2019 Motor Simulator, Mid-drives and other updates

[justin_le]

Sharp eyes may have noticed that we added a 'mid-drive' option to the online motor simulator last month and have been incrementally flushing out the details. I alluded to this plan earlier here
https://endless-sphere.com/forums/viewtopic.php?f=2&t=89877&p=1386574#p1386574
and it's finally live now. If you've used the simulator recently on your browser you will probably need to clear the cache by pressing Cntrl+F5 to have the new model load. Then if you check the "mid-drive" checkbox then the following parameters show up in the drop down:




Efficiency Is just an estimate of the mechanical drivetrain efficiency from the motor to the wheel. We have it defaulting as 97%, which is reasonable for a chain drive, but if you are running multiple reduction stages it could be lower than this.

Gear Ratio This is a gear reduction ratio from the motor to the crankset. It is useful if you are using a hub motor as a mid-drive motor through a separate chain and sprocket (stokemonkey style). In the drop-down selector for the BBS02/BBSHD motors, the gear ratio is left as ONE since the gearing to the crankset is included in the motor model.

Front Chainring / Rear Cog These fields are just the teeth counts of the sprockets on the cranks and on the rear wheel. At the moment they are numeric entries requiring a keyboard but we'll be updating this to be a drop-down selector soon so that it's easier to select from touch screens or with a mouse.

Actually all the info about each parameter has now been included as tooltips if you hover your mouse over any setting, and that should help provide in context info


The net effect of using this mid-drive checkbox is just as if you had input a wheel diameter that was scaled like
Wheel Diameter = Actual Wheel Diameter * Front Chainring / Rear Cog / Gear Ratio
And the output torque supplied to the wheel is scaled similarly and also multiplied by the drivechain efficiency term.

The idea here is to simplify the process of seeing how mid-drive setups will behave in different gear combinations.

 

[flat tire]

Yeah I just saw this. Thanks for the motor simulator. Use it all the time.

 

[Chalo]

Efficiency Is just an estimate of the mechanical drivetrain efficiency from the motor to the wheel. We have it defaulting as 97%, which is reasonable for a chain drive, but if you are running multiple reduction stages it could be lower than this.

Efficiency is definitely lower for the 11t sprockets that are popular among outlaw e-bikers around here.

 

[justin_le]

One of the steps to make this new simulator feature useful was getting the two most popular after market mid-drive systems modeled, namely the Bafang BBS02 and BBSHD drives. At our shop we don't deal much with mid-drives for the reasons outlined here but we were able to receive some sample units from Bafang with our last hub motor orders.

I was expecting the BBSHD to be a much more powerful motor drive than the BBS02, because you know that's how it's all hyped and marketed, and it's physically much wider/heavier drive too. However, when I characterized the motors as 'black box' drive systems it turned out that their normalized performance was nearly identical!

This graph here shows the unloaded crank RPM as a function of the power supply voltage.


The BBSHD is a faster motor winding at about 3.27 RPM/V, compared to ~2.71 rpm/V for the BBS02.

Meanwhile, the winding resistance for the BBSHD was 0.086 ohms, while the BBS02 measured 0.131 ohms.

On a normalized basis, we would expect that winding resistance decreases with the square of kV, so if the BBS02 was rewound to have the same RPM/V as the BBSHD, then it's winding resistance would be:

0.131 * (2.71/3.27)^2 = 0.089 ohms

That is almost exactly the same as the BBSHD, even though it's a smaller/lighter motor. In general, if two motors have the same KV and the same winding resistance, then their performances and power outputs will be similar too. We would expect that if the BBSHD is a more powerful motor then it would have a correspondingly lower winding resistance.

If you plot the BBS02 and the BBSHD with the same controller, in the constant power region (controller current limit) the BBS02 actually has slightly higher power output than the BBSHD

https://www.ebikes.ca/tools/simulator.html?motor=MBBS02&mid=true&gear=1&tr=15&cont=C25&cont_b=C25&motor_b=MBBSHD&mid_b=true&gear_b=1&tr_b=15&bopen=true

If you use the KV Adjust slider to normalize the winding KV's, then the two motors are virtually indistinguishable from one another

https://www.ebikes.ca/tools/simulator.html?motor=MBBS02&mid=true&gear=1&tr=15&cont=C25&cont_b=C25&motor_b=MBBSHD&mid_b=true&gear_b=1&tr_b=15&bopen=true&kv=3.316

I was not expecting that result at all.

 

[justin_le]

One thing that you do notice from the normalized graph above is that the BBSHD has a slightly higher efficiency. This is because when we look at the motor core losses, the BBS02 has higher motor drag than the BBSHD at the the typical pedal cadence range.



They both have similar drag at low pedal RPM's, but on the BBS02 it increases more rapidly with RPM. A higher rate of core loss increase with RPM means that eddie current losses are more significant, which is normally the result of a motor with a higher pole pairs and electrical frequency.

Anyways all of these observations started to make more sense once I opened up the motors to look at their internal structures.
The BBS02

And the BBSHD


The BBS02 caught my eye since its inrunner motor was made using the exact same laminations and 16pcs magnet geometry as the G310/G311 hub motors that we're quite familiar with, only it's about twice as wide:


On the BBSHD motor, the stator and rotor are about 30% wider than on the BBS02, but rather than having 16 external magnets on the perimeter of the rotor, it's using just 8 embedded magnets inside the rotor itself, hence it having a lower eRPM and lower hysteresis losses in the motor core. This rotor design is mechanically more rugged since it's not relying on any adhesives to hold the mognets in place, but it also means a lower torque density for the motor. There is less total circumference for the magnets, the magnets themselves are thinner (just 2.5mm thick vs 3.5mm on the BBS02), and there are fewer pole pairs.






It seems all the potential power benefits of having the BBSHD motor with a thicker and heavier motor were mostly negated by switching to a rotor design that has less flux and fewer poles.

 

[j bjork]

Thanks for the great work you are doing :thumb:

 

[neptronix]

Super interesting stuff. Thanks for continuing to make my favorite tool even better.

The stuff with the BBS02/HD is really interesting. There are a lot of design 'bugfixes' that bafang is leaving on the table with that drive and that is rather sad.

 

[drbenjamin]

Justin

Just want to take the opportunity to thank you for all you've contributed to the ebike community. The motor simulator (and the trip simulator) were invaluable to me in selecting components for my 1st ebike. As well as this forum!

Thanks so much -

 

[justin_le]

The stuff with the BBS02/HD is really interesting.

It really is. Of course the other part of the motor performance is the thermal side of things. If you have two motors with identical KV and winding resistance then their off-the-line performances are mostly indistinguishable, but if one motor has better thermal conductivity and/or a higher thermal mass will be able to run at higher power levels for longer before overheating.

Even though the BBSxx motors have a built in thermistor, this temperature is on the hall PCB rather than actually being in contact with the stator windings, so for measurement consistency I glued in a fresh NTC temp sensor directly on the windings of both the BBS02 and BBSHD and hooked them up in the wind tunnel like so:



This probably exposes it to higher air flow than you would get on a bicycle cruising at a corresponding speed where there would be wind shadow from the front wheel, cranks and rider legs etc but at least it's a consistent and reproducible. I set the test parameters for an 80oC constant stator winding temperature and stepped through from 40-120 rpm on the crank and 7-45 kph wind speeds



For the motor shell temperature, I had an IR sensor facing the left side plate of the BBS (Labelled Shell O), the edge of the BBS (Shell R), and the right side where the controller would normally mount (Shell I). You can see that the left side plate where the motor stator is pressed in was much hotter than the rest of the enclosure, and so this is the only temperature sensor that I used for the thermal modeling as a representative shell temperature.

A best fit model for the thermal conductivity parameters is shown here, and you can see that the resulting Calculated Stator and Calculated Shell temperatures match very closely in their time evolution with the measured stator and shell temps over the entire experiment.



Ran the same test on the BBSHD, extrapolated the net thermal conductivity terms at each steady state datapoint and the results are as we would expect:


The larger BBSHD drive has more surface area for shedding heat, and a larger contact patch between the motor stator and shell (being 48mm long rather than 29mm long) for conducting heat internally to the BBS housing, and so it has somewhat better thermal conductivity. I'm not quite sure what cased the blip at the 15 and 20 kph points on the BBSHD and will need to look into the cause of that, but overall you see like a ~15% improvement in thermal conduction with the BBSHD.

The other factor is the thermal mass. The 48mm deep BBSHD stator weighs around 1.8kg with the shell, while the smaller BBS02 stator is 29mm deep and weighs in at 1.13kg. That's 60% more thermal mass to heat up and hence 60% more time for for the BBSHD to overheat for a given amount of watts being dumped into it.

 

[justin_le]

As a result of this, (higher mass and larger size for shedding heat), once I added the thermal model to the simulator properties then we do see a difference in what the BBS02 and BBSHD drives can cope with. For instance, climb a 9% grade hill with the following gearing until steady state and the BBSHD sits about 23 degrees cooler than the BBS02 motor

https://www.ebikes.ca/tools/simulator.html?bopen=true&mid=true&motor=MBBSHD&gear=1&motor_b=MBBS02&mid_b=true&gear_b=1&tr_b=15&tr=15&grade_b=9&grade=9


Or, by similar token load the motors even more so that they overheat and you'll see that the BBS02 will overheat faster than the BBSHD all else being equal:

https://www.ebikes.ca/tools/simulator.html?bopen=true&mid=true&motor=MBBSHD&gear=1&motor_b=MBBS02&mid_b=true&gear_b=1&tr_b=15&tr=15&grade_b=12&grade=12


But what's interesting is you could have all of this exact same improvement by simply starting with the BBS02 motor and then clamping on some additional passive heatsinks to achieve better thermal conduction and heat capacity! The additional size and mass of the BBSHD motor is not resulting in any improved motor performance, what gains they could have had from the larger size seem to be lost because of the lower pole topology and a lower flux internal magnet design. Instead the only benefit comes from the additional metal and surface area which is silly.

If instead Bafang had kept the same rotor/stator design of the BBS02 and made it 48mm deep instead of 29mm deep for their HD model, then there would be an actually substantial performance improvement, with way better loaded efficiency and torque producing capabilities. As it is now, in terms of torque to weight ratio and power to weight ratio, the BBS02 is hands down a much better motor than the BBSHD.

 

[Deafcat]

This is super cool, and confirms some performance implications that others have found with these BBS motors, Roshan at Biktrix has been telling me for years that the BBS02 really is the better motor all around, especially considering the lower stock weight and cost... Seeing this all summed up with good testing really makes sense.

Kudos for adding these (and mid-drive calculation) to the simulator, I really dig it! I'd love to see how the G510 (Ultra) and the newer M600 perform on your test bench when you get around to them, I'm a big fan of the Ultra motor and its quality construction. Unfortunately it can't be easily retrofitted to a bike but it's certainly a solid motor and reduction assembly, and it has put up no complaints with me pushing it to nearly 60 km/h speeds.

 

[justin_le]

This is super cool, and confirms some performance implications that others have found with these BBS motors, Roshan at Biktrix has been telling me for years that the BBS02 really is the better motor all around, especially considering the lower stock weight and cost... Seeing this all summed up with good testing really makes sense.

Hey sweet and thanks for confirming that with some field experience from someone who's been dealing with these motors in a significant capacity. That's more or less what I was wanting to hear by posting these analytic test results, as these aren't a motor system that we deal with first hand and I was curious if people's experience would validate what is indicated by these lab results.

Looking at the motors themselves and these test data there's no way at all I'd pick a BBSHD over the BBS02. It's heavier and bulkier for no meaningful extra performance, and has an extra wide and uncomfortable Q factor on the cranks in order to clear the motor.

I've got other priorities now but I'm tempted to experiment to see what lightweight cooling mods and bolt on heatsink fins could be done to a BBS02 in order to make it meet or even exceed the thermal characteristics of a BBSHD. Here's what the breakdown in conductivity from stator core to shell and from shell to ambient currently looks like:


I imagine that with some thermally conductive potting compound or grease to better link the stator laminations to the casing, and with large surface area fins clamped onto the shell, you could make a BBS02 outperform the BBSHD with a much lower total weight. Possibly even at lower total cost too.

 

[keithd]

really, really great info

Thanks for your continuing commitment

 

[justin_le]

Kudos for adding these (and mid-drive calculation) to the simulator, I really dig it! I'd love to see how the G510 (Ultra) and the newer M600 perform on your test bench when you get around to them,

Well unless one just shows up on our doorstep don't hold your hopes for that yet (the simulator focus of course being on items that are DIY and aftermarket accessible)

But in the meantime, there are dozens and dozens of other new motors that I've tested and added to the simulator database over the past few years, and that's not been documented anywhere to give an explanation on what's what. So the next priority we'll be working on is including a photo and description of each motor in the dropdown list so that you can have context of what specific motor is being modeled, like this:



The idea will make easier for people to identify suitable motor substitutes when trying to figure out the expected performance of a hub motor that isn't on the simulator. You'd scroll the list to find one that is similar in size and weight and total gear ratio (if geared) and then use the KV adjustment to match the unloaded RPM.

In the meantime though I'll post some details of new motors on this thread (and some that have been added in the past) to explain what is what...

 

[DogDipstick]

This is awesome.

 

[neptronix]

Very nice!

 

[lantice13]

Subscribed. makes me want to donate one of my mid drive that is in mothballs for the sim. you know one of those popular GNG gen 1.

 

[qwerkus]

It really is. Of course the other part of the motor performance is the thermal side of things. If you have two motors with identical KV and winding resistance then their off-the-line performances are mostly indistinguishable, but if one motor has better thermal conductivity and/or a higher thermal mass will be able to run at higher power levels for longer before overheating.

Fascinating take. It seems it takes a hub motor apostle to finally have a neutral analysis of the 2 most popular mid drive conversion kits.
So the next question will probably be: do you plan on carrying a bbs02 with a custom "Grin tec" cooler that actually beats the HD ?

EDIT: On a sidenote, it seems bafang did exactly the same with the g311 and the g360 hub motors. First one with a small rotor and more poles, second one with a bigger rotor an less poles. Logically the g360 should not perform much better than the g311, except for thermal management.

If I may, I'd suggest similar analysis of the tsdz2. It s become quite popular especially with the open source firmware that has been born in this forum and one of the recurring question we see is how it fares against the bbs02.

 

[justin_le]

One of the motors that I was sooo anxious to get my hands on for characterizing was the G360 motor from Bafang. I've made no secret of my love for the smaller G310/311 motor series with the inrunner design and helical cut gears with a high ~11:1 reduction ratio, and the G360 motor promised to be a more powerful version of this same idea.

When the sample finally arrived a few weeks ago I was a bit surprised at the weight. It was a little over 4.0kg, which is consistent with Bafang's website at the moment but from my recollection on ordering it late last year I thought it was gonna be more like 3.5kg.

Anyways, it still had all the features that I was hoping for as a stepped up version of the G310, easily disassembled and running with a larger rotor and larger gears:



The main difference as you see is that the rotor uses embedded magnets rather than surface glued magnets like on the G310, which seemed like a great design upgrade given that part of the problem we've had with the G31x's has been magnets coming loose at high RPMs.

The actual rotor is the same ~15mm thickness, but it's a larger 80mm diameter vs 59mm for the G310.


Installed in the stator, it looks like this allows Bafang to have a smaller and more controlled airgap too, although I haven't quite quantified that.



The total gear stack is 12 tooth on the stator, planets with 42/18, and 65 teeth on the ring which provides the
(42/12)*(65/18) = 12.63 : 1 gearing.

 

[justin_le]

Qwerkus, you jumped the gun on me!

Anyways, onto the G360 motor performance. What I got was an 11T winding, which had a winding resistance of 79mOhm and a kV of 9rpm/V.

Core Loss Data

The no load core drag varies from 2-2.25 Nm at ebiking speeds, and doesn't increase too much with RPM which we can attribute to the lower pole rotor and hence less of an effect of hysteresis losses (similar to the BBSHD vs BBS02).



Here is the G310 by comparison. You can see it's about half the core losses of the G360, so we'd better hope that the G360 is twice as powerful to score the same 'performance metric'.



Dyno Test Data

And here are the dyno test results running with a 19A trapezoidal drive motor controller at 36V. You can see the controller phase current limit kicking in below 130 rpm which is why the graphs diverge more there.


And at 48V


Thermal Test Data

The thermal testing was done the same method as usual, running the core to 80oC with field weakening with IR temperature sensors on the motor shell and a 10K thermistor glued directly in contact with the stator windings.



We measured 0.8-1.4 Watts/Degree from the motor core to the motor shell, and 3 to 7 W/K from the shell to ambient air with increasing air flow. So as is usual with the hub motors most of the thermal bottleneck is getting from the stator itself to the motor casing:


And here is the total net thermal conductivity (stator to ambient air), compared with the G310.



The G360 is a fair bit better as one would expect from a correspondingly larger motor with more surface area move heat.

 

[justin_le]

With all this empirical test data and a nicely fit model we could get to the real point which is to compare the this new G360 to other motors in the same power class, and this is where things got interesting and a little bit disappointing.

In order to facilitate doing those side by side comparisons, we've changed the nature of the KV Adjustment slider in the simulator so that instead of it being an normalized adjustment ratio, it now displays and lets you type in the kV directly. So you can just type in the kV of motor A into motor B's field and then get a comparison graph that shows the results if they were wound to the same speed



So that's what we've done here, where you can see how the G360 stacks up side by side with a MAC motor:
https://www.ebikes.ca/tools/simulator.html?motor=MG360_11T&hp=0&grade=0&motor_b=GMAC10T&hp_b=0&grade_b=0&bopen=true&cont_b=C25&cont=C25&kv_b=9.182

The Bafang G360 is just over 4.0kg, while the similar sized eZee and MAC motors are already a bit lighter at 3.8kg. That would be OK if the G360 had higher sustainable power and torque capabilities, but when you do a normalized comparison between the two, you see that across the board a MAC motor has slightly better efficiency, lower heating, and higher output torque.



And if you look just at the thermal data, lets take these motors up a 10% grade hill with no pedaling, the G360 will get to 150oC in 10 minutes, while the slightly lighter weight MAC motor takes 17 minutes and would climb with better mileage too (43.4 wh/km vs 45.0 wh/km)
https://www.ebikes.ca/tools/simulator.html?motor=MG360_11T&hp=0&grade=10&motor_b=GMAC10T&hp_b=0&grade_b=10&bopen=true&cont_b=C25&cont=C25&kv_b=9.182



Our own conclusion is that mechanically the eZee, MAC, and G360 motors are all great, perfect for fitting in 135mm dropouts, using cassette freehubs, and having proper disk rotor compatibility. From a performance and weight side of things, the eZee and MAC motors are slightly better, even though the G360 seems at first glace to have the best engineered inside. The only thing that the G360 has going for it in this comparison is the potential for the smoother/quieter operation due to the helical cut gear in the first stage of the gear reduction.

Before doing these tests and looking hard at the results I was just about ready to jump in whole hog and make a huge commitment to offering kits based around the G360, but now I'm having 2nd thoughts if it really makes sense. It's an absolutely fine motor, but given that we already have eZee and MAC in this power/performance category it doesn't fill a new application space.

 

[qwerkus]

Before doing these tests and looking hard at the results I was just about ready to jump in whole hog and make a huge commitment to offering kits based around the G360, but now I'm having 2nd thoughts if it really makes sense. It's an absolutely fine motor, but given that we already have eZee and MAC in this power/performance category it doesn't fill a new application space.

Makes sense, though the G360 has 2 strong arguments on its side:
- easy oil cooling mod. Given the o-ring sealing, the cable potting and the use of an inrunner. Would be interesting to see the same analysis with 50ml ATF.
- price: the Mac has a retail price in china between 210 and 300USD, while the G360 only cost 130 to 200USD. Also, the high poles count in the mac requires expensive controller, while the G360 can work with lishui or even open source KT controller. So even cheaper.

Not sure if this justifies a new kit, but worth a thought.

 

[justin_le]

While bafang is mostly known for geared motors, they do have a few direct drive models for scooter wheels and I got my hands on a D410 in an 8" wheel.

https://www.bafang-e.com/en/oem-area/components/component/motor/fm-d410250d/

Which is this one in the dropdown menu




This is a pretty standard DD hub motor construction inside, 30 magnets, 105mm stator core 30mm thick with 27 teeth. Winding was 0.21 ohms with a kV of 22 rpm/V, and the thermal conductivity data from core to shell and shell to ambient looked like this:


Given the massive uptick of electric kick scooters especially with the scooter share fad from Lime and the likes I thought this would be an interesting motor to have modeled on there, and it's one that we might stock too because there are some pretty fun DIY projects you can do with small cast hub wheels of this size.

I plan to reseal the side plates and so another test on it with Statorade added too to see how much of an effect that might have for those wanting to do scooter wheel mod upgrades. I'm not expecting as much improvement as on a normal hub motor since the rubber wheel is cast right around the rotor ring and would block the most effective path for heat transfer, forcing most of the heat shedding from the side plates which doesn't have as much of a thermal bridge to the stator. Still should be interesting to see.

There's also another DD scooter motor on the simulator which I put up a few years ago and that's the one labelled XOFO 6" in the dropdown menu. That motor is the same as what we saw on most of the hoverboard wheels from a few years ago, but it hasn't had thermal testing yet so you won't see the overheating details on the simulator.

https://www.ebikes.ca/tools/simulator.html?motor=MD410&wheel=8i&bopen=true&motor_b=MXOFO6&wheel_b=6i

When you choose a scooter motor like this, be sure to manually set a custom wheel size of 6" or 8" since those fields aren't populated automatically and the 26" default will be way off.

 

[Deafcat]

Kudos for adding these (and mid-drive calculation) to the simulator, I really dig it! I'd love to see how the G510 (Ultra) and the newer M600 perform on your test bench when you get around to them,

Well unless one just shows up on our doorstep don't hold your hopes for that yet (the simulator focus of course being on items that are DIY and aftermarket accessible)

Well, I'm fairly certain that can be arranged... Particularly with the G510 Ultra motor since it's become a very popular choice for the high-power crowd. Actually I'm considering buying a second one of these motors myself for R&D with two aftermarket controls in particular (EXESS out of Germany, and of course Phaserunner/Baserunner). I'll check with Roshan, he may be willing to contribute one for SCIENCE.

 

[lionman]

Looks like someone needs to develop a custom rotor for the BBSHD...