alternators: multi phase. what would be better?

[brandnewb]

Please consider a quarter of a 3 phase alternator.



Assuming this winding is valid, it leaves space for 3 extra phases but opposed.

Is it?

What about if we keep the winding very simple?


rant start
seriously I'd like to donate to better forum software.
rant end

I'd like to hear any and all thoughts on the options we have,

[brandnewb]

I realize that my inserted images are less than useful. I'll try again for the latter one.;

[brandnewb]

this thread focuses on the winding of the central coils. The magnets are not in question in this instance

[MagnetJuice]

IMHO you could have better luck posting that question this on this forum.

https://electronics.stackexchange.com/

However, you will have to describe the image to them using words because they don’t accept uploading images.

I must say, your ability to use that graphic program has improved a lot. Congratulations.

Ed

[MagnetJuice]

However, you will have to describe the image to them using words because they don’t accept uploading images.

I take that back, the do accept images, you lucky man! 

Ed

[joestue]

That topology has been done before. You take a coil of electrical steel, then wire edm cut the slots and teeth out of it. another option is punching the tooth out.. but in order to make it from one single strip of steel you have to precisely punch the teeth out such that when coiled up, the teeth are straight.

Basically its like taking a traditional radial motor and turning the stator core inside out.

Without an iron core to make the flux flow where you want it to go, its of no improvement.

You could try the method with a coil of pallet banding strap, then mount your 3 phase coils in a 3d printed jig to hold them in place.. You have to wind them like a toroidal transformer, but even more complicated.

read this paper https://www.mdpi.com/1996-1073/11/4/774

[Adriaan Kragten]

Axial flux PM-generators normally have a 3-phase, 1-layer winding. Such a winding is given in figure 5 of my report KD 341 for an 8-pole generator. In the upper picture, the legs of the coil bundles are laid on the ideal positions. This means that there is an angle of 15° in between the legs of adjacent coils of different phase and an angle of 45° in between the two legs of one coil. So within one coil, there are two positions which are not used. These positions can be used for a second layer with also six coils, so with also two coils of each phase. But this second layer doesn't make the winding a 6-phase winding. It is still a 3-phase winding as the coil laid in the grooves at 90° and 135° is the third coil of phase U and the coil laid in the grooves at 270° and 315° is the fourth coil of phase U. So you can get much more copper in a 2-layers winding than in a 1-layer winding. However, the amount of copper in a 1-layer winding can also be increased if the winding is modified such as given in the lower picture of figure 5. A disadvantage of using a 2-layers winding is that now you get crossing coil heads and this makes the winding thicker. It is possible to realise that the thicker coil heads are lying outside the path of the magnets and this is done for certain Chinese commercial generators. But a 2-layer winding is mostly too complex for non professional manufacture.

A 2-layers winding as described above, results in a 3-phase winding if all twelve coils have the same winding direction. However, if the six coils of the first layer are wound right hand and if the six coils of the second layer are wound left hand, the winding becomes a 6-phase winding. This is because change of the winding direction results in a 180° phase shift. Assume that the three phases of the first layer are called U, V and W and that the three phases of the second layer are called X, Y and Z. The sequence of the phases is now U, Z, V, X, W and Y and the phase angle in between adjacent phases is 60°. The three phases of each layer need their own 3-phase rectifier and both rectifiers must be connected in series. A 6-phase winding may have an advantage if the generated AC voltage isn't sinusoidal but if each AC voltage has a block shape with a rather large distance in between the blocks.

I have investigated a 5-phase and a 9-phase winding in my public report KD 712 for a radial flux PM-generator. A 9-phase winding (with a phase angle of 40° in between the phases) has as advantage that there is almost no fluctuation of the rectified voltage and current. Another advantage of a 9-phase winding is that it can be laid in a stator with 36 slots. A 6-phase winding (with a phase angle of 60° in between the phases) has no advantage above a 3-phase winding because a 6-phase winding has the same fluctuation of the voltage as a 3-phase winding if the voltage varies sinusoidal. This is because there is a peak in the DC voltage every 60° for a 3-phase winding (see report KD 340 figure 9).

[Adriaan Kragten]

I doubted what I wrote in my previous post about flatting of the DC voltage for a 6-phase winding. So I have investigated if using of a 6-phase winding is useful if there is a strong pulsation of the DC voltage generated by the three phases of the first layer. It appears that the peak in the DC voltage of the three phases in the first layer starts at the same time as the peak in the DC voltage of the three phases of the second layer. So both peaks are strengthening each other. Therefore making a 6-phase winding is not effective to flatten the DC voltage fluctuations. If an almost flattened DC voltage is wanted, one has to make the distance in between the magnets at the pitch circle, half the magnet width.

[joestue]

I could re-wind my 36 slot, 30 pole 56 frame motor. the magnets are 2 inches long, 1/4" wide, 1/8" thick.. the stator stack is 2" deep.

If i were to wind the coils 3 in hand, the motor could be switched between 9 phases connected star, and 3 phase connected star and produce very close to the same voltage output. -a lot of reconnections have to be made.

testing would be pretty simple to show which is more efficient, but the motor has bushings instead of bearings and the rotor isn't perfectly centered in the stack, so the 30/36 position cogging preference greatly exceeds the 180 that it should have.

i suspect there wont be much difference.

[Adriaan Kragten]

I could re-wind my 36 slot, 30 pole 56 frame motor. the magnets are 2 inches long, 1/4" wide, 1/8" thick.. the stator stack is 2" deep.

If i were to wind the coils 3 in hand, the motor could be switched between 9 phases connected star, and 3 phase connected star and produce very close to the same voltage output. -a lot of reconnections have to be made.

testing would be pretty simple to show which is more efficient, but the motor has bushings instead of bearings and the rotor isn't perfectly centered in the stack, so the 30/36 position cogging preference greatly exceeds the 180 that it should have.

i suspect there wont be much difference.

If your stator has 36 slots and if you want a 9-phase winding, the armature must have 32 poles (or 40 poles) and not 30 poles (see KD 712 chapter 7 and figure 10). For a 9-phase winding there must be an angle of 40° in between the phases (see KD 712 figure 4) and you get an angle of 60° if the armature has 30 poles. If your armature has 30 poles, you can only use it for a stator with a 3-phase winding. However, in this case, half the number coils of a certain phase must be wound right hand and half the number must be wound left hand. Changing the winding direction results in a change of the phase angle of 180° and this is the way how to change a phase angle of 60° into a phase angle of 120° which is required for a 3-phase winding. I have used the same trick for the 60-pole generator of the VIRYA-12 as described in chapter 9 of report KD 727. The stator of this generator has 72 slots.

A stator with 36 slots, and so with 36 stator poles, has a pole angle of 10°. An armature with 30 poles has a pole angle of 12°. The difference is 2° and this means that there is a preference position every 2° and so you have 180 preference postions per revolution. This is a rather low number and therefore you will feel a rather large cogging effect.
An armature with 32 poles has a pole angle of 11.25°. So the difference with the stator pole angle is 1.25° and so you will get 360 / 1.25 = 288 preference postions per revolution which is much better to reduce cogging.

You can also make a 3-phase winding if the armature has 34 poles (see report KD 580). This gives an armature pole angle of 10.5882° and so the difference with the stator pole angle is 0.5582°. So the number of preference postions per revolution is 360 / 0.5582 = 612. This is very high and so there will be almost no cogging.
You can also make a 3-phase winding if the armature has 38 poles. This gives an armature pole angle of 9.4737° and so the difference with the stator pole angle is 0.5263°. So the number of preference postions per revolution is 360 / 0.5263 = 684. This is even higher than for a 34-pole armature and so there will be almost no cogging too.

[joestue]

You are right 30 wont work for 9 phase system. I could grind the magnets to a trapezoid shape and fit 32 in the stator.

But 30 poles is too many for a 56c frame motor, for use as a wind turbine alternator. The inductance and frequency are too high, and the motor can be shorted out at 1750 rpm and not overheat producing very little drag.

I have some 10pole, 12 slot hvac fan motors that are rated for 1hp at 1250 rpm, half inch diameter shafts 4 or 5 inches long. I think they could be directly driven with a wind turbine but no one seems interested in them.

Ever thought about designing a centrifugal clutch to let the turbine free spin until it has enough inertia to overcome the cogging torque?  The clutch could be designed with ratcheting mechanism rather than friction pads.

[Adriaan Kragten]

There can be another reason for preference positions than the difference in between the number of armature and stator poles. This is when the stator stamping is provided with four outside grooves in which thin iron strips with bent ends are laid to connect the sheets of the stamping together. About twenty years ago I have tried to make a 4-pole generator using such a stator which was used by an Indian company and this failed because the four outside grooves resulted in four very strong preference positions per revolution. Four magnetic loops are coming out of a 4-pole armature. The loop at the right side of the centre line of a north pole turns right hand. The loop at the left side of the centre line of a north pole turns left hand. So at the centre line of the poles there is an area in the stator stamping where the magnetic flux is almost zero. If the outside groove in the stator stamping coincides with the centre line of the magnets in the armature, the grooves will almost be of no hinder to the flow of the magnetic flux in the stator and therefore this will be a strong preference position. Therefore a 4-pole armature can only be used if the stator stamping has no outside grooves.

So I decided to use the same stator stamping (with 24 poles) for an armature with 22 poles. The flux density in the bridge in between the bottom of the slots and the outside of the stator is much lower for a 22-pole armature than for a 4-pole armature and so I thought that the four outsides grooves would no longer have an influence on cogging. But this isn't the case in practice! A 22-pole armature and a 24 pole stator should have 22 * 24 / 2 = 264 preference positions per revolution. However, the real number of preference postions was much lower and 44 if I remember well. This is because there is always a position for which two opposite outside grooves coincide with the centre line of two opposite magnets and this is an extra preference position. If the armature has rotated 360 / 44 = 8.182°, this will happen for the two other outside grooves. These extra preference positions increase the peak of the cogging torque which is caused by the difference in between the number of armature and stator poles.

If a stator stamping with 36 slots and four outside grooves is used for a 30-pole armature you will get 60 extra preference positions. If the a stator stamping with four outside grooves is used for a 32-pole armature you will get 32 extra, much stronger preference positions because now four centre lines of the magnets will coincide with the four outside grooves at the same time. So the cogging problem will increase if you change from 30 to 32 armature poles and if your stator has four outside grooves. Some stator stampings have no outside grooves but four 4 mm holes. These holes can have a similar effect as outside grooves if they arn't filled with 4 mm mild steel bars.

[brandnewb]

Thanks all for pitching in.

once again I have totally missed the conversation going on here because I am not receiving emails about it.

The one in where MagnetJuice suggested to ask on stackexchange I did get so.....

Anyway I'l try rephrasing the question as to keep all eyes on the ball.

With a dual magnet disk in attracting configuration there are several ways of winding coils.

If one would wind serpentine style like the second image I shared we have a lot of unused space for coils.

What do we do with that space? leave it empty doing nothing?

If one would wind "closed loop" style, like the 3th image I posted, then we will end up with no unused space.

I have new insights in the meantime; If wound correctly then whether serpentine with no unused space or "closed loop" we could still end up with 3 phases in total.

I am leaning towards the "closed loop" because now I can have my coils in quarters and connect them together after they have been installed.

What are your thoughts?

[Adriaan Kragten]

A serpentine winding is often done for car dynamos. The advantage is that you make one big circular coil for every phase and then push it into a serpentine by a special tool. As only the radial part of the coil is effective, there is no difference in coil resistance if the coil heads are going from one coil to its neighbour or if all coil heads belong to one coil. However, a serpentine 3-phase winding has as disadvantage that you will always get three layers and so many crossing coil heads. If you look at the winding of a car dynamo, you will see that the slots are not maximal filled with copper and that is because the coil heads prevent this, especially if thick wire is used. The advantage of single coils is that a 3-phase winding can be laid as a 2-layers winding if the number of armature poles is devidable by four. A 2-layers winding isn't common for axial flux generators but very common for radial flux generators with an iron stator stamping. Assume that you have a 12-pole radial flux generator and a stator with 36 slots. 36 slots means six coils of each phase. The sequence of the coils in the first layer is U1, V1, W1, U2, V2, W2, U3, V3, W3. Within each coil, there are two empty slots and these slots are used for the second layer after bending of the coil heads of the first layer to the outside.

[brandnewb]

Thank you Adriaan, and all of you pitching in, for pitching in.

Let's keep it simple stupid for the time being.

So what would you do? "closed loop" or serpentine?

And if serpentine, what would your do with the empty space?

[joestue]

Why cant you put 3 serpentine coils in and just press the coils into position?

[Adriaan Kragten]

Thank you Adriaan, and all of you pitching in, for pitching in.

Let's keep it simple stupid for the time being.

So what would you do? "closed loop" or serpentine?

And if serpentine, what would your do with the empty space?

If you are lying a 3-phase serpentine winding, you don't have radial wire positions which are not used. You only have non used radial wire positions for a 1-layer, 3-phase winding with normal coils as for this winding, you don't use the two empty slots within each coil.

For a 1-phase winding, the number of radial positions is equal to the number of armature poles. A 1-phase winding is always a 1-layer winding whether it is a winding with normal coils or a serpentine winding.

A 1-layer, 3-phase winding with normal coils uses a number of radial positions which is a factor 1.5 higher than the number of armature poles and so a 1-layer, 3-phase winding has a coil density which is a factor 1.5 higher than for a 1-phase winding. An example of a 1-layer, 3-phase winding for an 8-pole axial flux PM-generator is given in figure 5 of my public report KD 341. In the upper picture of this figure you can see the two free positions within each coil. In the lower picture these positions are not fully available because the coil legs are made thicker.

You also have non used radial wire positions if you lay a 1-phase serpentine winding. A 1-phase serpentine winding for a 12-pole armature uses 12 radial positions. A 3-phase serpentine winding for a 12-pole armature uses 36 radial positions. So the winding density for a 3-phase serpentine winding is a factor 3 higher than for a 1-phase serpentine winding. But if you have a 3-phase serpentine winding, there are no radial positions which are not used and so there is no empty space if the wire bundles are made thick enough.

The fact that you have crossing coil heads for a 3-phase serpentine winding should not be a problem if you put an isolation sheet in between the phases, then press the coil heads together and then wind a rope around all coil heads to keep them together and to prevent vibration. The total thickness of the coil heads will still be larger than the thickness at the radial part of the coil but if the coil heads are lying outside the path of the magnets of an axial flux generator, this should be no problem if the magnets are thick enough.

[brandnewb]

thank you Adriaan.

It took a while and an experiment but I finally understand what it is you meant and you are correct it seems.

I wound both coils in my test setup and the closed loop so to speak does not work. There is some voltage generation but that is because the magnets not have all exactly the same field. Btw I realize now I should have known when using the right hand rule then the flow is in oppositie directions in both legs of the coil.

The serpent coil generates voltage really well.

Regarding no unused space. when winding diagonal closed loops we get best of both (serpent and closed loop) worlds. No unused spaces and (relative) ease of winding.

[MattM]

Those serpentine layouts look nice when it comes to simplicity.  The radial layouts look easier than axial to make. 

There is one part of axial serpentine layouts that make me nervous.  Unless you guess perfectly for length you risk having run short where you probably have to completely start over, or excess which can promote your coil rubbing.  The excess seems less of a concern with radials, but running too short is definitely possible.

[Adriaan Kragten]

Those serpentine layouts look nice when it comes to simplicity.  The radial layouts look easier than axial to make. 

There is one part of axial serpentine layouts that make me nervous.  Unless you guess perfectly for length you risk having run short where you probably have to completely start over, or excess which can promote your coil rubbing.  The excess seems less of a concern with radials, but running too short is definitely possible.

On YouTube you can find videos in which it is shown how a serpentine winding of a car generator is made. One starts with a circular coil and then one uses a rather complex machine to make the serpentine by pushing the wires to the inside at twelve positions. One also uses a machine to push the serpentine into the stator stamping. As one uses a machine, this goes always in the same way. So only once you have to determine the required diameter of the circle. But if you make the serpentine by hand from a circle, and if you have no special machine, it is not guaranteed that you will always get the same shape of the serpentine and that all lobs of the serpentine are the same.

Aonother way to make a serpentine winding is to wind the serpentine directly around a pattern of round pins. However, this has a big disadvantage. Assume that you have twelve pins lying at a large diameter and twelve pins lying at a small diameter. The bundle of wires of the first winding will make contact with the pins and all radial parts of the winding are in parallel to each other. But ones the pins are covered with one layer of wires, the second layer is laid on the first layer. This means that the second layer will lay at the outside of the outer pins but at the inside of the inner pins. So in between the outer ands the inner pins, the wires of the second layer are crossing the wires of the first layer. These crossing wires make that the winding will become much thicker than for a serpentine winding made from a circle for which all wires are in parallel for the radial part of the serpentine. This problem becomes bigger as more layers are laid.