The limits of torque production for a given weight of motor

[speedmd]

Max torque for how long. Sustainable? 20 minutes? 5 minutes? Headroom will vary with weight /size /package configuration greatly.

 

[toolman2]

Interesting simulation

Comparing to the Astro:

We can run the Astro 3220 at 10,000 rpm.

It can make about 5Nm of torque continuously.

The simulation has 23Nm at 400rpm for 1000W input (I assume that is continuous, as based on 80°C winding temp?). The Astro would be 5.2Kw output.

Thats good Miles, and fine work not getting derailed, yet.
The Astro is as good an egsample as any for this job so its an interesting challenge to make 20 times the torque from the same amount of active material at 20 times less rpm as the first challenge, I think it makes sense theoretically that it can be done and it would clearly have more surface area for cooling from the bigger diameter, and the structural challenge is a separate one.

So am i right in saying that currently we are at 5 times the diameter of the Astro's stator od at present? and is the stack length is 15mm vs 60?mm for the Astro?
100Nm for this design might be getting into ir loss or saturation? be interesting though to compare to the Astros losses at 5Nm, probably not to bad until over 7.5Nm.
And are you then thinking of optimizing this size a bit, or going bigger diameter and shorter stack?

 

[Miles]

Here is exactly the same configuration/parameters apart from increasing the current by 100%

NB:

The voltage is floating and speed constant.

Emetor doesn't simulate for magnet losses. They shouldn't be that significant for this pole/slot combination and frequency, though.

Emetor doesn't simulate for coil interconnection losses.

See the "Average conductor length" figure in "Axial parameters" and compare it to the "machine length" figure, to get the ratio of active to inactive copper.

 

[Miles]

So am i right in saying that currently we are at 5 times the diameter of the Astro's stator od at present? and is the stack length is 15mm vs 60?mm for the Astro?

Astro 3220:
Stator OD 76mm
Airgap dia. 43.6mm
Stack length 50.8mm

Simulation 1.001:
Rotor OD 400mm
Airgap dia. 388.6mm
stack length 15mm

 

[Miles]

And are you then thinking of optimizing this size a bit, or going bigger diameter and shorter stack?

I suggest that we optimise this size and find its limits first.

 

[Miles]

I'll calculate a figure for the mass of active material in the Astro.

 

[Miles]

I'll calculate a figure for the mass of active material in the Astro.

It's around 1.6 kg

 

[speedmd]

Great work Miles. We have a bit more active material but some rather excellent performance estimates. I am struck by the copper losses. At the expense of KV should we be looking at lower turn counts to further optimize?

 

[Miles]

I am struck by the copper losses. At the expense of KV should we be looking at lower turn counts to further optimize?

I'm not sure that's going to make much difference..... If you do gain in fill factor, from that (not simulated for) you lose from the interconnection loss factor (not simulated for). Pretty marginal.

We can set the fill factor to whatever seems plausible. I used 0.6, which is about as good as you can get with hand winding in the conventional way. Non-circular wire might push it up to 0.7 ish, depends on the scale, thiough....

 

[speedmd]

Would you not loose some resistive load in the shorter wire runs? Assuming that going from four turns to 2 turns would be significant.

 

[Miles]

Would you not loose some resistive load in the shorter wire runs? Assuming that going from four turns to 2 turns would be significant.

Sorry, you've lost me.......

The nominal halving of length would balance the fact that losses go up with the second power of current. Same losses per unit torque......

Or, you mean something else?

 

[speedmd]

I am assuming the shorter wire run is always better. Understood that the current would need to go up accordingly. Same fill. What am I missing?

 

[Miles]

What do you mean by "shorter wire run"?

If you assume the same copper fill, there is a slight advantage (for the motor) to having a higher turn count, rather than a lower.

 

[speedmd]

It takes a longer wire to wrap a tooth eight times than 4 times. Roughly double the length. Assuming the same fill you would have something like 8 turns of 4 strands vs 4 turns of eight identical strands.

 

[Miles]

It takes a longer wire to wrap a tooth eight times than 4 times. Roughly double the length. Assuming the same fill you would have something like 8 turns of 4 strands vs 4 turns of eight identical strands.

Half the number of turns means double the cross sectional area and half the length. Resistance is halved and halved again.

Half the number of turns means double the current for the same torque. Losses go up with the second power of the current.

Where's the advantage, there?

On the other hand...

The length of wire connecting coils and phase groups is the same, no matter what the turn count. So, the relative resistance of it diminishes as the turn count increases.

 

[speedmd]

I was assuming the current rise loses (resistance) were with respect to the same wire resistance. Was also assuming inductive losses would be better with less turns. Understood on the connecting wires which makes sense unless they are also fattened up. Thanks for the clarification.

 

[Punx0r]

Here is exactly the same configuration/parameters apart from increasing the current by 100%

Interesting

Torque has doubled to ~46Nm (so still linear). If it doubled again it would be on par with the Astro with a 90% efficient gear reduction. Granted, with 0.4kg more active material.

One thing is clear, with the rotor iron representing 16% of total mass, a Halbach array wouldn't save much as the motor would still require and outer supporting ring of something.

 

[Miles]

Ok. here's the simulation for 80 amps.

This is how we're doing, so far..

20A  23.24Nm  1.162Nm/A   59W  94.1%
40A  45.79Nm  1.144Nm/A  142W  93.0%
80A  87.47Nm  1.093Nm/A  473W  88.5%

 

[speedmd]

WOW. Over 60 foot pounds of torque! This puppy certainly does not need a gearbox unless your hauling lumber.

 

[Miles]

So, any volunteers to work on the structural aspect?

I haven't got into structural FEA yet...

 

[parabellum]

So, any volunteers to work on the structural aspect?

I haven't got into structural FEA yet...

If the idea is to bring it to the shaft in the middle, could something like tension only spoke support be considered? Rotor spoked over stator spokes on static cylindrical or tubular shaft looks very light, but aligning and gap symmetry may get one issue.
P.S. Just to make a comparison: Turnigy 80100 concept with tension spokes only.

 

[Miles]

I think the first thing we need to do is make the active materials as rigid as possible. At the moment we've, effectively, got two hoops of 2mm thick steel 400mm and 370mm in diameter..... We need to turn them into better structures.

 

[speedmd]

Something like what you posted in the ultimate hub thread could work. Disc brake mount suitable?

 

[Miles]

I've done this one as an outrunner. We could certainly invert it. At this diameter, it doesn't make a lot of difference either way.

 

[Punx0r]

Looking good Miles Looks like winding losses are starting to become significant, but at 89% it's still better than the Astro (say 90%) plus a 20:1 gear reduction, torque output is now comparable. Yes, it's a little more active material, considering the Astro is an excellent motor and this is a first go...

As you say, now just for the small matter of the structural stuff...

Large diameter hollow axle (possibly aluminium) with thin ball bearings (ceramic may be appropriate here if we're gram-shaving). For side covers I see discs with tapered radial ribs, effectively emulating a spoked wheel in carbon fibre/magnesium/aluminium depending on desired lightness-cost. Heck, would it be feasible to actually support the stator and rotor rings with actual spokes? A thin, lightweight polymer could provide weather sealing. It might be an old-tech construction but a bicycle wheel is damn strong considering it's size and lateral & radial stiffness.

Just a thought: Is the magnet thickness optimal? Any reasonable ways to increase winding fill factor?