Keita K, Khanh V, Jennie W
Summary: JAC angular control loops work now. We had some set backs this morning and eventually had to change some wiring at the IOT1 table but we have re-measured the phasing, re-measured the sensing matrix and now locked the 4 asc loops.
ASC attempt 1:
We tried to get our loops to converge this morning but we were seeing not much change in the loop outputs for large WFS gains (40000 eventually) and then eventually we could not see the two loops we tried (DOF1_P which feeds back to pitch and DOF2_P which feeds back to JM1) makign any improvement to the JAC transmitted power even if we changed the sign.
JAC Rephasing
Keita and I checked the JAC phasing with the template at userapps/ioo/h1/templates/dtt/JAC_WFS_phase.xml and after checking the phase of the TF between the JAC length servo error point and the wavefront sensor error signals he realised that some of these were phased with the wrong sign. I did not think to check these in our previous round of phase measurements last week.
JAC QPD A had wrong input matrix
As a check of the signs in the ASC loop we slewed the beam to different quadrants on WFS A and B using JM1 and realised that the degrees of freedom on WFS A were swapped. I found a mistake in the WFS A DC input matrix where pitch and yaw were swapped. This is now fixed and I double checked these input matrixes were correct for both WFS sensors and the WFS B DC screen.
JAC WFS quadrants 2 and 3 were swapped.
Using our previous phasing template JAC_WFS_phase.xml we used the picomotors in front of each wavefront sensor to check the wavefront sensor quadrant response to the 8Hz length signal. That is, we checked which wavefront sensor quadrant showed the highest magnitude response at 8Hz to the QPD quadrant the beam was in.
Quadrants 1 and 4 in each wavefront sensor show a maximum response to the beam being in quadrant 1 and 4 respectively. Quadrants 2 and 3 appear to be mis-matched between wavefront sensor and qpd. This holds for both WFS A and WFS B.
Keita and I went to the IOT1 table and unplugged WFS A segment 2, the channel H1:JAC-WFS_A_I2_INMON went to 0. The same was true for segment 3 and the channel H1:JAC-WFS_A_I3_INMON. This test was the same for WFS B also.
Since the table feedthrough was the easiest place to swap cables (it is hard to reach the back of the demod boards in the rack so we can't easily do it there) we went ahead and swapped them here.
Checks using the previous method with the 8Hz signal on the length and slewing the beam only into one quadrant of the WFS showed that this cable swap has fixed things.
I want to go back later to try and trace this DOF flip down, wondering if it might be in the jac simulink model or in the demod board connections.
After this swap we re-tuned the WFS phasing after recentering the WFS (with JAC locked). Then we re-measured the input matrix using the method from this (alog #91189), except we did not divide the matrix by the largest coefficient.
ASC filters changed
Keita and I checked over filters that Sheila had put in for PZT and JM1 feedback. We changed the poles at 0.8 and 0.85 HZ in pitch and yaw to be 0.83 Hz to match the zeros in the JM1 locking filters. We also tuned the gain of the exitsing filters to give a UGF of around 0.1 Hz.
After loading the new coefficients Khanh and I closed each degree of freedom in the ASC in turn. We kept tuning the gain up as the loops were converging very slowly. The final loop settings are shown here.
Here is the ndscope showing the loops converging and making the transmitted power higher.
