I wanted to see if it was possible to spin a crankshaft indefinity with nitinol coils. This is the 3rd attempt, with previous builds only able to complete ~1 rotation and were not autonomous. Integrating the fan into the design proved beneficial, as it allowed the coils to quickly relax.
The coils are controlled with large transistors driven by a quad channel opamp. The fpga outputs a lose pwm signal to the opamps which use a voltage divider as a reference. This allows control of the pulse length, intensity, and pattern to any degree desired.
With that, It can spin the crankshaft in perpetuality.
Nitinol is a unique shape memory alloy that can be commonly formed into coils to produce mechanical motion. The actuators can be very cheap and light, however it is difficult to get any large amounts of mechanical work out of them.
I am currently working on a hexopod driven by coils in a similar configuration.