Example here:

Does this mean that there is a manufacturing defect in the spindle/BB/crank? Or damage? Or is it just the nature of these things that there will be one best orientation and in the other positions there will always be some amount of chainring wobble?

If it is a 'damage' thing, how could this damage be caused?

  • I'm wondering how much of the fix is just due to removing the crank arm and reseating it. Since he only seats it using a rubber mallet, and not by using the bolt, it could just be coincidence that it is more square in some positions that others. There could be some issues with manufacturing tolerances that cause it not to sit flush and wobble, but it seems odd that rotating it would make a difference since it seems like the wobble would just move to a different park of the chainring.
    – Kibbee
    Feb 17, 2022 at 21:22
  • I've read it other places too, so it's not just RJ rotating the crank and hitting it with a hammer, other people have had success with this technique too by fully belting on the cranks in a new position.
    – Wilskt
    Feb 17, 2022 at 21:35
  • My guess is that some wobbles may have a compound cause, being that the square hole is somewhat skewed and also the spindle is bent or worn out asymmetrically. By rotating the crank relative to the spindle the two deformations cancel out. But it may also be possible that it rather increases wobble too, that is, this is not an always true solution, it's kind of a gamble.
    – Jahaziel
    Feb 17, 2022 at 22:06
  • ^^^ Bolting, not belting.
    – Wilskt
    Feb 18, 2022 at 0:05

1 Answer 1


The spindle is trying to be perfectly straight with the crank interface surfaces all aligned perfectly to the center. The plane of the spider is trying to be perfectly flat and perfectly perpendicular to the spindle interface (the taper bore on a square taper crank). The chainring is trying to be perfectly flat, with perfect alignment between the plane of the tooth centerline and the plane of the spider interface.

Cranks and bottom brackets of other designs that have some elements different will have their own set of tolerance-sensitive areas.

None of the above will ever be completely perfect because nothing is. This is true regardless of whether it's damaged or new, but damage can make it worse. So, for example, if you were to measure the total vertical runout of the spindle, there will always be a worst spot. Same with measuring the plane of the spider's mounting surface for the worst amount of horizontal runout compared to the imaginary line running through the taper bore.

Testing each position works because in some positions the runout present in each element will stack up to the most pronounced total runout possible with that combination of parts, and in others it will do a better job of canceling itself out. That is always true regardless of how perceptible the difference is.

Acquired damage that causes chainring wobble usually comes in the form of damage from side-loading or impact that distorts either the chainring or the spider tabs. Most metal cranks can have the spider tabs trued some amount safely, although it's possible to imagine an exception. I've usually done it by bolting a cone wrench or other flat tool to the lower bottle cage bolt to make an indicator and using a 6" adjustable on the tab (with chainrings off). Low-quality cranks on cheap fixed gears is a good application for this (especially since they're made of playdoh-like weak, ductile aluminum).

Most of the time, with good quality parts in good condition there is no need to try the different positions. If none of the parts are overtly damaged, than the benefits gained from it will be subtle, and if parts are overtly damaged they should be repaired or replaced. But when working with poor quality parts and/or in situations where you've got more time than money, then trying the different positions on for example a riveted triple crank can easily make the difference between the worst runout spot causing unsolvable chain/FD rub or being pretty good.

  • 1
    Wow that's a great answer, thank you, so one position will sum all of the errors closer to zero than the other positions.
    – Wilskt
    Feb 17, 2022 at 23:09
  • 1
    Yes. Barnett's Manual has a good section on this topic. I haven't read it in a long time but most of the above is informed by using its guidance. Feb 18, 2022 at 1:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.