Delta, star, Jerry riged?

[Jerry]

Thanks to all how have entered into this debate. That even includes my wife (yes dear).

It took me 1 month to build this littel 8" dual rotor. I am now inspiered to build the 10" I've spoken of recently. This could take another month or 2 or maybe 4?

I'd like to build several stators for this one to make the nessesary test to answer some of the questions here.

Its very evedent that I'm a stray dog on this one. I think my presentation may be hard to follow...sorry about that.

If and when I do make the compairison between star and Jerryriged you bet I'll post the results. Ofcource if I'm wrong I'll just keep it to my self. LOL.

But if I'm right??  I think the test will requier some help from yoooz guys.

I will most likley be using those 1.5"X1/4" round NEOs. It will be the standard 9-12 arangement.

I wanted to do this at 10" rotors but is that large enough diameter? What would be the sugjested wire gage for normal 3 phase star? And the half gage size for delta to make the same voltage as star.

Then I will build several stators. One will be the sugjested star wire size turns count and all that. The other will be 9 coils with 18 coil ends available for diferant configurations.

I've had much fun with the littel 8" and it works far behond what I expected.

I intend to fly it soon. I plan on flying a Garbogen next to it and track there preformances.

Thanks again guys and get back to work.LOL

                     JK TAS Jerry

[Flux]

Just to digress a little, for those interested in star /delta switching, it may be easier to do a star/jerry switch with just a 2 pole relay rather than a 3 pole change over.

Flux

[Jerry]

Nice computer work Flux. I'm thinking of trying this in the lathe. This switch idea will make for a very quike compairison.

Is this still equal to all coil ends being conected to all coil ends?

What is the frequency of the 14 magnet GM car alt at 500 rpm? I read in a post from 4 years ago the HZ at 600 rpm is 60.

I have both a 12 volt (actuly a bunch of 12v) and a couple 24 v GM alts of the same physical size. Stators are inter changeable.

I'm thinking of doing the normal star 3 phase 12 volt stator load, no load, amp and volt test on it. Then seperate the phases on the 24v, wire them each to there own bridge perelell the dc outputs on these and do the same test.

I'm hopeing to see half the volts per phase in the 12v VS the 24 volt? Same with impeadance?

This is for all the above debate. I asked about the frequecy to determin the impeadance in ohms of each alt.

I'm thinking the impeadance per phase should be twice as much in the 24 volt as the 12 volt.

I will use a sine wave audio oscilator. 1/2 ohm stable car audio amplifier.

I'll conect 1 phase of the alt to the output of the amp in sires with ? resistance until voltage reads the same accross the alts phase and the resistor. Once the voltage is equal then the ohms of the resistors is the impeadance ohms of the alt.

This must be done in ac because the dc resistance of these alt is extreamly low. The ac impeadance is higher. Both measured in ohms.

I have tuns of electronic parts so finding the apropriate resistor will not be a proble either in ohms or wattage. I do have a good collection of non inductive resistors for this test.

The purpus of this test is to check to see if the impeadance of the 12v is half that of the 24v?

If I send the right frequecy into the phase/coil this should duplicate the lathes 500 rpm?

I'm also thinking of looking at the ac riple thats produced from both arangements on the scope. I'll do this with a resistive load. Also want to measure the slight voltage diferance between coils in the delta mode.

I thought of doing all this with these GM alts because I have them and it will be much less work then building another dual rotor alt and 2 more stators.

There may be some diferance between an air core alt and an iron core alt?

But I think it still may shed some light on the subject?

                   JK TAS Jerry

[DanB]

that would be a fun test Jerry!

To be fair - I think you'll want to be sure that the resistance of the load is 1.73x higher for the  'star' test than it is for the delta or 'jerry rig' test - because the voltage will be higher.  That is - if you're checking for power output into a load at a given rpm which I think is the bottom line here.

[Flux]

With switch closed it becomes star, with it open it reverts to Jerry rig.

It would not be fair to star to do a direct comparison as you are not changing the turns ratio. If you do delta or Jerry at 12v you should compare star at 20v, which you may be able to do if you have 3 x 6v batteries well charged.

A 12 pole alternator at 500 rpm is 500 Hz. 14pole will be 14/12 x 50 = 583Hz.

Now I am afraid the next bit goes completely off the rails, the impedance of an alternator seen by the load consists of a combination of a resistance and a LEAKAGE reactance. This is not the same as the reactance you would measure if you feed it from an ac source. The reactance you get in that case would be many times the leakage reactance. There are ways of measuring the leakage reactance by separating the components of synchronus impedance but this is no place to discuss that.

The only sensible way you can determine the reactance is to load it with a resistor until you reach half the open circuit voltage, the value of the resistor will be the

impedence in ohms, then you find the reactance from Z^2= root(x^2 - R^2). With neos the answer may be near enough. If you do it with the wound field the error from armature reaction will make it inaccurate to the point where it may be useless.

Air gap alternators don't behave that way and that is one reason you managed to squeeze more out of the axial than the garbogen.

Don't worry if you cant follow this, it's getting quite involved.

Flux

[Flux]

My mistake

Z^2 = X^2 + R^2       X = root(z^2 - R^2)

I wish computers would have roots and squares . These ^ things are confusing.

[ghurd]

Hi Mrs. Jerry!

It's nice to see other wives also give some support to such endevors.

No matter what it is, I say "Look honey! Magnets!" and she pretends to be excited.

G-

[Jerry]

Hi Dan B.

I think and correct me if I'm wrong? An alt wire star will have twice the voltage and half the current of an alt wiered delta, yes, no, ?

So my proposal is to use a 24v GM stator that is normaly wired star and wire it delta.

This should make this 24 volt stator do 12 volts but at twice the current it would normaly do star.

Next remove the 24v now star stator and replace it with a normal 12 v star stator.

I acctuly did this today and here are my results.

Laithe rpm again 500. Delta wired with normal 24v star wired stator no load 18.61 volts.

Star 12 volt stator no load 8.881 volts.

I used a 1 ohm 460 watt resistor as a load. The delta 10.40 amps at 10.45 volts.

Star 5.97 amps at 6.1 volts.

I think what is sposta hapen is both stators should preform this same? Therefor the load should be the same.

If I wrong on this one, then that info about delt and star on the same stator must be incorect.

Basicly the 24 volt phases were wired perelell and the 12v phases were wires sires.

The turns count on the 24v stator volt looks to be twice that of the 12v and the wire on the 12v stator looks to be twice the size of the 24v.

I took pix of the 2 stators I'll load them up.

                   JK TAS Jerry

[oztules]

Danb i'm not sure that the arguments presented by hugh or yourself regarding the slight coil variations giving effectively a push pull effect are relevent in the practical world. In reality we generate say 20v or more open circuit, but then drag it down to 12 v (or thereabouts) by loading it up. Now it's beyond being likely that under these circumstances, any one coil within a paralled phase (delta) would be below 12v while the other two in the phase are trying to achieve twenty but been pulled down to 12 by the load.  In no load condition, your argument makes sense, one may be 20.3 one 19.2 one 20.0 here there would be some too'ing and froeing, and who cares, we are not using any power coz we're open circuit

As soon as we load it ,all coils  get pulled down to the charging voltage. Now if we rectified each of this phases three coils independantly you and hugh  would be happy that the diodes prevent any leakage back into a slightly weaker coil within that phase. It is however probably a moot point in practice, as when under load, the lowest potential is the load itself, and all three coils are at a higher potential ( or at least trying to be) than it, so all will contribute to the load as best they can. Any counter circulation currents would at best be miniscule by comparison as all three coils see each other at the same potential ( the batteries potential at any given time plus the diodes interface voltage drop.

In truth, it may well be shown that under load there may be a more measurable effect on output with star configuration and non identical coils. Here the weakest link would indeed be the weakest link. and its lowered impedance will contribute materially to the other coils overal voltage. There surely will not be circulating current problems, but in practice, they may not materially exist for delta either.  

So maybe avoiding delta as a dirty second cousin to be avoided at all costs, may be a little unkind to this configuration.

ps i'm wiring mine up in star as a matter of interest .....more flexible in its output

and still havin fun..........oztules

[DanB]

Hi Jerry -

no - Star (assuming the coils are the same) would have 1.73x the voltage as Delta (or 'Jerry Rigged') at the same speed.

[Jerry]

So your telling me even though the coils are wired in sires (star) I only get 1.73 times the voltage that I'd get wired delta.

You now whats real disapointing about that. If I have 2 single phase cois wired in sires I'll get twice the voltage. I don't loose .23% of my voltage out there in never never land.

Thats why I keep saying. If I double the voltage in the delta configuration. And since delta dose more current then star I don't loose that .23% And the current is the same. Because basicly delta is perelell and star is a missmached out phase sires wired.

Like I said. Things in electronics. Wire 2 10 ohm resistors in sires you get 20 ohms,

wire 2 12v bateries in sires you get 24 volts, wire 2 12 volt (same) light bulbs in sires they work on 24 volts.

Just think if batteries were 3 phase and you wired 2 of them in siries you'd get 1.73X 12v = 20.76v not 24v. To me thats a ripe off when the 24 volts is there and 3 phase is stealing some.

Help out here you've said that if the coils are the same the voltage diferance between delta and star is 1.73%. Thanks

What is the amperage diferance if the coils are the same?

                    JK TAS Jerry

[Jerry]

I goofed up above there some where when I wrote star rather than delta. I was talking about the 24 volt now delta stator but wrote star.

Heres the pix of the GM stators.

This one is the stock 12v from the top.

This is the 24 volt GM alt stator from the top.

This is the stock 12v from the side.

This is the 24 volt from the side.

                        JK TAS Jerry

[windstuffnow]

  Decided to take a break with other projects and get mildly involved...

  As a hypothetical lets say each phase is 1 ohm the alternator is running 500 rpm and in star the open voltage is 30 volts... coils are all identical...

Wired in star you would have 2 ohms, from the 30 volts you'd see 9 amps (no losses included)

Wired in delta open voltage would be 17.32, and you'd see just over 8 amps and the resistance would measure .66 ohms

The way I see it using the "jerry rig" you've paralleled all 3 sets which would be the equivilant of .33 ohms which would give you 16 amps at 12volts.  Minus the phase difference which would bring it down to around 10.

I've experimented with one built like this with some pretty impressive results... wind the coils with smaller wire to get close to the voltage you need at a given rpm from "each" coil, wire all the coils parallel in each phase. If each coil was 1 ohm ( as an example ) using the 12 pole auto alternator that has 12 coils per phase then the resistance of each phase would end up being .08 ohms.  If its wired in delta you'd end up with a resistance of .05 ohms or in star .16 ohms.... works very well...

Take that and use your "jerry rig" rectifiying each coil... what would that come out to in resistance?  I've built a couple units using the 8 inch base, axial flux using slotted stators and my wedge magnets one making 3500 watts and the other (which I still have) making 2800 watts on a test stand.  I think your on the right track... below is a pic of the 2800 watt stator...





.

[oztules]

Nuts.... when you end up arguing with yourself, you know the end is near,

 it would appear that my above jibberish would almost hold water, but I plum forgot to consider the part of the cycle where any in phase coils are BELOW the battery/diode voltage. Here I must concede circulating currents could create some minor havoc. Once the voltage within a phase group drops below charging potential, (between say +12v to 0 to -12v,) we will have a period where unequal voltages will be able to shunt between coils as their differences will be exposed to each other now). This will not detract from the output, (as the diode/s will not even be conducting at this time) but will introduce heat and therefore lower efficiency, so I guess all isn't rosy with mismatched coils. The loss will be minor, but it will be there in the real world, which is contrary to what I was espousing initially.  

next I'm going to prove black is white, and get run down on the next zebra crossing

sighhh

..oztules

[ghurd]

This thread is the thinking that lead to some of my stranger questions!

I had some better results rectifing seperately, but assumed it was due to high reactance in the alternator.

This whole idea always reminds me of some of Ed's older work, somehow.

http://www.fieldlines.com/story/2004/6/7/21190/34164

My take on it has been the resistance is always adding together. At best the resistance the amps see is 1.5x each phase. At worst it is 2x.

I^2 x R shows losses increasing with amps and ohms.

Leading to seperated phases, with lower amps and lower ohms, has less losses.

A few more turns per phase when rectified seperately, a tad higher cut-in maybe, but less losses.

Therefore, as the amps go higher the losses go lower, percentage speaking. The road test could back that up?

Jerry, Maybe what everyone is saying is the reason the 1/2 hp GE ECM would not get past 100W?

I can't understand how there are delta style parasidic losses in a "Jerry Rig".

Anyway, I understand the line of thought.

G-

[wooferhound]

Try this page

http://www.lookuptables.com/

press the alt-key and while holding it down key in the code on the Number pad.

here are many codes I use

 248 ° -  degrees

 236 ∞ -  infinity

 155 ¢ -  cents

 246 ÷ -  devide

 253 ² -  Square (I think)

 227 π -  Pi

 171 ½ -  1/2

 172 ¼ -  1/4

There are many more, just click the link to see them

[Flux]

Jerry  For a direct comparison, if the 12v star produced 8.9v then your 24v star stator should have produced 17.8v. When delta connected it should give 17.8/1.73 and that is 10.2v, not 18.61 as you measured.

The stators are not sufficiently similar to do the comparison. Dan is right that if they were identical except for a 2:1 difference in winding it would be still a disadvantage to compare with the same load resistance as you would have 10.2v for delta and 8.9 for star, but it would have been much nearer.

The most useful thing you can do at this moment is forget the star and compare that modified stator in delta and Jerry rig with the same load resistor and into a battery.

Flux

[Flux]

Thanks Woof, I suspected it was possible. I will have to see if it is easier than the other way that everyone else seems to understand but I find confusing.

Flux

[zubbly]

hi Jerry!

quite a debate going on here. isn't it amazing all the thoughts and ideas that come out once our interest is captured.

just a small input of my own, and may possibly be an intersting experiment since at the moment you are experimenting quite a bit with different ways of cutting up or combining of the phases to search for the best overall connection method.

years ago, i was faced with a little dilema where a customer needed to operate his 230/460 volt motor on 575 volt, and there was basically no time for the stator to be rewound to that voltage. we came up with a little trick that seemed to work and never saw the motor suffer any magic smoke, lol!

what i am wondering, and never seemed to dawn on me at the time, was to try this conection on a 3 phase stator to work as a genny. it seems we are either stuck with less current in the long run and sooner cut in, or later cut in and more current with the star delta connections that we use now. here is a little diagram that you may want to try before rectification.

you would have to add a lead wire at the centre point of each phase. this way you can maintain "i think" the larger current carrying capability of the regular delta connection, but also gain the sooner cut in rpm. for now i will call it an "extended delta".

in the diagram, you will see that it connects as a regular delta, or the end of a phase taps into the centre point of its mating phase. i "think" this will give an approx 25% voltage gain over the regular delta, but with more current than a star connection would supply.

its just some thoughts and only presenting it to you if you wish to try it.

have fun!

zubbly

[DanB]

We'd wind the coils with the voltage in mind...

The voltage isn't 'lost'.  Its lower because the phases in series are out of phase.  But - it's not lost, it's 'shared' with another phase.  There are times in the 3 phase arrangement when coil voltage is low (below cutin) but current would still be flowing because that voltage is adding to another phase where current is perhaps above (or near) cutin.  

[Flux]

Just done a test comparing 19v delta with 33v star.

This is biased slightly in favour of star as the diode drop is more significant at 19v than at 33v. The fact that we need slightly higher speed for cut in in delta will be explained by this.

Green is star watts in

Blue is star watts out

Dark blue is delta watts in

Magenta is delta watts out.

Flux

[Gary D]

Flux, for the color name challenged (like me) is the magenta the purple (rightmost just over 300@ 520 rpm)? I assume it is, however you know what they say about assuming... Gary D.

[Flux]

Yes Gary, purple, magenta, lilac call it what you will. I had more trouble deciding what the dark blue one was.

I just wonder if different browsers alter the colours, be awkward if they do.

Flux

[Jerry]

"The voltage isn't lost. Its lower because the phases in sires are out of phase."

So help me out here again. If delta is produceing 10 volts. Star will produce 17.3 volts.

If the phases were in phase the voltage would be 20v (10 + 10) no phase diferance. You've said the phases are shereing the voltage.

Where did the 2.7 volts go and who are the out of phase phases shereing it with shairing with cause it must not be the batteries?

Like I said. If I wire two 10 volt 10 amp single phase transfomers in sires I get 20 volts 10 amps. But if the same 2 transformers are 120% out of phase I get 17.3 volts at 10 amps. Wheres the addvantage in loosing that 2.7 volts?

As you've stated. Star makes 1.73 X the voltage of delta.

Delta or perelell does not have this voltage loss because there is not out of phase condition that exsists.

So lets take those same 10 volt 10 amp phases this time and wire them star. The amperage should remaine 10 a because 10 amps in sires with 10 amps is still ten amps but the voltage goes up.

Example if I had 2 10v 10 batteries wired in sires the voltage would be 20v but the amperage would still be 10a. star is sires.

However since were talking about 3 phase or 120% out of phase the voltage is only 17.3v. 17.3v X10 amps 173 watts.

Now lets rewind some coils and make each phase do twice the voltage 20 volts using wire half the size of star. The amperage will drop to 5 amps per phase. Half the size of wire twice the turns. So we end up with three 5 amp phases for a total of 15 amps but the voltage is now 20v. because in delta the phases are in perelell so the voltage remains the same.

15 amps X 20 volts or 300 watts.

If the voltage isn't lost in star then it must be 20v or higher. But you said its not 20 but 17.3. If the amperage isn't the same as the amperage of one phase then what is it?

Even if the voltage is 20v or higher what is the amperage? I don't think the amperage of  phases wired in sires can be higher then phases wired in perelell through the delta conection.

Is the amperage of phases wired in delta  1.73X the amperage of star? What is the amperage compairisons? I know its said to be more, how much more?

For star to do the same wattage we need the same amps and volts. That means the 10 volt phases will have to produce 20 volts wired star and the 10 amp phases will have to produce collectively (star sires phases) 15 amps?

Again Dan if the star phases are shareing the 20 volts and only giving us 17.3v  were is the 2.7 volts going. I think thats about 1/8 lost or a littel over.

We start out with 20 volts and end up with 17.3v I think that a loss. And if it isn't lost how can you put in 20 and get out 17.3?

I prefure to put in 20 and get out 20.

                              JK TAS Jerry

 

[Flux]

Just checked Jerry rig against delta at 19v.

Jerry connection cuts in at lower speed, between star and delta. Jerry efficiency is better than delta but not as good as star.

Graphs would not show much , the loading is not very different, just a slight shift in speed.

Here are the efficiencies at various power out, this is the significant bit.

I never get figures to line up so I will try it this way.

Star  watts out 40 eff 65%. watts out 180 eff 73%.  watts out 330 eff 80%

Jerry watts out 40 eff 45%. watts out 180 eff 68%. watts out 330 eff 70%

Delta watts out 40 eff 37%. watts out 180 eff 61%. watts out 330 eff 65%

This could be completely different for iron cored machines with rectifier overlap.

Flux

[oztules]

something doesn't feel right here. As best as I can make out, all comparative tests on star/delta on this forum, have all had like results.

This can be summed up as Star works well at lower currents  and delta is lousy, delta works well at higher currents, and star is found wanting.

 In your graphical analysis, the more power in the system, the more divergent are the output lines, star continues to streak away at a very tidy clip, and delta is at best parallelling the slope of the star plot.

In real life, the two output plots converge, and then delta's gradient should markedly be steeper than star gradient (with both gradients becoming less steep as more coil losses are introduced).

how might this be explained, as it seems to go against everything we have gleaned thus far?, or is it a case that the resistance in the two windings has been declared equal, even though star contains twice the turns of delta, or have I just got it all wrong??

perplexed.........oztules

[Jerry]

I just mentioned a delta stator having twice the turns for a compairison to a star stator having half the turns. To end up with an alt that should have the same output but doesn't.

Acctuly in my 8" dual rotor jerry-riged alt the coils average 2.4 ohms each. Basicly all 9

of the coils are are wired in perelell. Isolated however my the 9 bridges but effectively .26666 ohms. These coils are 250 turns each.

It apears I will have to build some form of a dual rotor alt after all.

I'm thinking of 5.5" rotors with 4 each 1" X 1/2" X 1/8" NEOS. This will be a 3 coil 4 magnet per rotor dual rotor alt.

I will then wind 6 coils. 3 will have small wire 3 will have wire twice that size.

The small wire will have twice the turns count as the large wire. Maybe I should do the 2 in hand for the large wire. For the test the small wire will have a fullwave bridge for each phase/coil. The large wire will have the normal 6 diode bridge for star wireing.

May the best configuration win. Then for the volts, amps and whatts at a give rpm.

                      JK TAS Jerry

[oztules]

You are on the right track with the two in hand.

A wire twice the size has 4 times the current capability (area=pi x radius squared)  ie double the diam= 4times the cross sectional area), and so would not reflect the results properly.

2 wires in hand will give you an accurate base for your comparison.

you really opened a can of worms here, may the best system win....

ps "May the best configuration win. Then for the volts, amps and whatts at a give rpm.

                      JK TAS Jerry"

at a given rpm, may not quite solve the problem, as different systems seem to have advantages at different rpm thats why guys want to switch star delta in the first place, not any one system wins on all fronts (at least not yet till the tests are done). It will have to be done through the entire range, other wise you may not uncover the frailities of a particular system until it is too late (you build a biggun, only to find at higher or lower rpm, things didn't work out

.............oztules

[electrondady1]

what i'd like to see is a test that monitors stator/coil temerature for the different  layouts . it would require enough run time for the  devise to reach it's operating tem. i would think that the amount of waist heat generated by an alternater would be indication of overall eficiency, without geting sucked into the numbers. in the case of a star /jerry rigged comparison ,it would be good to get comparative readings at at specific rpm  as well as at specific power levels. per haps the temperature of the rectifyers could be monitored as well.

[electrondady1]

 jerry, i was thinking that by building a 3/4 test geni , while  i can see the puity of it it might be harder to monitor the advantage of spliting the phases. as far as i can follow the discusion, the main advantage is in eliminating the resistance of the additional phase and perhapse eliminate some form of inter coil modulation.  i would think the advantages of a jerry rigged system might be better displayed in a stator with a larger number of coils.  my particular interest is in slow rotation , large diameter rotors ,lots of coils , a vawt type.  you have other criteria perhaps.
 

shawn

[Flux]

OZ

I think from your later post that you may have had more thoughts but the real issue is that in reality this is a stupid way to use an alternator, its voltage should be allowed to rise with speed. We tie it to fixed voltage for convenience but at a serious price for alternator efficiency, we are prepared to do that for simplicity and reliability.

Within this restraint star is better at all speeds but the main advantage is at low speed where we want best efficiency. Efficiency in higher winds is not of much importance because we deliberately make it low anyway, knowing that we still get plenty of power. If we do something to change the alternator voltage so that we have a far better match in the higher wind, the star connection is still better than delta, but tap changing is complicated and most people never venture beyond star/ delta. The higher output and stiffer load in delta is not derived from delta's better performance, it is derived from a voltage ratio change of 1.73 and a reduction of the effective resistance to 1/3. If you rewind the alternator for this voltage change and use star you will do better.

This is all very interesting but we are splitting hairs to some extent, with air gap alternators under similar conditions no magic connection is going to make fantastic changes to output so why not use the best in low wind (star). It remains to be seen what effect Jerry's connection has on alternators that are limited by leakage reactance, but a modified car alternator with neos may never show this in a normal working range.( with the claw rotor and wound field it certainly will).

[hvirtane]

In real life, the two output plots converge, and then delta's gradient should markedly be steeper than star gradient (with both gradients becoming less steep as more coil losses are introduced).

This is the case.

When this pattern is not observed in the tests,

I think that something is very wrong.

More or less the correct output curves are to be

seen at that file by itdg/hugh, which address I posted earlier.

The other thing is that you can get higher

voltages with 'star' with low RPM. So the overall

performance is often good with 'star'.

So the question is, if we can somehow, using

'Jerry riged' or 'delta' with more windings

in the coils, get a good ouput with low RPM also.

So that it would be the winner with higher speeds, too.

Another possibility is to use 'Jerry riged'

and then connect the coils in series.

But then we have got the losses

with the diodes. Cf. above.  

- Hannu

[Gary D]

 This has opened a can of worms! People can choose Three phase Star, Delta, single phase, two phase, five phase (Hugh's), and now Jerry rigged!

 As you said, holding an alt. to a locked voltage is a poor way to harvest the wind's potential. During winter at least, adding resistance to the line at a point that could be usefull as heat(inside the house perhaps) could let the alt. see higher voltages and run nearer the desired TSR. As the machines seem to be getting larger swept areas, this will become more important to keep stator heat down. This is diverging from the testing tho...

 I think I can see advantages in the microhydro area in at least the star, if not the Jerry rigged area with an overshot/breastshot wheel. In a 9/12 arrangement using 1 coil per phase per set, you could go with 3 rings of separate 3 phase star, delta, or Jerryrigged wired to reach voltage at your preferred speed (smaller gauge wire of course to phisically fit), with knife switches to cut in (or open eash set separately inside a control box) depending on water availability ... It would make life easier than j trying to match a stator to a set amount and sacrifice during low water/ drought times... Ahh the possibilities and options, keep them coming... ;-) Gary D.