Transforming a PC fan into a generator: what do I do with the phases?

[Martini]

I modified a 12v DC motor (a computer fan) to make a generator today. I want to help a friend build a wind power generator model to teach kids about science but I have to be taught myself beforehand 

There are four coils, which have three connections on the PCB. I soldered wires to these three connections, bypassing the rest of the PCB.


Since it's an AC output, I fired up the old Tektronix to display some waveforms:




I have a polyphase system and I'd like to rectify it but I do not quite understand the wiring.
I have:

• 4 volts between yellow and black
• 4 volts between red and black
• 6 volts between yellow and red

And:

• 36 ohms between yellow and black
• 36 ohms between red and black
• 72 ohms between yellow and red

So I wonder if I have two phases and a neutral or three phases...or something weirder.



I made a three-phase full-wave rectifier with LEDs to see what happens:
(I was surprised it worked but happy I tried)


The green LEDs (connected to the black wire) don't light up. So I thought it was just a neutral and connected it directly to my DC ground: it makes it incredibly hard to lit the blue LED and the two pairs of the bridge lit up even when there is nothing connected to the DC (i.e. no blue LED).
I'm lost there



I'd appreciate very much to read some theory about the coils and phases.

[homebrew]

Hi Martini,

luckily Atmel has a Application Note http://www.atmel.com/images/doc8005.pdf on how to drive such a fan with an ATtiny13.

As you can derive from the schematic, you probably have a two-phase system. Maybe you can just find out the center tap by using your multi-meter in ohms range.
As far as I understand, the phases should be offset by 90 degrees.

[Martini]

Great, so I have two phases. The schematics lost me (the point of the zener diode?) but I don't need to run the fan, quite the opposite.


So What should I do? Use a four-diode bridge rectifier with the two phases and leave the other wire alone? Or....?

Maybe you can just find out the center tap by using your multi-meter in ohms range.

I put the resistance values in the op post but the ohm symbol has been replaced by a question mark

As far as I understand, the phases should be offset by 90 degrees.

...or 180° ?

Got this on site this afternoon:

[Zero999]

I can't see any of the images you've attached.

The zener just provides a regulated power supply to the Hall effect sensor.

The application note is relevant to your question because in order to use the fan motor as a generator, you need to understand how it works.

Going by your measurements, it sounds like you've got two coils in series. You need to separate out each coil.

[Martini]

Pity you can't see the images, I have attached them.


The pdf is great. I meant I didn't need to understand how the hall effect sensor and transistors worked because I'm not using them. Seeing how the coils are wired, though, was very helpful.


So I wonder if I don't have two pairs of coils connected together at one end with the black wire connected to that point. Like so:



If this is the case, what is the best rectifier to maximise the power?
(Keeping in mind the target to a few mA to light up some LEDs)

[homebrew]

So I wonder if I don't have two pairs of coils connected together at one end with the black wire connected to that point.

Yes, I think that is exactly like that.
And the waveforms you got from your oscilloscope look quite similar to the ones presented in the application note. As described there, one period would correspond to a 180 degree turn. Hence I still think that they are 90 degrees phase shifted.

I think you could apply the same four diode rectifier circuit that you would use on a symmetrical supply, e.g. like so: http://circuitswiring.com/basic-symmetric-power-supply/

Probably there are more efficient ways ...

Or just use a half rectifyer for each winding and define the center tap as ground.

After all, you said that it is only for some demo - hence efficiency would not be a prime concern ...

Tip: Use shottky diodes to reduce the voltage drop across the diodes.

[Martini]

So I wonder if I don't have two pairs of coils connected together at one end with the black wire connected to that point.

Yes, I think that is exactly like that.
And the waveforms you got from your oscilloscope look quite similar to the ones presented in the application note. As described there, one period would correspond to a 180 degree turn. Hence I still think that they are 90 degrees phase shifted.

90° with respect to the fan, alright.

I think you could apply the same four diode rectifier circuit that you would use on a symmetrical supply, e.g. like so: http://circuitswiring.com/basic-symmetric-power-supply/

I did that, got 5V between V- and V+ (still with the LEDs as rectifiers) and a bit over 20 mA when blowing like a maniac on the thing.



I'd like to understand why it is (or isn't!) important to connect that center tap?

Probably there are more efficient ways ...

Or just use a half rectifyer for each winding and define the center tap as ground.

After all, you said that it is only for some demo - hence efficiency would not be a prime concern ...

Yes but I'd like to demonstrate the right way of doing things. Otherwise, I could skip the rectifier altogether and that would still light the LED

[homebrew]

I did that, got 5V between V- and V+ (still with the LEDs as rectifiers) and a bit over 20 mA when blowing like a maniac on the thing.

I'd like to understand why it is (or isn't!) important to connect that center tap?

Because of the same reason as you wouldn't need to connect the center tap of the symmetrical PSU example if you would just be interested in the total voltage difference between V+ and V-. You could directly connect your load there without any ground reference at all.

As another idea ... Just connect your scope to the inputs of the rectifier, letting the center tap unconnected. I'm pretty sure that you'll find a nice sine wave between them.

Using LEDs as rectification diodes has some drawbacks, especially the relatively high forward voltage. A full-wave rectifier will drop twice that. Hence using red LEDs you will drop at least 3V. Green and blue LEDs are even worse in that respect.

[Martini]

As another idea ... Just connect your scope to the inputs of the rectifier, letting the center tap unconnected. I'm pretty sure that you'll find a nice sine wave between them.

Yes, and I get 6V AC (compared to the 4V between the black and any other wire).


About the LEDs, I'm aware of the problem, we'll probably use diodes (maybe Schottky diodes) in the final model but I thought it was very interesting to see the bridge rectifier work

I'd like to understand why it is (or isn't!) important to connect that center tap?

Hum...the center tap would allow me to display both an AC and a DC signal on the scope, wouldn't it?