Jhula principle - increase speed by applying brakes.

[prajna]

I've updated http://declarepeace.org.uk/jhula/.  Playing around with new simulations has produced a variety of very interesting models.  Some show chaotic run-away acceleration.  It looks like this is a runner.  Is anyone building out there?  All you need are a couple of bicycle rear axles and some weights.  I would advise adding a braking system as well since this machine does appear to accelerate to destruction.

[barbosi]

It looks good!
I would encourage you to give some construction details. I'm sure there will be people interested in builing it. It would be valuable for you their feedback.
Regards.

[prajna]

barbosi, there is nothing particularly important in terms of construction; weights and measurements are not critical.  Some first approximations are avaiable from the WM2D models (you can download a trial version from their website, linked from my page).  The device simply consists of two pendulums, one connected to a fixed point via a one-way hub (like you find as the rear axle of nearly all bicycles) and the second pendulum similarly connected to the first.  I believe that there is enough information on my website for any moderately competant mechanic or engineer to assemble a device.  Obviously some experimentation will be needed to get the best out of any device.

[prajna]

I'll do it later this evening, mramos.  I look forward to hearing how you get on.  Meanwhile, if you look at the two connections on the model (marked contraint motor) they are actually freehubs (bile back axles).  One connects a pendulum to a frame and the other connects the second pendulum to the first.

[prajna]

Cool.  Sure it runs in the simulator but the weights are set extremely high to overcome air resistance.  Quite how it will work in practice I am not sure yet.  The first version (really xnonix's version) seems to be a little too stable but we haven't really found out what makes a successful version yet.  Most versions will run if you take air resistance out or increase the weights so that it is not a significant factor.  All we can do is experiment.  I have been experimenting with sims and most of them run; some go completely off the scale and some run even with a really high air resistance (think of it in terms of load on the system).  I can't say what are the optimum ratios, weights or measurements, it will take experimentation in real life to find out.