A Look Inside Bicycle Gearboxes

A line art schematic of a bicycle CVT drive. Two large green circles at the bottom have the text "1. Increases speed" where the crank arm would enter the system. A series of cam arms highlighted in red say "2. Converts from rotary to reciprocating motion." Finally, a blue highlighted bearing says "3. Converts from reciprocating back to rotary motion."

While bicycle gearboxes date back to at least the 1920s, they’re relatively unseen in bike racing. One exception is Honda’s race-winning mid-drive gearboxes, and [Alee Denham] gives us a look at what makes these unique drives tick.

Honda has developed three generations of bicycle gearbox as part of their company’s R&D efforts, but none have ever been released as a commercial product. Designed as a way for their engineers to stretch their mental muscles, the gearboxes were only used in bike races and seen at a few trade shows. In 2004, the third gen “derailleur in a box” led to the first gearbox victory in the Downhill World Cup Circuit.

The third gen gearbox differs significantly from the CVT drivetrains in the first and second generation gearboxes, but it is unclear why Honda abandoned the CVT. [Denham] has a nice animation detailing the inner workings of these CVTs based on information from the original patents for these rarely seen gearboxes.

Derailleurs remain the primary drivetrain in racing due to their lighter weight and higher overall efficiency. While still expensive, the decreased maintenance of gearbox drivetrains make a lot of sense for more mundane cycling tasks like commuting or hauling cargo, but only time will tell if the derailleur can be supplanted on the track and trail.

For more on bicycle drivetrains, check out this chainless digital drivetrain or the pros and cons of e-bike conversions.


23 thoughts on “A Look Inside Bicycle Gearboxes

  1. Honda’s entries were all in pure downhill racing, where bike weight doesn’t matter and pedaling efficiency almost doesn’t matter as you’re only pedaling for maybe 5-10% of the race.
    There are a fair number of gearbox shifters out there. Rohloff has been making fantastic ones for probably 20 years, and people doing epic offroad races like the Tour Divide use them because they won’t gack a derailleur on a rock 50km from the nearest road. But they also cost as much as a kinda ratty used car.
    Similarly, the NuVinci rear hub is a really amazing piece of engineering, an infinitely variable drive relying on changing the rotational inclination of some bearing balls. Its operation is not obvious, but it works quite well… except that it slips if you push too hard because it relies on static friction to transfer torque.
    And both the Rohloff and NuVinci are *heavy*. The Rohloff hub weighs more than my race bike’s frame, fork, stem, and handlebars combined, 8x what my rear hub weighs. For a commuter or cargo bike that doesn’t matter, but unfortunately for all of us, consumer bike design is primarily driven by how close they are to full race bikes, unless you go ahead and commit to a cargo bike or comfort bike.
    I think that conventional gear-to-gear gearboxes buried in the bottom bracket are the somewhat distant future of at least mountain bikes. But it’s going to be a long time before they’re competitive, if ever.

    1. And of course Sturmey-Archer has been at this for longer than any HaD reader has been alive, with their hubs remaining serviceable for nearly as long, though I suppose the jury’s still out on the aluminum-shelled ones of the last twenty years under sunrace.

    2. My bike has a Nuvinci Enviolo rear hub. I’ve been riding it for 2 years and it has never slipped. It doesn’t rely on friction to transfer drive energy. It uses a non newtonian fluid as a “lubricant” that turns solid when subjected to the pressure between the rotating balls and rings inside the hub. The hub is sealed so there’s zero maintenance, and my bikes version has a 380% gear range. It’s main drawback is the weight- it’s heavy due to the 8 steel balls and input and output rings. The bike (Priority Continuum Onyx) has a belt drive, too, which has also been zero maintenance for 2 years, and still going strong. It’s silent, clean, and reliable.

      The Nuvinci hubs are not as efficient as a chain/derailleur drive, especially as you get to extremes of the gear range, so probably not for racing, but for commuting and touring, it’s hard to beat. The only noise comes from the freewheel ratchet in the rear wheel.

        1. A shaft-drive that does not also act as the rear swingarm (as with many shaft-drive motorcycles) seems a missed opportunity for mass reduction in a system that’s otherwise adding more unsprung mass.

      1. It’s also substantially less energy efficient. Deforming a belt takes considerable work, not to mention the fluid work. And yes, weight matters too. This stuff stacks up.

        A properly waxed chain can sap less than 0.5% of the energy being put into it, compare that to your cvt.

        380%? My 1-by MTB manages 520%…

        Now I won’t dispute the maintenance benefits, but the effeciency is a big deal – how far you can go, how long it takes, and how sweaty you are at the end of it.

        I *like* CVT tech. But I also understand why we don’t use it on most bikes.

        1. I dont know about that but chain drives have ben used to transfer power is some super high powered application like industrial equipment and 4×4 transfer cases forever. If your 750w drive is eating chains and clusters then the engineering sucks.

          It is always going to be really hard to beat the derailuere system because it imposes hardly any additional drag over a single chain sprocket drive and is very light weight. Parasitic drag and weight are two main things a bike tries to avoid.

  2. I wonder if the Gen-1/2 units had a tiny bit of backlash with each stroke of the reciprocating mechanism that means it always slips by some amount.

    $100,000+ does seem quite expensive, I assume that’s for the entire program and includes the development costs rather than per-unit.

  3. It was maybe 40 years ago that I realized that you could turn rotary motion into reciprocal motion (using a crank), and then turn this back into rotary motion (using a rack & pinion and a one-way ratchet), and be able to vary the output speed continuously from 0 by controlling the amount of offset of the crank pin.

    However, at the time, it just seemed too complicated and rife with issues that I just put my paper drawings away and forgot about it, only to see the same idea pop up now and again as time goes on.

    1. I enjoy thinking about transmissions from time to time too. The fundamental issue is that gear ratios work using integer math, and any attempt to make a continuously variable or automatic device requires an unreliable departure to belts or hydraulics. The easiest way around the integer problem is to use a secondary powerplant to take over while the other is changing gear.
      I am unsure if toyota’s prius uses this approach, but i do know it uses a motor and engine at the same time sometimes.

      1. Toyota has had various approaches over the years, but a key thing that they’ve always done is to use a planetary gear system in combination with a “small” electric motor.

        In a normal planetary gear setup in a regular automatic transmission, one part or another is held stationary by a brake band, thus choosing between a couple different gear ratios.

        In Toyota’s system, the electric motor is used to control the speed of one component, thereby producing a variable ratio between the other two components based on how fast that motor is turning. The motor can also act as a brake, preventing its controlled component from turning. It can also turn in reverse for additional gear range, even allowing for reversing the drive.

        Since Toyota’s hybrid system involves a gas engine as well as a large electric motor, there are many permutations for how it is driven beyond just one input and one output. Most of their setups are more complicated, with dual planetary gear systems.

        But such a hybrid system could work well for a bike, with an electric motor that could just control the pedal’s gear ratio, or could drive the bike as well if the rider stops pedaling, or assist by providing more output than the rider is giving. With a planetary setup, it seems like it could also offer a no-resistance pedaling mode if the battery dies, but you’d also lose the ability to change drive ratios in that case. (Normal electric bikes add some pedaling resistance when not powered.)

    2. If you change rotary motions into reciprocating motion and back again you are probably not being efficient. Reciprocating motion comes with accelerating and decelerating a mass in the opposite direction where rotary motion preserves power in the form of momentum.

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.