An Engineer in Wonderland – Roller clutches

I am currently driving to work as my bicycle has a fault which is in danger of becoming a saga.

roller-clutch.jpgI am currently driving to work as my bicycle has a fault which is in danger of becoming a saga.

A long long time ago far far away, I bought a back wheel.

And within the back wheel was a novel freewheel mechanism based on a roller clutch rather than the standard pawl system that has served cycling well for a century.

A roller clutch is a bit like a roller bearing: It has two concentric cylinders, one inside the other, and between them are metal rollers.

The cunning bit is that, unlike a roller bearing, ramps cut into the surface of one of the cylinders – one ramp per roller- make the gap between the cylinders vary slightly above and below the diameter of a roller.

As such, in one direction of rotation the rollers are jammed by the ramps onto the other cylinder where they consequentially transmit torque and rotation.

In the other direction the rollers roll freely around transmitting no torque or rotation.

Sometimes springs help the rollers engage.

All this I have learned trying to make my bike work.

The advantages of the roller clutch compared with a pawl type are twofold: silent operation and instant take-up.

The disadvantage, as far as I know, is lower torque capability for a given size.

And without spring assistance, my bicycle experience tells me, they also seem to be a bit dependent on the lubricant used.

After years of faultless operation, mine started to drag when it should have been free.

I don’t have the special and unobtainable Shimano FH-40 tool, and without this they appear to be impossible to dismantle – I have two mangled home made tools as Exhibits A and B.

So in the first instance I dribbled oil into every little gap and the thing started to work – except that it occasionally and unpredictably slipped when it should have been driving.

Back to the drawing board. Thicker grease?

Through its narrow gaps I washed it out with white spirit. At this point it seemed to be working perfectly.

I then sprayed in motorcycle chain lubricant – a sticky grease with a lot of solvent that allows it to flow in, then stick.

Initial results were promising but, I assume once the all solvent had gone, it now slips all the time and transmits no torque at all in either direction. Luckily for me, the walk hone wasn’t a long one.

I now have three choices:

  1. Wash it out and try another penetrating lubricant.
  2. Have go three at making a dismantling tool.
  3. Cheat, follow Shimano’s advice, and replace it.

Any suggestions?


If you wish to respond: do it below, or to



  1. Finally, after some more failed attempts, I took option 3 and bought another one.
    Then the opportunity to take it apart came when someone else with a heap of old cutting steel made up dismantling tool that would not fail under pressure.
    It came apart in a shower of tiny roller bearings.
    Inside it showed little sign of wear, so I suspect the correct lubrication would have sorted it.
    I lost enough of the rollers – ie, more than none – to make it not worth putting together again.
    The assembly is considerably more heavy that the standard pawl-based free-wheel mechanism, but swapping s impossible because the mounting is different – the roller-based design is much longer and mounts right inside the hub.

  2. I had a smilar problem with the pawl system that Shimano uses . I was not able to see Exhibit A and B so I would like to suggest the tools I use. I have used a beefy stubby long nose plier. Recently after looking on my damaged plier, I made a special tool which was made from a piece of 1 inch x 1/8 inch thick steel bar stock I had laying around.
    Cut off a 1 1/8 in long piece to make a 1 x 1 1/8 ‘washer’. Place the 1 1/8 side into the end of the free hub body so that it engages the 2 slots in the retaining race. For the handle I use a Crescent wrench, sometimes called by its generic name of adjustable wrench, which has been adjusted so that its jaws close on the 1/8 in thick washer. Turn or attempt to turn CLOCKWISE.
    It probably will be easier to leave the freehub body attached to the wheel while attempting to turn the bearing race.

  3. I suggest using a thin oil. It probably started missbehaving when the original lubricant got dirty and acted more like grease. The ramp is the secret to the operation so the rollers must be free to go up and down the ramps not be stuck in one place with grease.

Leave a Reply

Your email address will not be published. Required fields are marked *