Generator hub design - why clawpole?


1 µW
Sep 28, 2022
TL;DR: generator (dynamo) hubs use a super old arrangement called a "clawpole generator" and are way more inefficient than you'd expect, annoyed local attempting to understand why.

I've been looking at generator hubs (a.k.a. dynamo hubs) for a bit, and I've noticed that the measured efficiencies of these hubs are generally quite low - a 50% capture ratio of mechanical to electrical energy is common. Here's an efficiency study that I've been referencing (apparently from a Blogspot user called Skjegg, repackaged in English by CyclingAbout).

This got me thinking about the design of the generators themselves- they basically all appear to be clawpole generators, that is, they have a single central winding, with claw shaped flux guides that route the magnetic flux to the circumference of the stator. With the stator consisting of a single winding and a couple of stamped pole pieces, I can see the value engineering side of the equation, but surely someone would be trying to improve on the dismal 50% efficiency?

I have a couple of naive theories about the current state of the art:
  • Clawpole generators are cheap and fast to make, and nobody is using their wallets to vote for advances in efficiency
  • The German StVZO standards for bike lighting require that the generators not exceed 6V/3W at a nominal speed, and so generator manufacturers handicap their units to prevent them from exceeding their design output at higher speeds
  • The problem of cramming a generator into the space available in a bike hubshell isn't conducive to a more performant design, the state of the art is actually as efficient as we can get given the mechanical constraints, low rotor speed, small machine radius etc.

Anyone have insight on why the clawpole generator is still on top? Why not use a multi-winding radial flux arrangement as commonly seen in modern BLDC motors?


Staff member
Aug 17, 2009
Phoenix, AZ, USA, Earth, Sol, Local Bubble, Orion
Almost certainly is "money".

If it works well enough, and nobody is "asking" for something else....

A good active self-adaptive design (with electronics) would be significantly more complex and expensive (including costs of failures over warranty lifetimes, QC, etc).