To maximize resolution (and minimize stroke), we are also considering alternative magnet arrangements. I wrote a very simple (and slow!) 2D B field finite difference simulation, which is in <ahref='as5013-test/sim/'>sim</a>. It uses successive overrelaxation to compute the spatial distribution of the z component of the magnetic field. Below we should output of the simulation for a single magnet, as well as for a pair of opposing magnets. The question is whether we can create a region of high field gradient over each of the four sensors. I think a central rod magnet in one orientation surrounded by four with the opposite orientation would work nicely for this.
