The attached plots compare data of yesterday night and data of this night, after the work on alignment loops.

fig 1 - fig 6: in loop alignment error signals

fig 7 - fig 9: powers

fig 10 - fig 12: longitudinal error signals

The main improvement comes from the engagement of COMMp. The use of a better driving for DIFFp_TY (a combination of WI and WE, instead of WE only) makes a difference too; this driving did not work without the engagement of COMMp, due to an oscillation below 100 mHz. As a general remark, the increase of power and consequent increase of the optical spring makes the operations to reach the AA final state more critical. Considering that the power in the cavities has been increased only by a factor of 1.5, an hypothetical passage to the ADV final design would probably produce complications not easy to face.

A small remark: the peak at 50 mHz in fig 12 was due to a high gain of the slow loop wich controls the transversal position of PR.

Using B1p_DC (figure 7) as figure of merit for the quality of  angular control, we have to say that in some small region now the behavior is worsen. To be investigated.

Concerning the improvement of DARM (figure 10) we have to underline the reduction of the 9.8 peak (roll of the mirror). This is due to the engagement of NI and NE roll dampers, which is linked to the engagement of AA (COMMp) of north arm. At the same time the roll local control is moved to a less noisy filter, which should have reduced a bit the noise around 10 Hz. The last step of the low frequency optimization, the gain reduction of F7 TX/TZ dampers, has been tried, but it was clear that the gain used in june is too low, because the 0.415 Hz pendulum mode started to grow. This is probably one more side effect of the increase of power.