Sheila, Jenne, Nergis, Daniel, Stefan

Most of the things we tried today we tired today were goose chases, but here are the things that we left in:

- We moved SR3 mack to where is was before we turned on the cage servo.
- We (for now) switched back from the cage servo to the OPLEV servo.
- We switched back to the old SRC1 yaw input matrix:
   AS_A_RF36_I_YAW to SRC1_Y : -3
   AS_B_RF36_I_YAW to SRC1_Y :  1
  That's the matrix for which we had done the careful SRCL FF tuning.
- We scratched our heads for half a day about why the MICH_P loop goes unstable at high power.
- We then threw our hands in the air, looked at what other signals see MICH_P (i.e. BS pitch motion) - turns out to be AS_A_RF36_I, with a gain that is roughly -3/2 of that one in AS_B_RF36_Q (at least at low power)
- We punched in an input matrix that includes both error signals: 
   AS_A_RF36_I_PIT to MICH_P : -0.666 (remember that AS_A was never phased for PIT, so don't wonder too much about the I...)
   AS_B_RF36_Q_PIT to MICH_P :  1.0
  With those signals we were able to hold lock at 23W, although we still could improve the ASAIR_90 buildup by slightly offsetting the error point
- This also significantly reduced the drop in recycling signals right after power-up.
- After about 30min at high power, we stepped the MICH_P offsets by 100cts, and noticed that there was no signals left in AS_B_RF36_Q_PIT - no surprise that we lost lock on this signal before...
- After 3h of locking I revisited the MICH_P offsets, and found that no offset, but completely switching to AS_A_RF36_I_PIT gave us the best performance:
   AS_A_RF36_I_PIT to MICH_P : -1.333
   AS_B_RF36_Q_PIT to MICH_P :  0

After than we damped some violins (see above), and turned on whitening.

All this stuff is not in the Guardian yet - to be done tomorrow.