Star delta comparison

[Flux]

No idea if this will work, I have never posted anything here before, just replied to others.

I promised Jerry that I would look at the various connections and how they behave.

I have used the rotors of an 8 pole 6 coil machine and wound new coils. 3 wound for star and 3 for delta.

Star are wound with 100t of .9mm wire and star are wound with 170t of .71mm

Nominally the resistances are .6 ohm for star and 1.8 ohm for delta so they match the 3:1 requirement.

The 2 windings are in the same gap so there can be no differences there.

I have no bias in this discussion so I shall present results as I find them. I couldn't see how to be fair with a single speed test so I have done a range of speeds.

Both are with a 3 phase rectifier feeding a dc load at a constant 20v.

At the moment I have not looked at the Jerry connection, that will come later.

When I post lots of figures they seem to go all over the place so the results are in the form of a graph.

I have obtained similar results to Jerry in that the output in delta is higher even though the cut in speed is the same. The difference is not that great but it is real.

I will check the resistances more accurately but I doubt that there is much there.

At present I have no intention of commenting or forming any opinions, I shall just give the results as I find them for now.

I have no idea how typical this is and various forms of construction may behave quite differently. I have a feeling that jerry's new machine will show much more difference.

The curves are:-

Blue    input delta

purple   input star  

Green    output delta

dk blue  output star

[SamoaPower]

Many thanks Flux for a nice objective test. I note the somewhat better efficiency for star in the midrange but don't see the generally contended improvement at the low end - interesting.

I intend to do a test series myself and am curious as to how you accomplish the constant voltage DC load. At 20V, it's obviously not a battery and at least my power supplies won't sink current. Also, what is your prime mover and the method used for measuring input power?

Again, many thanks Flux.

[kitno455]

flux- can you plot the efficiency of star and delta as two more lines on the chart? since delta has higher input power, it is hard to tell the difference...

allan

[oztules]

Well your diary posting did work, and it is a worthy first posting in this section. The first of many "flux endeavours " I hope.

"At present I have no intention of commenting or forming any opinions, I shall just give the results as I find them for now."

Sensible stuff I feel. helps with the objectivity. Will hopefully overcome some of the incomplete albeit tantalizing partial results the Jerry had time to post.

Have you postulated any reasons for stars input power "bow" in the midrange..possible resonance or something?

Do you have any thoughts as to why delta is dying at the top end as compared to star.

"I have no idea how typical this is and various forms of construction may behave quite differently. I have a feeling that jerry's new machine will show much more difference."

If you were to extrapolate the performance to jerry's multicoil approach ie nine coils in parallel as in his original 1kw axial flux, would you expect the differential of the  "delta effect" to translate into a cumulative difference between the star and jerry-9

Not taking into account the circulating current problems which should be negated by seperate rectification..... or too many variables to make a prediction at this stage.

No-one has yet done any exploratory work on the multicoil approach of his first axial flux. ...another job for flux you think?

Thanks for your efforts thus far in unravelling the truth however it may fall

........oztules

[Flux]

Can I reply to both here.

Efficiencies seem higher than I normally get, but there is not as much loss in delta with this one as when I did the tests with the 12 pole machine. Having only 3 coils may mask errors in spacing with more coils.The coils are mounted on plywood and are not that close to the magnets so there may be more induced loss in the magnets with normal spacings. I have often wondered how conductive neo is.

Also this thing vibrates much more than a complete winding and may affect torque readings.

Measured efficiencies for what they are worth are as follows

Star  96%  91%  84%  76%

Delta 92%  89%  82.7%  76%

The top figures are odd I think the star is low by the shape of the graph.

Voltage is held by a linear dump regulator, batteries are hopeless for testing without a dump regulator to hold volts constant. I now find the dump load works perfectly well without the batteries and I can set any voltage.

The thing is mounted on a dc drive motor with the stator arm restrained by a spring balance to measure the torque.

I have just done some tests with the thing loaded as an ac machine into 3 x 2 ohm star connected resistors. The line voltage at 200 rpm was 8.3 v for star and 8v for delta. This is almost certainly due to the fact that the wire size is a bit more favourable to star, so the fact that star and delta should be the same on ac is not in doubt. It is a rectifier issue and I suspect that jerry connection will show a bigger difference.

Flux

[Jerry]

I would like to see a compairison made with with the standard 12 magnet 9 coil dual rotor disc alt. That will be a major chalenge. I think all of the preliminary testing will help set guidelines for such a test alt.

My feeling is the results will be the same. My goal has allways been and still is. using the exact same amount of copper and all other things being the same I will get more power per $ spent then star.

I know more power requiers a larger blade more tuorque. But its equal to bilding a larger alt without the larger price tag.

I also believe the test results will be the same with all types and formats of 3 phase alternaters no matter how many coils or magnets, iron core, air core, single rotor/lamination, car alt or even a funny littel 4" science fair alt. results is results.

If I had the time and money I would build an exact coppy of 2 of Dans smaller dual rotor alts, parts for part and piece for piece. All I would do is use half the wire size but same exact wieght in one as he did and of course wire one delta/lerry riged the other star. This would be a long exspencive undertaking. I looks like I may be pushed to that extream. Hope not LOL.

So far as far as I can tell from my test the power will be greater and have sooner cutin.

This would requier a larger blade but so would a larger built star alt (more money) to equal the same upgrade in power..

                    JK TAS Jerry

[RP]

"I have never posted anything here before, just replied to others."

Dang, I just looked.  1485 comments on other stories from Flux!

[Shadow]

1482 of those were helping me build my Wind Turbine!

[willib]

when you did the loaded test , was that rectified?

did you notice an increase in torque on the stator  without rectifiers?

oh yeah , welcome to  diary posting!!

[Flux]

Just a couple of updates for now.

I have found an error in the radius that I measured torque at. I knew the efficiencies were too high. The corrected figures wouldn't show on the graph but I will try and re run the tests when I do the jerry rectifier.

Waveforms    Both star and delta line waveforms are near perfect sine waves.

Star phase voltage and one coil of the delta winding alone show a small in phase (peaky)3rd harmonic. This is small and no doubt why my open circuit volts are equal and why Jerry has a low star voltage. He has an out of phase 3rd that is probably much bigger. I am not sure whether he saw any improvement in line waveform over phase, there should be.

When I close the delta the circulating 3rd harmonic can be felt as a vibration with no load.

Flux

[DanB]

Hi Jerry - I was pondering spending time on this today...  maybe still will although I think Flux has done a better job than I could here.

I dont see the need to build whole alternators  to test this.  If I do it I'll probably build one stator with 6 - or 9 coils in it.  Any 3 adjacent coils could be wired up as their own 3phase section - so 3 could be in Star, 3 could be Delta, 3 Jerry rigged if we wanted to.  Then Id be certain that airgaps and everything were the same and Id not have to be taking it apart/putting it back together so much.

It would be fun to get another such test though - but I think Flux did a very intersting one here and I doubt I could do better than that.  Maybe if I have the right sizes of wire here and all we'll mess with it today though - it would be fun!

[Flux]

Mental blockage over we can continue, I knew there was something seriously wrong with those last figures, little machines like this don't manage those efficiencies.

I have had a 12 pole machine on the rig for months and for that frequency is speed.

I forgot that this is 8 pole so speeds are 50% higher. In the graph multiply figures by 1.5 for speed. Input power curves are drastically wrong.

I have done all three windings now and there is no drastic difference and things swop about at various points so a graph would not be very useful.

In general star is a bit lower out and more efficient.

Delta generally gives most out and is least efficient.

Jerry is a good compromise between the others.

I will try the figures but if they don't make a readable form I shall have to try again. In preview there are so many funny red things about that it gives no idea what will happen.

Here goes.

Star first

RPM    watts out  watts in   efficiency

375.    14         26.7       54%
     
400.    71.8       122.4      58.5%
     
525.    129.6      257.1      50.4%
     
600.    199.7      526.5      38%
     
675.    255.6      757        34%
     
     

Delta

RPM    watts out  watts in   efficiency

375.    14.9       28.7       52%
     
400.    69.8       134.7      52%
     
525.    135.7      278.6      48.8%
     
600.    211.7      566        37.45
     
625.    275.4      826        33.3%
     
     

Jerry

RPM    watts out  watts in  efficiency

375.    15.9       34.4      46%
     
400.    69.7       134.6     51.7%
     
525.    129.5      305.3     42.4%
     
600.    199.3      520       38.3
     
675.    263        757       34.7%
     
     

[Flux]

Nearly worked. For some reason it lost the first figure of the speed readings.

speeds are 375, 400, 525, 600 and 675 rpm.

There you have it for this machine. I wouldn't want to guess what jerry will find with his odd thing, he has a knack of choosing rogue designs that show big differences.

I have a feeling that if you build a 12/9 machine with sensible proportions you will find much as I have, if you mess seriously with coil size spacing and hole size you may get something very different.

I would love to have the time to do the car alternator using neos, I am not sure jerry's stators were close.

Flux

[Jerry]

Hi Dan B.

I think the 2 in hand is the best way to make this compairison. That way the same coil is being used. I agree that 3 coils are enough to evaluate the diferances that would exsist in any number of coil/magnet count in a 3 phase configuration.

The 2 in hand will permit testing as normal star, normal delta or as I have sugjested a delta coil with twice the turns count at half the circular mills wire gage at the exact same onces of copper. This is equal to 100 turns of 18 gage for normal star or normal delta and 200 turns of delta/jerry riged all in the same coil. The compairisn only requiers sires or perelell conection of the tandum wound coil. My diagram will help show the configuration in a day or 2.   No cheater coils to help star voltage come up to were delta/jerry riged is. If star can't do this on its own and you use smaller gage wire with more turns for star then you've sacrificed some star amperage with the smaller wire to bring its voltage up. As we know star has less amperage anyway why penelize it further to help it compete when it still won't make as much amperage even with this help. Just use the copper in the gap and see wich coil/phase arangement produces the most power with the same componets. It my 2 onces copper against your 2 onces copper kinda thing. The 2 in hand coils will do this fairly.

But I'm even going to give star that handicap it needs anyway in my nest test as I did in the other test.

I plan on doing the rpm test at 100,200,300,400 and 500 rpm. Also a resistive load, an inductive load, votls per phase ac, volts per coil, battery load, wave form on scope ect,ect.

Well best get to it.LOL

                           JK TAS Jerry

[Gary D]

Flux, what blade size would you think matches the rpm power curve for about a 7 to 9 mph cutin? Somewhere in the 4 to 6 foot range? Nice to see more/different testing being done. Thanks for posting.... Gary D.

[willib]

http://nces.ed.gov/nceskids/graphing/classic/line.asp

[Flux]

Sorry Gary I don't follow the question can you have another try.

[Gary D]

My apologies Flux, I was looking at your rpm's on your 8 magnet/6 coil arrangement. This seems like the Hugh Piggott's 4 foot wind machine setup. I think, perhaps mistakenly, that you or someone else adapted a second magnet rotor/or blank steel plate to make it a duel rotor machine (with different winding counts/blade diameter)? Thus my query as to what diameter blades might work best. I understand that this temporary stator was only made for testing purposes, not to be a wind genny... Gary D.

[Flux]

With you now Gary.

You are right in some respects. I certainly did add another rotor disc without magnets to one of Hugh's 4 ft machines and made a nice 5' 6" machine for a low wind area. If you look through the recent posts I described this to someone( Marco I think).

This particular machine is a true dual rotor version intended to be used with a 6ft machine using a mppt circuit. The machine on its own would stall badly at 6 ft.

I just used the rotors for this test, the windings are temporary coils stuck onto a piece of 1/8 plywood and only have a tiny percentage of the copper of the proper winding.

Flux

[willib]

i put them all together for easy comparison.

http://nces.ed.gov/nceskids/graphing/classic/line.asp

[Flux]

Thanks for doing the graphs. I think it is the only way, all together would be on top of each other.

Flux

[Flux]

I have corrected the original graph.

Flux

[oztules]

Danb some thoughts thus far on your appraisal of this flux test, and some urging that you should listen to Jerry's original premise.

"Hi Jerry - I was pondering spending time on this today...  maybe still will although I think Flux has done a better job than I could here."

Flux has done an excellent job of testing the difference in rectification of equivelent electrical circuits here. The Delta/ star winding ratio's were well designed to reflect the differential for the three systems.

However, they did not reflect the Jerry premise of "bang for buck" The Flux test ran the Jerry coils at only 85% wind ratio. ie pound for pound, the Flux "delta" will be only 85% of the weight of the star coil .

Jerry has contended from the start that pound for pound is what he is getting at. Thus far no-one has successfully demomstrated this one way or the other. Jerry crippled his test by listening to outside influence which turned it into a test of something other than his contention. (116 turns for star, not 100:200 as he originally intended) little odd perhaps, but now a part of history.

"I dont see the need to build whole alternators  to test this.  If I do it I'll probably build one stator with 6 - or 9 coils in it.  Any 3 adjacent coils could be wired up as their own 3phase section - so 3 could be in Star, 3 could be Delta, 3 Jerry rigged if we wanted to.  Then Id be certain that airgaps and everything were the same and Id not have to be taking it apart/putting it back together so much"

In reality, I think that it will be problematic NOT to do the tests with multiple coils. The only person who has built a multicoil stator thus far is jerry (the 1kw model on the car test). This paralleling of like phase coils is where the difference may (/not) show up the most, holding the test to three coils only, may well paper over the systems greatest differences, and so should not be discounted as a useless part of the test until tested. (although I appreciate the difficulty level goes up dramatically).

"It would be fun to get another such test though - but I think Flux did a very intersting one here and I doubt I could do better than that.  Maybe if I have the right sizes of wire here and all we'll mess with it today though - it would be fun!"

Flux did indeed do an exemplary job of doing his test. But he was not testing Jerry's theory, mostly just testing the diode configurations  and efficiencies on different systems. This needed doing... but.... He has at my figuring, proven that jerryrig can hold its own at higher power, but not at lower, but that not in the efficiency figuring is the 15% less copper in the jerryrig system.

Now this does deserve more thorough examination. Does this mean you could have wound your latest stator with only 85% of the copper, and achieved the same power out (or thereabouts), and will this ratio get better or worse with multiple parallel coils (as in 9/12 machine).

Flux's test does not address that problem, which is at the heart of the jerry supposition.

Winding two in hand will help solve the similarity problem.

So I contend you could do " better than that", by testing that which jerry contended in the first place. Flux's test will be excellent first step in understanding what the results of the next actual jerryrig test mean, and I have no doubt will become the standard for referencing the differences between star/delta/jerry-rectification, but because of the 15% wire deficiency in his delta platform, does not test jerryrig, as espoused by jerry.

With the reinstatement of the extra 15% that jerryrig is missing in Flux's experiment, the lowspeed wind harvesting may improve markedly for no increase in the mass of copper currently used in star...who knows... thats why it's worth a look. The more testing done by more diverse people, the more believable the results (whatever they are) will be. and only then may we proceed to chjange our designs, or remain happy in the knowledge that no better one are currently on offer. To big to miss really.

What we do know, is that Jerry produced over 1kw at an indeterminate speed with 9 coils parallel jerryrig, and he claims the stator was bone cold... that still needs explanation. Good cooling, cold air, beats me...

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

[Jerry]

Well I did get around to assemling the 3 phase 6 pole(with 3 poles) motor conversion.

I'm not sure when I'll get bact to more testing. While chucking the alt into the lathe I smashed a finger prety bad. So I might be on the sidelines for awhile.

Here are a few #s I was able to record today. This is a 500 rpm lathe test.

The other RPMs I want to do will be with a 2 hp dc motor and a varyable dc power supply (back burner now, bad finger).

I wound 3 coils 2 in hand on each pole this equals 6 coils. Open voltage ac each coil 3.64. 2 coils of 1 pole perelelled 3.64 volts.

2 coils of 1 pole wired sires 6.729 volts.

I'll call the coils as follows. Phase one has 2 coils, coils 1a & 1b. Same for phase 2, coils 2a and 2b, phase 3 coils 3a & 3b.

Wired as star with A & B coils wound in perelell. Phase leg to phase leg voltage (star) 6.149 volts.

Star open voltage DC 6.21v.

Delta/jerry riged open DC 8.965v.

6. volt golf cart battery load star volts 6.071v at 1.09 amps for 6.62 watts.
   
6. volt golf cart battery load delta/jerry riged 6.085 volts at 1.97 amps for 11.99 watts.
   
   

From the above data we should be able to figure out the additional turns star would need to equall deltas higher voltage. However in the previus tests the added turns count did not help star very much and that dose not agree with my premiss pound for pound

If I had left the coils a and b  perelelled as in star but wire them delta/jerry riged the power at the same rpm would have been much less for delta/jerry riged.

But again thats my point. Same amount of copper more power.

When I heal up I'll get back to the other test. Its a good thing I didn't hurt my one and only typing finger. LOL.

This is the 3rd style alt same results.

                      JK TAS Jerry

[oztules]

Hi Jerry, Lifes a bitch sometimes, My thumb has just recovered it's nail from belting it with a hammer against a bull gear on the rear drive of a david brown tractor, so my sympathies are with you.

From your post  

  "From the above data we should be able to figure out the additional turns star would need to equall deltas higher voltage. However in the previus tests the added turns count did not help star very much and that dose not agree with my premiss pound for pound"

Flux has done that test quite well by the looks of it, so extra energy expended in that directioon is probably not warranted.

You should stick to proving your own theory first and foremost. Only those parameters pertinant to your idea should be incorporated. Until your ideas are tested thoroughly, there's no point in muddying the waters with compromises that do not directly relate to your primary objective. Your two in hand directly allows for the testing of your hypothesis.

Here it cannot be in dispute that your primary point (ie pound for pound) is satisfied.

Now only more thorough testing of different speeds, loads, efficiencies, can we get a clear picture of where and if Jerryrig is a suitable alternative. It may turn out that for different objectives, different systems are naturals, but my humble opinion is .....dont deflect resources away from what you are primarily trying to prove or disprove.

That all being said, it would appear that if you were trying to charge those batteries at those rpm, with that alt, jerry looks the goods.  Until we can get torque figures for this test, we wont know what efficiency you are getting. Star may be high efficiency at these power loads, and jerryrig may be at the lower end of its efficiency...who knows at this stage.

You will have more time to figure out new ways to confuse us all with the extra time on your hands now.

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

[Flux]

Sorry I have made another error. When I compared my corrected graph with Willib's there was obviously something wrong.

For the power input calculations there are several factors converting frequency to speed, torque at an odd radius to lb ft etc, and I stick these in the memory of the calculator so that for each reading I multiply the memory figure by lb and freq.

I seem to have done something to corrupt the figure in memory or pressed M+ with something else in it.

The results is that the input figures of the last tests have the wrong torque figure and are high.

I left it over night to see if anyone spotted it. It doesn't affect the comparison in any way but we might as well keep things straight and come clean about mistakes.

Sorry Willib, any chance that you can alter your graphs and re post them with the same name.

I wish it was possible to go back to the original figures and alter them but I can't and they will have to stay to confuse people.

Corrected input figures frequency then power in.

Star    375 17.8    450  91.8   525  171.4    600   351   675  505

Delta   375  19.1   450  101   525  185.7    600   377.5  675  551

Jerry   375  22.9   450  101    525  203      600  346.9   675  505

Even if they jump about it should be possible to sort it this way.

Flux

[Flux]

Sorry about the finger Jerry, giving blood over this is beyond the call of duty.

Did you do my trick and leave the key in the chuck. Hope it soon heals and you can carry on with the tests.

I also wish replies to these things would end up where I try to put them and not at random . If this is not under your reply I am sure you will see it.

Flux

[spinner]

re the torque thing:

i think jerry has indicated that more prop will be needed with the jerry rigged unit...interesting to see how much tho, it would be a useful rule of thumb, perhaps.

spinner

[spinner]

hafta ( from one who knows far less than many of you ) agree that deviation from jerrys origional type of unit, on which the JR rig is based, seems to introduce more and more variables each time it is done.

spinner

[willib]

done.

if the graphs look the same as before click refresh when viewing them..

[SamoaPower]

Flux,

I took the liberty to assemble your corrected data into one place and graphed with the data tables. I included an efficiency comparison graph since this is what I'm most interested in. Hope you don't mind. Parallel = Jerry. Interesting stuff.

[Jerry]

Hi Flux.

Thanks for taking the time to participate in the testing. I wish we could get a few more DIY folks involved.

You mentioned batteries would be hopeless as a test load. I did both battery load testing and resistive load testing.

My results in resistive were exactly the same as yours. I used a 1 ohm 400 watt none inductive load here are the #s.

Star 4.6 volts at 4.57 amps for 21.02 watts, delta/jerry riged 4.55 amps at 4.561 volts for 20.75 watts for a diferance of .27 watts.

This is real close to the ratio in the data you've collected. However in the battery as a load test delta/jerry riged did 11.99 watts and star did 6.071 watts.

I see this diferance every time I use a battery load. If you could please do a battery load test. I now this is an unstable load but a diferance this much will be quite evident even if the #s are a littel flutery.

Today at the store I did another test with the alt. No lathe at the store so I used a large drill. I'm not sure what the drill rpm is, I think its less then 500? and the 6 volt golf cart battery I used was full charged so its load resistance then the 6 volt cart battery I used at home.

The results were the same as far as the diferance between delta/jerry riged and star both in the resistive and battery alone.

I did a load test with a 62.5 amp load tester accross the battery. I wanted to see if the resistive load accross the battery would act like resistive or battery load. It acts like the battery load. The delta/jerry riged had the same increase of power as per the battery only load.

I think you'll find when using a battery as a load the delta/jerry riged version makes much more power.

I also beleave that these alts that we build are most likely used 99.9% of the time for battery charging. Some for heating air or watter but most of the time for charging batteries. When the dump load is ingaged the wind gen should be disconected from the battery and then connected to the dump load what ever it may be and I think this is done automaticly. If in the dump load mod either star delta/jerry riged make the same amount of power into a ristive load nothing is lost but while charging batteries delta/jerry riged will make more power just seems like an addvantage?

At any rate if you could please try a battery load inplace of the dump load. I know as you test the battery voltage will be climing but you'll still see the great diferance between the 2 phase scheems.

                               JK TAS Jerry

[Jerry]

Hi Dan B.

I know your realy good at 2 in hand. I think this is an easy way to test my my pretence. Then if you didn't likre the results just wire the bifeler (SP?) windings in perelell and go star, no harm done and in the end a usefull alt.

                      JK TAS Jerry