As GP pointed out, the problem is that if the source is inside the can, the walls are near the source - where the EM field diverges - whereas with an EMP the source would be far away (hopefully) so the EM field lines would be roughly parallel over the surface of the cage