Chainrings are sold as having a specific compatibility for only certain speed setups, some overlapping across many. What physically makes this the case?

From what I understand, 5/6/7/8/9 speed chains all have the same internal width, 3/32", and the only difference is the external width of the chain. Does this also extend through 10/11 speeds as well, or do they have smaller internal widths which would necessitate skinnier chain rings as well? Or is it the case that the external width just keeps shrinking?

IF the difference is only the external width, then it would seem the only compatibility issues would not be in the chainrings themselves, but the spacing in between each one. If it was too small/too wide a chain, the chain would hit the next chainring, and if it was significantly too large the chain may fall in between rings.

However, it would seem that you could use the same rings with a wider chain (9 speed rings with a 5/6/7/8 speed chain) if you put appropriate spacers between the rings, such as these. The opposite may be possible if you're willing to grid / mill the appropriate amount off the mounting pad.

What this would also imply is that with the right width spacers (and bolts that can accomodate them) you could always go down speeds with a given set of chainrings.

If that is not the case (say 11 speeds have a different internal width), then they might be too wide for the chain an you'd have to remove material near the teeth to make them fit inside the chain or conversely that you'd be wearing on less of the chain than you should so your chain wouldn't last as long.

Furthermore, some rings are marketed as 5/6/7/8 speed only, some 9 speed only, and some 5/6/7/8/9 compatible. Is this some sort of blending of the two specifications (the spacing being some compromise between true 8/9 speeds) or just that they're made for one but tested to work on the other (or just a marketing trick)?

  • Or maybe the spacing is set by the cranks?
    – Ehryk
    May 7, 2012 at 5:28

1 Answer 1


The internal width is obviously one important feature. But for indexed shifting the ring has a pattern embossed on the side facing the smaller ring that catches the pins of the chain when shifting. The combo of the embossing depth/pattern and the ring-to-ring spacing affects what width chain (and brand chain) will work best with the ring.

I've no idea what combos can be expected to work with what, though. I lost track when they went to 8 speeds, and now it's a total free-for-all.

  • Hmm, then I think that leads me to another question about the differences in 7/8/9/10/11 speed chains then. Would it be possible that if the pin diameter and 'stickout length' are the same, you could use the same chain-lift groove design, and the difference would then collapse down to just differences in spacing? If this is not the case, then perhaps the grooves are specific to the individual pin diameters and 'stickout length.'
    – Ehryk
    May 7, 2012 at 18:02
  • It's possible. But there's a certain "art" to the whole indexed shifting thing, and making it work smoothly. I'm sure hardly any dimension (other than the basic cog profile) is a product of pure mathematics, but rather they've all been "artistically" fudged until they work. For instance, as the chain rides up on the grooves it will twist slightly, and thinner chains will likely twist more. Compensating for the twist would be a part of the "art". May 7, 2012 at 20:06
  • Bah, I detest all that art nonsense. I'm sure it can be modeled mathematically somehow. This would be why I pursued a degree in mathematics and not art. (The math equivalent of 'getting a bigger hammer' is 'adding some more dimensions'). Thanks for the insight though, I hadn't considered the lift patterns/grooves.
    – Ehryk
    May 7, 2012 at 22:07
  • 1
    Ehryk: yes, but as with much engineering the gap between "the maths that correctly and completely describes the situation" and "what we can actually do" is substantial. It's like the trip from Newtonian physics to Heisenberg's uncertainty principle. Eg, think about the variables for a single joint in the chain. Manufacturing tolerances on every part, lubrication, temperature (and variation across parts), force + torque vectors on the 4 "rigid" parts involved etc. You end up with so many interdependent variables that the problem is intractable.
    – Kohi
    May 7, 2012 at 23:51
  • Yep, and consider the dynamics. Have you ever watched a rear derailer shift at high speed? The chain "dances". (It's really quite artistic.) Yes, it can probably be modeled, but even if it can and were worthwhile for the component designers to do so, the parameters would still be inaccessible to you and me. May 7, 2012 at 23:57

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.