Being the geek that I am, I decided that the revolving magnet was
- Too imprecise
- Too environmentally effective
Basically both of these came from the use of a single magnet. It would take an average of half a revolution to register the fact the wheel was moving (actually in practice it would be half a turn, more later) and a turn later before a speed measurement could be made. At the end of the running episode the hamster could have run nearly a whole turn without it being measured. This could be solved by mounting the sensor at the "six o'clock" position.
The other problem comes from the fact that the wheel is now statically (and dynamically) off balance. This probably wouldn't have the effect of trying to shake the wheel apart as she is unlikely to hit resonant frequency, but it would make it harder for her to run as the magnet went up, and could cause the wheel to "overrun" her as it came down. It would also ensure the wheel was 180
In true to form fashion, I chose to over-engineer the solution. I picked an idea from my robotics stuff and decided to make an "optical quadrature shaft encoder". "Wotizit" you might ask, well, it is a method of measuring the speed and direction of rotation of a wheel attached to an axle (the shaft) using light.