Hub bearing preload approaches

Question

I am the owner of a bike with a DT Swiss RWS thru-bolt (kinda like a thru-axle but it fits in regular 9/10mm dropouts) skewer system. I saw in their marketing fluff that one of the supposed advantages was that you could put way more tension into the axle than you could using a QR skewer. Leaving aside any discussion of whether it's an advantage at all (indeed some users report that it's actually quite easy to strip the alloy threads on the skewer if you overtorque it). I was curious how this would affect hub bearing preload.

With most of the other wheelsets I own, skewer tension directly contributes to bearing preload and anyone familiar with the correct way to adjust classic cup-and-cone bearings will know that it's important to check the bearing clearance with the skewer/axle bolt tightened, if not it's easy to put too much preload onto the bearings.

Looking on the DT Swiss website, their approach to preload seems to be different. The inner bearing race fits flush to shoulders on the axle on both ends of the axles and the shoulders on the hub shell for the outer race are spaced slightly narrower. This means that the outer race of one bearing sits flush against the shoulder on the hub shell but the other outer race floats slightly off the shoulder on the other side of the hub shell and is kept in place with an interference fit within the hub. Through some physics that I can't quite picture, this means that before the wheel is put into the frame the inner axle exerts outwards force on the inner races, preloading the bearings. When you put the wheel in the frame and tighten the skewer, the inner races (and axle) get compressed, the preload is reduced to the correct amount (that is apparently independent of the skewer tension) and you're good to go.

I guess my question is why do they do it like this and is it actually any better for ensuring correct bearing preload vs. traditional systems?

Answer 1

There's a couple points being covered here.

Classical cup and cone hubs use standard axles (M9, M10, 3/8" etc) that flex under compression from the QR. The cones are getting a little closer to one another.

Once you get into hubs that don't use standard QR axles, most of them are much stiffer and have little or none of the same effect. Almost all higher end hubs from recent years are like this. Even Shimano, who still uses cup and cone, use nonstandard axles on most of their nicer hubs and have for a long time now. You can adjust those to have some play and test how the bearing responds in the bike. Most of the time you won't be able to make the play go away or even change appreciably with force from a quick release, meaning you should adjust it to have no play off the bike.

Your second question is one of those topics that bike people might say whatever about but is more properly the domain of bearing engineers, but I'll try. Hub designs that aren't like you describe are then by necessity trying to play the game of having the shoulder to shoulder distance on the axle be exactly the same distance as the bore to bore distance on the shell. If they fail at this by too much you get excess preload. A lot of hub designs do work this way, but not all. Some designs eschew this by allowing the outer race on or both bearings to not actually be bottomed in the bore. BMX three-piece crank cartridge BB bearings for American shells are a poster child for this, although a lot of people don't realize it. Those shells are made with zero uniformity or precision of the exact face to face dimension, yet the BB kits come with a tubular spacer that's machined to be a certain length. The trick they are playing is that when you press it all together, one or both bearings unseat themselves in their cup. It works by the cartridges being overbuilt enough to handle being loaded this way without premature wear.