APL's DIY axialflux motor
Re: APL's DIY axialflux motor
Got it. So it could be one turn, and we can use use that to calculate the rest. Since it's just voltage I assume that the air
gap will have no effect.
gap will have no effect.
Re: APL's DIY axialflux motor
For measuring kV, it will be the number of turns regardless of length.
The air gap will make a difference. The closer the gap, the more voltage you'll generate, but for typical motor gap sizes, I don't think it will make a huge difference. You will need both rotors to get a good measurement. The idea is to get the flux close to what it will be in the final assembly.
Once you know the voltage at a given RPM for a single core and knowing how many cores are in series, we can do some math to get a pretty close estimate. Having close to the optimum number of turns for the configuration will really help efficiency.
The air gap will make a difference. The closer the gap, the more voltage you'll generate, but for typical motor gap sizes, I don't think it will make a huge difference. You will need both rotors to get a good measurement. The idea is to get the flux close to what it will be in the final assembly.
Once you know the voltage at a given RPM for a single core and knowing how many cores are in series, we can do some math to get a pretty close estimate. Having close to the optimum number of turns for the configuration will really help efficiency.
"One test is worth a thousand opinions"
Re: APL's DIY axialflux motor
Air gap will have a substantial effect as it’s directly connected to the flux. And wind some turns at least, single turn will generate low voltage so you’ll have a bigger effect from any measurement errors. diameter of wire is less important since it’s very low current.
Ride on!
Re: APL's DIY axialflux motor
Right, I'll do a test with ten turns, and one with 20 turns. Then spin it at 100 & 200 RPM etc. I'll have to come up with
a way to spin it yet, since I traded my little 20" scooter off,.. maybe a drill or something. I'll have to make a decent
motor spinner one of these days.
Lots to do. I have two cores mounted, and I'm working on the air gap at the moment, which requires machining the axle
and spacers, along with taking it apart and putting it together a hundred times.
Looks like the fiberglass stator carrier is going to be strong enough though, since it hasn't bent with the force of the
magnets yet. That's a big relief.
I should be getting the 3D printed coil forms back in another day or so, and we can see how thats going to work out too.
a way to spin it yet, since I traded my little 20" scooter off,.. maybe a drill or something. I'll have to make a decent
motor spinner one of these days.
Lots to do. I have two cores mounted, and I'm working on the air gap at the moment, which requires machining the axle
and spacers, along with taking it apart and putting it together a hundred times.
Looks like the fiberglass stator carrier is going to be strong enough though, since it hasn't bent with the force of the
magnets yet. That's a big relief.
I should be getting the 3D printed coil forms back in another day or so, and we can see how thats going to work out too.
Re: APL's DIY axialflux motor
Plenty of energy,.. it's just that things keep getting in my way, last week was crazy.
The closer I got to the air gap, the slower I had to go, but I finally got it close to 1mm. Once all the cores are in place
it will probably change a little with all the magnetic attraction, and I'll find tune it then.
I rigged up a really crappy drill driver, but it did the job. I think I'm going to send off for one of those MY1016Z gear drive
motors to make a driver out of,.. they look perfect for the job.
Anyway, here's the specs with ten winds, five on each side. (series) Took the voltage at four different RPM's.
(Somewhat approximate. as it's hard to hold the rpm steady) The 1.0v reading is a solid figure though.
200 rpm  .4v
400 rpm  .7v
450 rpm  .83
550 rpm  1.0v
This motor is a 18S/20P each side.
I got the 'split core' forms back, and they work perfectly, except the coils wind up being to big, so I had some stock size
cores printed, and I'll just deal with it. I can wrap a layer of tape around it, and use the poly bag to pull the coil off,
maybe sand a very slight taper. We'll see how it goes.
The stock size plastic cores were very useful while doing the air gap, as I didn't have to fight the magnets upon assembly
and disassembly.
The closer I got to the air gap, the slower I had to go, but I finally got it close to 1mm. Once all the cores are in place
it will probably change a little with all the magnetic attraction, and I'll find tune it then.
I rigged up a really crappy drill driver, but it did the job. I think I'm going to send off for one of those MY1016Z gear drive
motors to make a driver out of,.. they look perfect for the job.
Anyway, here's the specs with ten winds, five on each side. (series) Took the voltage at four different RPM's.
(Somewhat approximate. as it's hard to hold the rpm steady) The 1.0v reading is a solid figure though.
200 rpm  .4v
400 rpm  .7v
450 rpm  .83
550 rpm  1.0v
This motor is a 18S/20P each side.
I got the 'split core' forms back, and they work perfectly, except the coils wind up being to big, so I had some stock size
cores printed, and I'll just deal with it. I can wrap a layer of tape around it, and use the poly bag to pull the coil off,
maybe sand a very slight taper. We'll see how it goes.
The stock size plastic cores were very useful while doing the air gap, as I didn't have to fight the magnets upon assembly
and disassembly.
Re: APL's DIY axialflux motor
Nice work.
I guess I'm not understanding the winding configuration. We'll treat both sides together as one coil. How many stator poles are there total?
I guess I'm not understanding the winding configuration. We'll treat both sides together as one coil. How many stator poles are there total?
"One test is worth a thousand opinions"
Re: APL's DIY axialflux motor
Sorry to make it confusing,.. yes, one coil at ten turns, 18 coils. And 20 pole magnets on a rotor,.. so 20 poles.
Don't mean to jump ahead, but I'll make a foolish attempt at figuring this out.
At 550 rpm, theres going to be 6 volts per phase, and 18v for three phase. Inversely proportional means that at 18v the no
load speed would be near 550 rpm. (if I'm getting this right) Since I'm using a 36v supply, (40v full charge), I would need
around 20 turns? (Maybe a few more, since I plan to upgrade to a 48  52v system a little later.)
Am I even close?
Don't mean to jump ahead, but I'll make a foolish attempt at figuring this out.
At 550 rpm, theres going to be 6 volts per phase, and 18v for three phase. Inversely proportional means that at 18v the no
load speed would be near 550 rpm. (if I'm getting this right) Since I'm using a 36v supply, (40v full charge), I would need
around 20 turns? (Maybe a few more, since I plan to upgrade to a 48  52v system a little later.)
Am I even close?
Re: APL's DIY axialflux motor
Never calculated this myself but i think it's like this: assume the coils are connected in Y: two phases / 12 coils will be connected in series and active at each commutation step (not all) so it will be 12v out at 550rpm
you can then calculate the kV here:
http://www.bavariadirect.co.za/constants/
It will be kV 34 in Y and kV 59 in delta for 10 turns, 20 turns give you kV17 and 29.
at 52V 20turns Y you'll get 884 noload rpm,
I'd go higher in rpm if you think balancing is ok for more?
you can then calculate the kV here:
http://www.bavariadirect.co.za/constants/
It will be kV 34 in Y and kV 59 in delta for 10 turns, 20 turns give you kV17 and 29.
at 52V 20turns Y you'll get 884 noload rpm,
I'd go higher in rpm if you think balancing is ok for more?
Ride on!
Re: APL's DIY axialflux motor
I'm not sure how to calculate it exactly either, but each phase will be 6 coils, so 6v at 550rpm sounds right there. I'll assume phases are connected in wye configuration, but they will be out of phase at any given time, so not a simple sum, but more like 9v between phases.
The other thing to consider is the gearing, so we have a target RPM for the desired top speed. For optimum efficiency you want the noload RPM to be about 10% higher than the target top speed.
Seems like you might want more turns.
The other thing to consider is the gearing, so we have a target RPM for the desired top speed. For optimum efficiency you want the noload RPM to be about 10% higher than the target top speed.
Seems like you might want more turns.
"One test is worth a thousand opinions"
Re: APL's DIY axialflux motor
Basicly the generated voltage should be two x phase voltage as you’re measuring between two phases.
Voltage calculated here supports this:
Voltage calculated here supports this:
Ride on!
Re: APL's DIY axialflux motor
I forgot about the two phases in Y again, definitely plan to wire in Y. 20T with a Kv17 is in the range, albeit not perfect.
Thanks for doing the calculation, and setting me straight. And thanks for the Bavaria constants link, I missed that page,
for doing Kv and efficiency work.
I can do 20T easy enough, even though I'll have to machine the bottom of the spacers a little to clear the end windings,
but any more layers will be more difficult, and not be covered by the tooth brim. Should have done more research in the
beginning, but we're not that far off.
The other possibility is to use smaller wire, like 15ga to gain more turns and a lower profile, but with a max current loss.
I'm using 14 Ga. because it represents whats on the Crystalyte, but maybe I don't really need it.
As for gearing, I can go a little lower in back, the current Crystalyte motor will run it at roughly 20mph with a 28 or 30T
cog on the freewheel. (I'll have to check that T#)
The front cog is a 17, and if I go smaller there, the chain tends to 'power jump' when a der is used. Too much torque.
I have a clutch der on it now to help with that problem.
I have thought about going single speed, since I don't use the gears much anyways,.. in which case I can go as big as I
want on the hub, I think even into the 50T range. But with that, I'd need a righteous top speed, for the fun factor.
So we have a lot to work with, and even the 884 rpm  52v figure may be workable as well. What I should probably do
is wind the 20T and retest, to see what the real world voltage is, and go from there.
Any input on wire size? I always figured it to be a 2kw motor, but I'm certainly not using that much, and I think 1  1.5
would be more realistic with my battery plans. I always want more power, but I should come down to earth,.. I can
always upgrade later.
If a 2 or 3kw motor is wanted, then the core needs to be redesigned to accommodate the wiring, and saturation needs
to be addressed more, along with magnet T.
Thanks for doing the calculation, and setting me straight. And thanks for the Bavaria constants link, I missed that page,
for doing Kv and efficiency work.
I can do 20T easy enough, even though I'll have to machine the bottom of the spacers a little to clear the end windings,
but any more layers will be more difficult, and not be covered by the tooth brim. Should have done more research in the
beginning, but we're not that far off.
The other possibility is to use smaller wire, like 15ga to gain more turns and a lower profile, but with a max current loss.
I'm using 14 Ga. because it represents whats on the Crystalyte, but maybe I don't really need it.
As for gearing, I can go a little lower in back, the current Crystalyte motor will run it at roughly 20mph with a 28 or 30T
cog on the freewheel. (I'll have to check that T#)
The front cog is a 17, and if I go smaller there, the chain tends to 'power jump' when a der is used. Too much torque.
I have a clutch der on it now to help with that problem.
I have thought about going single speed, since I don't use the gears much anyways,.. in which case I can go as big as I
want on the hub, I think even into the 50T range. But with that, I'd need a righteous top speed, for the fun factor.
So we have a lot to work with, and even the 884 rpm  52v figure may be workable as well. What I should probably do
is wind the 20T and retest, to see what the real world voltage is, and go from there.
Any input on wire size? I always figured it to be a 2kw motor, but I'm certainly not using that much, and I think 1  1.5
would be more realistic with my battery plans. I always want more power, but I should come down to earth,.. I can
always upgrade later.
If a 2 or 3kw motor is wanted, then the core needs to be redesigned to accommodate the wiring, and saturation needs
to be addressed more, along with magnet T.
Re: APL's DIY axialflux motor
Or you spin it faster.
Power=torque x speed so if you spin it at 5000rpm you’ll get 5x more power out of it
It all depends on balance, integrity and eddy losses.
Power=torque x speed so if you spin it at 5000rpm you’ll get 5x more power out of it
It all depends on balance, integrity and eddy losses.
Ride on!
Re: APL's DIY axialflux motor
OK, I'm going to have to take some better measurements on my current motor RPM's and speeds per gear. Batteries are
getting charged, and I'll have more on that tomorrow. I think the Crystalyte is turning slower than I thought.
Big single speed cog in back is probably going to be the solution. Like you say, spinning it faster will get more torque,
and it seams that I have plenty of top end speed to work with. I don't mind tossing the der out, other than it makes it
hard to take the rear wheel off, and keep the axle from pulling forward, but I can deal with that.
I looked at wire sizes a little, and see that 15 awg will get me 24 turns. 16 awg will give 28, but thats getting
too small.
14 awg = 1.63mm
15 awg = 1.45mm
16 awg = 1.29mm
I have 9.5mm core length.
getting charged, and I'll have more on that tomorrow. I think the Crystalyte is turning slower than I thought.
Big single speed cog in back is probably going to be the solution. Like you say, spinning it faster will get more torque,
and it seams that I have plenty of top end speed to work with. I don't mind tossing the der out, other than it makes it
hard to take the rear wheel off, and keep the axle from pulling forward, but I can deal with that.
I looked at wire sizes a little, and see that 15 awg will get me 24 turns. 16 awg will give 28, but thats getting
too small.
14 awg = 1.63mm
15 awg = 1.45mm
16 awg = 1.29mm
I have 9.5mm core length.

 Posts: 33
 Joined: Dec 07 2016 7:01am
Re: APL's DIY axialflux motor
Sorry to interrupt so late in the discussion, but to my understanding passing from delta to star configuration is not a x2 increase in phase voltage but x √3 so x 1.73 because when one phase is producing his pic voltage the magnet are 60 electical ° off for the other one hence producing less. Maybe i'm wrong about this so feel free to correct me !
Another thing to consider is that because coil number doesn't match pole count, for reducing torque ripple, each coil pic inside one phase will also be a little off by some ° further decreasing the produce voltage (not by much i think but it should be calculable).
Another thing to consider is that because coil number doesn't match pole count, for reducing torque ripple, each coil pic inside one phase will also be a little off by some ° further decreasing the produce voltage (not by much i think but it should be calculable).
Re: APL's DIY axialflux motor
It is a 3 phase motor and assuming it produces a sinusoidal back emf an alternative approach to calculating Kv would be to look at the DC voltage required to produce 3 120 degree separated sinusoidal voltages with a magnitude of 6V as generated by the motor at 550rpm. The answer is 6 x SQR(3) = ~10.4V so the Kv is 550/10.4 = ~53
I found this paper to have some great content for designing axial flux motors without going too deep with the math.
https://pdfs.semanticscholar.org/96e7/c ... 1bbd05.pdf
I found this paper to have some great content for designing axial flux motors without going too deep with the math.
https://pdfs.semanticscholar.org/96e7/c ... 1bbd05.pdf
“You never know if quotes on the internet are genuine.”  Abraham Lincoln
Re: APL's DIY axialflux motor
Thanks guys, that changes things a bit. Kiwifiat, I'm not sure what '6 x SQR(3)' means. But 10.4 Kv comes out to
416 rpm at 40v, and 540 rpm at 52v.
(Using a Kv/rpm/volt calculator; http://www.rcdronegood.com/brushlessmotorkvtorpm/)
I found this handy dandy calculator, found here on Spinningmagnet's post on wheel rpm, viewtopic.php?t=16114
Electric Scooter Parts calculator; http://www.electricscooterparts.com/mot ... ratio.html
Which calculates motor rpm directly into speed. Very Nice! Now I can get rid of all the piles of paper.
According to the calculator, 410 rpm = 20 mph with a 28 rear cog. and 540 rpm = 26 mph. Which is right where I
need to be. (In theory,.. theres still the needed 10% higher rpm's, and mechanical losses, etc.)
So with all that, you have shown how to convert one cores turns into mph/speed, which is really amazing.
416 rpm at 40v, and 540 rpm at 52v.
(Using a Kv/rpm/volt calculator; http://www.rcdronegood.com/brushlessmotorkvtorpm/)
I found this handy dandy calculator, found here on Spinningmagnet's post on wheel rpm, viewtopic.php?t=16114
Electric Scooter Parts calculator; http://www.electricscooterparts.com/mot ... ratio.html
Which calculates motor rpm directly into speed. Very Nice! Now I can get rid of all the piles of paper.
According to the calculator, 410 rpm = 20 mph with a 28 rear cog. and 540 rpm = 26 mph. Which is right where I
need to be. (In theory,.. theres still the needed 10% higher rpm's, and mechanical losses, etc.)
So with all that, you have shown how to convert one cores turns into mph/speed, which is really amazing.
Re: APL's DIY axialflux motor
SQR(3) means the square root of 3. What I am saying is that to produce 3 phase voltages of 6V per phase you need 10.4V DC. Since your motor will produce 6V per phase at 550 rpm your Kv is 550/10.4 = 53 rpm/VAPL wrote: ↑Jun 11 2020 9:56pmThanks guys, that changes things a bit. Kiwifiat, I'm not sure what '6 x SQR(3)' means. But 10.4 Kv comes out to
416 rpm at 40v, and 540 rpm at 52v.
(Using a Kv/rpm/volt calculator; http://www.rcdronegood.com/brushlessmotorkvtorpm/)
I found this handy dandy calculator, found here on Spinningmagnet's post on wheel rpm, viewtopic.php?t=16114
Electric Scooter Parts calculator; http://www.electricscooterparts.com/mot ... ratio.html
Which calculates motor rpm directly into speed. Very Nice! Now I can get rid of all the piles of paper.
According to the calculator, 410 rpm = 20 mph with a 28 rear cog. and 540 rpm = 26 mph. Which is right where I
need to be. (In theory,.. theres still the needed 10% higher rpm's, and mechanical losses, etc.)
So with all that, you have shown how to convert one cores turns into mph/speed, which is really amazing.
And yes you can calculate motor Kv from a single turn but as has already been noted errors are reduced with more turns. Have a read of the paper I linked, the math is outlined there and it isn't too complex unlike some academic papers.
“You never know if quotes on the internet are genuine.”  Abraham Lincoln
Re: APL's DIY axialflux motor
Still there are 12 coils active in the stator at the same time, not 6
Ride on!
Re: APL's DIY axialflux motor
Right, but not all 12 are aligned with the magnets at the same time so I think what kiwifiat had sounds correct.
If you decide more turns are better, then use smaller wire. Another technique is to use multiple parallel strands of much skinnier wire. Sort of a pain to deal with the end connections, but much easier to wind due to less stiffness and the smaller strands reduce eddy currents in the wire.
"One test is worth a thousand opinions"
Re: APL's DIY axialflux motor
First off, awesome build!
Second, why not go all out and use Litz wire (if your budget allows)? Lots of strands of 3642AWG wire would pack a lot of copper in there.
Second, why not go all out and use Litz wire (if your budget allows)? Lots of strands of 3642AWG wire would pack a lot of copper in there.
 Attachments

 Skin Effect.png (151.96 KiB) Viewed 294 times
Re: APL's DIY axialflux motor
Litz wire would be nice, but the motor is still in the experimental stage, and design changes happen almost every day.
I'd prefer to save the good stuff for later on, when things are more solid. I'm still not sure that the 18S/20P config. was
even such a good choice yet.
Using multi strand wire is an option too, although not one I like too much. I thought that being able to use single strand
wire was an advancement, and a nice feature of the axial motor. Especially if square wire is used.
But we gotta do what we gotta do,.. so multi strand is not out yet.
Now I have to decide on either 14 or 15 awg wire. I think larsb was saying that its better to go larger, (or shorter) wire if
posible, and run faster, for less copper resistance losses.
The axial link that kiwifiat posted is a good one, but it will take some time to digest. It gives some relevant info. in the
#10  #13 formulas. I'm surprised at the 1hp output of that little air coil motor, although it has to wind up to get it, that
thing has got potential, and many other advantages. Halbach arrays are not so easy to make though.
I was never any good at formulas, and my eye glaze over pretty quickly. But I think I finally get the square root of 3 at least.
Like Thecoco974 stated, the square root is 1.73, and six times that is 10.38, or 10.4, as kiwifiat stated.
Although I'm still not sure why were using the square root, but if it works, then I guess I don't need to.
If this is correct, then we have it figured. I'll have to wind the motor to find out for sure.
I'd prefer to save the good stuff for later on, when things are more solid. I'm still not sure that the 18S/20P config. was
even such a good choice yet.
Using multi strand wire is an option too, although not one I like too much. I thought that being able to use single strand
wire was an advancement, and a nice feature of the axial motor. Especially if square wire is used.
But we gotta do what we gotta do,.. so multi strand is not out yet.
Now I have to decide on either 14 or 15 awg wire. I think larsb was saying that its better to go larger, (or shorter) wire if
posible, and run faster, for less copper resistance losses.
The axial link that kiwifiat posted is a good one, but it will take some time to digest. It gives some relevant info. in the
#10  #13 formulas. I'm surprised at the 1hp output of that little air coil motor, although it has to wind up to get it, that
thing has got potential, and many other advantages. Halbach arrays are not so easy to make though.
I was never any good at formulas, and my eye glaze over pretty quickly. But I think I finally get the square root of 3 at least.
Like Thecoco974 stated, the square root is 1.73, and six times that is 10.38, or 10.4, as kiwifiat stated.
Although I'm still not sure why were using the square root, but if it works, then I guess I don't need to.
If this is correct, then we have it figured. I'll have to wind the motor to find out for sure.
Last edited by APL on Jun 12 2020 1:39pm, edited 1 time in total.
Re: APL's DIY axialflux motor
It’s not used because
litz is expensive
It lowers copper fill percent due to all wire crossings and insulation thickness
and
frequency of operation is so low that it gives minimal benefit.
litz is expensive
It lowers copper fill percent due to all wire crossings and insulation thickness
and
frequency of operation is so low that it gives minimal benefit.
Ride on!
Re: APL's DIY axialflux motor
I'll rewind the core with 20 turns and check the voltage again, since I have the 14 awg., and then go from there.
I was also thinking, (theoretically), that if 20 turns equaled 100% of no load speed, then 2 turns should be 10% more.
So 22 turns would give the 90% of no load efficiency.
I was also thinking, (theoretically), that if 20 turns equaled 100% of no load speed, then 2 turns should be 10% more.
So 22 turns would give the 90% of no load efficiency.
Re: APL's DIY axialflux motor
Fewer turns will make the motor go faster. So if you're getting the desired speed at no load, you want about 10% fewer turns to keep the same speed under load.APL wrote: ↑Jun 12 2020 8:22pmI'll rewind the core with 20 turns and check the voltage again, since I have the 14 awg., and then go from there.
I was also thinking, (theoretically), that if 20 turns equaled 100% of no load speed, then 2 turns should be 10% more.
So 22 turns would give the 90% of no load efficiency.
If you can get the desired number of turns with a single strand then that tends to be the easiest to build and the coils will have more mechanical strength.
I agree that motor frequencies are so low that the advantages of Litz wire are minimal but not zero.
"One test is worth a thousand opinions"