Hub Motor Design Simulations

[anon94428844]

Hello All,

I’m here to bring forward some simulation data on the available hub motors. These simulations assume that the VESC is a Sinusoidal, Space Vector Modulation @ 6 Khz or 20Khz. I am going to present the information available for @Hummie and @jacobbloy’s hub motor designs. Because of the limitations of the software I am using, I cannot simulate their temperatures. That said, the temperatures I do simulate do not account for passive air cooling through vents or shaft/hangar heat sinks (@carl.1).

Target RMS Current is 30Amps for these simulations. I do account for the voltage difference between Jacobs (10s) and Hummie’s (12s). I ask that everyone STOP TALKING ABOUT KV. Focus on Temperature and Torque because those are the variables we need to be more careful about - not speed! Scaling can happen later. If you can respect that, then please add your commentary.

I’ll update this as I get around to it. If people want me to produce more accurate results, I do not have any version of Hummie’s motors so please tell me the specs.

Jacob’s V2 Motors (24 Pole, Wye)

@jacobbloy

0.19mm (36 gage) wire… strand count should be between 18 and 23… 8 turns… 0.480ohms per phase.

6Khz SVM Freq.

20Khz SVM Freq.

Winding Specs.

Hummie’s V2 Motors (12 Pole, Wye???, 80 kv)

@Hummie

80kv have 33 turn and the 90kv have 30 turns. Single strand.

I interpert this to mean that his 80 kv model is 24.25 Awg, 1 Strand, 33 Turns.

6Khz SVM Freq.

20Khz SVM Freq.

Winding Specs.

Hummie’s V2 Motors (12 Pole, Wye???, 90 kv)

@Hummie

80kv have 33 turn and the 90kv have 30 turns. Single strand.

I interpert this to mean that his 90 kv model is 24.75 Awg, 1 Strand, 30 Turns.

6Khz SVM Freq.

20Khz SVM Freq.

Winding Specs.

Hummie’s V2 Motors (12 Pole, Wye???, 100 kv)

@Hummie

27 turns and 100kv.

I interpert this to mean that his 100 kv model is Awg, 1 Strand, 27 Turns.

6Khz SVM Freq.

20Khz SVM Freq.

Winding Specs.

Reference:

VESC: Sinusiosal SVM 20,000hz switching

Torque for 80mm Wheel to stall/equal to gravity with a 100kg load: 05% - 3.9Nm total (or 3.9/2 Nm per motor) 10% - 7.8Nm total 15% - 12Nm total 20% - 15Nm total 25% - 19Nm total

RPM/Speed for 80mm Wheel: 0560 - 05 mph / 8.1 kmh 1100 - 10 mph / 16 kmh 1600 - 15 mph / 24 kmh 2100 - 20 mph / 32 kmh 2700 - 25 mph / 40 kmh

For the same max current, Jacob’s Hubs are superior. That does not mean that they are without fault. The losses are significantly lower however. -KM

[Pedrodemio]

Amazing data, the program you use is specifically made for simulating motors or just a FEA with your own modeling?

i read in another thread that you are developing your own hub’s that are simulation optimized, any more info on that? any idea when they will become available?

[johnny_261]

I have no idea what any of this means, but looks cool! Maybe at some point, you can give the “for dummies” run down at some point of your findings!

[anon94428844]

This is MotorSolve by Infolytica. Its designed to simulate BLDC motors, a real handy program but its hard to get a hold of. University IT people might help get an EDU discount, but its about $500-700 a copy. Very useful tool though if you’re planning to build motors.

I am building my own motors and also an ESC specifically for them. If and when they get to market, they would be released as a set because the drive signal would not be SVM but Current Hysteresis. This allows me to have better control over the motors temperature and torque.

[anon94428844]

Compare the Torque of each model’s graph to the RPM, Compare the Current of each model’s graph to the RPM, Compare the Efficiency too.

Thats the simplest way to compare them.

Changed by 10 char to make this a reply

[Hummie]

This is a simulation based on assumed resistance and inductance numbers and the simulation doesn’t take into account magnet strength, air gap, and permeability of the back iron which will affect the inductance of the whole motor beyond the winding What size stators are they in the simulation? I can give u real inductance and resistance numbers when I plug back in the vescs as I forget the numbers but there are other variables

[anon94428844]

False, it does take all of that into account. These are finite element simulations. I have been working from the specs I have found online and have been shared by you.

Here is an example from your 80 kv design, I only changed the winding between these simulations. I have done my best to recreate your designs. I have Jacobs and can verify the accuracy of these simulations. I cannot accurately represent yours without a sample.

Full Hummie Motor Specs

[Hummie]

You can’t know the inductance in the airgap without knowing the airgap width, magnet strength and the permability of the back iron. The winding inductance is just one side of the gap and the other will affect it hugely. No? It might as well be iron less in this simulation. You could compare iron less to a huge back iron and get the same numbers here but have drastically different performance in the motors

[anon94428844]

It’s specified in the above specifications. Jacobs was 0.1mm. Yours was somewhere between 0.5 and 0.1mm as I recall. I specified the magnet materials as Neodymium Iron Boron 38/15 which translates to N35 (Remember, there is no N52 magnet that is rated to survive 200*C. N35s have that rating.). Iron losses can also be calculated from the 0.2mm laminations of M-19 29Ga silicone steel. I did not include simulations with losses because it would distort the graph in an attempt to display the kW losses (apparently the rubber-melting heat can be identified my friend).

I am not here to attack you, but I need you to help me out here. I can help Jacob patch up his motors to some degree with the CAD and numbers I already have from him. I have none of yours so I’ve had to speculate based off an old schematic you released (and have continued to use, just with varying winding):

I have reached out to you before to get official specs so I could crunch numbers. As you told me, the majority of your electromagnetic work is being done by a company and not by you (determining the windings, for example) which is why you can’t tell me details like how many turns of what gauge with how many strands. If I have those numbers, I can make some sense of it. But this is what I have gleaned.

Also, this is based of a max current of 30Amps. I know you push yours much higher, but I’m looking for an efficient board.

I do my best to represent the science here. Can we work together and move past the blame game and criticisms? We all just want something we can reliably ride and not have to wear insulating gloves (flesh meltingly hot hangars? I’ve been there using Jacob’s Hubs with a Seismic G5 truck as a heat sink. It hurts. @carl.1 #raptor2)

[Hummie]

No one has a .1mm airgap. You’ll be lucky to find .4mm. There are now N52 that can do much higher than in the link here that scorpion motors use. I’ve been using n45sh. Check here for common high heat stuff https://www.kjmagnetics.com/specs.asp

Iron losses will be much smaller than copper with small hub motors

As far as “electromagnetic work” it came down o the manufacturer trying to see what was the fatest single strand they could fit for the amount of turns. 30 for 90kv. Really they used the fatest they could do for 33 turns and used the same wire so the 33 turn 80kv has greater copper fill and less resistance

Not feeling attacked but making sure if you’re doing a sumulation with results it’s accurate

I think there is not much that can be done to optimize other than getting as much copper in and as strong a magnetic field as can be built with strong magnets, powerful back iron, and thin gap. The stator shape and size is staying.
I was interested in looking into different magnet shapes and sizes and strengths and temp ability but it’s a toss up and bigger isn’t always most efficient and seeming pretty insignificant

At the end of testing or simulations those motors I was selling were decided as too small and I don’t sell them anymore. I feel like simulations might tell finer details, possibly innacurately, and real world testing is easier and really the simple solution is just to get a bigger motor if you want greater power or efficiency. Until the iron losses balance with the copper the motor is smaller than it should be for efficiency. Likely every little motor we are using. Bigger solves all problems But I’m interested in what u can come up with and will get u real resistance and L numbers tonight

[anon94428844]

Jacobs are designed with a 0.1mm airgap. The result is that the 51.8mm diameter stator is surrounded by magnets with an ID of 52mm. I’ve measured mine with mistsuboyo calipers and they’re basically spot on with the drawings.

Yeah, but you’re cooking them. Jacob tried N42SH for the anti-corrosives. I cooked half of my magnets. Have you seen the scorpion factory in china? They hand wind all of their motors. They run them at insane temperatures and I don’t know if I would trust their numbers consistently. “Patent Pending” is always written on their sales pages, but I have yet to see any submitted applications.

You can use TWO wires to accomplish the same task as one. They should know that too. 12 Gauge wire is basically impossible to use in a stator because of how thick it is. I used 6 strands of 20 gauge wire to get 14 gauge equivalent. I would have used fewer strands with 18, but the 20 was cheaper!

Copper fill doesn’t always directly influence the motors performance is the iron is already saturated. Jacobs have less than a 10% fill and half the resistance of the designs of your motors that I simulated (the winding factor was basically the same though.

The results are accurate for Jacobs. Thats why I want your gauge, turns, and strand count. I got them for the most part. Again, I used a 30 Amp current because I know both motors can sustain that for 120 minutes max torque at max rpm.

Using two magnets per pole, for example, is a great way to increase torque without working too hard. I’ve done it with great success for small BLDC motors in the past.

Real world is great, but spending thousands on motor development for each production run seems like a bad idea. Better to do the math a bunch, build it a few times, then release it (Boosted really does win in the product design and testing department. Using PIC controllers for their motors is brilliant, professional, and cheap. We all need to learn from them to a degree.

Get me the Phase Resistance, Strand Diameter/Awg, Strand Count, and Turns and I’ll run it for you. If you could also send me a 2D cad of the stator just so that I can clean up the tooth thickness and what not that would improve the windings as well. Magnet widths/lengths help too (I know they’re 2.65mm thick, but idk how wide they are/the gaps between the magnets are.

[Hummie]

I can do that but it would be a simulation between two motors that aren’t available anymore. More interesting would be simulations of the the motors I’ll be getting with 4770 stators and enertions hub motors which are said to be the “most powerful” and I believe the same stator as the 6372 motor they sell. I’ve never seen a motor with a .1mm airgap for sale anywhere and scorpion and hacker will do at most a .4mm

[anon94428844]

Sure.

If you send me details on the new one I can tell you what to expect.

[Pedrodemio]

For anyone wanting to jump on the boat, here is trial version

http://www.infolytica.com/en/products/trial/motorsolve.aspx

[anon94428844]

I would instead suggest people use the online trial if they’re not willing to wait for simulation results. Their servers are faster than most peoples computers.

[Pedrodemio]

Just started my trial so i can play with it on the weekend, fantastic tool, but too expensive, i will start an academic study on research around hub motors and will try to contact then to see if i can get a good discount or something like that, with the current political and financial situation here no way i will get my university to pay for the license

@anon94428844 , can you give some guide lines on how to achieve a relative flat torque curve? i tried lots of combinations and none seen to be going on the direction o flatness, i saw some torque graphs from @Mellow updates on kickstarter and they have an almost flat curve until max rpm

Edit: After a lot of tests I accidentally discovered that using hysteresis current control the torque goes up significantly and becomes flat, so thats one of the secrets of a good permanece hub motor Based o what I read this is somewhat similar to FOC but no quite, so using VESC on the actual stage there’s no way to get the best performance out of a hub?

[Trillium]

Hi all, I’m never on here but I think I might be doing some beta tests through this forum in near future. I simmed the Trillium motor and its a bit of a lower performer torque-wise than hummie (30A limit, however) but it is in a 75mm wheel and has better thermal management. Glad to see people doing new stuff but at the same time frustrated! Would be nice to see some of my other competition before I waste any more effort with my motor development!

[evoheyax]

@Trillium

Would be nice to see some of my other competition before I waste any more effort with my motor development!

I felt the same way when I was a building my VESC iphone app. But hummies have been around since February or so of this year. Idk if I would say he’s that new…

[Trillium]

I was talking about kmeyerson

[anon94428844]

If you plan to use a VESC or similar ESC, or motors aren’t comparable.

I am using DTC, not SVM so my windings are different and yield different results. For the record, the transition from DTC to DTC-SVM is not easy. Simulink hasn’t been great help here.