Building An Automatic Bicycle Transmission In A Week


Every year, the ECE department of Carnigie Mellon University hosts Build18, an engineering festival intended to get students out of the classroom and into the workshop. [Andrew Toth] along with team members [Jenna MacCarley], [Peter McHale], and [Nicolas Mellis] have been busy this last week putting together an automatic bicycle transmission.

Most cyclists agree that a cadence of 80 RPM is just about right for most cycling. The team’s transmission uses Hall effect sensors to sense the cadence of the rider and will change to a higher gear if the cadence drops below 60 RPM and a lower gear if the cadence is above 100 RPM.

One of the requirements of the Build18 festival is the completed project must cost less than $250. By using an Arduino Mega and a servo to change gears, the team has a fairly low cost solution to automatically changing bicycle gears.

It’s a very cool project, and hopefully we’ll see a video once the competition is over at noon, EST today.

58 thoughts on “Building An Automatic Bicycle Transmission In A Week

  1. “The team’s transmission uses Hall effect sensors to sense the cadence of the rider and will change to a higher gear if the cadence drops below 60 RPM and a lower gear if the cadence is above 100 RPM.”

    I think that should be the other way round…

    1. It’s just a convenient AVR development platform; *which* Arduino they used doesn’t really matter, if it were built into a final product they would surely make a custom board.

  2. But when you change gears on a bike, you have to temporarily stop pedaling or pedal slower than neccessary. I wonder if it has a warning alarm for that. I don’t think I would like this on my bike at all.

    1. It would probably be unpleasant to use, and cause extra wear and possibly damage to the chain since shifting and pedaling won’t be coordinated.

      However, it looks like a fun project to build. Maybe the creator will decide it is not a practical way to operate the gears on a bike and take it off after a few days, but he will probably learn something in the process, and maybe some other fun project idea will come out of it.

    2. I’ve seen this project before I think and I thought the same thing. Chain doesn’t mesh well with teeth when there’s no change in torque on the chain. I was thinking the best remedy for this would be to mechanically separate the rear wheel from the rear cassette with a fluid coupling of some kind. Not sure

    3. When Jo says “But when you change gears on a bike, you have to temporarily stop pedaling or pedal slower than necessary.” [b]THIS IS NOT TRUE[/b]. The way a bikes rear and front derailleurs shift requires the chain to be moving at a decent speed to change gears in any direction.

      1. Anyone who’s ridden a bike knows what he means. You can’t be applying torque while shifting, so you either stop “pedaling” or you “pedal slower than necessary” in the sense that you’re no longer putting energy into the drivetrain. It’s a pretty bad feeling when the chain starts shifting on it’s own while you are biking along. Again, most anyone who’s ridden enough bikes will at some point encounter the same thing when the derailleurs come out of adjustment.

        1. If you have to change your pedaling when you shift, you were in the wrong gear to begin with. There should be almost no pressure on the pedals when you’re in the right gear and you can shift seamlessly, assuming you’re only going one gear up or down.

          Of course, if you’re riding a Huffy, that might be a different story… all bets are off there.

          1. As the people who ride bikes have pointed out, you need to easy up a little to relieve the tension on the chain before shifting.

            The old advice has always been to stop pedalling for a brief fraction of a second, it’s easy enough to do when a pedal is at the bottom of a stroke.

            Generally speaking it happens more when shifting to a lower gear (going uphill rather than down) as the chain tension will be higher.

            If you have “almost no pressure on the pedals” then you’re probably not moving.

  3. Is the mega even needed? Just go with an ATMEGA328, 2 caps and a crystal – then add the servos. If the servos yank too much juice add some transistors and run it directly from the battery? Depending on the type of gear one could also play around with interfacing the transmission directly in the box.

    Very nice project, i could use something like it myself on my foldable.

    1. – agreed, seems like overkill, but probably what was available and easy to work with. -Could probably use an ATTiny and might even be able to cram it into a servo case, or a slightly extended one.

  4. I have a feeling that using servos to drive a spring loaded mechanism is a very bad idea. They will constantly draw a high current and probably will have a short life. A self locking linear actuator would work better in this application.

    1. You could actually hack the servos a bit: use some strong enough to prevent the springs from changing position even while unpowered. The the micro culould also use a transsitor to cut the power to the servos to save power.
      Im offering this solution because servos big enough will probably be cheaper and easier to find compared to actuators…

    1. The problem is efficiency. Humans make about a tenth of a horsepower sustained, and the the typical chain & derailleur system makes the most efficient use of that power. Belts aren’t terribly efficient (v-belts especially so), and even shaft drive isn’t quite as good.

    1. Seems like racing would be the last place you’d want the system as envisioned. A racer needs to be able to shift gears in a heartbeat for higher or lower RPMs to suit race conditions – a system that was trying to maintain 80 RPM would work against that.

      Fast electronic shifting could be great for racing, but I’m scratching my head a little as to why they’d pick a target RPM and actually make it an automatic transmission in that sense.

  5. Re-inventing the wheel………

    Nuvinci internal cvt bicycle hub.

    Shimano DI2, an electronic shift system on the market currently which is sensitive enough to automatically trim the shifting and has cadence sensors, easily a software upgrade could make it all automatic. There are systems also for internal gear hubs with 7-8 gears coming. Plus these parts are race proven.

  6. Seriously? Electronics? I would think this could easily be controlled by some type of centrifugal force mechanism.

    Or even better, a centrifugal force mechanism that worked a CVT type of transmission arrangement… would probably be far less shock to the system and keep you pretty damn close to an ideal pace all the time, it would likely take less shock loading, though, so you might not be able to pull wheelies and whatnot.

  7. Admittedly, we could’ve been a lot more cool with the mechanical part, but all of us (on the team) are ECE (Electrical and Computer Engineering) majors, so while we had mechanical ideas, no one had the experience on how to design or manufacture it.
    Also, while there are many other and arguably better ways to do this, we only had a week. On top of going to classes and such.

    1. It’s always good to get a project up on its feet and out there for people to see, so you all should feel good about that.

      At this point you might as well go whole-hog on the automatic shifting and just program it to shift as needed to wear the rider out as fast as possible. :)

  8. I would not ride such a beast. Random shifts mean trouble! I commend the effort and the ingenuity, but not the product. Now, a continuously variable bike transmission would be pretty exciting, like the one made by that aussie with planetary gears.

  9. As many posters have stated before, you have to remove torque from the pedals(pedal softer) while maintaining pedal speed.
    Much the same way you let off the gas when upshifting in a car.

    Why not just make the system a manualy-controlled automatic transmission?
    As in the electronics are there to shift the gears but YOU control when the shifting happens by hitting one of two buttons on the handlebars.

    An automatic transmission only works when it has control off, or can at least sense, torque being applied.

  10. I have a mountain bike that shifts gears on its own. It typically happens when I am pedaling it really hard. The jarring to the knees is really extreme and damaging to them.

    So a system that automatically shifts must somehow avoid that.

    I much prefer concepts where I control the shifting, but it is done via servo. An auto option might be cool.

  11. Why would anyone want to put something on a bicycle that has almost no durability? Efficiency means nothing if it breaks within a year…

    To all people including myself who want mechanical auto for the durability:

    Companies have been doing those hubs for a long time, seems you could make it work with a lot of gears like 18 or 21…

  12. New transmission CVT design by BitRaptor. Is a continuously variable transmission CVT gear only (the only one functional in the world), very compact and lightweight, and which could replace the current systems both for efficiency, simplicity and not least the costs.

    Because this CVT work only with pinions is better the all other systems by efficiency and high torque transmission.

    In the web page you will find more explanations, drawings and a short video of a basic prototype.

    The first prototype will be ready for tests during this year.

    Other possible applications CVT design by BitRaptor are: bicycles, motorcycles, cars, automobils, boats, gearboxes, electric motors, steam and wind turbine, as well as a large number of industrial or agriculture applications whenever is necessary some adjustment of the gear ratio.

  13. I’m just curious if it is possible for a student to design an automatic gear transmission in bicycle which shifts gear in both front and rear derailleur?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.