I was looking at the correlated spectra of the IMC WFS segments and minimizing the q-signals.
Attached: screenshot of new values and spectra.
I made a template to check this phasing, it's in userapps/ioo/h1/dtt/IMC_wfs_phase.xml
Also, we used to use LSC_EXTRA_AO_2 for putting excitations into the common mode or IMC board, that channel still exists but it is now used for JAC locking. So, instead we should use EXTRA_AO_3 from now on.