Multistage axial flux motor design and prototyping

Thecoco974

10 W
Joined
Dec 7, 2016
Messages
94
Hello everyone,
I've been learning a lot on this forum that is, as his name suggess, an endless source of knowledge on electrical vehical. I've stayed mostly on the Electric techhnology Discussion part of the forum and have been inspired by awesome threads like Lebosky's and APL's on motor building and by good general motor and controleur design discussions (most recently with HalbachHero's thread on his toroïdal axial flux design) .
I think I should share my build/design even though it might not work as intended.

After building a motor from hoverboard hubmotors and being disappointed by the high losses linked to too thick laminations i've looked for other motor desigh that looked somewhat DIY'able and could help me with my goal of reliable 5kw continuous power (and cheap to build, but i've come to realise that with prototyping it doesn't work at all).
Like other i've come to realise that the easiest way to diy a motor was to skip the steel core part and ended up designing an axial flux coreless permanent magnet motor.
A good inspiration for me (has you will see) was Ben katz pancake motor that is very well documented here : https://build-its-inprogress.blogspot.com/2015/02/coreless-axial-flux-motors.html

As stated above my goals with this design are the folowings :

- Continuous power of at least 5Kw
- Peak power of close to 10Kw would be fun
- Running voltage <60V (so 13S li-ion in my case)
- Scalability
- cheap materials
- Doable with small machine tool (a mini lathe bought for the occasion)

I haven't talk about the use case, for hub motor and mid-drive afficionados it might sound wierd, but the first vehical to receive this motor( if it fulfill the requirements) would be a front-wheel friction drive racing two wheeler. We have races where I live with those kind of stuff named solex that are quite fun.

Without further do here is the simplified design process i've followed until where I am now :

I've firstly looked for the widely available magnet dimensions (going for the cheap ones) and ended up buying 100 of the 40 x 20 x 5mm N37 neodymium magnet.

Starting with those magnets I then started doing a lot off FEMM simulations to find the optimal airgap based on the FOM procedure explain very well on this video : https://www.youtube.com/watch?v=AWJYCzGvYzs
The result ended up being somewhere around 9-11 mm airgap with twin 5mm back-iron rotor.

Being limited to a turning diameter of 205mm with my newly bought 8x16 lathe I ended up with a rotor diameter of 177mm and 7 poles pairs for an inrunner design and an outer shell diameter of 195mm. Here is how it look :
rotor1.PNG

The stator as been inspired by ben katz design but without spreading each turn over the full phase pole occupation and keeping turn section going radially (copper section is a funtion of inner radius, number of poles and stator thickness). With a goal of increasing airgap copper density I have designed a 3D printed winding core that permit paper thin central membrane and after a lot of though gave me on option for watercooling (when going for inrunner type), here is my idea :

stator.PNG

The skewing of the winding althought hurting the motor constant by decreasing flux linkage/angling the Laplace force vector/spreading current over an entire pole (pick the one you dislike the less) decrease the winding resistance enough to give an overall beter performing winding as found by Ben Katz (looking forward to do the calculus and find the same result). And the bonus being giving a way to watercool that thing, each pocket overlap two on the other side giving a circulating pass around the stator like this :

cooling flow2.PNG

The pass might be to much restrictive but multiple inlet/outlet is posible if the flow is too low.

Being a flat design, stacking is good option for multiplying power capabilities here is my first design with 3 stator and 4 rotors :

moteurV2.2.PNG

V1_cut.PNG
(spacing messed up because of later part revision for another assembly)

This particular design, with single magnet thin inner stator, after motor constant calculation proved itself to be not much better than a two stator design with two magnet thick middle stator. This is mostly due to too low magnetic flux through the middle stator. Here the axial flux density through one pole with previous presented design and after the one with the double magnet center rotor :

flux_cross.PNG
flux_cross_2.PNG

The later version will be chosen because the first stator prooved itself to be time consumming to manufacture to say the least :lol:

A prototype is beeing build and I will share with you the state of it and the building technique used in a later post. It's coming quite slowly because of a lot of reasons (covid induced shortage on material/time lacking/me just being lazy/being a beginner machinist :roll: )

(by the way sorry for my bad english, beeing too lazy to correct it here ...)
 
Very excited about this. It's almost exactly how I'd do this. So close. The only differences I'd have is 1) skip the water cooling until later. This is going to be a pain in the arse and 2) not have a friction drive. This motor is too good for that :lol:

My initial impression is that i agree with all the design decisions until your testing shows otherwise. The diagonal winds should make it really easy to get great packing density and the way you've attached the rotor means it would be hard to make it much stiffer...

Keep posting. If this works, it'll become my next ambition for my ebike... Even at 200mm diameter, with no iron it probably won't be especially heavy.
 
ben katz couldn’t reslly drive his motor due to too low inductance. Guess that’s an effect of the coil sides cancelling inductance due to the thin disc.

No matched controller, no drive. Think about that before spending a gazillion hours designing this.

More steel, more turns, more cigarettes. Any way to make it easier to drive is a good way :D
 
What's this,.. Thecoco974's thread'n it? AWESOME!!! :bigthumb:

Love to watch your progress, and can't wait to see how it turns out! Continuous 5kw is a hefty goal and has me drooling
with anticipation. :)

I think the friction drive is great, and allows the motor do it's high RPM thing,.. feel sorry for the tire though. Sounds like
an interesting race... those little Solex motors don't stand a chance against this guy! :lol:
 
Nice to see I have some people interrested in that to discuss with :D

mxlemming said:
1) skip the water cooling until later

The first stator I made is using the 3d printed carrier I designed for watercooling.I'm using it already because it's actually just empty "in between" with through holes that helped me aligned the two print when gluing it together. For a first test of course i won't bother with that and see what it can do without it, but with nearly 600g of copper I can add it to it without doing another one :lol:

mxlemming said:
not have a friction drive. This motor is too good for that

Don't worry about that I don't like that either but it's a mechanical challenge in it's own so interesting to do. If it deliver as I anticipate it a 4 stator version with a chain drive for motorcycle conversion is in the box.

mxlemming said:
The diagonal winds should make it really easy to get great packing density and the way you've attached the rotor means it would be hard to make it much stiffer...

Yes and also a flat stator because there is no overlapping on the end turn, it was more desirable when i was designing an outrunner but I never build it for a few raisons including as you mentionned the better rigidity a inrunner setup add to the stator mounting (that you may loose on the rotor but with proper design it should be good )

mxlemming said:
Even at 200mm diameter, with no iron it probably won't be especially heavy

It could be, but won't in my case because of material choice (based on price because i'm cheap) and rotor design. Going to composite hallback array would light it up a lot( details on weigth coming soon).

larsb said:
No matched controller, no drive. Think about that before spending a gazillion hours designing this.

More steel, more turns, more cigarettes. Any way to make it easier to drive is a good way

I'm not really agreeing with that, I know that low inductance motor are hard to drive and require high switching frequencies but i would think we are now at a point in time where those controller are doable and actually exist. Benjamin vedder prooved with his vesc 75/300 a 70khz switching frequency was achievable and drove à 1.3 uH motor with it ... I hope mine won't be that dramatically low but if that's the case and prooved to be challenging adding external line inductance even with an efficiency hit is a non issue.

APL said:
What's this,.. Thecoco974's thread'n it? AWESOME!!!

Love to watch your progress, and can't wait to see how it turns out! Continuous 5kw is a hefty goal and has me drooling
with anticipation.

I think the friction drive is great, and allows the motor do it's high RPM thing,.. feel sorry for the tire though. Sounds like
an interesting race... those little Solex motors don't stand a chance against this guy!

I't been a while since i was planning of starting my own ever since I folowed yours with great interest and start designing this. I finished school and started to work few month ago, so just starting going forward to the manufacturing of this in the past few weeks when time is found.
Mine clearly won't be as beautiful as your anodised aluminium outrunner build :roll:
Solex race have got some seriously prepared motors (50cc) that can go close to the 100kph mark so 5kw is a minimum and tires are swapped pretty regularly on this for sure :lol:
 
I'm not really agreeing with that, I know that low inductance motor are hard to drive and require high switching frequencies but i would think we are now at a point in time where those controller are doable and actually exist. Benjamin vedder prooved with his vesc 75/300 a 70khz switching frequency was achievable and drove à 1.3 uH motor with it ... I hope mine won't be that dramatically low but if that's the case and prooved to be challenging adding external line inductance even with an efficiency hit is a non issue.
I'm not sure i saw this, might have. Did he drive it for some hours in a traction EV application? I guess not. If it was noload spinning it basically doesn't mean anything.

Adding outside coils to solve a low inductance issue is just too poor if you're not only doing this motor as a "college project" / just for fun. Weight/size/efficiency downsides are too high.
 
Here is the video in question :

[youtube]fbK2dcoYS7g[/youtube]

Running a 10kw EDF isn't an unloaded test so it seems stable enough for me :wink:
It was done 2 yeard ago so not really fresh news in that regards so i'm pretty optimistic about being able to run it without external coils.
 
larsb said:
Propeller application is also easy vs traction/EV

High constant load / high current at high rpm / high BEMF

(Not full current peaks at zero rpm, current peaks and reverse load from jumps and chain slack etc)

It actually take 300A at low speed from inertial loading at first, I agree this is not a heavy load though. But is low inductance challenging only at low speed ? I would assume that if you can keep ripple current in control even at medium speed where BEMF isn't helping to reduce it at all it would be the same at low speed with the same current assuming the control loop isn't going crazy.
Do you have constructive informations to give us about that ?
 
No, just anecdotal. It’s not only challenging at low speed though but the reason the RC ESC:s don’t immediately explode with a propeller but explode in traction application is just that: a propeller provides a nice constant load and the needed current is exponential vs the rpm.
 
Thecoco974 said:
larsb said:
Propeller application is also easy vs traction/EV

High constant load / high current at high rpm / high BEMF

(Not full current peaks at zero rpm, current peaks and reverse load from jumps and chain slack etc)

It actually take 300A at low speed from inertial loading at first, I agree this is not a heavy load though. But is low inductance challenging only at low speed ? I would assume that if you can keep ripple current in control even at medium speed where BEMF isn't helping to reduce it at all it would be the same at low speed with the same current assuming the control loop isn't going crazy.
Do you have constructive informations to give us about that ?

I wouldn't worry about this too much. You'll be able to drive your motor. Hallbach hero has been driving his with a crappy 10$ ESC and a prop. VESC and many others do cycle by cycle current regulation, at 20khz+. This is incredibly fast compared to any mechanical phenomena a bicycle can possibly experience.

My ESC happily runs a 10uH motor at 24V. I have a 2uH motor that it runs... Very badly... But it runs.

RC ESCs tend to have no current feedback atall, so they don't suit traction applications; they'll sit there and try to push any unknown amount of current into a motor. I find it frankly perplexing that they don't instantly explode all the time for traction applications.
 
I’ve gotten provoked cutouts on FOC sensorless controllers from ice patch skidding back to asphalt and especially from jumping. Wheel loses traction, spins to max speed, smack! onto ground. this is equally fast process with high g-forces acting as a whip through the chain.

Not to bash the 75/300 vesc but people on the efoil forum have had quite a few broken. More than i would expect since that also is propeller drive.
 
larsb said:
I’ve gotten provoked cutouts on FOC sensorless controllers from ice patch skidding back to asphalt and especially from jumping. Wheel loses traction, spins to max speed, smack! onto ground. this is equally fast process with high g-forces acting as a whip through the chain.

Not to bash the 75/300 vesc but people on the efoil forum have had quite a few broken. More than i would expect since that also is propeller drive.

Interesting. I've not experienced this problem though I have had all kinds of VESC problems with hall sensors.

I think in general, VESC doesn't do the best job of protecting itself. I can't find anything in the VESC fast loop that says "hey, have we reached dangerous current or voltage? Whoopsie, let's turn it all off." Maybe it's there but it isn't obvious.

A chain whipping and wheel smacking to ground is slow compared to 20khz. Very slow. This is probably related to the phase angle observer, which can only work on the time scale of full electrical revolution for sensor less and per hall change with hall sensors (generally less than 1khz changing times). Possibly the motor gets a complete stall or even goes backwards?

Have you considered an encoder for your bike? Is that a possibility? Hypothetically this 100% takes care of the above issue, though I doubt encoders are as well tested as hall sensors.
 
larsb said:
Not to bash the 75/300 vesc but people on the efoil forum have had quite a few broken. More than i would expect since that also is propeller drive.

Unfotunatly vesc are quite expensive and I know they are good at self destructing. That's why I hope to find a diy controler with separated and cheap to fix powerstage that is capable of 60+khz. I can not afford to burn one of those 75/300 for sure at 450€ a piece :?
mxlemming said:
Have you considered an encoder for your bike? Is that a possibility? Hypothetically this 100% takes care of the above issue, though I doubt encoders are as well tested as hall sensors.

Firstly thanks for your support mxlemming. I have used magnetic encoders on slow (very slow) generator and if it's the solution vesc already support this. For now i'll try sensorless but hall sensors where planned to be installed.

To cut on controler and inductance problem (will come back soon for sure if the results are as intended) I will keep you up to date at where i am in the prototype building.

- A little more detailled look at the simulations on wich the design is based first :

As said above the goal is a 2 stator stacked motor so here is a FEMM sim of it with 2 turns per stator per phase and the stators in series :

2_rotor_femm.PNG

With 100A peak in the windings (correctly 120° spaced and clocked for best torque) Femm output a linear force of 144N that corrolate to 8.61Nm with a medium radius of 60mm so a derived Kt (rms) of 0.1217 and a kv of 96. That is if the winding were radialy oriented since they are skewed over a pole width we can (as I understood, maybe it's not correct) multiply those value with a factor of 2/pi that gives a kt of 0.04476 and a Kv of 150.8.
The winding max cross section is 32.8mm². I'm using a 0.14mm enameled copper to cut on eddy current losses at high speed so the maximum number of strand per slot is theoretically 1673. With two turns it gives 836, assuming I can press 90% of that in there it give 753 strands in parallel (I have been optimistic here). So a 11.59mm² copper cross section. With extra, a turn length is approximatly 1m so giving us a phase resistance of 8.12 mOhm. (Sorry for the borring maybe not accurate math :roll: )

- Now doing it for real :

750 parallel strand is pain and after some thoughts here is how I did it :

winder1.jpg

The wheel has a slot as wide as the one in the stator and has a 1m perimeter. With the handwheel it take approximately 10min to do the 750 turns.

after that the copper is glued with cyanoacrylate based glue (to avoid wires being pulled in the later step) and tape in a notch in the wheel then cut :

notch.jpg

Once cut it gived me a mess of wire that was kinda hard to work with so i find the solution of wrapping it with a single enameled wire wich actually gived a satisfying phase bundle :

bundle.jpg

5 bundles later and here we are with the winded stator :

stator1.jpg

the stator carrier isn't ideal, switching from one side to another i modeled an axial path wich was the easiest in CAD but when winding wasn't the natural direction the wire wanted to take. It lead to some wire passing over and not wanting to stay in the slot
Next time i'll make that pass follow the skewing angle (it will slightly increase flux linkage so that's a good thing):
voids.jpg

I then proceed to impregnate the winding with HT epoxy resin and squeeze it between two glass sheets with the help of a mdf "press" to keep it flat and as thin as possible :

press.jpg

I ended up with a 10mm stator thickness instead of the 9 i was hoping for because of me beeing optimistic with copper fill and the carrier problem mentionned previously. If I can reduce the rotor-stator clearance to 0.5mm instead of the 1mm i was planning on we are good, if not the airgap will be increased and Kv further increased.

After 16 hours of cure time at ambiant temperature the stator is then post cured at 60°C for another 16 hours. This will bring the epoxy to it's rated glass transition temperature of 115°C. Laking a heat controled oven, i did it on my 3D printer heated bed with insulating sheet on top to limit power consumtion :

heat_curing.jpg

For watercooling the next step would be to lay a fiberglass cloth on to close each individual pockets but i'll keep that for later. As it stand the stator is rock solid, I feared that a vacum would have been needed to make the epoxy (wich is quite thick) impregnate the strands good but it appears that capilarity is slow but eventually does it thing (even to much since it has found his way through the short ends of turn further that i expected, should have tin them first ... Now I hope I can do it anyway).
Each 1m turn has 100g of copper, the 3D print is 40g and 40g off epoxy gives a 680g stator (feels substancial)

Now for the rotor building, due to shortage of stock (covid19) I had to buy some not ideal steel plate and do a lot of cutting to fit it on my small lathe :

rotor _stock.jpg

(sorry for the machinists out there, i know the stock mounting is sketchy, but whatever it take to get it done right ? :lol: )

After a LOT of passes :

rotor_machined.jpg

Magnets will later be glued to it using the same HT epoxy with a bit of thickening agent I supposed (I have not really looked into heavy duty magnet securing. The rotors has a 4mm machined lip so I supposed i'm good with centrifugal forces but who knows at 15000RPM what could happen :?
The second rotor is almost finished as well as the aluminium spacer (just needing parting at length) and the axle too.
As a firts test, since I haven't received my aluminium stock (2 months since I ordered it :evil:), I have 3D printed some outer shells and will test to see what it does with a single stator pretty soon I hope (Kv should be 300).

outer_shell.jpg
 
That 4mm lip on the rotor will be forming a magnetic bridge at the ends of the magnets. If you're planning on 15k rpm then you might have to live with the outer one but the inner wouldn't be hard to correct, I doubt it will be well enough balanced for 15k rpm though, usually even professionally CNCd parts would need some balancing for those kind of rpms.

The face the magnets mount to could probably benefit from a fine facing cut too, it looks like there's a step there. If you're having trouble with swapping tools for the inner and outer corners, you get used to it with practice but a bit of cigarette paper on the face you're trying to match up to can help. Damp the face, stick a bit of paper on and when you bring the tool up until it touches the paper it will be somewhere around 0.001" to 0.002" away from the face depending on the thickness of the paper.

No worries on pulling it into the chuck with a bolt btw, I'm not sure about "approved" practice but it's certainly common practice ;) Usually you'd machine a mandrel for the centre hole out of a bit of scrap bar and maybe add a drive dog. Worth playing around with centre mandrels, they're the usual method of accurately machining bored parts with a 3 jaw chuck.

Looking forward to seeing how this turns out :) :bigthumb:
 
stan.distortion said:
That 4mm lip on the rotor will be forming a magnetic bridge at the ends of the magnets. If you're planning on 15k rpm then you might have to live with the outer one but the inner wouldn't be hard to correct, I doubt it will be well enough balanced for 15k rpm though, usually even professionally CNCd parts would need some balancing for those kind of rpms.

Actually haven't tough about that at all :shock: ! Maybe reducing the lip heigth to half the magnet thickness will be sufficient for keeping the magnet from flying and cut the bridge ?
I don't think i'm going to cut on balancing for sure, static first and dynamic if needed (but for that i'll need an osciloscope and some IMU). The two stators version is expected to run at 8-9K max so a bit more realistic.

stan.distortion said:
The face the magnets mount to could probably benefit from a fine facing cut too, it looks like there's a step there. If you're having trouble with swapping tools for the inner and outer corners, you get used to it with practice but a bit of cigarette paper on the face you're trying to match up to can help. Damp the face, stick a bit of paper on and when you bring the tool up until it touches the paper it will be somewhere around 0.001" to 0.002" away from the face depending on the thickness of the paper.

Yeah it was shatering pretty badly when doing this cut, to much leverage from where the 3 jaws was holding and 5mm start to be a bit too thin to have enough rigidity I suppose. There is a step of 0.5mm near the center but it's there mostly because of me changing my mind during the cut between having a 5mm deep groove or having a 5mm back iron (as I should), and that because my stock beeing 10mm thick for a 10mm machined dimension :roll:
The next one is a bit better in that regards, I started with the face groove when the part still had rigidity and it went much better.

stan.distortion said:
No worries on pulling it into the chuck with a bolt btw, I'm not sure about "approved" practice but it's certainly common practice Usually you'd machine a mandrel for the centre hole out of a bit of scrap bar and maybe add a drive dog. Worth playing around with centre mandrels, they're the usual method of accurately machining bored parts with a 3 jaw chuck.

Good to know :)
That's actually a better way for sure ... That way everything is concentric with the bore and true to the first face reference. I just can't figure how would you fixtured the part for machining the bore in the beginning ? maybe a fixture plate ?

I plan on truing everything once mounted on the axle before gluing the magnet to increase my chances of having it balanced.
 
Thecoco974 said:
...
That's actually a better way for sure ... That way everything is concentric with the bore and true to the first face reference. I just can't figure how would you fixtured the part for machining the bore in the beginning ? maybe a fixture plate ?

I plan on truing everything once mounted on the axle before gluing the magnet to increase my chances of having it balanced.

Repeatable too, if you need to take the workpiece out it can be re-mounted with everything still centred. Usually you'd use a faceplate for the initial bore, a flat face with slots for clamping. I was going to say one would be worth getting if they're the right price but in all honesty it's usually years between uses here, they're good to have but they'll spend most of their time gathering dust. The easiest option is probably to temporarily weld a piece of heavy tube or drilled bar to the back and cut the welds after boring, if you're facing both sides then distortion shouldn't be much problem.

Another option you rarely see used these days is to mount the workpiece to the saddle, tool post or wherever convenient and put the drill in the chuck. Usually the spindle has a morse taper, mostly that's just used for turning between centres but there's no reason you can't use it for a drill chuck or other tools, lathes being used for milling operations used to be common practice. Adjustable boring heads are a great thing to have if you go that route, not crazy expensive and they're ok as fly cutters for facing. Adjustable reamers can be worth getting too, with practice they give very good results.

Not sure what would be the best option for dealing with the chatter, didn't realise it was that thin! A faceplate would be a big help to you there, pull the workpiece up against it either with a drawbar or revolving centre in the tailstock and it's a solid face to work on. If you think you'll be doing a lot of thin, large diameter parts then it could be worth considering one but you could knock something up for one-offs, maybe a disk of heavy plate welded to a bar or an old car brake disk, etc.

Could be worth trying those magnets before doing anything about the bridge, you could get some solid figures on what kind of effect it has before and after.
 
stan.distortion said:
Repeatable too, if you need to take the workpiece out it can be re-mounted with everything still centred. Usually you'd use a faceplate for the initial bore, a flat face with slots for clamping. I was going to say one would be worth getting if they're the right price but in all honesty it's usually years between uses here, they're good to have but they'll spend most of their time gathering dust. The easiest option is probably to temporarily weld a piece of heavy tube or drilled bar to the back and cut the welds after boring, if you're facing both sides then distortion shouldn't be much problem.

I actually did use your advise to true up the seconde face of one rotor, mandrel is the way to go for this :thumb:
Actually the lathe even if it's kind of cheap for this kind of tool was a lot of money to spend on a hobby like this for me, it might be a while before I pull the trigger on big accessories like this (might even not be worth it like you said, even more for this low quality chinese lathe). When I'll get a welder this trick will for sure be used :lol:

stan.distortion said:
Another option you rarely see used these days is to mount the workpiece to the saddle, tool post or wherever convenient and put the drill in the chuck. Usually the spindle has a morse taper, mostly that's just used for turning between centres but there's no reason you can't use it for a drill chuck or other tools, lathes being used for milling operations used to be common practice. Adjustable boring heads are a great thing to have if you go that route, not crazy expensive and they're ok as fly cutters for facing. Adjustable reamers can be worth getting too, with practice they give very good results.

I was actually planning on doing something like that to mill keyways in the axle, but it's to much work for a prototype and the probable small torque I will get out of this. You are mentionning a lot of cool tools I will surely want sooner than I think !

I haven't made a lot of progress the last 10 days, time is hard to find.
But i've managed to finish the rotor machining :
rotor1.jpg
rotor2.jpg

stan.distortion said:
Could be worth trying those magnets before doing anything about the bridge, you could get some solid figures on what kind of effect it has before and after.

Before I glue the magnet in I'm not sure how I could see if the bridging is as bad as we think :? I don't have a teslameter and even if I had one I don't know how i'd quantify the effect....

I also have dificulties soldering the enamaled copper wire even though it's the "solderable" kind. I can solder a single strand with my soldering iron but when putting a few strand twisted together in a makeshift solder pot nothing happen even after a few minutes. here is how I tried to do it :

solderpot.jpg

Maybe the temp is to low but seeing how the flux is turning black I don't think so ... Any ideas ?
 
Plumbing flux, most hardware shops will carry it. A good polish first with wire wool, emery, scotchbrite etc followed by a good smear of flux and it should tin easily, that torch will have more than enough heat.

Good job on the rotors :) Looks like the indexing came out well on the bolt holes too, always a pain to get right by hand. Not sure if that lathe has any gears driving the headstock, might not with the variable drive but if it has they can be used as a basic indexing head, just needs something to lock between the teeth. For ex, if it has a 24 tooth drive gear it can index every 15 degrees, score a mark with a tool or even make a mount to put a hand drill in the toolpost, ymmv with the accuracy of drill bearings and rigidity of the body but it can often give very good results.

Tooling is the real killer with all machine tools, a bottomless pit that can never be entirely filled. Figuring out how to get the most out of what you have is a big part of the challenge and figuring out what tooling is worth getting can be just as hard. Every long established engineering workshop has lots of equipment that should have been extremely useful but is just gathering dust and lots that was bought as a one off and turned out to be indispensable. Worth keeping an eye of the classified adverts, bargains often come up and it can often be worth buying a machine just for the tooling that comes with it and sell the machine on again.

Nothing wrong at all with that lathe, tried similar before and they're not bad at all and spent hundreds of hours on what was essentially the same machine scaled up. Is that one 550w or 750w? Either way, that's an awful lot of power for a small machine. If it has a drive belt I'd suggest slackening it off to the point where it has just enough drive for your needs, it can save you a fortune in tooling in the long run (probably saved me over $100 in parting off tools on a similar sized machine just today).

I'd also suggest keeping an eye out for a coolant pump, feed etc, ideally a proper pump and tank but cheap 15$ fishtank pumps can work ok. It's too easy to get into the habit of not using coolant and it's a very bad habit to get into, tool life, time taken, finish quality, they all take a step backwards without coolant.

A welder is definitely a good thing to have. If you know anyone with a stick welder, give one a try. Maybe you'll get on well with them (some don't, I've never been any good with stick) and if you do you'll save yourself quite a bit in both money and headaches. Mig welders are great but they can be a pain in the neck sometimes, wire rusting when standing, liners wearing out, tips need replacing and gas... and more gas... If you get one and it proves useful, get a rent-free bottle asap, the disposable bottles are a curse! Maybe consider tig too, extremely versatile and it doesn't have to be expensive, scratch start with a tig torch on a regular stick welding unit works ok, certainly well enough to see if an AC/DC inverter unit is worth getting (you can weld aluminium then :) ) Oxy-acetylene would be well worth considering too, possibly the most useful addition to any workshop. Cutting, welding , brazing, freeing up rusty things... not good for heavy welding but fine for up to around 3mm (far more expensive than electric at those thicknesses though).

Sorry, I'm rambling. Hard not to in this place though, there's some absolutely stunning workmanship here and judging by your progress so far, it looks like there will be plenty more to come :)
 
A good paint stripper will take enamel off after a few minutes,.. could be another thing to try?
 
First off, this looks awesome. I really like you method for making the multi-strand bundles.

As far as the wire enamel goes. I found that color enamel is typically polyamideimide. I have not found an easy way to work with it. I found that a lighter will cook it off, but too long and it will melt the wire. I found that using a weaker enamel was worth the trouble it saved. If you do figure something out, I would love to switch back to that type of wire for the improved heat tolerance.

Keep it up!
 
I just stumbled across this video on Youtube. thought it might be helpful.
https://www.youtube.com/watch?v=dlfzlzulb54
 
Reviving this old thread, hard to realise more than a year as passed since my last post ...

Sorry for not a answerring to all of your great suggestions back then, I had been caugth in my professional work without motivation left for this diy motor adventure.

To respond to stan, It is a 750W lathe with direct belt drive to the spindle but has a gear drive for the lead screw so indexing is possible !
When I get more time for myself I fear I will be dropping very quicly in this infinite tool buy and collection habit ^^
I still don't have a welder but I have a little experience with stick welding and will surely have one at some point, possibly a multi-process unit for tig capability and aluminum with mig (AC/DC welder are crazy expensiv ...)

Getting back on the build itself, I never got the first stator phases to tin properly and decided on building a second one, this time with splitted individual length of 236 stands each. here find some pictures of how it had gone :
IMG_20220312_131150.jpg
IMG_20220320_173344.jpg
This time learning from previous troubles I pretined all terminaison before winding. But as you van see on previous pictures I still got troubles tinning 236 stands tight together at one go, the cotting seems to tin very good on single strand but when tight together I suspect the heat being wicked away to much. Having the tin to hot seems also to put the vernish on a "ccoked" state sligthly brown and it is then not solderable at all. I got away some time with the right temp tin bath (I have a cheap solder pot now) and preading the stands help also a bit.
IMG_20220322_191955.jpg
IMG_20220327_151216.jpg
I tighten the press a bit to much this time and the glass didn't survive. I need to find stronger material to replace the mdf board.
IMG_20220829_084829.jpg

I machined the casing pieces out of 10mm aluminium plates. But have taken them a bit to thin and not as rigid as I would like. Also I haven't found suitable stock to do the outer parts so they are 3D printed for now. I don't want to machine this on a big aluminium block because of cost and all the lost material. For the final parts I might try to cast and them.

On this stator I took 75% of the initial stand count to be sure to have it to the right thikness once pressed. The split individual turn length is so that I can configure it either as a 3turn/phase ore 6 turn/phase. for indivual testing or multistage connexion. It inevitably ended in a mess of wire once connected :lol:

I had some issue centering the stator inside the rotor but enventualy got it right with 3d printed axle spacers. I hot glued the hall sensors inside the cavities between phases and the glue was what was draging.

It is now turning freely and I got it running with a Vesc 75100.
Detection come out as 70mOhm and 10uH, I got it to 19825 ERPM (so 2832rpm with a 7 pole pair rotor) with 70% dutycycle and a 23.5V battery. If I'm correct that's around 170kv. It's a bit higher than the 120 I got with a test 6 turn stator. I suspect this is because when guing the magnet I did'nt got them perfectly lined up with my previous bolt placement and the magnet are a bit skewed. This should be fixed by elongating the rotor bolt holes.

As of now I haven't got it to properly start sensorless, it just wondered back and forth. With hall sensors it run quite good, until the transition point to sensorless were, if the current is to high, the vesc throws a ABS overcurrent code. I will post a video showing the motor running and this issue soon.
 
Here is a video showing the VESC configuration and issue with sensorless :

https://youtu.be/IblrYzZZGLE
 
Hi, beautiful work on that motor.

Best guess to why it won't spin in sensorless is the lambda is wrong. You can calculate lambda from the kV using
lambda = 60 / (sqrt(3) * 2 * pi * kv * Npp) = 0.046. lambda is the most important parameter for sensorless.

At the moment there are a lot of people struggling with hall to sensorless transition.
 
Back
Top