The paper sheet feeding mechanism in my retired Canon Pixma MX340 multi-function inkjet is a mechanical marvel. There are details I didn't understand until a second or third look, like the spring that was installed on top of the big black gear driving the output cam. When I first saw it I thought it wasn't a big deal. It looks like a spring, probably there to absorb some kind of shock to the system. No big deal, except I was wrong. The real story was actually much more interesting: it's a freewheel mechanism.

My lesson came from a different part of the gearbox. This pair of gears were mounted on a static shaft that did not turn, so this pair exists to convey rotational power from one gear to the other. But if they were a straightforward direct coupling, they could have been injection molded as a single piece. Their multi-piece assembly hinted at something more, so I picked it up and started playing with it. I quickly found that I could rotate the top gear (as viewed in picture above) counterclockwise while holding the bottom gear static, but if I try turning the top gear clockwise they would lock up and turn together.

Aha! It's a way to ensure something only turns in one direction. A concept implemented several different ways (example) all inside this gearbox. I also knew this concept from the rear wheel of my bicycle, allowing me to coast without having to pedal in sync. I knew there had to be a term for this common concept but my search efforts came up empty. I ended up asking my friend Emily Velasco (who has bike hacking among her many talents) for help and she told me it's called a freewheel. Unlike the rear wheel hub of my bicycle, this pair of gears didn't make the clicky-clack noises of a ratchet mechanism. It was smooth and quiet so I had to see how they implemented it. Maybe it's like one of the illustrations on Wikipedia, a series of spring-loaded ball bearings all around the perimeter? Maybe a clever arrangement of many layers of friction material?

I popped the two gears apart and between them I found only a single metal coil. Wow. How did this work?

I first focused on this detail: both ends of this coil stuck out beyond the coil diameter. I believed it would allow them to smoothly coast along a surface in one direction, but dig in when moved in the opposite direction. A little bit of this dug-in force would expand the diameter for this coil, relaxing its grip on the inner cylinder (half from one gear and half from another) and allowing gears to turn independent of each other. As soon as the direction reverses, the coil contracts back down to its normal diameter and its grip keeps the two gears moving in sync. Torque transmission would have been limited by the friction of the metal coil against slick white plastic, but it's more than enough to resist my finger strength and evidently enough for this application.

On further examination, I changed my mind. I looked inside the coil housing for any marks of surface damage from coil ends digging in, and found it completely smooth to my eyes. Also, for long-term durability, it would make sense to avoid any mechanism that destroys the surface over time. Perhaps friction against the coil interior, without any wedge dig-in action, is enough for this design to work. [UPDATE: Indeed it is! Emily Velasco told me this is an example of capstan effect.] If so, the fact that the coil ends stuck out beyond its diameter may merely be an artifact of its manufacturing process.

One data point supporting the "friction is enough" hypothesis is the fact this coil is wound from thin metal wire with a square cross section instead of the typical round wire. This would help maximize contact surface area.

Another supporting data point can be found on the big black output gear, where one end (the <1cm length of metal stub) is held static at all times and the other end has nothing to dig into. This is enough to allow counter-clockwise (as viewed in this picture) rotation but resist clockwise rotation.

I found a slot for that metal stub on the gearbox lid that holds the stub in place.

Getting this functionality from a single precision-manufactured coil of metal is a feat of mechanical engineering and manufacturing that impressed and amazed me, and I almost missed it entirely. Given that, I'm sure there are many other details in this gearbox that has gone completely over my head without me noticing, because they are designed to priorities different from mine.

This teardown ran far longer than I originally thought it would. Click here to rewind back to where this adventure started.