DIY Toroidal Axial Flux PM

mxlemming said:
TorontoBuilder said:
fechter said:
The shape of that curve will change dramatically when there is back iron or stator iron.

Correct, timely and succinct point. :bigthumb:

If I am not mistaken, much of the cutting edge electric motor research is in relation to hybrid rotors and complex 3D flux paths which is only possible because flux can be manipulated by components external to the magnets.

now HH's reading load has increased exponentially :p

I'm pretty sure most cutting edge motor research is related to bogus claims of micro optimisations and creative accounting of mass, power and cooling mechanisms.

:lol:
 
Hmm... well then.

Either way it certainly has me thinking. I recall hearing that deep inserts for the magnets as long as you don't create a magnetic "short" can improve things. maybe this is because it takes advantage of the increase in flux at the edges. I'm sure attaching back iron changes that curve, but does anyone know how? could be some interesting data to collect.

Also Mxlemming, I saw in one of you're posts, you were measuring harmonics with the peaks setting, I tried that as well. Here are the results.
resized-image-Promo - 2021-11-11T211929.522.jpeg

As you can see, it has some peaks showing, but they are not showing in the table. Not sure what to make of this
 
You can see some little blips at the harmonic frequencies but they look pretty small. I think that's good.

The flux across the face of a magnet will be much more uniform if there is an iron path on both sides. The flux always takes the path of least resistance and iron is way more permeable than air or copper.
 
I found it to be true when measuring a magnet with the magnometer, I could not find any consistent readings anywhere,
and the most flux was at the edges.

But can a magnet have any more flux anywhere, or is it the same throughout it's mass, and only appears to be more at
the edge because of the area being measured? (with a hall sensor)

The flux on the face is spread out, but will stick to a metal surface seemingly more because of the area. The flux on the
edge seems more concentrated, but sticks to metal with the same force as the face. :?:

The flux on the end of a long flat or round magnet seems to be stronger, but the area has been reduced, and most of the
magnets mass is behind it. (what happens if we have a sphere magnet?)

( Don't wade to deep in magnet creek, what is in between the north and south pole anyway? (the phantom zone? :lol: )
 
Top and bottom

“ . By simply adding an iron ring with a rectangular cross-section to the inner yoke at each end of the air gap, the effects of the fringe fields can be counteracted, and, hence, the length of the region, where the flux density remains within a given tolerance band is increased.”


https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8268960/#!po=0.510204

Not intended for motors but still maybe a help. Be an easy test: glue them in and see if the kv or no-load current drops.

..then why do they always make these magnet spacers in aluminum instead of steel?
 

Attachments

  • 1A4839AC-A785-4F69-963C-831066925233.jpeg
    1A4839AC-A785-4F69-963C-831066925233.jpeg
    58.8 KB · Views: 631
Hummina Shadeeba,
I like the idea, but I dont have the capability to manufacture a single piece like that. Do you think that putting the inserts in as separate pieces would still have the same effect? Otherwise I could try to get something milled, but it would be super tight tolerances. On the outer edge, the gap is only 1.66mm, and only gets smaller as you move in. I think it would be difficult to make a part like this. I imagine it would be a perfect job for a metal printer though...

This concept seems to introduce design and manufacturing complexity which will ultimately add to cost, and potentially more issues. and it seems like the only real benefit, is a bit of weight savings. Each of those iron backs I have weighs about 50g, I am curious how much of that I could ultimately shave off, even if it was 50%, which I think is unrealistically high, I would only save 50g from the total weight. Granted that something like 10-12% of the total weight, im just not sure its worth it.

very curious about the results though. I would love to set up an experiment.
 
modeled up the new rotors, it has a golden ratio impeller for a spacer. I already have some other changes I would like to make, but I have to sort of work around the metal pieces I have now.
Screenshot 2021-11-12 232702.png
 
HalbachHero said:
Hummina Shadeeba,
I like the idea, but I dont have the capability to manufacture a single piece like that. Do you think that putting the inserts in as separate pieces would still have the same effect?

The paper describes using just a steel ring not this toothed thing I posted which is common in cylindrical motors to space the magnets and is so similar yet always done in aluminum.

Wonder how effective your quasi hallbach is and how small the pole would need to be to have a true hallbach and like 99% of the field trapped. I had a long list of all the halbach variations but dont know where it went but it needed like 4/5 of the magnets being used as sideways magnets to reach its full potential. Maybe hallbach would be best suited with a unique stator with thin teeth and signing up with emetor again and figure stuff out instead of many vague unknowns and remember it would model hallbach arrays.)

(Id skipped to the end of your thread abs just now read most.. why use litz wire? I thought it’s value is in avoiding skin effect at very high frequency. Or maybe that’s the wire you have so that’s what you use.)
 

Attachments

  • 39BBF657-3F53-4324-9FBF-F7FD4D92C569.png
    39BBF657-3F53-4324-9FBF-F7FD4D92C569.png
    1.4 MB · Views: 595
Hummina Shadeeba said:
The paper describes using just a steel ring not this teethed thing I posted which is common in cylindrical motors to space the magnets which is so similar yet always done in aluminum.

right, I could certainly test the ring idea. but again, I wonder how much weight savings is really attainable. but I'm willing to try for science if nothing else


Hummina Shadeeba said:
Wonder how effective your quasi hallbach is and how small the pole would need to be to have a true hallbach and like 99% of the field trapped. I had a list of all the hallbach variations but dont know where it went but it needed like 4/5 of the magnets being used as sideways magnets to reach its full potential. Maybe hallbach would be best suited with a unique stator with thin teeth and signing up with emetor again and figure stuff out instead of many vague unknowns and remember it would model hallbach arrays.)

I have seen variations of halbachs as well. but it seems you incur more weight in magnets that you would otherwise just using steel. once you need 4 magnets for a single pole, it seems like its not worth it, also, this limits the pole width, based on the size of the magnets you can get, and if you are looking for segmented arc magnets like in the image you provided, they can get pricey really quick, and tolerances are more important.
The concept of mine was to keep it relatively cheap. I was able to buy 1000 of each type of magnet for like $300 USD. This gives me enough to make ~20 motors with the current pole count.
Earlier in the thread I posted about my findings with the simulations that I was running, and on the outer diameter, the spacing of my magnets did leak some out of the back, but adding 1.5mm of steel was all that was needed to prevent leakage.
Also I made a graph that shows permutations of iron thickness vs halbach/no halbach using the same bar magnets. I found that you could get the highest flux density using a combination of iron and halbach.
Granted if not using halbach, you could fit more poles, but this also means each phase needs to be packed together tighter. So I feel as though there is an optimization to be had in the balance of all these factors, and I think that what I have now is fairly optimized.

Hummina Shadeeba said:
(Id skipped to the end of your thread abs just now read most.. why use litz wire? I thought it’s value is in avoiding skin effect at very high frequency. Or maybe that’s the wire you have so that’s what you use.)
Your understanding is the same as mine, its to prevent skin effect at high frequencies. but it also provides flexibility when working with it. Its definitely not something that I had and decided to use, I actually spend a lot of time making it from one large spool of 32 gauge stuff. I would really like to see a chart or find some details about where litz is applicable. I assumed with a high pole count I would be dealing with higher frequency, but perhaps its really not that high (Keep in mind. I have no formal mechanical or electrical engineering education, and really had no clue what I was doing when I started this project, and litz wire was a design decision that I made early on and just never strayed away from) I have recently been thinking about using something with a larger gauge so I can play with higher current, and likely less strands to keep the equivalent gauge the same.
 
Keeping the same general shape as the current back iron I removed a chunk from the back of the magnets leaving it only near the edges? I figure by keeping the inner and outer ring attached its easier to make and position correctly. and leaving iron completely for the sideways magnets since they are so thin. Am I on the right track?
Screenshot 2021-11-13 110343.png
 
You would save the weight of the cut out parts, but it seems like a waste of effort unless weight is critical. You will get just as good a result without the holes and it would be stronger and a lot less work to make. Using a halbach array will allow the back iron to be much thinner than would be needed otherwise. It's easy enough to test whether you are saturating the back iron.
 
This is the best info I've found on litz wire.
https://www.elektrisola.com/en-us/Litz-Wire/Info
I've yet to find / understand if the wire size is by carrier frequency or ac frequency - carrier frequency seems right...
 
fechter said:
You would save the weight of the cut out parts, but it seems like a waste of effort unless weight is critical. You will get just as good a result without the holes and it would be stronger and a lot less work to make. Using a halbach array will allow the back iron to be much thinner than would be needed otherwise. It's easy enough to test whether you are saturating the back iron.
I would tend to agree with you, but it seems like being able to laser cut things makes adding extra holes like this nearly zero extra effort. And since I have to order laser cut parts from a service, its only extra money at the moment. I am curious what the weight difference would be, I don't think it would be significant, and I agree, it would take away some if its integrity.

Jrbe said:
This is the best info I've found on litz wire.
https://www.elektrisola.com/en-us/Litz-Wire/Info
I've yet to find / understand if the wire size is by carrier frequency or ac frequency - carrier frequency seems right...
This is a great resource, thanks for sharing. I realize that its a bit specific to the circumstance on when to use litz wire, but I have not really had a great understanding on when it becomes applicable. based on some of the charts that I have seen on that site and others. It actually seems that 0.2mm diameter is pretty damn close to ideal, and that happens to be what Im using.

The rationale is this.
absolute max RPM is likely close to 10k RPM. given 12 poles, this would result in 12Hz per rotation, so ~2kHz at max RPM, and 1kHz at the upper end of what it might actually run at.

Given 1-2kHz, it seems like you start needing to be concerned about skin effect. I may be able to get away with using larger gauge given that I would generally be operating lower than 1kHz, but it seems like what I have going now is close to ideal. I could adjust strand/turn count to vary Kv and max current.
 
Just crunched some quick numbers
The area of the full back iron: 4560.8mm^2
area of the "optimized" back iron: 3297.4mm^2
thickness = 1.5mm

4560.8 * 1.5 = 6,841.2mm^3
3297.4 * 1.5 = 4,946.1mm^3

if the steel weighs 0.0079g/mm^3

Then 6841.2 * 0.0079 = ~54.0g
And 4946.1 * 0.0079 = ~39.1g

this assumes that the back iron could in fact stay the same thickness. This would only provide a weight savings of ~15g (~28% less than the original back iron weight. This would add up to 30g or <7% of the total motor weight. I dont think its worth it
 
HalbachHero said:
The rationale is this.
absolute max RPM is likely close to 10k RPM. given 12 poles, this would result in 12Hz per rotation, so ~2kHz at max RPM, and 1kHz at the upper end of what it might actually run at.

I think 10k rpm with 12 poles would be .. 60hz and count complete polarity back and forth switch?
 
Hummina Shadeeba said:
I think 10k rpm with 12 poles would be .. 6Hz per rotation and count complete polarity switch
Sorry yes you are right, I said that wrong. There are 48 magnets, 24 poles, 12 pole pairs. I believe that would be 12Hz per rotation, please correct me if im wrong
 
Hummina Shadeeba said:
10k rpm? 10,000/60 =166 /12=13Hz when doing 10,000 rpm

Would it not be 166 rotations/sec * 12 as there would be 12 oscillations in the one rotation?
 
Hummina Shadeeba said:
looks close to mine! but im no expert. should be simple. i fixed mine after thinking about it.

No worries. two minds are better than one!

The PWM frequency I am not sure of. is that frequency experiences by the phase coils, or only inside the controller. if the phase coils do in fact see that sort of frequency. I think that validates the argument for using litz wire, and may even make the argument for using even thinner stuff....
This goes beyond my knowledge at the moment.
 
Hummina Shadeeba said:
10,000 rpm. times 12 pole pairs. 120,000. divided by 60 seconds. is 2000Hz. i get something different every time

this is because you rounded 10k rpm to rotations/second
10,000RPM = 166.666667 Rotation/sec* 12 pole pair = 2000 Hz
 
Back
Top