Kiwamu, Lisa, Matt, Stefan - First we verified that the CARM to transmitted light transition still works today (arm power 0.32*single arm). - Next we repeated the transition to AS_45_Q, and turned on the gain scaling (arm power 2.0*single arm). - We also immediately turn on AS_45_Q gain scaling with TR_X (arm power 2.0*single arm). - Then we turned on the DHARD WFS (arm power 2.0*single arm): whitening gain of 9dB: H1:ASC-AS_B_RF45_WHITEN_GAIN 3 H1:ASC-AS_B_RF45_Q_PIT_GAIN and H1:ASC-AS_B_RF45_Q_YAW_GAIN gain to 0.1 input matrix AS_B_RF49Q to DHARD H1:ASC-INMATRIX_P_8_6 1e-4, same for yaw H1:ASC-DHARD_P_GAIN 0.3 , same for yaw control filter modules FM1 (integrator) and FM5 (Low pass LP0.2) output matrix: H1:ASC-OUTMATRIX_P_7_8 = 1 , H1:ASC-OUTMATRIX_P_8_8 = -1 , H1:ASC-OUTMATRIX_Y_7_8 = 1 , H1:ASC-OUTMATRIX_Y_8_8 = 1 - Then we went to a hgher arm power (arm power 32*single arm). - We noticed that TR_X starts saturating right around there, so we intended to switch to the QPDS, but had an unreasonable amount of trouble for this simple operation. - Still trying... Attached is a plot of the CARM transfer function. It also includes a transmitted power ratio of the QPD's and the LSC PD's.
The message is that we increased the build up in the arms up to ~30 times the single arm power (CARM offset ~ 40pm). The first plot shows a trend of the arm power transmission. In the first (incredibly stable!) lock we made several attempts to reduce the CARM offset further, but the IFO started to become unstable, so Stefan was quickly going back to a larger offset to prevent unlocking. It turned out that the problem was a saturation in the X TR diode (clearly visible in the second attachment). The following locking attempts were killed by the (in principle) straightforward task of switching to the in-vac TR QPDs. The problem turned out to be that the signals that we were carefully matching were one before power normalization, and the other one afterwards..
Later, we reduced the CARM offset further down to approximately 20 pm at which point the arm build up was 100 times higher than that of the single arm.
As reported in Stefan's previous log, we were having a difficulty in switching the TR sensor from TR_X(Y)_A to ASC_QPD_Bs. Keeping using TR_A actually introduced a broad noise peak at around 70 Hz which showed up all the LSC loops. Stefan then carefully adjusted their relative gains and therefore the transiotion went very smooth. We did not see an obvious transient in any of the LSC signals during the transition of the TR sensors. Then we further reduced the TR_CARM offset to -15 counts without a problem. We did not see a gain peaking or any sign of significant change in the optical gain of DARM. So we did not have to chage the DARM gain. At this point we saw REFL_DC almost going down as low as 50% and also POP_DC increasing significantly. We kept running the DHARD ASC loops.
We stayed at this point for a while, but then it seemed that the DRMI dropped the lock. Before the lock loss, RMS of the SRCL control signal slowly kept increasing. It could be a coupling from CARM which was contaminating the SRCL control. We may want to decrease the SRCL UGF in order to prevent the SRM suspension from saturating.
Nice going, you are almost there. (The old grouch (for Lisa's benefit)