Jhula principle - increase speed by applying brakes.

[Dingus Mungus]

I am trying to reproduce the system from scratch but the new one doesn't seem to be OU.

Heres a new example:
Only 6 feet tall and it averages 32 RPM...
I'm currently working on an even further scaled
down realistic version for physical replication.

[aleks]

Heres a new example:
Only 6 feet tall and it averages 32 RPM...
I'm currently working on an even further scaled
down realistic version for physical replication.

Great model! Slow increase in rotational speed is immediately visible on the top graph. For constant mass of the model and absence of driving motor it means torque is increasing as well.

[prajna]

That version stops after about 18 minutes, Dingus.  This could either be because your locking joint on the counterweight 'slips' (rv<0 rather than rv=0) or because the pendulum joint 'drifts' after a time (I have never worked out why this happens in WM2D). I have attached the graphs.

I have also attached a modified version of the previous model, which runs fine, albeit slightly slower. The pendulum pin also drifts in this version but not so much and it doesn't seem to effect the speed; it is still running fine after 1 hour.

[Dingus Mungus]

DOH! 
I'll need to go rework that model anyway... Too light weight.

Well we just need to keep working on models untill we find the answer. I think the "energy tap" is coming from loading the spring when it aligns with the axel,
and releasing the kinetic energy on both the pendulum and the counter wieght.

~Dingus

I really think this pendulum driven gravity wheel could work if built well.
Maybe not perpetual motion but hopefully overunity...

[prajna]

Don't panic, Dingus; I have found the sweet spot with the mini version.  I changed the counterweight to steel, which is handy since it gives us a known material for prototyping, reverted the freewheel gear for the counterweight axle to the old formula and rebalanced the pendulum.  Now it runs great and the freewheel gear only engages during the first few seconds before the system gets in sync.

I've attached the graphs and the sim.

I think the spring may bias the pendulum to the left but I have yet to investigate properly.