The plot below shows the sideband powers during lock acquisition and power up.

The top panels shows the sideband powers from the OSA scans. The middle plot shows the power transmitted by the X arm (I shaded part of it, since during the power increase the normalization of this channel is not quite right). The bottom panel shows the set point for the input laser power.

As a note, right before the input power increase, you can see that the arm transmitted power increases (750 s < time < 790 s) by about 1.5%. This corresponds  to the time when we are reducing the 45MHz modulation depth.

The reduction of transmitted power around t = 500 s is due to the engagement of the soft ASC loops, which are optimized for high power.

The second attached figure is a zoom in the vertical axis, to show that the 9 MHz sidebands are unbalanced even at 2 W.

The 45MHz sidebands change slightly in amplitude after power up, and their imbalance increases. The 9MHz sidebands instead show a large increase of amplitude at the antisymmetric port after power up (about a factor of two). We don't know if the 9MHz sidebands at the antisymmetric port are composed mainly of TEM00 mode or some high order modes. The fact that the transmission to the dark port increases seems to confirm that we have mostly differential lensing developing during power up. It would be useful to have an OMC scan to see what is the mode content of those sidebands.