[Yuta, Evan]
We have implemented a servo that uses the tip-tilts RM1 and RM2 to keep the REFL beam on the in-vacuum WFS.
By driving RM1 and RM2 in pitch with a sine (20 cts amplitude, ~2 Hz frequency) and monitoring the pitch signals from REFL_A_DC and REFL_B_DC, we were able to extract the following coupling matrix which takes RM1/RM2 pitch to REFL_A_DC/REFL_B_DC pitch:
| 0.097 | -0.0068 |
| 0.024 | 0.0079 |
From this we computed the following normalized inverse matrix:
| 1 | 0.86 |
| -0.24 | 1 |
We entered this matrix into ASC_INMATRIX_P_FULL as taking REFL_A_DC and REFL_B_DC to DC1 and DC2. We repeated the sine wave injections and found that the responses of DC1_P and DC2_P were largely orthogonal, indicating that this matrix gives good diagonalization.
We repeated the above for the yaw signals. The measured matrix is
| 0.055 | -0.0085 |
| 0.019 | 0.0092 |
and from this the appropriate normalized inverse is
| 1 | 0.92 |
| -0.34 | 1 |
We again verified that this gave good orthogonalization of DC1_Y and DC2_Y.
We then used the DC centering filter banks to make a loop. We use a pure integrator, constructed from a pole-zero stage at 0 and 0.05 Hz respectively, and a pole at 0.05 Hz. For DC1_P, we use a gain of 200; for DC1_Y, we use a gain of -200; and for DC2_P and DC2_Y, we use a gain of 4000. We measured the OLTFs of these loops and found that they had a UGF of 0.4 Hz, with at least 30 degrees of phase margin.