(Chris, Kiwamu, Stefan) We started off by relieving all MC offsets to alignment sliders, and setting some electronics offsets on the MC board. Next we aligned the whole HIFO-Y (green pointing into the arm, PR3 for green transmission, PR3 & IM4 for maximum IR fringing, PRM for maximum power recycling fringing). Next we peaked into HAM1 and identified the REFL beam - it hits the vacuum chamber right next to the view port. So tomorrow we will have to do another incursion to HAM1 to get that beam out. Next we touched up the beat node on ISCT1. With 0.1mWatt into the beam combiner from the SHG, and 1.4mWatt from the y-arm, we expected a maximum of 220Vpkk (assuming 1kOhm transimpedance and 0.6A/W). we got 150mVpkk. Next we checked out the electronics. The cable from the common PD went to the differential PFD (Phase Frequency Discriminator), and vice versa. So we switched the cables. Next we noticed that the Bekoff control between COMM and DIFF is also switched - this needs to be fixed. Next we started playing with the feedback to the IMC and PSL. We were able to get the PFD off the rails by some slow feed-back to MC2, but unfortunately we have no way of doing this without already putting a signal into MC2 length. Next we tried to simply increase the the REFL_SERVO_SLOW to MC2 length gain, but we get an oscillation at 6Hz to 8Hz (depending on the slow integrated). It seems like we are missing some filters in the chain... Looking at DTT spectra of H1:IMC-L_OUT and H1:LSC-CARM (both frequency sensors) we find that the PFD is still in the non-linear regime. The 8Hz signal upconverst noise everywhere, dominating the frequency noise in H1:IMC-L_OUL (the two signals look identical, with opposite sign). So we have a chicken and egg problem: no signal from ALS_COMM without feed-back, and no feed-back without checking that the signal makes sense. In other words: It is time for a complete IMC and ALS model in MATLAB - otherwise we are running blind.
I wrote the contribution of mine alluded to above up as https://dcc.ligo.org/LIGO-T1300534-v1 . It addresses the problem that if you want to drive the optic in L (longitudinal) from the top mass then there is a significant and annoying coupling to P (pitch). It turns out that in the crucial low frequency zone up to about 1 Hz, the most bang-for-the-buck thing to do is ignore P and simply notch out the 0.434 Hz fundamental L mode in the L drive. This was tried yesterday and made a useful improvement.
It would be good to try the rest of the recipe, which involves some P actuation to cancel the remaining L->P cross-coupling. (The filter for this is much simpler when the L peak is handled by the L filter.) However it turns out there's no convenient place to inject this. The L filter is currently in FM1 of the H1:SUS-ETMY_M0_LOCK_L. However we don't want to put the P filter in H1:SUS-ETMY_M0_LOCK_P, we want to fork the ISC L signal, put a different filter on it and inject it as P. That is, we need a mini-DRIVEALIGN just for the LOCK. We plan to test the idea with the H1:SUS-ETMY_M0_TEST_L and H1:SUS-ETMY_M0_TEST_P filters and see if it works well enough to warrant topology changes.