OK, here would be my definitions for discussion:
Valve float: At high-RPM, the momentum of the valve is too great for the spring to control and it leaves the prescribed cam lobe profile ballistically over the nose.
[/b]Valve train seperation[/b]: Whenever any of the valvetrain units - cam, (lifters, pushrods,) rockers, valves, or springs - lose firm contact.
Valve bounce - type I (lobe bounce): After an instance of valve seperation, the spring will be forcing the components back in to contact in an uncontrolled manner. The impact of the contact will compress elastic components. As they re-expand, they will impart an opposite force to the valve, causing it to 'bounce' open again.
Valve bounce - type II (seat bounce): If the lash is too great or the closing ramp too steep, the valve face will hit the valve seat with too much velocity, elastically compressing the valve diameter and the valve seat. As they re-expand, the valve will 'bounce' off the seat and open again - also causing seperation.
I'm also under the impression that valve bounce type I is usually caused by valve float.
For example: At an excessive RPM, the opening velocity is too great for the spring to control. The momentum of the spring/valve/rocker mass continues to compress the spring (continuing to open the valve) even after the rocker has passed the nose of the cam. This is the primary cause of 'valve float'. i.e. the spring does not have enough force to overcome the momentum of the valve assembly at high speed, thus the rocker leaves the surface of the cam lobe.
Now, the cam continues to rotate, rapidly increasing the distance between it's lobe face and the floating rocker. Eventually the spring force overcomes the momentum of the valve, and it reverses direction and begins to close, and therein lies the danger of valve bounce. The spring is now accelerating the valve closed as fast as physics allows - far faster than it would if it were following the cam lobe profile. At some point far down the closing ramp, perhaps even on the base circle, the rocker slams into the cam lobe. The elasticity in the cam, rocker, valve stem and spring absorb that impact and then reverse it right out, pushing back on the valvespring - once again causing seperation as the valve assembly 'bounces' off the cam lobe. Often times it can actually bounce several times as there is little damping in the system - keeping the valve intermittently open far past it's intended closing point.
From my personal experience, the biggest problem for street/strip engines is usually valve float - which then causes seperation and often valve bounce (type I). Although interestingly enough, in serious money racing, where valve lift and duration are restricted, some cam designers are now modifying the profile to
intentionally cause valve float! A steep ramp right up to the nose will launch the valve right off. It'll open the valve further and keep it open longer, and the closing ramp is intended to 'catch' the valve smoothly to prevent bounce. But the mechanical dimensions of the cam lobe are still within spec. Tricky, eh?
