success ! boost convertor works

[jacquesm]

So, after a two day break fiddling with coils and fets I have it working. It's still a little 'rough' around the edges, and I faked the windmill coils with a small transformer.

Thanks to Victor for an excellent idea.

The principle is very simple, and it turns out the practice is too, the coils in the alternator have enough inductance to 'reuse' them as boost coils in a boost convertor as seen in this schematic.

So all you really need is a way to short out those coils...

The trick is to have one fet per phase, each fet handles a 'leg' from the coils. The fets then replaces the lower half of the diodes in the full waves bridge.

The gate gets driven with a very low duty cycle (about 3%) signal of 5 Khz.

These spikes will push all the fets into conductance at the same time, effectively shorting out the coils (mosfets have very low 'on' resistance). Upon the release of the gate voltage at the end of the pulse the current in the coils will induce a huge voltage in the other direction, and this is enough to overflow into the batteries. The effect is the same as a 'regular' boost converter, you save yourself the cost of the coil, and you're switching the AC side rather than the DC side, which makes everything very elegant.

The practical upshot of this is that you can start charging at ridiculously low wind speeds.

One issue with more powerful windmills is that the parasitic diode in the fets is used as the rectifier, which will work but on a more powerful windmill will probably fry the fet sooner or later (likely sooner). To remedy this you'll have to place a schottky diode parallel to the fet, the schottky diode has a forward voltage low enough to relieve the fet from the current.

The cost of the circuit is somewere around $5 in parts, one 555 a bunch of fets and some cheap passive components. You could feed the 555 from an extra set of diodes on the coils, a cap and a small supply, to make the circuit non-parasitic.

Tomorrow I'll draw up a schematic and give a proper explanation of this, right now I'm just happy.

[RatOmeter]

To be optimal (or maybe just to work at all), wouldn't the switching need to be timed to be in phase with the AC coming from the alternator? Say, switched on during a portion of the '+' half of the sine and switched off to dump into the '-' half of cycle?  If so and the control circuit were to remain purely analog, perhaps it could be triggered from the down-going zero crossing (easy for me to say, harder for me to do).  Or maybe made easier with a once-per-pole pulse from  a single hall effect sensor.

Anyway, I like where this is going!

[jacquesm]

that's what I first thought too Rat !

Took me two days to get over that one, the coils aren't shorted when you do that, you only hit one half and while there is a spike it's nowhere near enough to get you in to a high enough boost factor to be worthwhile.

I was looking at spacing the 'firing angles' of the fets 120 degrees. Turned out that wasn't neccesary after all.

The schematic is extremely simple, simplistic almost. Now I have to test it on a real windmill to see if it still works as advertised.

In the end there are only four wires to connect, battery ground and one lead to each phase.

You should be able to retrofit it onto any existing windmill.

A more 'fancy' version would include a shunt (then you have two grounds, one in and one out of the circuit) to switch itself off once the winds have pushed the alternator on it's own well above charging voltage.

You can sense that with the shunt because the current flowing will exceed some preset value above which the pulses get suppressed.

[ghurd]

'Some Dude' down under had a setup to do something similar with broken solar panels that would not reach 12v, or to use 12v panels for 24 or 48v charging. If the panel reached high enough voltage, there was a bypass provission (just a diode?). And bypass transistors if the current exceede the other ratings. The details were online, but I can't find them or him. He was in Australia for sure. (.au ?)

Thought something he did could smooth out some of the rough edges.

Not much to go on I know, but worth a shot?

G-