Motor comparison spreadsheet

[crossbreak]

ok if i google transverse-flux this is another story

anyway... any ideas about the hyst/eddy loss thing?

yep, telsa does not use hub motors, they use single-speed reduction inrunner bldc drives. .....

Tesla uses Induction Motors, unless they have recently changed.

they have https://www.teslamotors.com/blog/induction-versus-dc-brushless-motors

ups it's induction. like all the others dont do with reason. i never would have thought that i ever drove an induction EV. ok i was so wrong as it was quite a lot of fun driving it ... so now: why not induction? just because it may be less power dense?

 

[Punx0r]

I just did a search for more information on the relationship between pole count and hysteresis losses and came across the following document:

Achieving High Electric Motor Efficiency by John Petro: http://www.novatorque.com/aboutus/white-papers/060_John_Petro_final_paper.pdf

It won't tell the experts much and it's biased towards axial flux motors, but I found it a helpful short summary of the various aspects affecting motor efficiency (core/copper losses, axial Vs. radial flux, inducton Vs. PM, surface Vs. interior magnets, winding shapes, thermal considerations etc).

Two quotes jumped as being unexpectedly relevant to this thread:

"Motor Design: Radial and Axial Torque Scaling": Axial motors:

Because the radius of the motor increases directly with motor diameter, and the force
increases with the square of motor diameter, torque of an axial motor increases with the cube of
diameter. Notice that since only the diameter is changing, the axial motor volume only increases by
the square of the diameter, even though the torque increases by the cube of diameter. This leads
to the fact that for larger motors, an axial motor can use less material to achieve the same torque,
compared to a radial motor. Please note that the above is a theoretical approach, and in actual
practice the length of an axial motor does need to increase somewhat with increasing diameters,
since the air gap dimension, strength of materials and realistic sized bearings must be considered.
This still leaves the axial motor design at an advantage over radial motor designs when compared on
the basis of torque generated per unit volume of active material required.

For a Radial flux motor torque increases with the square of the diameter, but for an axial flux it increases by the cube! The torque Vs. volume/amount of material also appears to increase non-linearly.

Also: "Motor Efficiency Improves with Increased Size":

Efficiency improves as radial motors get larger, because motor torque increases directly with
motor volume, while losses increase at a lower total rate. While iron losses scale directly with motor
volume, conduction losses proportionally decrease with larger motor sizes. This is the result of the
increased area having more flux carrying capability. High flux means that for the same operating
voltage, fewer turns of conductor are needed to achieve the same output torque. Given that the
larger size has also increased the area available for conductors, fewer turns of heavier wire can be
used, which reduces the motor resistance in two ways.

For an axial motor, the output torque increases faster than the motor volume increases, so
the iron losses are proportionally smaller as motor size increases. In addition, axial motors also
experience reduced conduction losses proportionally as motor size increases. This gives the axial
motor design an even better increase in efficiency with increased motor size.

 

[speedmd]

It seems to me that Tesla and others have gone with the induction motor is to get much higher RPM's and much better freewheeling. The little extra size and weight are no issue in such a big car.

Not so sure the switched reluctance types are just good for printers and servo motors. There is a whole range of hybrid designs that take advantage of both higher torque density than induction and larger rpm ranges - lower hys. losses than PM motors. I could see some sort of hybrid design coming about for a wheel motor.

 

[major]

Achieving High Electric Motor Efficiency by John Petro: http://www.novatorque.com/aboutus/white-papers/060_John_Petro_final_paper.pdf

.....it's biased towards axial flux motors,....
"Motor Design: Radial and Axial Torque Scaling": Axial motors:

This still leaves the axial motor design at an advantage over radial motor designs when compared on
the basis of torque generated per unit volume of active material required.

For a Radial flux motor torque increases with the square of the diameter, but for an axial flux it increases by the cube! The torque Vs. volume/amount of material also appears to increase non-linearly.

Mr. Petro overlooks the fact that in the axial flux motor, the air gap current density decreases with radius. This can be compensated by increasing the armature length or increasing the inner diameter of the armature (which increases inactive material). I think this negates his d-cubed vs d-squared torque ratio.

Achieving High Electric Motor Efficiency by John Petro: http://www.novatorque.com/aboutus/white-papers/060_John_Petro_final_paper.pdf

.....it's biased towards axial flux motors,....

Also: "Motor Efficiency Improves with Increased Size":

....fewer turns of heavier wire can be
used, which reduces the motor resistance in two ways.

Mr. Petro forgets that larger motors need longer wires.

 

[Arlo1]

Mr. Petro forgets that larger motors need longer wires.

Seeing as the slate is blank you can chose any magnet count/poll pair number you want and any stator tooth count/arrangement you want you can at least keep end turn losses to the same percent of the smaller motor but I think in reality you will have more copper and more stator iron and more magnets available to do work making the power rating higher....

 

[major]

Mr. Petro forgets that larger motors need longer wires.

Seeing as the slate is blank you can chose any magnet count/poll pair number you want and any stator tooth count/arrangement you want.

But more coils means longer wire total, for all the coils combined. Yeah, we can argue conductor size and turn count, but I'm speaking of, as Mr. Petro was, or should have been, two different size motors of similar design type, yielding the same torque. The bigger versus the smaller. I agree that the bigger will be more efficient, but not as much as Mr. Petro appears to infer.

 

[crossbreak]

+1, the major end turns of an axial flux motor are tangential on the outer diameter. which rises for axial machines just like with radial ones

the connections of the coil groups can take place in the inside of an axial stator, quite sexy

 

[major]

end windings that dont produce torque?

Of course. Their claim of no end turns is completely bogus. As is their claim of no back iron.

 

[Punx0r]

Interesting feedback, chaps. From a layman's perspective axial flux seemed to be quite popular for the more advanced brushed DC motors before the sea-change to brushless. Now the market seems to have returned to radial flux but brushless.

Despite the shortcomings of Petro's paper, is the tenet of larger motors having disproportionately greater efficiency and torque per unit active material still valid?

 

[liveforphysics]

Interesting feedback, chaps. From a layman's perspective axial flux seemed to be quite popular for the more advanced brushed DC motors before the sea-change to brushless. Now the market seems to have returned to radial flux but brushless.

Despite the shortcomings of Petro's paper, is the tenet of larger motors having disproportionately greater efficiency and torque per unit active material still valid?

Yes.

 

[Nathan]

Crossbreak= the cyclone motor simply stays behind it's possibilities. As it's an 8 pole inrunner, it can run a lot faster than the 3500rpm cyclone proposes.

What does this mean that the motor can run at higher voltages and higher RPM then 48v ? or is it limited by something else in the design ? who proposed 3500 rpm was the limit ?

 

[crossbreak]

cyclone sells this as an 48V motor and runs it through the cranks with a 44/14T chain reduction plus the 9.33:1 planetary. So i come to this 3500rpm . No idea if the insulation, bearings and gear lubrication are good enough for 7000 or 10000 rpm. You are happily invited to try at least some 72V or so.

From this perspective, you could run the cyclone through another 1:2 reduction directly to the wheel without going through the cranks. Iron loss should still be acceptable, form what I can read from the measured data. There is not so much iron in this tiny motor though. Still i would recommend the cooling fins they sell with this motor plus oil lubrication of the gears (stock is grease) if you wanna try this.

 

[Nathan]

cyclone sells this as an 48V motor and runs it through the cranks with a 44/14T chain reduction plus the 9.33:1 planetary. So i come to this 3500rpm . No idea if the insulation, bearings and gear lubrication are good enough for 7000 or 10000 rpm. You are happily invited to try at least some 72V or so.

From this perspective, you could run the cyclone through another 1:2 reduction directly to the wheel without going through the cranks. Iron loss should still be acceptable, form what I can read from the measured data. There is not so much iron in this tiny motor though. Still i would recommend the cooling fins they sell with this motor plus oil lubrication of the gears (stock is grease) if you wanna try this.

So 72v is ok with this motor being 8 pole inrunner and its 0.5mm laminations ? Whats better in terms of efficiency to run higher 72v voltage and low current 70a ? or lower voltage 48v and high current 100a ? As someone with this motor has run at 48v with 100 amps i.e 5kw power ! but i dont know where this motor current saturation point is ,, can that be worked out from the data sheet ?

also how did Miles work out the efficiency of this cyclone motor is 86% by measuring the no load current? ? doesnt that depend on what controller you use ? i.e if you use a grintech phase runner that is FOC sine wave wouldnt it be more efficient then a normal trapezoidal controller ?

Also doensnt the voltage you run it at alse effect the effciency ?i.e 48 ot 72v ?

 

[crossbreak]

yep, sure there is an influence of the voltage on efficiency. there are some rows on the right of the datasheet. i made a version for you to play around a bit. It has a gear reduction row and an additional current row (this is phase current, not abttery current). 48V/100A should be less efficient than 72V and 66.7A, since it is closer to the sweet spot. As a rule of thumb, you can layout your system in a way that phase current equals battery current near the sweat spot.
i think you would overload this tiny motor with so many Watts. 72V and 50A sound more realistic to me. Just enter it to the spreadsheet and have look by yourself. Note that this is not an "official" Spreadsheet, just for playing around with figures.

also how did Miles work out the efficiency of this cyclone motor is 86% by measuring the no load current? ? doesnt that depend on what controller you use ? i.e if you use a grintech phase runner that is FOC sine wave wouldnt it be more efficient then a normal trapezoidal controller ?

The calced efficiency is close to the efficiency a motor would get with FOC involved. With a simple trapezoidal one, it will be less. The formulars can be found a few pages back, and of-course in the spreadsheet itself

 

[Nathan]

Not sure how to use your sheet their is so many columns which one shows the voltage you are running the motor at and the gear reduction ? and is their 2 voltage and speed options to compare 48v and 72v ? some of the columns are not labelled.

So on column R i can see the cyclone motor efficiency peaks at 89.95% and peak power at this efficiency is 5980w, is that running 72v ? wow thats a lot of power from such a small motor! and the efficiency is very similar to the ASTRO motors of 3210 8t has 93% efficient but at a low lower power of 1219 W and the 3220 4t has 93% efficiency but at low peak power of 2830w is that true ?

But also if you factor in the ASTRO motors have a higher 169 KV they will probably be running higher RPM and hence you would need more reduction and maybe another reduction stage which adds weight and hence the efficiency looses would be worse compared to the cyclone 122 kv. As others have said this can be 5 to 10% losses per reduction stage, so when used as a mid drive or direct to the rear wheel with one stage reduction the cyclone motor may have overall higher efficiency possible than these 2 common Astro motors if you add the gear reduction losses ?

 

[crossbreak]

which one shows the voltage you are running the motor at and the gear reduction ?

gear ratio is column 3, it is only used to play around, not official. There is no row for voltage, you have to take care by yourself that the motor can reach the entered torque and entered rpm. If you want to play around/compare, then just copy and paste the row you wanna compare.

The astro has 9000 rpm "sweet spot" near to lower torque. This is simply due it's better core. The lower core loss leads to lower peak efficiency power, due the 50% / 50% copper core loss ratio. A motor with no core loss (and bearing loss of course) would have it's maximum efficiency at Zero watt's output, since all that matters is core loss, which is also almost zero near no load. It would have no no load current, since all no load current comes from core loss (and bearings).

So this does not mean that astro is not capable of much more power than the cyclone. If you compare cyclone and astro at 7000rpm and 5Nm, astro gets 92%, cyclone only 88%. The opposite is the case.

Yep serious power from such a small motor. But as luke objected, you will have a hard time getting rid of the heat. That is more than 500W @ 7000rpm @ 5.5Nm @ 4000W. It reality this will be even more, as the motor wont stay cool and copper gains resistance by 40% per 100°C. The smaller the motor, the more you have to keep an eye on it so it does not overheat. A cycle analyst + probe inside the motor is a must.

I had a look at some pics of the cyclone. If you have a look at those tiny gears, i highly doubt they stand 5 Nm at the rotor for long. 50Nm is already a bit much for the 14T freewheel this motor comes in stock, this is over 1200N which is ok for a normal bike chain. This all sounds like something that costs more time in the garage than you get to actually riding it.

 

[crossbreak]

made a voltage row, it's read only. simply input speed and torque you want into rows AG and AH

also added some "explanation" which rows are meant to be input or output (read only)

 

[liveforphysics]

yep, sure there is an influence of the voltage on efficiency. there are some rows on the right of the datasheet. i made a version for you to play around a bit. It has a gear reduction row and an additional current row (this is phase current, not abttery current). 48V/100A should be less efficient than 72V and 66.7A, since it is closer to the sweet spot. As a rule of thumb, you can layout your system in a way that phase current equals battery current near the sweat spot.
i think you would overload this tiny motor with so many Watts. 72V and 50A sound more realistic to me. Just enter it to the spreadsheet and have look by yourself. Note that this is not an "official" Spreadsheet, just for playing around with figures.

also how did Miles work out the efficiency of this cyclone motor is 86% by measuring the no load current? ? doesnt that depend on what controller you use ? i.e if you use a grintech phase runner that is FOC sine wave wouldnt it be more efficient then a normal trapezoidal controller ?

The calced efficiency is close to the efficiency a motor would get with FOC involved. With a simple trapezoidal one, it will be less. The formulars can be found a few pages back, and of-course in the spreadsheet itself

With a well designed controller, there is essentially no relationship between input voltage and motor efficiency.

 

[Nathan]

Cool thanks i will play with that updated sheet more So which column do the bearing losses come in ? is that part of the no load losses, i wounder how much is iron losses and what percentage is bearing losses at 7000 rpm.

 

[crossbreak]

you can't tell how much of the calculated hysteresis loss is bearing loss actually

 

[Arlo1]

Cool thanks i will play with that updated sheet more So which column do the bearing losses come in ? is that part of the no load losses, i wounder how much is iron losses and what percentage is bearing losses at 7000 rpm.

Your bearing losses in the typical motor are uber small like .01% or less of the losses. Your Eddy current increases as the square of the increase of the rpm.

 

[crossbreak]

maybe true for large hubbies that can do only few rpm. not so true for motors that do like 10krpm+

calced 12.24 mNmm for the 6001 bearing from the forumla given here http://www.skf.com/group/products/bearings-units-housings/ball-bearings/principles/friction/estimating-frictional-moment/index.html

This 8% of hysteresis loss or 2% of total loss at the sweet spot for the cyclone

 

[Punx0r]

If the inaccuracy is only 8% for a worst case (high speed) motor then I would be pleased, considering how simple the measurement method is!

 

[crossbreak]

that is no in accuracy in that manner, since bearing loss is included already in hysteresis torque you measure/predict from doing that no-load test

inaccuracy depends much more on your measuring setup

 

[crossbreak]

continued from here https://endless-sphere.com/forums/viewtopic.php?f=30&t=74526&start=25#p1125777

... nice job they stuffed so much copper into it. As much per Weight as a HS4080, but into lesser teeth

With the 2 motors being so drastically different in air gap radius, design, structure, efficiency, etc, how did you come up with that comparison? One is made as cheap ($0.02 halls, mystery metal covers, etc) and light as possible (without increasing cost), and the other overbuilt to the extent that kilos can be removed to get it down to it's basic form as a hubmotor, so copper/weight may very well be comparable. I don't understand how you are getting to the copper content to make the comparison.

specific Km² is the about the same, same copper loss per weight. While they are totally different as you objected, there are different ways to archive high Km² figures. Lower pole count does not automatically mean higher copper loss (due less space for copper). But this is what most people think. Or how do you explain why most hubbies have these 46 poles?

seems as: my favorite motor must be a half as big one, like a 20T24P 1/2-MidMonster to gain a Km² figure of about 4 times the MXUS3000 (that is quite good on core loss too here, still not s close on being equal though) using just a 2:1 reduction to gain that figure. This motor shall be running a 26" wheel but like running like a half-midmonster in a 13" wheel and very little reduction loss, almost none. 2:1 is very efficient for chain reduction, like ~97%+ and very little vibration when using a good chain.

Anyway i dont want more than 6 maybe 7kg of motor weight in my bike. I want headroom for battery.