The Paradox Engine

[Tusk]

the power off spin down is very fast and all by itself will move the motor in the other direction

Sounds like you have the 'rotor arm reversal' or as I like to call it, the 'third bite of the cherry' 

[telecom]

another version of the same idea

[Tusk]

Your first thoughts on that second version were correct telecom; the disk rotation is disadvantageous if you are going to drive the rotor arm. But make those outer black rods EM drives and provided both disks rotate CW together or CCW together the secondary reaction will be advantageous to the rotor arm rotation. But then you would need to brake the disks from the centre axis to avoid rotor arm reversal (and increase rotor arm motion) if that was your intention.

Keep looking, there will be many viable configurations, no doubt some far better than my own 

[Tusk]

On your testbed are you getting the spin down issue??

I assume you are referring to the rapid spin down you spoke of earlier webby1; the spin up and down seem to take about the same time on my apparatus. There is quite a strong attraction since the embedded magnets are neodymium, but the disk functions well as a flywheel with all that mass around the outer edge, and let's not overlook the advantages of a greater radius 

 

[Tusk]

I do not think that that displays your effect effectively.

Fair enough webby1. When designing the PE apparatus I resisted the temptation to do an easier build and complicate the dynamics (e.g. by using an electric motor to drive the disk via cogs/drive wheels or whatever) because the device had to demonstrate the phenomena as clearly as possible while minimising the possibility of raising concerns over complex interactions between unnecessary elements.

This may seem more complicated than your present line of experimentation but you might want to consider, instead of trashing those electric motors mount one over the central axis of a freewheeling rotor arm (but in this instance not on the rotor arm, rather bench mount it) turning a simple disk (or two disks, even better) by means of a rubber drive wheel. I haven't tried this configuration but I see no problem with it other than :

1. Having an additional element which some might imagine contributes to the additional motion. 

2. Since the drive wheel must take up some space the point of applied force will be slightly off the central axis; again there might be a claim of significant interference from this.

I may even knock one of these up myself, it doesn't get much simpler and the data from a twin disk system would be useful. Bench mounting the drive motor over the central axis calls for a serious piece of supporting framework but the thing is straightforward enough. It can't go on the rotor arm due to the counter torque, we don't want the rotor arm going the other way. See I said it would cause trouble