Modelling aurorae, we asked what boundary condition (BC) to use for the E field on the upper boundary. Typically a Dirichlet BC is used, since processes above the domain generate E. But then conductivity structures trigger FACs driven immediately by magnetospheric convection, even though it is a finite energy source, delayed by the Alfvén speed. If the BC is not ideal, then E x B drift in the ionosphere depends on the plasma's properties. So we investigated.