When you have a rotating reference frame and a mass on it whose distance from the center of the rotating axis changes, angular momentum J is not conserved. This is because somehow, some place a torque (force x distance) is applied:
J = I x omega. Then J-dot = I-dot x omega + I x omega-dot
The confusion, or even scam, is that you need to either apply a force to change the distance of the mass from the rotating center or split the mass in some way to reduce moment of inertia. When doing that, the torque applied on the rotating frame axis due to a non-conservation of momentum generates a "jerk" due to Coulomb non-linear friction coupling. If you could have a perfectly frictionless ball bearing no such force can be transmitted to the frame and result in a linear momentum. In essense, the ball bearing acts as a converter.
A couple things are tru then:
1. The linear pulse or momentum is due to the coupling of the rotating frame and the enclosure frame caused by non-linear friction. This results in a "knee-jerk" reaction of the frame and when it is placed on a frictionless surface like ice, the frame moves.
2. The constant application of a jerk at the ball bearing connecting the rotating frame shaft with its enclosure causes the bearing to break or lock after some revolutions depending on the magnitude of the torque applied. This is the reason demonstrations of such machinery is not made often and is not reproducible in any way.
3. One cannot control neither the magnitude not the direction of the linear momentum vector for obvious reasons.
4. Everyone talks about a linear momentum vector perpendicular to the angular speed vector of the mass but no one dares to even mention one parallel to it. You need a paralle vector to defy gravity and that is impossible for anyone who took 101 can understand why.
Peace brothers.
