Motor comparison spreadsheet

liveforphysics said:
crossbreak said:
This way it's no motor constant anymore. Lets call it drivetrain constant


It's a worthless misleading metric that would lead one to believe a pocket watch is the optimal drivetrain choice.

That said, enjoy life and have fun with math as much as you like. :)

oh luke, why again such undifferentiated argumentation?

Think we are in common that there are 5 fundamental constants that characterize a motor in an adequately simple model:
-weight [kg]
-heat dissipation [W/°C]
-Eddy current torque constant [Nm/rpm]
-Hysteresis torque constant [Nm]
-Copper loss constant [Nm²/W]

the first three were easy to find. the last one...not so much. For the drive train it would be:
-weight [kg]
-heat dissipation [W/°C]
-Eddy current drag constant [N/kph]
-Hysteresis drag constant [N]
-Copper loss constant [N²/W]

if you model a pocket watch as a drivetrain, you will find that it really sucks in 4 of those 5 disciplines. Even a 1:200 reduction cant conceal a tiny Km². For a drivetrain, drag and transmission loss shall be included in the drag and copper loss constants for simplicity/comparableness. A model for transmission loss has to be included, i'll start some literature research about a viable transmission loss model for belt, chain and gear transmissions. Anyway this discussion is not about motors themselves. I'd like to start a new thread, call it "Drive Train Comparison Spreadsheet"
 
bunya said:
I notice the no load current isn't included in Column AJ . For example, with the cyclone figures I uploaded no load is approximately 2.85A at 6000rpm (Sinusoidal) and at no load efficiency is 0%. Yet the table shows that for 6000rpm efficiency is ~40% at 1.92A, I'm guessing this must be 1.92+2.85A for 4.77A.
exactly. this is an error in the spreadsheet. thanks for that heads up!

Edit: Corrected playaround sheet
 

Attachments

  • play_around3.xls
    250 KB · Views: 149
Crossbreak, my old versionof Excel won't open your .XLS file. I want to see the motor details of this 2.9A no-load at 6krpm, or is it a small motor already pushed to it's limits, ie Rm sucks.
 
I asked myself a similar question about the high losses in no-load condition. I don't think Rm has much to do with it though, copper losses are less than 1W at 6000rpm whereas total losses are approx 150W. My guess is the IPM design with the embedded magnets results in more hysteresis losses (97w) when switching at 400Hz then most other designs. Eddy losses are also high 0.5mm laminations etc. I don't think the poor fill factor has much effect on no load current either but am not sure.
 
bunya said:
I asked myself a similar question about the high losses in no-load condition. I don't think Rm has much to do with it though, copper losses are less than 1W at 6000rpm whereas total losses are approx 150W. My guess is the IPM design with the embedded magnets results in more hysteresis losses (97w) when switching at 400Hz then most other designs. Eddy losses are also high 0.5mm laminations etc. I don't think the poor fill factor has much effect on no load current either but am not sure.

No-load current is how you measure the iron losses of a motor, which are rpm related. It's a good measure of them, because the copper losses are virtually nil at those current levels.

I wanted to look at Crossbreak's sheet, since I always assumed those Cyclone motors were just crappy cheap China motors. The good iron core info surprised me, so now I want to know the resistance. Unlike Crossbreak, I don't need a bunch of easy to pick apart formulas any more. Kv/Kt, Rm, no-load current info, dimensions, and weight, tell me everything I need to know about a motor other than heat dissipation.
 
Measured across two phases with the third open Rm is 0.060 Ohms, inductance is 154 uH if that's any use.
 
I have some pretty accurate data on the slightly bigger jm1 JOBY motor (40mm wide stator) that might be worth adding to the spreadsheet,
and for that matter maby gng type "1000w" big block, colossus, revolt 120 etc.

The JOBY has a kv of around 98, phase resistance (as seen by the controller) of 9.7 m ohms,
no load Watts are 93w for 1000 rpm 286w for 5000 rpm and 565w for 9000rpm, and these include about 40 to 100w in switching losses/phase current ripple, and (somewhat incorrectly) all tested at 88v.
driven it at 33khz pwm, theres 22 poles so at max rpm its 1650hz and the laminations are now down to .12mm, split magnets and they are wrapped in kevlar with around a 1mm air gap, it totals 2.7kg.

im genuinely impressed with how well put this argument is and gave some thought as to how it applies to an off road enduro race bike .

liveforphysics said:
Somewhere we may have a conceptual issue. Let's look at what gearing does.

Picture a statically loaded cog/cog interface. Trim away all the teeth that aren't engaged right now (typically just 1 tooth at a time is carrying torque). So it looks like a pair of lever arms interfacing with each other, whatever ratio between lever arm lengths happens to be is the amount of additional rotation needed by the shorter lever, and also the amount of force increased by the leverage on the longer lever. In a gear box with a bunch of stages, it's a bunch of lever arms pushing on lever arms, so we can make a bunch bunch of input motion on the lever become a tiny amount of motion on the output lever with proportionately increased force.

All I'm advocating, is rather than getting your 'gearing' through a bunch of external series lever arm arrangements (which each add mass that isn't copper or iron and hence not helping convert electricity into what you ultimately wanted, torque at the rear wheel over a speed range), to instead just use single stage lever geometry that permits achieving the same amount of leverage you wanted, but in a single stage that has no additional moving parts (failure modes) and weight.

So we need between 300 to 400NM at the tire to match the acceleration of most smoker enduro bikes and run between 8:1 and 12.5:1 gearing depending on the track with the above motor.

the magic pie headed us in the right direction over 5 years ago with a massive 265mm diameter 56 poles and only 18mm wide stator (they can do 120NM max saturated), so a double like falarfles would be about as good as good a china hub as any, but for now even the monster 10 to 15 kg hubs require over 10kw of waste heat to not really make the torque we need, ive tried a smaller rear (moto) wheel to get closer to the same newtons at the tread on same bike design and they seem impressive on their own but get cooked after a kw/hr or so and completely hosed by the joby that also uses about a third less power to do it (less than 1/5th the battery power needed to get the same off the line acceleration), so fine for commuting on smooth roads but fall well short in both acceleration, off road ability, and wh/km (efficiency) aand means we have to leave the battery behind to keep the same bike weight, seemingly not a win here.

Using an up-diametered joby (for 12 times the torque) about the diameter of the wheel would be cool, and about max out current high end and lightweight motor designs but there are inherent scaling problems involved.
-a bit like an insect strolling on the water but an elephant not so well. :wink: For this application the hub would ideally weigh under half a kilo and the motors weight be central to the bike, so even if it could be done to anywhere near the same 2.7 kg on the wheel i have gained little and it still triples the wheels unsprung mass..

to me it seems that we have found ourselves alive NOW to design our electric weapons, NOT at some point in the future -whilst its interesting to ponder which way things might go, its very difficult to design and ride machines with parts that are yet to, and may never exist.

Not much chop waiting if we can get the performance needed already, with over 3kw/hr onboard, more power and (real not just peak) efficiency and the total weight less than most that carry about 1kw/hr, we then have a platform to head in to battle with the smokers and cover heaps of off road miles and press on in the development gaining experience, understanding and good fun along the way.
Im pretty sure (unlike the big ICE manufacturers who will have us all old and grey waiting for the big breakthrough) most folks here are not prepared to wait, and want to get their high performance ev's builds happening now. :D

let me know if you need more info on the motor specs.
 
thanks for those specs. interesting motor for a single speed single reduction drive. A drive cycle that can feed a simulator for offroad and onroad application would really be good. Justin already has begun a drive cycle simulator and uploaded it to his page, thumbs up from here. Think we can then see that DD-hubbies are quite good for on road applications but lack of efficient high torque delivery (actually i do not need a simulator to tell you so...the figures posted yet are more than enough) Something we can archive in the near future if enough people contribute their free time. Next year will be challenging but it will be worth it... motor choices grow these times as well as drive train choices

@JohnInCR: Km² is 0.1 for the cyclone, hysterisis torque is 0.154Nm. All u need to know i guess?

@luke: do you have Km² and hyst torque figures for the Zero motor?
 
toolman2 said:
to me it seems that we have found ourselves alive NOW to design our electric weapons, NOT at some point in the future -whilst its interesting to ponder which way things might go, its very difficult to design and ride machines with parts that are yet to, and may never exist.
Quite.. Now you have experience of the JM1S and the JM1. How could they be improved on, for traction vehicles? As opposed to propeller driven ones.....
 
Miles I finally got the now load measurements at full voltage or close enough for the Nissan leaf motor.

7.1a at 444v for 3152.4 watts at 9090 rpm
It will spin faster with field weakening but I hit a rev limiter so I put it in the car and won't get a better no load number as I reinstalled the gear reduction and axles.
[youtube]ryCmrmIqqAI[/youtube]
 
Thanks for the data Arlo1 and toolman2

Any chance of another reading at half the voltage Arlo?

toolman, I've put in the basic stuff on the JM1 but not the no load values yet. Still working out how best to interpret them..

Anyway, I've uploaded a new version for now.
 
Miles said:
Thanks for the data Arlo1 and toolman2

Any chance of another reading at half the voltage Arlo? I've put in a second 'placeholder' value, for now.

toolman, I've put in the basic stuff on the JM1 but not the no load values yet. Still working out how best to interpret them..

Anyway, I've uploaded a new version for now.
Yes earlier in the thread I think I posted a test at ~ 110v and a test around 220v. If you can find it I will find my notes. :)
 
toolman2 said:
I have some pretty accurate data on the slightly bigger jm1 JOBY motor (40mm wide stator) that might be worth adding to the spreadsheet,
and for that matter maby gng type "1000w" big block, colossus, revolt 120 etc.

The JOBY has a kv of around 98, phase resistance (as seen by the controller) of 9.7 m ohms,
no load Watts are 93w for 1000 rpm 286w for 5000 rpm and 565w for 9000rpm, and these include about 40 to 100w in switching losses/phase current ripple, and (somewhat incorrectly) all tested at 88v.
driven it at 33khz pwm, theres 22 poles so at max rpm its 1650hz and the laminations are now down to .12mm, split magnets and they are wrapped in kevlar with around a 1mm air gap, it totals 2.7kg.

im genuinely impressed with how well put this argument is and gave some thought as to how it applies to an off road enduro race bike .

liveforphysics said:
Somewhere we may have a conceptual issue. Let's look at what gearing does.

Picture a statically loaded cog/cog interface. Trim away all the teeth that aren't engaged right now (typically just 1 tooth at a time is carrying torque). So it looks like a pair of lever arms interfacing with each other, whatever ratio between lever arm lengths happens to be is the amount of additional rotation needed by the shorter lever, and also the amount of force increased by the leverage on the longer lever. In a gear box with a bunch of stages, it's a bunch of lever arms pushing on lever arms, so we can make a bunch bunch of input motion on the lever become a tiny amount of motion on the output lever with proportionately increased force.

All I'm advocating, is rather than getting your 'gearing' through a bunch of external series lever arm arrangements (which each add mass that isn't copper or iron and hence not helping convert electricity into what you ultimately wanted, torque at the rear wheel over a speed range), to instead just use single stage lever geometry that permits achieving the same amount of leverage you wanted, but in a single stage that has no additional moving parts (failure modes) and weight.

So we need between 300 to 400NM at the tire to match the acceleration of most smoker enduro bikes and run between 8:1 and 12.5:1 gearing depending on the track with the above motor.

the magic pie headed us in the right direction over 5 years ago with a massive 265mm diameter 56 poles and only 18mm wide stator (they can do 120NM max saturated), so a double like falarfles would be about as good as good a china hub as any, but for now even the monster 10 to 15 kg hubs require over 10kw of waste heat to not really make the torque we need, ive tried a smaller rear (moto) wheel to get closer to the same newtons at the tread on same bike design and they seem impressive on their own but get cooked after a kw/hr or so and completely hosed by the joby that also uses about a third less power to do it (less than 1/5th the battery power needed to get the same off the line acceleration), so fine for commuting on smooth roads but fall well short in both acceleration, off road ability, and wh/km (efficiency) aand means we have to leave the battery behind to keep the same bike weight, seemingly not a win here.

Using an up-diametered joby (for 12 times the torque) about the diameter of the wheel would be cool, and about max out current high end and lightweight motor designs but there are inherent scaling problems involved.
-a bit like an insect strolling on the water but an elephant not so well. :wink: For this application the hub would ideally weigh under half a kilo and the motors weight be central to the bike, so even if it could be done to anywhere near the same 2.7 kg on the wheel i have gained little and it still triples the wheels unsprung mass..

to me it seems that we have found ourselves alive NOW to design our electric weapons, NOT at some point in the future -whilst its interesting to ponder which way things might go, its very difficult to design and ride machines with parts that are yet to, and may never exist.

Not much chop waiting if we can get the performance needed already, with over 3kw/hr onboard, more power and (real not just peak) efficiency and the total weight less than most that carry about 1kw/hr, we then have a platform to head in to battle with the smokers and cover heaps of off road miles and press on in the development gaining experience, understanding and good fun along the way.
Im pretty sure (unlike the big ICE manufacturers who will have us all old and grey waiting for the big breakthrough) most folks here are not prepared to wait, and want to get their high performance ev's builds happening now. :D

let me know if you need more info on the motor specs.


Thank you for your kind input Toolman! I appreciate your shared wisdom and experience my friend.

When it comes to what is available today, remember not so long ago this was a state-of-the-art electric motor design:

dadc9d0796dea7adff63eca99ac0566f.jpg



It functions, and it would have been possible to conclude further motor development had everyone simply accepted the state of design as 'good enough' at that time.

Lucky for us, there were some folks who weren't content to just work with what was functional and available, and instead pioneered new bold motor designs and worked through the associated engineering challenges. When folks stop being willing to do make this continued commitment towards maturing electromechanical drive topology, then EV power-train development will stagnate at each choice to settle for making whatever is available work.

ATB,
-Luke
 
Miles said:
Miles said:
crossbreak said:
+1 for the negative hysteresis torque constant
Yes, that is interesting :)
Please could you check your notes on the first test, Arlo? :) Thanks.
NL current can't be going up 3.7x for a doubling of rpm....
This is with different controller settings so maybe a bit of difference there but the same basic stuff just t different battery voltages.
Why can't it?

I am sorry but I now have a gear case with oil attached to it.

I think the amount of energy to turn the rotor at high speeds will increase drastically at higher rpm.
 
from the values given in this graph http://energy.gov/sites/prod/files/2014/03/f13/ape006_burress_2013_o.pdf no load current of this motor should be 6-7 amps at 9000rpm and around 4 amps at 4500rpm. The 1.9amps at 4400rpm seem to be very low.

file.php
 

Attachments

  • leaf_motor efficiency.png
    leaf_motor efficiency.png
    57.5 KB · Views: 941
Sorry guys. I will try to get batter test equipment.
Miles is the NM/Amp based on data I give you or from the very rough estimate I came up with from extrapolated data from all I could find online?
 
Arlo1 said:
Miles is the NM/Amp based on data I give you or from the very rough estimate I came up with from extrapolated data from all I could find online?
It's based on the Kv figure that you gave.

Arlo1 said:
Sorry guys. I will try to get batter test equipment.
No problem. Just another mystery.... If the no load current goes up by a factor greater than two, then the value for hysteresis torque becomes negative :) That's just because of the way it's derived, though. The only thing that I can think of that would cause that would be if there was an excessively high amount of windage losses, which seems unlikely....
 
It's a smooth rotor I would imagine at 9k pluss the the Eddy currents start to get pretty bad. But if anything it would be my current sensor which I used both times but its possible its not consistent.
 
Back
Top