As I understood it the input to the 'complete unit', initially runs through the planetary gears to give the torque split, transferring the 'turns' from the input shaft to the outer case. The outer case is toothed to the plates of the viscous and the other plates are toothed to the output shaft.
I theorise that I am only welding the torque split part of the unit leaving the engagement part of the viscous to still rotate. it would only completely 'lock' to permanent 50/50 when the viscous fluid got hot enough.