the entire frequency band.
Seems like this was puched in by hand last night.
Combination of ASC-REFL_B_RF9I and 45I was routed to unused DC7 as a fake PRC2 on REFLB.
I measured the phase between this DC7 and PRC2 (which is POPX_I) in nominal low noise, and the YAW was out of phase for microseismic peak but not above 3Hz or so (first attachment left) though PIT was pretty good.
By changing the mixing ratio of REFLB RF9I and RF45I for YAW, the coherence between DC7 and PRC2 improves and the phase becomes saner (but never as good as PIT). The second attachment shows when the relevant input matrix components were [0.14, -0.08], i.e.
DC7_YAW= 0.14*REFL_B_RF9I_YAW -0.08*REFL_B_RF45I_YAW.
(Nominally this was [0.14, -0.2].) The change was put in the guardian, I and Jenne were able to manually switch. Now PR3 is on REFL WFS instead of POPX.
This made ASAIR_B_RF90_I bouncier, but IFO seems to run. We'll run a2l and we'll also measure the sensing matrix