Miles' 90mm inrunner build thread

[Doctorbass]

Hmm interesting thread!!

Btw, That motor really look like the Zero 2013 75-7 new motor! same kind of inrunner desing...

The new 2013 70hp capable Zero motor cost 1200$.. not bad !

Doc

 

[Miles]

Be carefull Miles,
That 1st taste of the hard stuff msy get you hooked!

I'll soon be mugging old ladys to fund my Emetor habit....

 

[Miles]

Btw, That motor really look like the Zero 2013 75-7 new motor! same kind of inrunner desing...

Also the Joby motors.

 

[Miles]

Getting pretty good results with a 27t 30p version.......

 

[Miles]

Peak efficiency for the 27t 30p version is 95.6% at a torque of 3.5Nm...

We'll see how much of a hit there is when I switch on non-linear properties....

 

[liveforphysics]

Peak efficiency for the 27t 30p version is 95.6% at a torque of 3.5Nm...

We'll see how much of a hit there is when I switch on non-linear properties....

750hz doesn't sound like much for a commutation freq, I don't know what controller you're planning to use my friend, but many don't do 750hz nearly as well as I would like them to do.

The motor looks fantastic, I love the rotor design, I love the perfect BEMF sculpting, and I think the predicted performance looks nothing short of fantastic.

 

[Miles]

Thanks Luke.

......I don't know what controller you're planning to use my friend, but many don't do 750hz nearly as well as I would like them to do.

I'm hoping things will have improved a bit by the time this actually gets built.... LOL

 

[Miles]

[pre]Comparison between 24t 28p 27t 30p

Winding factor (2 layers) 0.933 0.945
Number of symmetries 2 3
Number of cogging steps 168 270[/pre]

 

[Miles]

27t 30p cross-section.

 

[Miles]

Comparing the results of my simulations, so far, there is little difference in efficiency between the two options. The most significant difference is in the amount of torque ripple at low ouput. At 0.5Nm, the 24t 28p has 15.1% torque ripple whereas the 27t 30p has only 2.4%. Obviously, at higher outputs the differences converge....

 

[Miles]

The most significant difference is in the amount of torque ripple at low ouput. Obviously, at higher outputs the differences converge....

[pre]. 0.5Nm 1.5Nm 3.0Nm 5.0Nm

27t 30p 2.4% 1.8% 1.7% 1.7%

24t 28p 15.1% 5.1% 2.5% 1.7%[/pre]

 

[Miles]

Regarding the relation between pole count and Kt:


[pre]. 1.5 Nm 3.0 Nm

30p (0.0241 Ohms) 11.0 amps 22.0 amps

28p (0.0193) Ohms) 12.2 amps 24.4 amps[/pre]

 

[bearing]

So, copper losses at the same torque is slightly worse with 30p?
You would expect them to be better, since the winding factor is higher. Are the end turn losses higher with 30p? or is the total copper area smaller?

 

[Miles]

So, copper losses at the same torque is slightly worse with 30p?
You would expect them to be better, since the winding factor is higher. Are the end turn losses higher with 30p? or is the total copper area smaller?

The 24 slot stator has a total copper area of 1284mm². The 27 slot stator has a total copper area of 1272mm².

Just done some non-linear simulations. Peak efficiency is up

 

[Miles]

For the 27t 30p design, eta max. 95.75% @ 3.5Nm (1100W output @3000rpm. 49W of heat....)

For the 24t 28p design, eta max. 95.87% @ 3,5Nm.

 

[bearing]

Just done some non-linear simulations. Peak efficiency is up

Nice!

That's what I saw too. It's probably because when you use non-linear materials, the peaks of flux are lower, so the predicted iron losses gets lower too. Which seems a bit strange/counter intuitive.

The 24 slot stator has a total copper area of 1284mm². The 27 slot stator has a total copper area of 1272mm².

What is the "Average conductor length" compared to the machine length of the two?

 

[Miles]

Used up all my credits already

 

[Miles]

What is the "Average conductor length" compared to the machine length of the two?

27t 30p version A.C.L. is 38mm
24t 28p version A.C.L is 39mm
machine length 30mm

 

[Miles]

Lots of things change slightly when non-linear properties are used. Back-EMF waveform, torque ripple %, etc.

 

[Miles]

The highest I got to was 7Nm output torque @ 93.4% efficiency. (2210W output power, 157W heat)

Kt down 12%

 

[Miles]

How much heat can an outrunner 92mm in diameter, 56mm long and 1.1kg in weight, disperse continuously? No venting (sealed case).

 

[bearing]

Lots of things change slightly when non-linear properties are used. Back-EMF waveform, torque ripple %, etc.

I can imagine that the results only changes slightly on your well balanced design. On the Colossus, there are quite big changes with non-linear materials. Probably because the iron is saturated pretty far, even at no load.

 

[bearing]

[pre]Comparison between 24t 28p 27t 30p
Winding factor (2 layers) 0.933 0.945[/pre]

The 24 slot stator has a total copper area of 1284mm². The 27 slot stator has a total copper area of 1272mm².

What is the "Average conductor length" compared to the machine length of the two?

27t 30p version A.C.L. is 38mm
24t 28p version A.C.L is 39mm
machine length 30mm

Based on these numbers, the 27t 30p should have less copper losses, IMO. Also, in pdf's you have posted, the 30p has more magnets of the same size (more magnet angle) which I think should give it a higher flux density. And that should also give it less copper losses at the same torque.

 

[bearing]

The highest I got to was 7Nm output torque @ 93.4% efficiency. (2210W output power, 157W heat)

Kt down 12%

Did you ever try with sinusoidal currents?

I have noticed that the peak flux in teeth is lower at the same torque, compared to star connected DC.

When running with sine AC, it also calculated the inductance.

 

[Miles]

Based on these numbers, the 27t 30p should have less copper losses, IMO. Also, in pdf's you have posted, the 30p has more magnets of the same size (more magnet angle) which I think should give it a higher flux density. And that should also give it less copper losses at the same torque.

Final magnet angles were:

30 pole 148 deg.

28 pole 150 deg.

This gives the same absolute distance between them for the upstand.retainer.

So, magnet area is slightly less for the 30p (same thickness and magnet strength).