Smart Drive alternator preliminary testing #2

[commanda]

I re-did the testing today, with the drill press set to 280 rpm & 350 rpm. I measured the open circuit voltage, the frequency unloaded, the loaded voltage, the loaded current, and the frequency whilst loaded. I used the frequency measurements to find the true rpm of the drill press, and corrected the no load voltage based on this. I was then able to calculate the internal impedance of the winding. It calculated out to be 11.25 for one case, and 11.3 for the other. Note that this is for one winding, from the common neutral point of the star connection, to one free end.

I then used a spreadsheet to do a maximum power transfer calculation at various RPM's, based on an open circuit voltage of 0.275 volts/RPM, and an internal resistance of 11.25 ohms. This yields the maximum power extractable at that rpm, if one could adjust the load to match it perfectly (3 phase motor driven auto-transformers anyone?)

I think 2 of these alternators on one shaft, geared up 2:1 (automotive toothed belt camshaft drive), would yield 1 KW with a decent breeze. Low wind performance, however, would be abyssmal.

Amanda

[commanda]

Oh yeah, I should add, just for reference. My coils are wound with 0.8 mm wire. The F & P units come in 3 flavors; 0.6mm, 0.8mm, and 1.00mm.

Amanda

[iFred]

Yep, that looks about right. You could try hooking the three phases together in a series connection for higer voltages and rectify, but it looks good as is. As for loading it, try a hysterisis load circuit, works well and controls well at any given rpm, (basically load match to rpm)

Good Luck!

[jacquesm]

I'm not so sure about the motor driven auto transformers (haven't seen one since I worked in a movie theater long ago), but what you *could* do is use switched capacitors after the rectification stage. It's tricky, but if the max 232 can do it in a dip then it's just a matter of scale ?

[commanda]

Getting capacitors big enough, with the current handling capacity, is pretty difficult. Let alone expensive. Max232's work at the milliwatt level. We're talking hundreds of watts here.

I was only joking when I suggested motor-driven auto-transformers.

Better solution is a switchmode down-converter, with a microprocessor driving the pwm. I'm working on a design. I've got a Picaxe18X, but I suspect it's faulty. It won't communicate with my PC. Have to get another one.

Amanda

[jacquesm]

Hi Amanda,

I've done a bit of micro processor work here (on a plasmacutter), and have found that it is very hard to make a design EM hardened enough to not occasionally crash
in a hostile environment.

I changed one cirquit that I could not get stable over to an analogue design and all the trouble went away, for a code jockey that is a really annoying thing to do.
(this circuit sampled the across-the-arc voltage of a running plasma cutter to do automatic height adjustment in order to minimize dimensional errors while cutting thin sheetmetal at fairly high speeds).

As to high current caps the biggest problem is probably inductance, most big caps are almost as much 'L' as they are 'C' if the frequency gets high enough... check out capacitors made for 'studio flash banks', they have very low inductance and will charge and discharge at phenomenal rates. They may be a bit more expensive than regular electrolytics with the same capacitance but not as much as you may think. Wiring is also a really big factor in a setup like that. I realise the 232 operates at the milliwatt level, that was a joke, but still if you could overcome the scaling problems then it might be possible to build a maximum power point tracker with nearly 100% efficiency using whatever windings a given generator has to perfectly match the impedance all the time. It would be a great thing, and a huge step forward from transformers and such.

Possibly a bunch of IGBT's could be used to do the switching, I'm not too current on those things but the on resistance is something to drool over. No heat dissipation whatsoever at 50A and more (tiny little heatsinks...). It would also save tons of money in transmission lines from tower to battery bank, and you could use it for photo voltaics as well as for windpower. Like a transformer for DC almost, without the problems inductor based dc-dc convertors have. (whine, 2-5 % loss, very bad EM behaviour and so on). At the current prices for cable a setup like that would probably be cost effective at distances > 100 ft or so, which is most of them, and it would be able to transform both 'up' and 'down' changing smoothly from charging in parallell to charging in series as conditions change.

[commanda]

I'm also thinking of playing with a circuit like this.

http://www.acs.comcen.com.au/buildppt.html

Modify the pot/resistor set point, so it is fed off the output of my tacho. Possibly with a logarithmic amplifier to get the correct curve. Something like this should be easy enough for at least most people to duplicate successfully.

Back to points you raised.......

Any analog inputs fed into micro-controllers need serious interfacing thought. You should never connect any line from the outside world directly to any sort of chip. Where I work now, we build belt weighers (conveyor belts). All our analog & digital I/O goes through plug-in opto-isolated modules. And the load cell inputs have R-L-C filters, MOV's, and diode clamps.

As for capacitors. The best are the flash discharge caps used to fire ruby lasers. These are an ongoing maintenance cost on these sorts of lasers, as they need replacing regularly. I still think the best converter is an L-C based switchmode device, even at 90-95% efficiency. The cost of increasing the blade swept area slightly to compensate for any losses more than makes up the difference. And by going to a high-voltage transmission, once you get past several hundred watts and some tens of metres, savings on cabling cost alone puts you way ahead.

Amanda

[jacquesm]

I never realised caps actually wear !

I tried just about everything I could think of to get rid of the spike that the plasma generates whenever it switches 'on', including almost all the things you listed other than the 'movs', but no success. A plasma cutter is a pretty 'dirty' device (almost as dirty as an arc welder) and I simply did not have the patience (or the time) to keep at it until I got the controller working so I picked another route and that worked. One added advantage that came out of all that is that the new circuit leaves the 'z' axis stepper motor wired into the main computer which allows much easier alternative uses of the table (such as a 3D router to mill blades with), in the end this is probably a better solution.

I read up a lot on the 'pic' chip (the micro controller that I was using), and I found that lots of people were having the same problems, appearantly the 'pic is quite difficult to make stable in a dirty environment. At a minimum a triple layer pcb with a groundplane under the chip. That's beyond my capability, so that's the main reason I gave up (and the time it would take as well). The new solution is a simple analogue comparator that tells the host if the torch is on, and if so if it is above or below a preset 'optimum' value of voltage across the arc.

I think the real culprit in interfacing that circuit was the 'pilot arc', a glorified electrical gas lighter that determines the proximity of metal by trying to jump a spark across the gap (up to .25" or 6mm).

Interesting motor/generator you've got there, what kind of rotor are you using ?

It looks like it has tons of space for windings, much more than most ordinary laminated designs of the same size.

[commanda]

Interesting motor/generator you've got there, what kind of rotor are you using ?

It looks like it has tons of space for windings, much more than most ordinary laminated designs of the same size.

It's a smart drive motor from a Fisher & Paykel washing machine. Completely unmodified. They're made in New Zealand. Been sold in Australia for years. Only recently starting to appear in the states. They make about 1 watt per rpm. Mine makes 0.48 volts per rpm with star wiring. There's several different versions available, with different wire gauges and number of turns.

Amanda