My FAILED attempt at an Induction Conversion.

[sPuDd]

This is embarrassing. Some time back I spotted this little unit on

eBay And picked it up for AU$50. It's a 3-phase 250W (1/3hp)

unit. Although It claims to be an Australian company, I suspect

it was made in China. The messy windings are a dead giveaway.

First look. The connection box is almost as big as the motor. Cute

Spec plate. It's a two pole, so the RPM is very high. Not looking good already.

Inside the connection box. Nice and tidy. Easy Star/Delta change.

Rotor being removed. I noticed the rotor actually rubbed against

the stator slightly when spun by hand. Also the IP65 seals on each

bearing made it tight.

As you can see, it's an 18 slot and the windings are very fine

magnet wire. I realized quickly that a two pole would need to

be rewound. Bad move sPuDd..

Rotor is only ~50mm (2 inches) diameter. One of the bearings

was also very stiff.

Here you can see some of the windings that have wandered

off on their own. Looks like a bad hair day for the windings.

Snippidy-snip-snip... Those windings are harder to cut than they look.

Destroying things is almost as much fun as making them

I found that putting the stator in the lathe chuck and milling off

the windings was a good way to do it. I did damage one or two

laminations, so I milled them off too. Hence the ring you can see

around the stator teeth. After that I used a pin punch to hammer

the copper out of the slots. Also made the insulation come out.

I was intending to use some 5mm (3/16 inch) thick x 13mm (1/2 inch)

round Neo's I had. I machined the rotor down and stuck some on to

see how it looked in the stator. Notice the damaged motor casing.

I did that when it came loose in the lathe chuck.

Machined rotor with 6x magnets on it. I figured I could rewind the

motor for 6 pole 3 phase. Tsk tsk tsk... biting off more than I can chew.

I figured I needed a 20 degree skew for an 18 tooth stator. So I

super glued three magnets for each pole for testing a new coil.

Here you can see the skew. There was about a 1-2mm gap

between the magnets and stator teeth.

The stator with insulation in the slots. I hand wound three test coils

of different wire diameters, and found that there was every chance

I could produce power from this thing. But...

Two problems:

1. I can't wind motor coils for s#$!  I envy Zubbly for being skilled

    enough to do this. Its just not my cup of tea.

2. When I spun the rotor, it cogged so bad I needed pliers to turn the

    shaft. When I used a drill - it sounded like I was hammer drilling

    a 2 inch hole in solid concrete. Damm awful.

So, its been scrapped for parts. We must all have one of these projects

to remind us of our strengths & weaknesses.

[dinges]

Hmmm... Curious too why it cogged so badly. When you say you skewed magnets 20 deg. I disagree, it's much less, but not relevant here as an explanation for the cogging, because as can be seen in the 3rd picture from below, your magnets are exactly on one line of your original rotor. Very strange indeed. Only 'explanation' I have is that you use few large magnets, whereas Zubbly usually uses many small ones (relatively speaking, relating to the size of the rotor).

I'm very curious what the guru has to say about it

The rewinding is something that can be learned and takes a bit of hard work, but seems doable. That cogging though really worries me... Because I can't explain it.

Peter.

[dinges]

Just thought of something else: before filling an entire rotor with expensive magnets, it could be an idea to just fill one 'line' (not even necessarily an entire pole) with magnets. So, in your case, just 3. Put it in the motor and turn it. If it coggs now, it will when entirely filled with magnets. If it doesn't, it shouldn't cogg with a full rotor. At least, that's my present idea

That way one would save on magnet costs. And frustration too, maybe.

Peter.

[dinges]

And while I'm at it...

In your case, the ratio of poles:stator-slots = 1:3 (6:18)

In Zubbly's case it's usually 1:9 (4:36)

Perhaps this could have something to do with it? Just guessing here.

If I were you, I'd knock off the magnets and try a 4-pole configuration, with same skew. If it still coggs, try 2-pole with same skew as you have now.

I'd be VERY interested in the results of such an experiment. Consider it a learning experience, and if it doesn't cogg anymore, well, then there only remains the hurdle of making new coils.

Guess someone sooner or later has to make such an experiment if we want to solve this cogging problem once and for all.

Peter.

[zubbly]

hi sPuDd,

lol!  welcome to the failure club. beleive me, i have had my fair share of them, but its all part of the learning curve.

it may not all be lost yet, it looks like it could still make a nice little unit.

reason i never liked the idea of using 2 pole motors is that the rotor is quite narrow in comparison to 4 or more pole motors. thats my main beef, but they still can be done.

your magnet width should be as close to possible as the width of 3 stator teeth (not the wide part at the slot opening but just down on the vertical section of the stator tooth) a little narrower is ok though. where i should sometimes have 1.25 inch diam mags, i can get away with using 1 inch wide without problems. your mags look a little narrow and suspect that even with the proper 20 degree skew, it is just working out that the width of the magnet just happens to be the same width as approx 2 stator teeth. you may be able to try and shift say the end mags slightly one way and the middle mag slightly the other way to try and find the ideal spot.

i suggest moving you mags around a bit. keeping them in a straight line and possible increasing or decreasing that skew may work out. if you can reduce the cog to an acceptable level, then go for it.

a few other pointers. when cutting your insulation for the slots, make sure to extend past the end of the laminations a bit so the magnet wire will not contact the lamination. you can wind the stator with only 9 coils, 3 per phase. just try winding each coil seperately around 2 wooden dowel pins. pick a dowel pin that gets as close to the coil span of 1-4 as possible.

actually sPuDd, what has happened may be the most valuable learning curve you will get. there is nothing more interesting when something does go wrong, you get to analyse it, find the problem, come up with a solution, correct it and have it work in the end.

anyway, i admire your efforts and look forward to your solution.  

zubbly

[sPuDd]

Dinges,

I see in hindsight (the act of looking out ones posterior (see politician))

that perhaps I should have experimented by using smaller diameter

mags, or different skews, etc. Your and Zubbly's suggestions tingled

the curiosity in me. But I have since used most of the parts for new

projects (I waste not The shaft, bearings and end plates are to become

a mini quad rotor experiment using 96 of those little mags you saw.

More on that soon...

Zubbly,

I gave the winding on dowels etc a go, but trying to fit the thing

back into the slots was a nightmare. I also realised I should have

made the stator slot insulation longer - after I cut it. I will learn

Just getting my fingers inside the stator was a band-aid commercial.

I realise that slot winding is not a skill I possess at the moment. So

I will wind coils from scratch for my flat rotor machines, and pick

4, 6, 8 & 12 pole 3 phase motors for conversion of the rotors only.

Don't despair, I have not given up - merely singled out my current

skill level and focused it in order to achieve the same end.  R.E.

[Gary D]

Is the cogging/resistance possibly due to the ends of the rotor not being turned down enough? Could be the flux is travelling the path of least resistance into the laminations (at the ends?).... Just a possible issue, please take it with quite a few grains of salt, or a whole container if you like?...Haven't tried this (no machinery handy) yet... Gary D.

[dinges]

Spudd,

Nil desperandum!

I've solved your problem. Expect a bill soon.

See this picture:

It's immediately obvious that there are 2 magnets (1 and 3) that are exactly over a stator slot, whilst there's only one (nr. 1) that's in between. So, 1 and 2 cancel eachother. But nothing cancels magnet 3! So now you have 6 magnets (1 from each pole) that are uncancelled. No wonder it coggs so terribly. Notice that the skewing angle (in one plane) is 13 deg, whilst in another plane it's 20 deg. Think about it and it will become obvious (sooner or later )

We can go to an even more extreme case: only 2 magnets. Now, the standard rule says that each magnet should be above a different stator slot. But in that case, there's no magnet in between! So instead of cancellation of cogging, you get amplification!

Now it also is obvious why Zubbly had so little problems with cogging: he uses many magnets, say 7 or 9. In that case, a small error (of the 8th or 10th magnet) has much less relative importance than in the (extreme) case of 2 magnets/row or, in your case, 3 magnets. Conclusion: the less magnets, the more important it becomes to skew correctly!

The case of 2 (and) magnets are limiting cases, which show the importance of correct skewing. When using more magnets, things seem to be less critical.

The solution? See picture below.

Basically I've added an imaginary magnet, nr 4. This is only done as an aid whilst correct skewing angle. Notice that magnet 4 and 1 have the same relative position over the stator, but that magnet 3 is a bit 'short' of the position it had in the picture above.

When you think about it and try to visualize why the last (real) magnet has to be a bit 'short' of position, it becomes obvious, eventually

If you don't understand this all, I'd say, don't despair. I've solved your case. One should skew a bit less than the original marks on the rotor indicate. Why? Because now we're using discrete magnets, whilst the rotor was 'continuous'. Like the difference between differential equations and difference equations, for the mathematically inclined. Discrete steps vs. infinite small steps of change.

The solution: when drawing, add an imaginary extra magnet which should sit at the same relative position as your first magnet. Use this as an aid to determine position of the in-between magnets. And you're finished, without resorting to specialized software (CAD).

I'd really like to know whether my solution (I'm 100% sure myself) is correct for you. So, I'd ask you to take off all the magnets, and put just one row back (3 magnets), but skewed at an angle of 6 (six!) degrees. This should give you your correct skewness. If you want to you can fix all your magnets too. Put it in the stator, rotate it. And I think (know) you will be VERY pleasantly surprized.

Hope all this makes sense to youse. I feel I've got it all figured out, it's just hard for me to explain.

It's all obvious now. Every problem is easy once it's solved.

Be expecting the bill soon

Peter.

[dinges]

Correction: make that 9 (nine) degrees of angle, not 6 (six). So much work to explain the nitty-gritty details and then you goof up in a major way: a 6 looks just like a 9, upside down...

9 deg. is very close to 10deg, which is half of your original 20 degrees... Coincidence or not? Don't know.

Peter.

[dinges]

And while I'm at it, why not completely take over your thread

You say in the first post 'this is embarassing'.

No it isn't. I think this is one of the best & most important postings I've seen w.r.t. motor conversions. And it's very interesting because it describes what went wrong! Which gives us an opportunity to IMPROVE without each one having to make the same mistake as you and having to learn the hard way.

THANKS FOR SHARING THIS 'FAILURE'.

(or, as my physics teacher used to say: experiments will never go wrong. It's just that sometimes, they have unexpected outcomes.)

[Flux]

Sorry I don't follow this concept of skewing 10 deg for 36 slot or 20 deg for 18 poles.

Those angles are between adjacent slots, I can't see what that has to do with the skew to reduce cogging.

With straight edged magnets the skew should be 1 slot over the length of the core, this angle will depend on the core length. If the core is a foot long it will be a small angle, if it is 2" long it will need a larger angle. I think Peter has got to grips with the idea.

What happens with round magnets I have no idea. If the magnet diameter is right for the slot geometry then it may work with no skew at all. I see no chance of calculating what you need.

You need the average reluctance torque to be zero, something very difficult to visualise with other than straight edged magnets. Seems to be down to trial and error to me. Once someone has settled it for one case then it should work for other cases of the same slot dimension and magnet.

Flux

[kitno455]

that is because there are two ways to think about 'degrees' here. you are thinking of a piece of paper, wrapped around the rotor, with an angled and a 'parallel to the axis' line drawn on it. when you unroll the paper, the number of degrees between the two lines will vary based on the length of the stator core.

the other way to think about it is a cylindrical coordinate system, look at the rotor from the end while the paper is still wrapped around it. draw a radial line on the rotor lam closest to you, from the centerline of the rotor out to the point on your paper where the skewed line meets the end lam. now do the same at the other end of the rotor. now pretend the rotor is transparent and look down along the axis. 10 degrees between the lines you drew on the two end lams.

allan

[Flux]

Allan

I follow your last paragraph, but from the pictures I see, I don't think that is what people are doing. I am sure that was what Zubbley intended and if you do it with a weighted line as he suggested then it will work but I think people are measuring 10 deg across the surface of the rotor. I may be wrong but many of the rotor pictures seem to be skewed far too much.

Flux

[kitno455]

agreed- i think folks have got the two concepts mixed up. i also think the focus on number of degrees clouds the actual point:

try not to have a situation where lots of magnet edges line up with any edge of a stator tooth at the same time (leading or trailing). based on the magnet size, this might actually mean that you have to deviate from a straight line even.

allan

[dinges]

Agreed with all except the deviation from a straight line.

Zubbly is correct in what he does when skewing by 1 stator slot. But, I think some people simply take the angle (360/#stator slots) and then project that in another plane, i.e. put that angle on a piece of paper and wrap it around the rotor core. And this is NOT correct. See my CAD drawings, how 20deg skew translates to 9 degrees in another plane. Can't put it any more clearer than in those pictures.

That's what I've been trying to explain for more than a week now and I don't feel I'm making much progress. It's been a long time since I've been this sure of something

But there's something else at work. One shouldn't put magnets exactly over the original skewing on the rotor. The best way to think about this is by looking at the limiting case: 2 magnets, each exactly over the middle of a stator slot. Because, hey, that's the way it's supposed to be done, right? Wrong. In that case you WILL have severe cogging. Now, take 3 magnets. The outer 2 are exactly over the stator slot, while the middle one is in between. Magnet nr. 1 & 2 'cancel' eachother, w.r.t. cogging. But what about magnet nr. 3? This will be the one that causes the cogg!

The solution, as I've put it, is to skew by a slightly smaller angle than the original one. The simplest way of doing this is by adding an imaginary last magnet, and making sure it ends up at the same relative position (but one slot away) as the first magnet. Draw a line between the first magnet and the imaginary last one. Put your 'real' magnets evenly spaced on that line. 'remove' your imaginary magnet, and you're done.

I determine it differently, with a CAD package (that's how I accidently stumbled upon this all). Easier for me that way, but I don't think everyone has access to such a program.

Hope I'm at least a little bit clear. To be honest, I've done my best trying to explain it both in my diary as in other threads. I think (know) that those that follow the corrected skewing-rule will have no cogging problems. And like I said, it's clear why Zubbly didn't have much problems with cogging: he usually uses much more magnets than those of the limiting case of 2 magnets. Except Spudd came dangerously close with his 2 magnets, and following the 'simple' rule WILL lead to lousy results w.r.t. cogging.

Peter.

[dinges]

Correction: the last paragraph should read 'with his case of 3 (three) magnets)'

Flux, you mentioned something about reluctance. Didn't understand that, nor how it's different for round magnets. I simply try to visualize what happens when the rotor rotates and how stator ribs are approached and left by the magnets. I can't see how the geometry of the magnet (round vs. square/rectangular) has to do with it.

I'm hoping that Spudd still has 3 magnets and the rotor left to do a quick experiment with a 9 deg. angle. If so, we would have conclusive proof whether my explanation is right or wrong (well, ok, it can't be proven right, but that's a philosophical discussion). I'm in the process of building 2 conversions, both according to the 'modified' rule. This should give more empirical evidence as to the correctness.

BTW, none of this is mathematically difficult stuff; just try to visualize what happens when the rotor rotates for the limiting case of 2 magnets and it should all become clear. Because, if one adheres to the rule that there should be one stator rib skewness, with 2 magnets you will end up with each magnet above a stator rib. And anyone can see that it will skew terribly. (just imagine it with only 2 magnets on the entire rotor; both magnets being part of the same pole, and on a line).

I hope Flux is wrong (that it is basically a trial-and-error thing). I'd hate to see guesswork playing such a big role in motor conversions. Because then it would be more or less a lottery.

The good thing is, nature behaves according to fixed rules (at least on this level). Only thing is to find out what these rules are

Peter.

[Flux]

Peter

I don't want to get too bogged down about this, but how can you even consider the issue without thinking about reluctance, surely that is where the whole thing starts.

If you have a smooth stator there is no issue. When you cut lumps out then the problem starts. If the lumps left are the same size and shape as the magnets then the flux will try to take the path of least reluctance ( they will shorten themselves like rubber bands). A large force will be needed to rotate the thing and once you have it moving the force will decrease until the point where the magnet is directly over the removed portion, beyond this attraction will occur to the next pole and increase rapidly as you pass the neutral point.

If the magnet is not the same size or shape as the slot then you will have a resultant pull that depends on where the flux passes. Even with straight sided slots ( Open slots, not normally used on ac punchings) and rectangular magnets it is difficult to define exactly where the flux will pass as the magnet will not be uniformly magnetised and the edges will produce only a fraction of the flux  due to fringing and self repulsion within the material.

When you consider the normal motor slot, the tooth is not always of constant section and neither is the bridge piece at the top. Throw in a magnet section where you have no idea of the flux distribution, then you have an interesting problem to solve.

Fortunately it seems as though this non uniformity within the magnet makes life a lot easier than with sharply defined steel pole tips, otherwise the failure rate would be very high.

Everyone here seems intent on a mathematical solution to every problem. Any mathematical analysis that fits the facts is fine and if it removes trial and error I am in favour, but as you grow older and wiser you gradually learn how little you really know and when you reach the point that you realise this then you have reached the point where you actually know something.

Machine design has been going on for over a century and it is only in the last 20 years that some of this has been used satisfactorily for computer machine design. Computer programs are excellent tools but are only as good as the basic knowledge fed into them.

They work well for designs that don't depart far from what has gone before but sometimes throw a big "wobbly" if you extrapolate beyond the truth of the basic assumptions.

Flux

[sPuDd]

Alright,

       Looks like my posting has really got the skew/cog think

tank into top gear. Although I'll admit it took a little while to

figure out what Dinges was talking about. Essentially what I

need to do is make another rotor & shaft and test first with the

same magnets at different skew angles, say 4-30 degrees in 2

degree steps. I'll just use 3 magnets all in a single row (1 pole).

Hey Dinges - cheques in the mail mate :-p  Honest dingo !

I'll get back to you once I have the results as I'm half way through

the heavy work on the tower & blades for the 3hp conversion.

Though I might have a chance to do a new rotor this week...

Dang !  I just seen some new fancy words used by you blokes !! Sweeet

OK, I'll do the experiment and everyone can play with the results.

How about you all call a degree # between 4 and 30 and see who

gets closest?? I may have to measure the shaft tension to break

first cog in order to be most accurate.

So much thinking on this forum - you should KNOW our

masters hate it when we think outside the square

sPuDd..

[dinges]

My money is on 9 degrees. Perhaps you could try 13 too, this is Zubbly's rule.

I'd be interested to see which coggs less, 9 or 13. Don't think there will be much difference between the two, but I think that when you go for the average, say, 10deg, you'll find that

Just my idea. Hope I'm right here.

But what triggered it all: you realize that, when you skew in one plane by the correct 20 deg, that you can't simply transfer that angle to another plane? I.e., when marking the magnet placement out on the rotor, it should be 13deg (if you go for Zubbly's rule, i.e. 20 deg).

Wish I could experiment myself some w.r.t. this matter. Still need to order the magnets and make an appointment with friends to machine the rotor. So far, all I've done is talking and making nice drawings...

Peter.

[dinges]

Oops

... you'll find that it coggs much much less than in your original 20deg configuration (i.e. 20 deg as you marked it is probably really 30-35 deg. skew).