That makes sense. I agree that this is probably not cost-effective, but it could still be fun as a hobby project. > Also it wouldn't be easy to keep something

That makes sense. I agree that this is probably not cost-effective, but it could still be fun as a hobby project. > Also it wouldn't be easy to keep something in contact with the capstan to extract its energy, since the capstan/pinch-roller pair has been designed to work with tape, not pulleys or gears. The force could be extracted using a tape/ribbon similar to a real cassette tape, but one that loops around internal pulleys with a tensioner. It would be probably not cost effective, but I would still be curious of the maximal power draw you could get from the capstan. I suppose we know the tape speed (2.4 cm/s according to Wikipedia), but we also need the maximal force the capstan can excess on the tape. I wonder if there are some specifications for maximal friction forces within a cassette tape to be conforming, so we can get a good guess at this. 2.4 cm/s translates to 41.7 mN / mW (millinewton per milliwatt). This is not counting for losses for converting the mechanical energy. edit: The force requires seems to be excessive. The lowest power draw I can imagine is ~0.1W for a BL device like this (probably generous), and then we would need 4N pulling force on the tape for that (still not counting losses), which is a lot. Anyway, this was a fun exercise.