Yes the surfaces are creating angles, but the angles are not sharp. The bad cells are located in the free space of the volumes too, which i dont understand.
Here i used element quality criteria.

i will try this mesh and see if the solution will converge.