The beginnings and ends of phase windings 2

[zlatkodo]

Distance between the beginnings of three-phase windings is , generally, two thirds of the full pitch.
But this may not always be so, it depends on which shortened pitch is used.
For example: where are the beginnings for three phase winding, 33 slots, 8 poles, double-layer:

- the beginnings of the first, third and fifth pole-phase group or
- beginnings as shown in the attachment?
Which option is correct? Which is better?
Is somewhere I can find a detailed analysis of this topic?
Zlatkodo

 

[zlatkodo]

Electricpete,
Thanks for the attachment.
I have had the opportunity to see several similar opinions, but also the opposite (one is attached). Unfortunately, it is not in english but there is a conclusion: if the distance between beginnings is not equal to n * 120 degrees, the motor has increased noise, reduced starting torque, and if we are talking about a generator, then it gives unbalanced voltages.
One thing I know, that I certainly will not make a mistake if I make a beginnings in accordance with n * 120th .... In addition, if we know in which slots phase starts , then we can more easily determine the winding arrangement for the case when we have no pre-defined templates.
By the way, I'll make a small program to determine the phase beginnings. Another, more important program to calculate all the winding data, I've already made (see:

http://megaswf.com/serve/25158/

this is not advertising - is not for sale.)
Zlatkodo

 

[electricpete]

Hi zlatkodo.

I respect your opinion in general - you certainly know a lot more about winding principles than me and there’s a lot I can learn from you.

But on this one specific point, I am 100% positive of the conclusion: The top and bottom diagrams will perform identically. It is a matter of physics, not opinion. The basic principles are illustrated in the mental thought experiment described above. I would be glad to discuss it if there is specific aspect that you disagree (do you doubt the conclusion for the box... or think the box does not adequately represent a motor?).

attached... there is a conclusion: if the distance between beginnings is not equal to n * 120 degrees, the motor has increased noise, reduced starting torque, and if we are talking about a generator, then it gives unbalanced voltages.

He is either talking about grouping of coils (rather than connection of T-leads) or he is wrong.

In addition, if we know in which slots phase starts , then we can more easily determine the winding arrangement for the case when we have no pre-defined templates

I agree. Determining coil grouping and determing location of T-lead attachments are two separate things. What applies to determining coil grouping (requires knowledge of slots that are exactly 120 degrees apart electrically) does not apply to determining locations of T-leads (T-leads need not be attached at location exactly 120 degrees apart... for single-circuit winding, any connection that includes all coils in proper polarity is equivalent).

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

 

[zlatkodo]

Electricpete,
Unfortunately I can not see a whole chapter from a book (Liwschitz) that you previously sent in the attachment, but there is one thing I'm interested in (see my attachment). This is ,the mentioned in the book, case q = 1 5 / 13, for example: 108 slots and 26 poles.
Does this mean that in this case we can not get unbalanced winding with the proposed order of beginnings?
If we define the beginnings from U -1, V - 37, W - 25 instead of from slots 1, 7, 13 (as suggested in the book), whether such a winding is to be balanced?
Note: I do not have this book and I am unable to read the entire article.
Zlatkodo

 

[zlatkodo]

Sorry, there is a typo:

If we define the beginnings from U -1, V - 37, W - 25 instead of from slots 1, 7, 13 (as suggested in the book), whether such a winding is to be balanced?
It should be:
U -1, V - 37, W - 73
Zlatkodo

 

[electricpete]

Analyse
26 poles, 108 slots
q = 108 / (3 * 36) = 18 / 13 = 1 5 / 13
Using Liwshitz-Garik notation
a = 1, b = 5, beta = 13 = number of poles in repeating unit
N = 18 = slots per phase in repeating units
8 * 1 + 5*2 in 13 slots 18/13
26/13 = 2 recurrent groups
P is lowest integer to satisfy d = m * N * P + 1 / beta is integer
P = 6 makes d = 25
d = m * N * P + 1 / beta = (3 * 18 * P + 1) / 13 = 25

Layout the slots in phase A:
=1
=1+d=26
=1+2*d=51
=(1+3*d - 3*N)=22
=1+4*d-3*N=47
=1+5*d-6*N=18
=1+6*d-6*N=43
=1+7*d-9*N=14
=1+8*d-9*N=39
=1+9*d-12*N=10
=1+10*d-12*N=35
=1+11*d-15*N=6
=1+12*d-15*N=31
=1+13*d-18*N=2
=1+14*d-18*N=27
=1+15*d-18*N=52
=1+16*d-21*N=23
=1+17*d-21*N=48
Thus the following slots belong in phase A: 1 2 6 10 14 18 22 23 26 27 31 35 39 43 47 48 51 52
Repeating for other phases, we come up with the pattern:
2 1 2 / 1 1 2 / 1 2 1 / 1 2 1 / 1 2 1 / 2 1 1 / 2 1 2 / 1 1 2 / 1 1 2 / 1 2 1 / 1 2 1 / 2 1 1 / 2 1 1
2 1 2 / 1 1 2 / 1 2 1 / 1 2 1 / 1 2 1 / 2 1 1 / 2 1 2 / 1 1 2 / 1 1 2 / 1 2 1 / 1 2 1 / 2 1 1 / 2 1 1

The distribution factor is 0.955 in all three phases. You could wind it in one circuit or two circuits.

Assuming one circuit, you could start the winding anywhere you want as long as polarity is not changed. Starting in U -1, V - 37, W - 73 would give the same results as starting in 1, 7, 13. The distribution factor for each phase can be calculated by adding up all the voltages induced in each coil within the phase... doesn’t matter what order we add them – the sum is the same. Both sums are the same and they are both balanced.


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

 

[electricpete]

Note when we talk about "starting the winding" in different places, I am not talking about changing the coil grouping, just changing the location where the T-leads are connected (but keeping the same coil grouping).

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

 

[zlatkodo]

Electricpete,
I also think this is a good winding arrangement but I would still use the U -1, V - 37 W – 73.
For others who are not familiar with this theme, some more informations.
There may be more proper arrangements for motor with fractional q.
If q = b + c / d, then we have a "c" proper arrangements.
In our case, q = 1 5 / 13, we have 5 of these arrangements, as follows:
211,211,212,112,1...repeat,
211,212,112,112,1...repeat,
211,212,112,121,1...repeat,
212,112,112,121,1...repet,
212,112,121,121,1...repet.

These arrangements can be defined in several ways. It may also be part of a computer program for three-phase motor design.
Zlatkodo

 

[electricpete]

Those are more compact ways to describe the winding.

I think it is all effectively the same winding. In fact we can generate 13 variations of the same winding.

The winding I described was 2 "recurring units": one on the first line and one on the 2nd line below:
2 1 2 / 1 1 2 / 1 2 1 / 1 2 1 / 1 2 1 / 2 1 1 / 2 1 2 / 1 1 2 / 1 1 2 / 1 2 1 / 1 2 1 / 2 1 1 / 2 1 1
2 1 2 / 1 1 2 / 1 2 1 / 1 2 1 / 1 2 1 / 2 1 1 / 2 1 2 / 1 1 2 / 1 1 2 / 1 2 1 / 1 2 1 / 2 1 1 / 2 1 1

We had N=18 and beta = 13. Each recurring unit is 3*N = 54 coils, located within 3*beta = 3*13= 39 groups. The two recurring units together give the required 108 coils in 78 groups (26 poles). As you observed, the sequence of numbers (coils per group) within a 39-group recurring unit is periodic at a frequency of 54/3 = 13. (i.e. the pattern of 39 numbers is really 3 repeating patterns of 13 numbers). Liwschitz-Garik doesn't mention this, but it makes sense that after we count off beta = 13 groups, we have traveled N=18 slots, and we have traveled an electrical angle of N * alpha = N*180/(3*q). Substituting q = N/Beta, we have travelled N*180*Beta/(3*N) = 60*beta or an exact integer multiple of 60 degrees. Since the Liwschitz-Garik's "slot star" represents all 3 phases in 180 degrees, the phases are considered 60 degrees apart (taking into account allowed polarity swap to keep the entire slot star in a 0-180 degree range). After we travel those N slots in Beta groups and exactly 60 electrical degrees, we must land on the "same position" (**) within the next phase. Since the phases are symmetrical (when beta is not multiple of 3), we must be starting the coil grouping pattern all over again at that point. Therefore the pattern must repeat itself when we land at the same position in the next symmetric phase after beta = 13 groups.

Since the pattern is periodic with interval beta = 13 groups, we can generate 13 variations just by starting at a different group within the 13-group periodic pattern each time.

Starting at the 1st group within my "recurring unit":
2 1 2, 1 1 2, 1 2 1, 1 2 1, 1, repeat....

Starting at the 2st group within my "recurring unit":
1 2 1, 1 2 1, 2 1 1, 2 1 1, 2, repeat....

Stargting at the 3rd group within my "recurring unit":
2 1 1, 2 1 2, 1 1 2, 1 1 2, 1, repeat....

And continuing starting at the 4th thru 13 group within my "recurring unit":
1 1 2, 1 2 1, 1 2 1, 1 2 1, 2, repeat....
1 2 1, 2 1 1, 2 1 1, 2 1 2, 1, repeat....
2 1 2, 1 1 2, 1 1 2, 1 2 1, 1, repeat....
1 2 1, 1 2 1, 1 2 1, 2 1 1, 2, repeat....
2 1 1, 2 1 1, 2 1 2, 1 1 2, 1, repeat....
1 1 2, 1 1 2, 1 2 1, 1 2 1, 2, repeat....
1 2 1, 1 2 1, 2 1 1, 2 1 2, 1, repeat....
2 1 1, 2 1 2, 1 1 2, 1 2 1, 1, repeat....
1 1 2, 1 2 1, 1 2 1, 2 1 1, 2, repeat....
1 2 1, 2 1 1, 2 1 2, 1 1 2, 1, repeat....

But, assuming that all coils are identical, and that there is nothing unique about the 3 phases other than a relative phase relationship, and recognizing that the location where we choose to attach the T-leads doesn't affect anything we care about electtrically or magnetically (as long as correct polarity is maintained), then these 13 variations are all effectively the same winding.

(** "same position" with respect to the pattern that allows flipping coil polarity to roll the slot star into 180 degree window instead of 360 degree window).

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

 

[koizumi]

Hi electricpete,

After a couple of days out of my office, back here seeing your discussions I myself am thinking of changing my concept of fractional q windings. My English is not very good, and I haven't got access to valuable sources of information, such as the book "Electrical Machinery", by Liwschitz-Garak and Clyde Whipple that you brought forward.

I agree that there are some arrangement variations in one given fractional q winding. In practical point of view, only one or two best option shall be chosen.

I'll try some practical experiment on a winding, at first about changing the position of T-leads, then file a record to you all. But it may take some time.

 

[zlatkodo]

Koizumi,
You wrote:
I'll try some practical experiment on a winding, at first about changing the position of T-leads, then file a record to you all.
Thank you for sharing your practical experiences with others. That is why I voted a star for your post.
I think that this post was useful for all, (there are lots of free, useful informations for rewinders, even for those who do not wish to share their experiences with others).
Zlatkodo

 

[electricpete]

This post seems to be dieing into the background.

I was a little disappointed that no-one else weighed in on the question asked 10 Sep 10 12:45 and answered by me 10 Sep 10 15:43.

It seems to me that in a forum full of electrical engineers, it should not be so hard to agree that from in an electric motor, there is absolutely no difference (*) if we change the order of coils in a series circuit without changing their polarity or physical location. This comes from basic analysis of the electric circuit and the magnetic circuits.

(* of course it is limited to phenomenon of interest to electric motor and does not include high frequency wave behavior.)

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

 

[zlatkodo]

Hi. Electricpete,
Also, I'd like to hear other rewinders.

Assuming one circuit, you could start the winding anywhere you want as long as polarity is not changed. Starting in U -1, V - 37, W - 73 would give the same results as starting in 1, 7, 13.

But, assuming that all coils are identical, and that there is nothing unique about the 3 phases other than a relative phase relationship, and recognizing that the location where we choose to attach the T-leads doesn't affect anything we care about electtrically or magnetically (as long as correct polarity is maintained)

You should clarify something. Does this mean that the phase-beginnings can also be the beginnings of the first, second and third group?
BTW, I just finish my program, which refers to the determination of the symmetrical winding arrangement for all the double-layer, lap windings (for q> = 1), for all slot-pole combinations in the range 12-300 slots, 2-60 poles.
For a few seconds, the program provides :
- choice of one of the proposed symmetrical arrangements,
- determination of symmetrically distributed phase-beginning and phase-ends,
- the possibility of parallel circuits,
- recommended step, etc. See an example for 75 slots, 14 poles here:

http://img412.imageshack.us/img412/6587/95904427.png

http://img151.imageshack.us/img151/725/76535586.png

If any of rewinders must verify the arrangement in a particular case, let's feel free to contact me.
Zlatkodo

 

[electricpete]

You should clarify something. Does this mean that the phase-beginnings can also be the beginnings of the first, second and third group?

I'm not sure exactly the distinction that you're drawing. The physical position of coils in a given phase wouldn't change and their polarity wouldn't change. As long as you meet those conditions, then you can connect all coils of one phase of a single-circuit winding in any order you want and performance will not be affected. If "beginning" refers to which coil is next to the T1, and "end" refers to which is next to T4, then yes the beginning and end can change.

One illustration is the top and bottom windings you posted 10 Sep 10 12:45... they will perform identically. If you have specific configuration to illustrate your question, maybe it would help me understand your question.

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

 

[rhatcher]

Pete is right about the lead locations and everything else he has addressed in this thread. As an example on the discussion about lead locations, the following windings are essentially equivalent except for the need to change the number of turns to match the number of circuits.

 

[rhatcher]

Only one attachment per post. Here is the second drawing with two windings included.

 

[rhatcher]

And the last attachment with two more connections shown. These are actual connection diagrams from motors we have wound in our shop.

The 1-10 jumper connection marked (A) is from the EASA connection book. The 1-10 jumper (B) is a modification of this winding. The 1-10 jumper (C) connection is my version designed to place all of the leads on the outside and in repeating order for ease of winding. The 1-7 Jumper marked (book) is another from the EASA connection book and the 1-7 jumper connection marked (mod) is my version of this connection, again designed to place the leads on the outside, in repeating order, and grouped together for ease of winding.

 

[zlatkodo]

Hi, Rhatcher,

Which is slot-number for the last example ?
Zlatkodo

 

[rhatcher]

Zlatkodo,
The connections are not dependent on slot numbers. The pole groups that are represented could be one coil (18 slots) or 100 coils (1800 slots - an unlikely number but not impossible).

 

[zlatkodo]

Hi, Rhatcher,
Please read my first post.
Here we talk about:
- what is the distance between the phase-beginnings (expressed in number of slots) and
- whether that distance is equal to two thirds of the full pitch ((expressed in number of slots).
What is the number of slots in your case ?
What is the distance between the phase-beginnings in your case?
By the way, the distance between the phase-beginnings is the same for both cases that you mentioned (see attachment).
Zlatkodo

 

[electricpete]

Ray
Thanks for weighing in. When I saw you showed back up on the forum in the 4kv/13kv thread, I had a feeling you might be interested comment on this, so I bumped it to the top. I vote you LPS joining the discussion. By the way, another discussion that I have some questions (for which I don’t have answers) is the discussion that emerged about wiring opposite poles in series or parallel here thread237-281260


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