The chuck pins thread in, compensating tolerance on location and depth from face of receivers in spindle. I've no recollection of springs or bearing balls. The notch in the pin, obviously retains an engagement the cam 'likes'. A simple test places a loose pin in a receiver and lock the cam. You'll see a slight negative position from spindle face to face of pin.
Compared to nuts and bolts, torquing cam-locks is different. Fasteners stretch intentionally to remain tight. Cam-locks have thread depth to pull the chuck in, but the cam ratio is about all there is to stay in place. Very dependable, it won't back out, kind of like a worm will turn a gear, but a gear can't operate a worm.
Cam lock is shorter. Less expensive to manufacture. Less dependant part inventory. W-I-D-E selection in aftermarket tooling. No cross threading or scarred keys. That's why of all the spindle mountings, cam-lock is king. Once reversing motors [hello metric threading] went de rigueur, game over for threaded and taper lock. I'm betting 'A' mount is next highest acceptance; but have seen many scarred short tapers. Cam-lock pins are guided into battery before the taper gets a hit.
I'm thinking I should post my chuck manipulation bar. Any interested?
Each cam/ pin combination should be set individually, so cam locks well short 180° rotation which otherwise will loosen at some inappropriate time. Always mount a chuck, what ever type; 3/4 jaw, collet closer, face plate etc, so a pin you designate #1 goes in same hole. If it is right, all cams will have nearly identical "o'clock" position when tightened for use. Tighten same as you mount an auto wheel, across the pattern, not around 1-2-3-4. . .
Far as designating #1; in a 2,3 or 6 jaw, use one closest to the master pinion usually marked l|l, just so it's easy to find. A paint pen spot on the chuck and side of the spindle, you're good!