I've read all the differences between PF30, BB30, PF86/92, etc. I understand that it's easier to make carbon frames with press-fit than bonding in a threaded shell, and you save some weight, but I don't understand how we ended up with so many different standards while BSA has always been BSA. Does anyone know anything about the history of press-fit and why it diverged so much?

  • 3
    I don't know much about the press fit standards, but "BSA has always been BSA" is really an oversimplification. BSA really only became "the standard" after a long time of competing BB standards (italian, french, swiss, etc). Instead of saying the PF has "diverged", maybe we can hope for them to "converge" some day. Probably not, but we can hope. Commented Feb 22 at 17:45

2 Answers 2


All else equal and up to a point, larger ball bearings and more widely spaced bearings will be more durable. Juhist alludes to this in his answer. Keep in mind that diminishing returns-type relationships are common in many settings (economics, business, sports science), and that there are other constraints (e.g. physical ones like the crank and chainstays), so we don't have to get the bearings infinitely big and the spacing infinitely wide.

Those facts drove at least some of the evolution of the press fit standards. BB30 was the first, pioneered by Cannondale, although I think the main advantage of BB30 was the possible weight reductions from eliminating the bearing cups and allowing an oversized aluminum axle. Note that BB30 has bearings pressed directly into the frame. I think PF30 was the second press fit standard, which has slightly bigger bearings than BB30. Both are relatively narrowly spaced - although you can design the crank and the BB such that the bearings sit outside the shell, which increases the spacing, subject to other physical constraints.

In addition to the bearing durability concerns, PF30 houses the bearings in hard plastic cups (or rather, it was designed with this in mind; you can and do find metal PF30 cups). I believe this may have let frame manufacturers use looser the tolerances on the BB shell, because the plastic cups can deform somewhat (looser tolerances = more allowed variation from the nominal standard = lower manufacturing costs, less QC rejections).

Both those BB designs anticipated people using 30mm aluminum spindles, although you can of course design your BB to take a 24, 25, or some other spindle diameter. Shimano's PF86 design was one of the next designs, and it improved the bearing size and spacing. However, because Shimano has only ever made 24mm steel spindles, the PF86 design wasn't designed for 30mm spindles. I know that PF86 BBs for 30mm or DUB spindles exist, but I think that they are less durable in that format. Trek's BB90 standard is going away, and it's only suitable for 24mm spindles. Anyway, to some extent, the follow-on PF standards probably were trying to improve on some aspect of BB30 and PF30.

Speaking of the DUB axle standard, SRAM tried a 28.99mm axle precisely to increase ball bearing size. I don't know if they really couldn't get away with 0.01mm wider to make it a nominal 29mm, so the .99 bit may be marketing.

Kogel Bearings says that Cervelo's BBRight standard was created to allow bigger chainstays. I'm not sure how the geometry works.

There are some minor standards out there. For example, Colnago may have had one at one point. Some of them may be proprietary, some of them might be technically open (like BBRight) but only used by one company. Whatever the engineering justifications behind them, they would serve to tie customers to the brand - only to be immensely frustrated if the brand later dropped the standard (Colnago) or if the standard had problems with durability or compatibility (Trek BB90). If you're going with your own standard, you had better think the engineering aspects through well and you had better have enough market share to be sustainable. Shimano seems to have done this, less the issues with >25mm spindles.

At the time of writing, the bike industry does not seem to be seeing new standards introduced, which many would argue is a good thing. We are seeing a number of minor variants on the T47 threaded standard, discussed below. It seems like the bicycle equipment has equilibrated at a point where we don’t need increases in bearing size or stance width - also, road cyclists tend to prefer narrow stance widths and they might complain if stance widths grew.

Obviously the post asked about press fit systems. However, we should mention T47. This is an open standard for an oversized threaded shell. I think that it uses the PF30 dimensions, only it’s threaded. Some mainstream manufacturers signed on to this. One is Trek, whose BB90 system wasn’t well received and wasn’t compatible with DUB axles. Another is Cervelo. T47 has a number of variations. Trek uses a slight variation in the shell width to get better tool grip on the cups. Cervelo has an internal cup on one side and an external cup on the other, similar to its press fit BBRight standard.

The obvious engineering intent is to allow for oversized BB shells that can accommodate internal routing and larger/more widely spaced bearings than threaded shells. BSA can actually accommodate external cups for 30mm axles, but I think that BSA shells in general may not have enough room for manufacturers to route the rear brake hose through the shell. Custom metal bikes may use T47 or BSA.

With carbon frames, T47 means you have to bond an aluminum shell into the frame. Aluminum and carbon have different coefficients of thermal expansion, which must be considered in manufacturing. Aluminum shells can disbond from the carbon in the worst case.

  • 1
    +1, the one major puzzle piece not mentioned is that in the Shimano paradigm, it's important to have your spindle ride on plastic to prevent creaks and wear at the dynamic surface that occurs in these designs at the spindle bearing contact area. Commented Feb 22 at 19:27
  • 3
    Regarding the ‘reduce tolerances’ thing, the usual verb I hear used for that meaning is ‘loosen’ (likely because ‘looser’ tolerances are less stringent and more forgiving of variation). Commented Feb 23 at 13:00

I think the idea is as follows.

Long time ago, bottom bracket spindles were not hollow inside. This created a heavy design and the spindles were too weak anyway.

Then, Shimano tried Octalink bottom brackets where the spindle was hollow inside. This proved to be a durable spindle design. Unfortunately, this left so little space for the bearings (given that BSA bottom brackets don't have much radial space), that the bearings failed early. Also, the Octalink crank-to-spindle attachment was flawed, leading to loosening crank bolts if the rider had a habit of standing on the pedals left pedal and crank in rear position, right pedal and crank in forward position, which created alternating reverse and forwards torque on the spindle.

Then Shimano introduced Hollowtech II. It made the bearings outboard since the bottom bracket shell had limited space. These outboard bearings allowed using a durable hollow lightweight spindle with big enough bearings.

So the BSA shell is too small. All of the bottom bracket standard designers for press fit realized that, and wanted to make the press fit shells bigger.

Unfortunately, there was no agreement what the crank spindle diameter should be. Maybe somebody wants to build a crank spindle out of aluminum (in which case it needs to be very big, much bigger than steel spindles). Maybe somebody else wants to make it out of steel but make it bigger than with Hollowtech II to further allow weight reduction while still maintaining the durability.

So all of the bottom bracket standard designers had the knowledge the BSA shell was too little, spindles need to be bigger than with square taper and therefore the shell needs to be big to have big enough space for inboard bearings as well (which are more ideal than outboard bearings which can be considered to be a hack to allow using the flawed BSA shell). But apparently they didn't have an agreement of what the ideal spindle diameter should be. If you look at the list of the different standards, some of the entries in that list tell a crank spindle diameter for which the shell is intended. These spindle diameters have a difference. So the different standards are designed for different crank spindle diameters.

Besides, we do have different standards in threaded too. BSA is just one of many standards that happens to be the most common. We still have Italian.

Maybe someday there will be an agreement on which spindle diameter is ideal, that day the number of press fit standards could be reduced. However, this would mean that crankset manufacturers need to agree whether steel or aluminum is the best spindle material. As long as aluminum is a possibility for spindle material, it requires a bigger bottom bracket shell.

Personally, I think that every existing bottom bracket shell standard is flawed. All threaded ones are flawed since threads are a poor retention mechanism in this application. Press fit is flawed too since it requires special tools. The bottom bracket shell should be like a threadless stem, so that it has a collar that is split. This split collar would then be tightened around the bottom bracket bearings. Just untighten the bolts, remove old bottom bracket (no press fit), insert new bottom bracket, tighten the bolts.

  • Why is threads a poor retention mechanism? It might be true, but for BSA the spindle/retention is ok in my experience, but they just wear faster than I'd like. So if a bigger shell with bigger bearings could help solve this I would be all for it.
    – WornChain
    Commented Mar 1 at 18:55
  • threads and carbon don't mix nicely for one thing. Commented Mar 3 at 5:29
  • Also, pedals move in pedal threads and cracks fail often at that interface. What should be done is a conical surface and the thread only provides tension, pressing the object to the conical surface. If you have a radial load and think that threads would work there, you're wrong.
    – juhist
    Commented Mar 3 at 5:48

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.