Originally Posted by
oneclick
Although manufacturing tolerances are (generally) better now than they were, in a no-load and zero clearance situation, the axle only contacts the cups and three points (each end), the three being where the three balls contact the cup race (it is is possible that more contact, but that requires the those balls have *exactly* the same amount of space, i.e. none, between their surface and the race, possible, but depends on the manufacturing variences.
With rotation, which three are in contact constantly changes (those tolerances...)
With pre-load, the cup and axle are forced closer together, the balls and cup elastically deform, and you have more that three points.
Too much pre-load and the degree of osculation goes up, and with too much preload, spalling occurs.
So clearly there is a compromise between pre-load and bearing life.
Cartridge bearings don't give the user a chance to make that decision, and for most users that is a good thing.
True. Proper preload adjustment is pretty tricky (even more difficult on quick release hubs). Turning spindle in fingers, any radial slack at all, or very easy spin, usually not enough preload. "Notchy" feel, too tight. Backing off just slightly from notchy and it smooths out, but you can feel all the balls "engaged", I call it when the bearing "sings". You'll know it when you feel it.
I used to run all bearings with just the slightest bit of slack, based on dad's saying that a little loose is better than a little tight. He was a mechanical engineer, but surprisingly, on this, he was wrong. No preload, and spalling comes much quicker, usually the inner race first. Had this happen in both bike and automotive until I read deep dive on bearing theory and manufacturer recommendations. Since using proper preload, drastic improvement in bearing durability, like night and day.