After seeing terrible glitches right before both locklosses today (see figure 1) I started digging around to see what the cause was. The noise looked similar to times of other PIs. Looking at the summary page which monitors the PI behaviour you can see the 28-32 kHz region is elevated right before both locklosses (see figure 2). After plotting (see figure 3) the timeseries of H1:OMC-BLRMS_32_BAND{1-7}_
Mode is ETMY 32761.5 Hz.
First attachment shows the mode as seen in H1:OMC-PI_DCPD_64KHZ_AHF_DQ while rung up before the first lockloss ~7 UTC today. Second attachment shows that this mode (in black) is from ETMY by following other known ETMY modes (in blue) during the 19 hr lock that ended in that lockloss. We see that it is indeed at 32761.5 Hz - as opposed to aliased down from 32775.5 Hz - since it follows the slope of the ETMY drumhead and 15kHz mode, contrary to the 18kHz mode which is aliased down from 47kHz and thus is seen to move opposite.
I will set up damping asap - it will drop us out of Observe and since we are currently in coincident time, I'll wait until one site drops or the mode starts to ring up. Currently monitoring.
Sheila, Terra
There's 4 modes in this 32.76kHz region, two on each end test mass: 32761.5 Hz ETMY, 32764.5 Hz ETMX, 32766 Hz ETMX, 32767.5 (aliased down from 32768.5) Hz. ETMY, as shown here.
The first and last have sat the highest over the past several locks (the first breaking a few locks), so we've set up damping loops for them:
MODE 23: 32761.5 Hz ETMY and MODE 22: 32767.5 Hz ETMY
Note that the Nyquist is 32768 Hz, so MODE 22 sits right there on it and none of this region can be seen in DTT (something something it cuts off before the Nyquist?). They can be seen in the OMC-PI channel and they've been added to the PI2 StripTool. Have added all 4 to PI Wiki.
Mode 22 and 23 are set up to damp on ETMY. both using channel 5 on ETMY. Both 22 and 23 are on the PI2 stip tool, but niether is added to verbal alarms yet so operators will want to keep an eye on the StripTool this weekend.
The damping gains and phases for these modes are unmonitored in SDF. If you see one of the modes ringing up, try damping it. Since we don't know what shape these modes have, Terra suggested that we try damping it with the same sign on UR and LL. If this isn't working, you can try flipping the matrix element for LL to -1, see the attached screenshot for how to do that.
