Hi,
In original n-machine the brush for sure had to be a stator because the moment/place the conductor leaves the area of magnet's ring the direction of flux changes and reverses and so currents would flow towards the edge of magnet both inside and outside the circle of magnet.
On the other hand, steel is not a superconductor that pushesh field outside, to the opposite it attracts the field even more so much that if we enclose a coper wire in a thick pipe of magnetic steel then it will flow mostly in the steel all over around the pipe, almost none in the pipe.
So if we let a coil through a center of a rotation axis and unshielded up to the edge of coil and outer radius of the rotation, then on the very edge of magnet, to the pipe, around the magnet, still shielded back to the axis, under the magnet, then back again, again unshielded, a couple of times and finally drive the ends of wires of the coil to separate ends of rotation axis on to two insolated (from each other) brushes then shouldn't we get the normal volatage from lorenz law (faraday disk/n-machine) multiplied buy the number of turns of the coil?
I have a model and tried to run it in comsol but it returns Inf/NaN in this case for computations. I can share a model, maybe someone will have more luck. Maybe someone may want to try it in some other magneto-electric computation tool?