Interleaved windings 1

[zlatkodo]

We know that in a standard three-phase windings, the maximum number of parallel circuits is equal to the number of poles. By applying the "interleaved windings," we can get more parallel circuits, which is important in high-power, low-voltage machines .
I'm interested in description (advantages and disadvantages) and rules for forming such windings. Also, maybe someone has an example of an internal winding diagram for the interleaved windings?
Zlatkodo

 

[starkopete]

Hi,
There is a real good article on the EASA website under Technical Articles titled: Interleaved Windings Provide Useful Alternative by Chuck Yung. It does a good job of explaining their application. The article is free if you are a member.
Thanks

 

[zlatkodo]

Hi. Starkopete,
Thanks. It is a useful article.
I'm not sure whether the conditions for interleaved windings can be met at least for some of fractional slot windings (not only for integral slot windings).
Does anyone have an example from practice of such a winding (internal connection diagram)?
Zlatkodo

 

[zlatkodo]

Attached is an example of interleaved winding for 24 slots, 2 poles and 4 parallel circuits , as I see it.
Comments and suggestions are welcome.
Zlatkodo

 

[electricpete]

That doesn't look right to me. Your figure shows the coils laying in the same slots in both cases.

For interspersed windings I thought it means there is an overlap between adjacent groups. For example the coils from a given pole phase group would be laid in slots 1, 3, 4, 6 and adjacent pole phase group would fill 2 and 5. That allows to increase the effective span for harmonic purposes on 2-pole motors (which require small coil span for ease of winding and economy of material).

=====================================
(2B)+(2B)' ?

 

[electricpete]

I guess I was describing "interspersed" rather than "interleaved"

I am looking at 2 articles in EASA Tech Manual..one by Chuck Yung and one by Dave Gebhart. (What is the author of the document you guys referred to?). According to those:

Interspersed refers to the fact that the pole phase group does not occupy consecutive positions. That is what I was referring to above : 1, 3, 4, 6

Interleaved has to do with connections of a single group in 2 parallel circuit paths.

In the articles I mentioned these two techniques are used together. Is there a benefit to interleaved connection for non-interspersed groups?

=====================================
(2B)+(2B)' ?

 

[starkopete]

Electripete,
You are correct, interleaved is when each coil group is divided to form two parallel paths. I believe that is what Zlatkodo is asking about. If I'm not mistaken, that would make the first & third coil in the group one path and the second & fourth coil the other path.

Thanks

 

[zlatkodo]

Hi, Starkopete,
If you choose the first & third coil in the group as one path and the second & fourth coil the other path then you will have an electrical angle of 15 degrees (8 %) between midpoints of two paths and consequently a large circulating current.
But if you chose the first & fourth as one path, and the second & third coil the other path, then electrical angle between midpoints of two paths is 0 ( no circulating currents).
Because, in that case, midpoint of main group coincides with both midpoints of two paths.
For this example, I think, this is the only possible solution , if we want to avoid the circulating currents.
Am I right?
Zlatkodo

 

[electricpete]

But if you chose the first & fourth as one path, and the second & third coil the other path, then electrical angle between midpoints of two paths is 0 ( no circulating currents).

fwiw, what you said makes perfect sense to me - that any parallel paths have to have the same induced volage.

But now here is something that confuses me attached - it is an excerpt from the EASA Tech Manual regarding interspersed winding. It represents a 2-pole 48 slot machine. There are 48/(2*3) = 8 coils per group. The interspersing part as we mentioned means that the bottom coils of a given group may lie in slots numbered 1, 3, 4, 5, 6, 7, 8, 10. (one slot near each end allocated to adjacent group).

For some reasons that I don't understand, it can be beneficial to interleave these 8 coils per group into two parallel sub-groups of 4 each. That is the figure shown.

Now the other part I really don't understand is these two parallel subgroups don't seem to have the same voltage based on their position in the winding (their centerlines are different) and I'd think we'd have circulating current problems.

Obviously I am missing one or two things to understand this picture...



=====================================
(2B)+(2B)' ?

 

[electricpete]

Now looking at my last attachment again, I see that there is a pretty good symmetry of both subgroups about the midpoint of the group... it is just a little harder to see because it is not a concentric pattern about that midpoint.

The midpoint is halfway between where terminals 7 and 8 are connected... so halfway between the 4th and 5th of 8 coil connections. If I call that 0, and everything to the right positive and everything to the left negative, then the red subgroup has -3.5, -0.5, +1.5, +2.5 and the blue subgroup has -2.5, -1.5, +0.5, +3.5. Each one has a sum of 4 on each side of the midpoint, which roughly (or exactly?) means symmetrically distributed about that midpoint.

So I have answered my confusion about the attachment and I agree with zlatkodo. If there were 4 consecutive coils of a group in sequence than the only way to form two 2-coil subgroups in parallel would be the way he showed (1,4 in one subgroup and 2,3 in the other... in that case each subgroup concentric about the group midpoint of the group).

=====================================
(2B)+(2B)' ?

 

[starkopete]

zlatkodo & electripete,
I stand corrected. You are both correct on the 1-4, 2-3.

Thanks

 

[starkopete]

Electripete,
Chuck Yung from EASA wrote an article in the January 2003 Currents that uses your example as his sample problem in the paper. The article is titled Interleaved Windings Provide Useful Alternative (the article can also be found on EASA’s website under Technical Article archive). It is not the same article as the one in the Tech Manual.

Thanks

 

[electricpete]

Thanks s-pete. I'll look for that.
[electricpete QUOTE]If I call that 0, and everything to the right positive and everything to the left negative, then the red subgroup has -3.5, -0.5, +1.5, +2.5 and the blue subgroup has -2.5, -1.5, +0.5, +3.5. Each one has a sum of 4 on each side of the midpoint, which roughly (or exactly?) means symmetrically distributed about that midpoint.[/QUOTE]
Hmmm, my coil numbering omitted the missing coil that belongs to another phase. Should've been more like -4.5, -0.5, +1.5, +2.5 and the blue subgroup has -2.5, -1.5, +0.5, +4.5. No longer seems centered. I'll have to think about that some more.

=====================================
(2B)+(2B)' ?

 

[starkopete]

e-pete,
Chuck uses a differant numbering system, he simply numbers them 1 - 8 from left to right. Using his system you'll end up with: red subgroup 1+4+6+7=16 and blue subgroup 2+3+5+8=16. I think you were right the first time.

"If I call that 0, and everything to the right positive and everything to the left negative, then the red subgroup has -3.5, -0.5, +1.5, +2.5 and the blue subgroup has -2.5, -1.5, +0.5, +3.5. Each one has a sum of 4 on each side of the midpoint, which roughly (or exactly?) means symmetrically distributed about that midpoint."

Thanks

 

[electricpete]

Hi pete
Thanks for responding

Chuck uses a differant numbering system, he simply numbers them 1 - 8 from left to right. Using his system you'll end up with:
red subgroup 1+4+6+7=16 blue subgroup 2+3+5+8=16. I think you were right the first time.

I agree it doesn’t matter where the counting starts (middle or any other convenient point) because if the sum counting from center is the same, then the sum counting from any other position will be the same (just take the first sum and add the slot-difference between ref points to each sum… equal sums remain equal).

BUT since we are adding voltage vectors associated with assumed sinusoidal flux, we need to make our count represent position around the core… and the only way to do that is to make our counter change with every slot whether or not associated coil belongs in our group. So I think Chuck must have just been using numbers to identify coils within a group without doing any comparison of voltages.


If slot position numbering includes all slots (including those in adjacent groups), counting from the left as shown attached you have
For the red subgroup: 1+5+7+8=21
For the blue subgroup: 3+4+6+10=23

Or counting from the centerline:
For the red subgroup: -4.5, -0.5, +1.5, +2.5 = -1
and the blue subgroup has -2.5, -1.5, +0.5, +4.5 = +1

Either way, the totals are not the same… doesn’t look anywhere close to being similar induced voltages (*). I’m sure I’m missing something because I have a hard time believing EASA would publish something that creates large circulating currents… just not sure what it is that I’m missing and open to suggestions.

(* by the way I think the sums being the same is a necessary but not sufficient condition for same voltages… if we have the same sum but in a pattern that is not concentric about the centerline then we might expect approximately but not exactly the same voltage when we perform the vector sum).

=====================================
(2B)+(2B)' ?

 

[starkopete]

e-pete,
Thanks for the numbering on the drawing, I now notice that you are both interleaving the group and interspersing it with the adjacent groups. I'll have to look at it a little harder.

Thanks

 

[starkopete]

e-pete,
If I'm looking at this right we are ending up with a 3.75 degree displacement between the subgroups. It appears that that is as good as its going to get, or am I wrong. I agree with you about EASA, when I get a chance I will contact them for an opinion.

Thanks

 

[starkopete]

Or is it 3 degree displacement?

 

[electricpete]

I agree with your estimate of phase angle difference between the groups.

The easy/ballpark estimate: When we added the slot positions of the 4 coils in each group, the total differed by 2 slot positions. That means the average slot position changed by 0.5 slots. Since each slot corresponds to 360 degrees/48 slots = 7.5 degrees per slot, the average slot position in these two subgroups differs by approx 50% * 7.5 = 3.75 degrees.

The exact vector solution is a 3.18 degrees difference between as shown in attached.

=====================================
(2B)+(2B)' ?

 

[electricpete]

I took a look to see if the EASA balance could be improved, including we wanted to fit a group of 8 coils into two subgroups of 4 within 10 slots (not allowed to go in the 2nd or 9th position).

The possible solutions could be identified by a sequence representing what goes in each slot: 1 for 1st subgroup, 2 for 2nd subgroup, X for neither (i.e. in the 2nd and 9th positions).

Attached are vector results for 3 possibilities in order from worst to best (according to the vector difference in induced coil voltages).

1 - The EASA approach: 1X221211X2.
Phase angle difference: 3.2 degrees
Magnitude difference: None
Vector difference: 0.21
(vector difference is expressed as multiple of single coil voltage, since single coil voltage was assumed to be 1.0).

2 - Concentric approach: 1X22112211X1
Phase angle difference: None
Magnitude difference: 0.20
Vector difference: 0.20

3 - Better approach: 1X222111X2.
Phase angle difference: 0.8 deg
Magnitude Difference: None
Vector difference: 0.05

So it seems option 3 is best from the standpoint of balancing the voltage among subgroups. Also it seems a little bit simpler because than the EASA approach (EASA approach broke each subgroup into 3 sub-sub-groups... option #3 breaks each subgroup into 2 sub-sub-groups) and i'd think maybe a little easier to connect since it uses more simple series jumpers (4) than the EASA approach (which uses 2).

I'm probably missing something.

Anyway, sorry to hijack the thread. In addition to not knowing why EASA presents the solution they do (which seems needlessly unbalanced), I don't understand why we would interleave. The best reason I can think of would be as a means to increase the number of parallel paths in order to minimize unbalanced magnetic pull, but I'm not sure it that's why it's used.

=====================================
(2B)+(2B)' ?