Some time ago I tried replacing the ball with a cylindrical magnet. IIRC it was 1mm dia x 1mm long. Didn't go well because the magnet was not constrained enough and tumbled in the cavity.

Having a magnet in there would allow for remote uncoupling using a simple electromagnet under the layout, a huge deal. The EM would provide a repulsion force upwards on the magnet. The EM would be much simpler than the KD electromagnet due to it providing vertical flux rather than transverse flux. Another advantage would be that the manual uncoupling tool could be a piece of iron rather than being magnetic itself and thus not get into mischief itself as mine frequently does.

Drawbacks might include the internal magnet binding if there were some ferrous material nearby like for example a steel car weight. Also it is likely that bad things would happen if passing over a KD between the rails magnet (somewhat unlikely amongst readers of this topic).

Since my experiment was crushed by the tumbing phenomenon I'm thinking that either a longer magnet is required, which may require a custom length, or a redesigned, longer cavity would be needed. If the latter then perhaps this scheme might be better suited for the brass version because that technique appears to be very precise and the  investment casting master shape may be easier to change than the die cast mold.

Anyways for those of you in contact with Erik please bring up the subject with him.

Thanks
George
Edgewood WA