My Axial Motor Project

[IanFiTheDwarf]

I think I will be pretty safe then, I hadn’t really thought about the gearing.

A lot of these type hubs apart from being a press fit do rely on the shaft end bolt to hold them together but some like this one are have a ring shrunk into a grove to hold them together. I think all bolt on rear hubs will be the same as there is no drive shaft to hold them together.

The ring marked by the red arrow is heated then dropper over the hub and shrinks as it cools into a grove. This hub can’t be disassembled without cutting or breaking this ring.

I will post an edited drawing showing the hub in two parts for clarity.

 

[IanFiTheDwarf]

Updated drawing showing fixed and rotating parts of the hub

 

[jaz1902]

Greetings All: I stumbled upon ES while researching applications of the Halbach Array. Glad to have found this thread, though activity seems to have fallen off for 2014... Cheers, Joe.

 

[IanFiTheDwarf]

Hi all, I haven’t given up on this project, I just have a couple of things to work out before I commit the £500+ in parts to build it. I have done most of the calculations and my plan to use magnetite resin for the coil cores just isn’t going to work. The magnetite will saturate way before the rest of the aspects of the motor reach their limit.

I still like the idea of using magnetite cast in resin to get the complex shape but I need the higher saturation of silicon steel laminations. I was thinking of casting a simple bar of laminations inside the magnetite resin giving the best of both worlds. I even thought of robbing the laminations from transformers for the first test motor. I know the losses will be high due to the thickness of the laminations but it’s just to test the theory before spending a lot of cash on thin laminations.

I also found out that mild steel mig welding wire actually has silicon in it to improve the properties of the weld. I couldn’t find a full brake down of the composition to see how close it is to electrical steel but it did get me thinking, if I were to spray 0.6mm mig wire with lacquer to insulate it then cut it into lengths it could be bunched together to make any 2D shape.

From what I’ve worked out the specs I’m hoping for are

Two rotors 256mm in diameter

16 magnet poles per rotor using 1/4" thick N42 arc segment magnets with rectangular magnets between them to make up the hallbach array

18 individual coils of 32 turns (can be configured 3S2P, 2S3P or 6P for different voltage/current)

Overall dimensions should be about 320mm diameter and 120mm deep

The power and torque from the motor is dependent on cooling which will be from oil pumped round with an electric power steering pump. I have taken a realistic value for the rate I can remove heat from the motor and it allows the following values

Maximum Continuous Torque 400NM
Maximum Continuous Power 103KW
Maximum RPM 7500

All the figures here are from a spreadsheet I made up with formula I found in various places. I couldn’t find a torque formula for an axial motor so I just used the median radius which will give a slightly lower torque figure than the actual as there is more magnet area on the outside half of the ring of magnets than the inside half. The magnetic field in the air gap came from a femm simulation which I did in slices and took the average.

 

[IanFiTheDwarf]

I have started assembling some of the parts together, so far all I have is the hub and discs but that will at least let me make the rest of the measurements needed. I’m thinking of making up some dummy magnets to try everything out before spending the money on the real magnets.

My next job is to find laminations for the coils and work out how I am going to mount them

Excuse the odd bolts. I have ordered cap screws from eBay but they haven’t arrived yet.

 

[fechter]

I also found out that mild steel mig welding wire actually has silicon in it to improve the properties of the weld. I couldn’t find a full brake down of the composition to see how close it is to electrical steel but it did get me thinking, if I were to spray 0.6mm mig wire with lacquer to insulate it then cut it into lengths it could be bunched together to make any 2D shape.

If the wire is round, there will be a good percentage of open space between them which would result in lower saturation flux. A hexagon shaped wire would be ideal.

 

[IanFiTheDwarf]

If the wire is round, there will be a good percentage of open space between them which would result in lower saturation flux. A hexagon shaped wire would be ideal.

I think I am better sticking to the laminations cast into magnetite, I just have to find a supplier.


I have been thinking about how I am going to hold the magnets to the discs and I am very wary of just gluing them on, no matter what glue I use. I can calculate the force on the magnets but I can never be sure how much force the glue can take and if there will be any weakening of the bond due to heat, vibration or anything else. I think some kind of clamping arrangement would give me much more confidence when the thing is spinning at 5000+ rpm as its strength can be calculated.

Something like this would work well

I have a little over 6mm space on the outside edge of the disc and 4.5 mm space between the magnets and the hub of the disc to play with, but any clamp will have to be partially over the magnets and this will require the coil core hammer heads to be shaped to match.
The worst thing about this idea is I am going to have to machine 64 clamps

 

[Lebowski]

Why not do like i did and use an aluminium sheet with magnet sized holes ? Mine are also not glued, i've had it upto 1500 rpm...

Big difference though i use aircoils, so without core iron close by my magnets really stick to the back iron

 

[Tench]

Your clamps could be full circle rings which would be simple to machine.

 

[IanFiTheDwarf]

I’m using a halbach array so my magnets will make a continuous ring and won’t be able to fit into holes. I did think about using a complete ring but it would be bigger than my mill travel.
I might have to buy a rotary table and dividing head to accurately drill all the holes for the clamps anyway so that would let me mill a complete ring.

While I say my magnets make a complete ring that’s not telling the whole story, the radially polarised magnets are much smaller and rectangular not arc segments so the outside shape is a little like the petals of a flower not a perfect circle. I was also going to use slightly shorter magnets than the arc segments to give me space to screw the clamps. Using narrower radial magnets to make the halbach array dose reduce the effect of enhancing and cancelling the field but not as much as you would expect, it is definitely not a linear relationship between the ratio of magnet widths and the effect.
Shortening the radially polarised magnet length dose reduce the effect by a large amount but I figured it was worth it to get a secure fixing and I still have a steel backing to complete the magnetic circuit.

 

[IanFiTheDwarf]

I’m thinking that machining 64 clamps from scratch is just too much work, so I’m going back to machining a lip in the discs and was thinking of using “T” nuts to clamp the magnets down. I will still need to take a few mm off each “T” nut but that’s just one operation and won’t need any setting.

It will also mean I need a larger area of the coil core recessed to clear the “T” nuts.

On the subject of clearance, how much air gap do you think I will need? I was thinking 1.5 to 2mm.

I have also added a spigot ring to the disc to keep it aligned.

 

[Lebowski]

I typically have an airgap of 1mm on either side of the stator. But I have aircoils meaning there's
no attraction between stator and rotor. In your case there will be a significant force, so when deciding
on your airgap you will also have to keep warping of the stator in mind..

 

[Miles]

Great to see this coming together..

On the subject of clearance, how much air gap do you think I will need? I was thinking 1.5 to 2mm.

You're using shims to adjust? I'd say 1.2mm to 2mm, loaded. That's one of the advantages of axial flux, having control over the airgap.

 

[IanFiTheDwarf]

Cheers guys, I think I will plan for 1.5mm unloaded, but as you say I can add shims to adjust the air gap as required. I’m going to start with a 1mm shim between the hub and first disc and no shims between the discs. From this point I can increase the left most air gap by moving the shim from between the hub and first disc to between the discs and add more shims between the discs to increase the right air gap.
The discs are 10.8mm thick before machining and I wouldn’t want to take them down to less than 8mm which gives me a 2.8mm lip to hold the magnets.

Gaps show where the shims woud go

My next problem is finding laminations to embed in the coil cores, I was hoping to find bare transformer cores that I could cut down to size but I can’t find any off the shelf.
RS components used to sell EI cores so that you could wind your own transformer but I can’t find any suppliers anymore.

 

[Lebowski]

looking at your motor picture, it reminds me of my thoughts on why a motor needs iron in the windings...

My own private thoughts on metal cores say that the only reason they are there is to channel the flux
through a small area, for the reason that then the wire length for X amount of windings (wound around this
small area) will be shorter (so lower resistance and better efficiency).

In the picture your coil core looks so fat, it kind defeats its whole purpose as the wire length for the windings
will not really be reduced. The magnetic flux should be channeled though an area so small the iron is almost
saturated. Then the wire length for the windings will be very short, giving you less resistance and better efficiency.


Second, make sure you can actually put the motor together and take it apart again ! Fully loaded with magnets
forces are going to be huge, make sure you add stuff for pullers to get a grip, so you can pull it apart and gently
put it together.

 

[IanFiTheDwarf]

When I first planned to use just magnetite for the cores I tried to maximise the area of the core to avoid saturation. Now that I am planning to imbed stacks of laminations in the magnetite cores I will be able to reduce the size of the cores somewhat, but it’s all a bit of a balancing act. The smaller the cores are the more efficient the motor will be, but will limit peak torque due to saturation.
I am building this motor to power a car at hopefully better than the standard performance so I don’t mind sacrificing a couple of present efficiency for more power.
I haven’t done any calculations with the laminations in the cores as I was hoping to find suitable transformer cores to use and do the calculations with their dimensions rather than just guess.

 

[Miles]

My own private thoughts on metal cores say that the only reason they are there is to channel the flux
through a small area, for the reason that then the wire length for X amount of windings (wound around this
small area) will be shorter (so lower resistance and better efficiency).

In the picture your coil core looks so fat, it kind defeats its whole purpose as the wire length for the windings
will not really be reduced. The magnetic flux should be channeled though an area so small the iron is almost
saturated. Then the wire length for the windings will be very short, giving you less resistance and better efficiency.

I don't see the connection between core area and its loading? A larger core area has greater flux capacity, which increases Kt - copper losses follow the square of the current.... It also raises iron losses, though. Larger copper area lowers resistance but copper losses are only directly proportional to resistance. What are your priorities?

 

[IanFiTheDwarf]

Correct me if I am wrong but Kt does not increase with the area of the core, it is constant but the amount of current that can be carried by the copper before the core saturates will increase, and therefore the maximum torque will increase.

I know the iron losses are going to be higher than ideal because I am just using what I can find, and not having very thin laminations stamped but I am just going to have to put up with it for this motor.

I have calculated the copper losses at about 3% at full load which I think is acceptable. I am designing the cooling to be able to move 3kw of heat continuously, so the motor can be run at full load for prolonged periods.

My priorities are just to implement the ideas I have into a motor I can use in a practical way. In this case to power a car to at least as good as its performance with the standard IC engine. I don’t need it to get 99% efficiency, or weight less than 20Kgs but near 90% efficiency would be good, and I think 40kgs is achievable

 

[Miles]

Correct me if I am wrong but Kt does not increase with the area of the core, it is constant but the amount of current that can be carried by the copper before the core saturates will increase, and therefore the maximum torque will increase.

Surely, as long as you maintain the same flux density in the teeth, Kt will increase with their cross-sectional area? Right?

I'd better test this :wink:

 

[kiwifiat]

Correct me if I am wrong but Kt does not increase with the area of the core, it is constant but the amount of current that can be carried by the copper before the core saturates will increase, and therefore the maximum torque will increase.

Surely, as long as you maintain the same flux density in the teeth, Kt will increase with their cross-sectional area? Right?

I'd better test this :wink:

Yes, and in order to maintain the same flux density in a larger core either N or I or a combination of both are required and of course a larger core can accomodate that.

Keep up the good work, I am working on a similar project to power a small car. I likewise am not using efficiency as a primary driver for the design as I am certain to beat the 30% of an ICE and loosing a few pounds of body weight will impact the overall system efficiency more than a few % in the motor design. That is not to say that we shouldn't employ best practice when possible but often to get the job done pragmatism wins over.

 

[parabellum]

Any progress?

 

[IanFiTheDwarf]

Hi, I haven't been able to do much for a while due to work and and not having my new workshop built yet but I haven't given up.

I'm planning a slightly simplified version with simple laminated cores (no hammer heads or magnetite) and a sandwiched flat plate design to allow me to build it without a mill, and just to get something working.

I just need to find the cash for the parts now, especially the magnets which are quite expensive.
I am going to buy the FR4 sheets off this month's pay if I can but my kids might want some Christmas presents so I'll have to see.