Originally Posted by
Kontact
You don't need to ponder anything. Standard bearings are not designed for side loading to remove play.
And you have completely misunderstood the cap torque. They list .7 to 1.5 Nm, but then provide a tool that you use by hand and is designed to slip from your grip when it is sufficiently tight. That is supposed to work out to .7 to 1.5, but functionally it is just enough to locate the crank are against the forces of shaft and thread friction. The cap is there to remove play, not preload anything.
You are one of those people who has convinced yourself that your narrow experience and engineering textbook knowledge cause you to have a better, smarter understanding of how bikes work. But bikes don't work like jet engines. They exist in a low torque, low rpm, low temperature world where virtually all the wear and tear comes from bad specs and dirt - not treating the bearings in some super special way.
If you were having problems with bearings wearing out quickly, it was almost certainly due to something you were doing wrong, or out of spec BB shells. So I wish you would stop inventing stuff and then telling people about your new wisdom. You don't even know how the equipment you already own works.
I wish, for the life of me, I could find where online, years ago, they had a superb explanation on bearing life versus proper preload. It was great, and has proven to be true for me. It makes sense, and explains why I was spalling cones and balls years ago when I would adjust the cones for just the tiniest bit of slack, based on dad's advice on bearings, "better a little too loose, than a little too tight". He used to snug tapered roller bearings on front car axle, then back off "two flats", so 120 degrees. I later learned this is way, way wrong, it results in bad roller edge loading, and he was an automotive engineer, but didn't have as much knowledge with regard to rolling contact bearings. Bicycle bearing loads are not huge, but the difference between loading almost half the balls per side and only a couple, makes a huge difference in wear, **especially during peak loading, like 10 events per life cycle, like severe pothole load**. Bearing loads matter for lifespan. This is something I have experienced. Perhaps a high preload is not needed, just taking out the slack, but either way, it's much better than leaving in the slack, which, by the way, also stretches the seals radially. I torque my cap to specs, just like you said in metric equivalent, I could have guestimated it, but I have a microtorque wrench which makes it easy. The crank maker did include a plastic wrench, but if they only wanted people to use that, they wouldn't have made the interface a standard Allen socket, nor labeled it with torque value.
The biggest difference in durability has been with pedals; Cup and cone with tiny balls, proper preload has made a huge difference in durability, versus intentionally leaving in the tiniest perception of slack.
Just because knowlege comes from outside bicycle mechanics, doesn't mean it's wrong. I appreciate you noting that Shimano externals are not angular contact. But evidently some externals are (in my revision above while you were posting). And taking out slack on straight radial bearings may help, I have to think about that, my guess is, with just a small preload, it still behaves mostly like a pure radial bearing, but the preload to one side, taking out the slack, makes it perform better.
I'll be curious about the wheel hub setup on my new bike, which is supposed to have cartridge bearings front and rear, so I wonder if the spindle just clamps axially to the inner races of each bearing, or if there is any side preload between inner and outer races across the two bearings?