Simple generator

[gyulasun]

.....
And repeating the experiment at this low currents indeed reduces the torque considerably as you can see, but the question is whether this just a linear relation ie. 10x less current = 10x less torque? 

I think it is a non-linear relation and the B-H curve is to blame for it. However, the B-H curve has a nearly linear section, in your case it covers input current range from say 20-30 mA to say 1 Amper, this is from your attached graph showing the field strength at different currents,  your file 2016-06-22_7-30-16.png

However this brings us to the next subject, the torque is almost CONSTANT until it reaches the edges and then it does an instant flip....it's the exact opposite of what I thought would happen. A single pole motor/generator usually has a sinusoidal curve for it's torque vs angle graph. I added the counter torque graph based on the current torque graph and previous induced current graph.

It is indeed interesting that at 0.5 A current levels there is still some interaction between the coil edges and the rotor magnets. This may mean that at 0.5 Amper the laminated core is still able to develop magnetic poles at the coil edge areas. The reason for this may be in the fully closed ring core shape whenever DC biased even to a small current value. There has to be poles created at the coil ends once your tests show interaction, whatever small it is. But this is now not a drawback.

Now let's go a step further and do some power calculation, the average counter torque seems to be 0,00165Nm at 1A output. If we want to keep the rotor spinning at 10000 RPM and extract 1A from the coil it will cost us 1.727W of mechanical power. This seems rather low. 

Well, friction and attract forces are the enemies in your setup which may increase the actually needed input power for a certain power output.  The attract forces may be balanced by design.  Your calculation gives a low input power indeed for a would-be prime mover but it has to have a >90% efficiency figure, the higher the better of course.

[broli]

Yes it all depends, this is also why I used the tapered roller bearings from SKF, these are truly amazing even if the attraction forces are large they keep on spinning pretty well (they are rated at 60.5kN),

http://www.skf.com/au/products/bearings-units-housings/roller-bearings/tapered-roller-bearings/single-row-tapered-roller-bearings/single-row/index.html?designation=32305%20J2

They are pricey but standard ball bearings would have locked up long before and the quality is well worth the money.

In the mean time I replaced the magnets with neo disc magnets. The voltage waveform is now much closer to what I see in the simulation as well. And again it's amazing to see it being constant for half a rotation and do a sharp flip as soon as the magnets cross the edges and continue being constant.

https://youtu.be/A9B9HgePgzo

Seeing how the voltage has increased considerably using these new magnets I got excited and just shunted the coil with a low resistance. At first I got disappointing as the current fell to near zero and power output was nowhere to be seen. After playing with a pot meter attached to it I figured what was happening. The inductance was pulling the current down. Even if the coil is loaded it's acting like a full blown inductor, another strange behavior for me. In a transformer for instance the inductance is dropped and the current rises, however this generator does not seem to lose its inductance, which I think might be a very good thing.
So in order to fix this the inductance needs to be balanced by a capacitance. Sadly I don't have an assorted set of capacitors lying around but I'll be ordering these. I'm curious to see what will happen when the LC tank hits its frequency.

[lumen]

It's possible the current output is low because of the alternate flux path. Once current is generated in the coil, Lenz pushes back and it becomes easier for the flux to take the other path which limits the current output.

Possibly the most efficient configuration would be a coil on four quadrants each with it's own full wave bridge so at the junction of the coils, the flux is forced to take one or both paths and generates full current.

It appears that even better efficiency could be achieved if each quadrants load could be controlled to increase just after passing each quadrant, which could push the flux ahead of the magnet rotor and reduce the forward Lenz pressure in the next quadrant.

You may be on to something here.

[broli]

Lumen, what other path? The coil is enclosed around a toroid, when you energize the coil the flux can only follow one path. The flux from the magnet is still splitting evenly across the core as long as you don't reach saturation that is. Femm also shows this.

So I think adding more coils/quadrants will only add more complexity while we lack the basic understanding of the current design.

I said that the circuit behaved like it had a very large inductor in series with it, but even when rotating the rotor by hand at perhaps <1Hz I can still produce an open circuit voltage of around 200mV P2P and STILL the current drops to a marginal low value, I measured 20mA across a 1ohm load, knowing that the resistance of the coil is 1.5ohm this means the voltage dropped from 200mV to 50mV when the inductance of around 80mH should not have act like such a big impedance at this frequency. I'm scratching my head over this. I'm veeeeery eager to see what happens when I attach a capacitor and hit some resonance frequency.

Ironically currently the design seems to be acting more like a back emf free motor, the complete opposite of what I set out to design.

[lumen]

Broli, The flux should split evenly and it will over time, but with loading on the coil the flux will increase in the side without the winding as it is blocked by Lenz on the coil half.

The effect is that only a small current output can be achieved or the flux will easily increase in the alternate path around the side without the coil.
Flux will still continue to push to get through the coil and balance it's path so it is like a loaded inductor for a short time but never a strong driving push.

I like your idea because it seems to be on the right track by letting the magnet fight Lenz and not the prime mover.