More kV = More power? What are the trade-offs?

[liveforphysics]

If you want peak performance, you should always choose to run as high of voltage as your controller permits, then choose a wind that allows the motor to reach it's mechanical RPM ceiling.

For just making a nice quiet bike to ride, still go with as much voltage as possible, but pick a wind that keeps motor RPMs lower, even at high voltages. You will make some sacrifices in power in favor of making it not sound like a turbine engine.

 

[mud2005]

excellent, thanks Miles and Luke.

 

[MitchJi]

Hi Mud,

I was about to post the same question, thanks!

And thanks Miles and Luke for the replies!

ok more dumb questions... I'm looking at the 3210 chart http://www.astroflight.com/pdfs/3210WEB.pdf and each motor has a voltage listed next to it. I'm assuming thats the "best" voltage for a given winding to keep the motor in the most efficient rpm range which seems to be about 7500rpm according to the chart.
also listed is the "best amps" which is for 93% efficiency at 170 in oz of torque and stays above 90% up to 375 in oz.

So then what is the best way to find the proper motor for a build based on this?

It seems to me that the 8 turn at 48V would be the best for most builds considering the voltage limit on the controllers being about 50V.
opinions, comments??

I'll post the charts from 3210WEB.pdf which might help someone understand the question without downloading the pdf:


It seems like another reasonable option would be to choose a motor with more turns for a low kv (quiet and easier to gear) and get the increased performance by using a bigger motor (an 8150 instead of a 3210 for example). Of course the larger motor has a lower kv for the same number of turns which reduces the kv even further (5t 3210 is 271 vs 197 for the 5t 8150). So an 8150 instead of 3210 seems like a good deal because for one hundred dollars more you can have a combination of more power/torque plus quieter and easier to gear.

 

[liveforphysics]

Yep, that's good advise. The bigger the motor, the easier it is to setup a simple quiet drive. It's also a lot more forgiving in regards to handling high power bursts.

It looks like he lists a mechanical RPM limit of 12,000. This would mean the optimal performance wind for folks running 48v systems would be the 5 turn, which once you factor in voltage drop would be just about perfect for extracting maximum power from the 3210. This would mean something like a 2.5turn for the 3220 would be about perfect for a 48v system. Likewise, if you're going to fit sensors and run 100v, a 10turn would be perfect for a 3210, and a 5turn would be perfect for a 3220. This is only the correct wind for folks looking to get maximum power from the motor.

You may find much better luck at making a nice quiet simple single-stage drivetrain bike by choosing a lower KV wind, and sacrificing a bit of the potential power the motor is capable of pumping out.

 

[katou]

It seems like there are two ways to pick this, one based on efficiency, and one based on mechanics.

Method 1: choose winding based on operating voltage of controller and best amps figure

Method 2: choose winding based on kv at operating voltage of controller required to hit motor rpm maximum

Combining these is too much for my poor brain. For the 3% efficiency loss, I'm not sure it's worth doing. There will be much larger things important for the system, than efficiency of the system than the motor itself. Like noise. That's one that concerns me a bit, but hard to deal with until it's built. Or durability, or easy access for repairs, or weatherproofness, etc.

I chose my winding by method 2. This means that my motor will hit 12,000 rpm at full voltage of 52v. This means that at full voltage, it will be making max noise as well. I plan to deal with this problem by enclosing the motor as much as possible, which is also important for durability and weatherproofing.

Katou

 

[swbluto]

Here's a question that's hit me. If changing the kV does not affect the continuous torque limit (The continuous torque limit is held the same for different kvs), does it change the continuous power limit?

I haven't thought this through completely, but I have a feeling someone has already done the work and can share their insight. I have a feeling this might have already been covered in this thread and I missed it or forgot.

I have a feeling that if the continuous torque limit is held constant, then if the motor torque happens at a higher RPM with a different Kv, then that would imply the continuous output power limit has increased. But, does that actually happen or is there some other sublime physics that inevitably prevents that?

 

[Miles]

Here's a question that's hit me. If changing the kV does not affect the continuous torque limit (The continuous torque limit is held the same for different kvs), does it change the continuous power limit?

Yes, if the motor runs at a higher rpm...

 

[swbluto]

Sweet! I was looking back at some of the earlier graphs I have that compares two "identical motors"(Same copper fill, same size, etc.), one with double the kv of the other. I then drew a continuous torque limit to see how their continuous power limits might compare.




It appears that the minimum continuous RPM correspond to "full throttle"/load is increased, meaning that an increased motor RPM and the same torque limit means a higher continuous power! Excellent.

However, it's interesting to note the relationship. If the continuous torque limit is half the starting torque of the "original motor", then the continuous torque limit corresponds to half the no-load of the original motor, and 3/4th the no-load rpm of the doubled-kv motor. That means the continuous full-load motor RPM of the double kv is 3 times the amount of the original one, meaning, the continuous output power limit of the double kV would be 3 times the amount as , "roughly", power = torque * rpm. That's freaking sweet!

But, more generally, it appears that the minimum boost in the continuous power limit is (kv_2/kv_1) and, in real life, it'll be more than that.

So, this implies you can get more and more continuous power as long as you gear it correctly?! So, what's the practical downsides? I suppose higher kv would require taller and taller gearing ratios, which would require more gearing stages which would reduce efficiency.

I think this might explain why RC motor's kick hub motors so badly. The smaller RC motor would probably have a smaller copper fill than a hub motor, true, but its far higher kV increases its continuous power limit per "copper fill" by *at least* kv-of-rc-motor/kv-of-hub-motor which errs significantly in the RC motor's favor in terms of power.

 

[johnrobholmes]

All else equal, a different wind of motor won't be able to produce more power before saturation. A 10kv motor on 40v or a 20kv motor on 20v, same power. Increase the voltage on either motor and you get more power. Basically, if you can spin a motor faster you have the potential for more power. Increasing the KV is essentially the same as increasing the voltage, but you are trading amps for voltage to hit a certain power range.


In the real world there are benefits to low amp/ high voltage setups, and there are wind counts with the "sweet spot" for copper fill. But any time you increase the voltage or KV on a motor you can get more power if the batteries and controller will allow it.

 

[recumpence]

This is a great thread! I hope to have some real world data soon. My yellow trike has two 3220s wound 4 turn Delta. They should rev 14,000 to 15,000 RPM. My final drive ratio is about 13.5 to 1 from motor to wheel. I am really excited to see what kind of performance I see! The only negative is, the heat from my motors will be roughly 250% of what the 6 turn deltas I used to run. Then again, these new motors achieve their best efficiency near 14k rpm at 7kw.

I am expecting these higher KV motors to run like hyper-high performace race engines.

Anyway, we shall see soon enough.

Matt

 

[swbluto]

In the real world there are benefits to low amp/ high voltage setups, and there are wind counts with the "sweet spot" for copper fill. But any time you increase the voltage or KV on a motor you can get more power if the batteries and controller will allow it.

True, however, it seems "easier" to increase the kV by 10-20 times and to gear correctly than the voltage by 10-20 times. With luke running the HXTs at 100 volts, an equivalent power-matching voltage for a typical hub motor would be something like 900-1300 volts. I don't think anyone's been getting close to that.

I think the main limiting factor right now are controllers, as higher motor RPM requires a higher operating speed and modern controllers seem to appear to be at the "edge" of sensored controller's electrical RPM limits, and so higher kVs would be impractical. The ESC sensorless have pretty high RPMs as they were designed as such, but that kind of sensorless isn't as well suited as larger controllers with better heatsinking.

 

[johnrobholmes]

Comparing the speed of an outrunner to a hub motor is apples and oranges, since the diameter and width of the motors are vastly different.

Sensored commutation doesn't have speed limits built in, it is the shenzhen controllers that are slow. My R/C controllers that are sensored will commutate a 12 slot 14 mag outrunner past 15k RPM. Since they can be run as hybrids they can be switched into sensorless commutation at certain points, and sensorless allows for higher efficiency and timing advance as you rev up. This could allow a hub motor to have an effective KV of 20 (or whatever) at startup, and 25kv at full throttle.

 

[swbluto]

Comparing the speed of an outrunner to a hub motor is apples and oranges, since the diameter and width of the motors are vastly different.

Sensored commutation doesn't have speed limits built in, it is the shenzhen controllers that are slow. My R/C controllers that are sensored will commutate a 12 slot 14 mag outrunner past 15k RPM. Since they can be run as hybrids they can be switched into sensorless commutation at certain points, and sensorless allows for higher efficiency and timing advance as you rev up. This could allow a hub motor to have an effective KV of 20 (or whatever) at startup, and 25kv at full throttle.

What matters in the comparison is the copper fill factor. Taking that into account, hub motors and RC motors are comparable and they are no longer "apples to oranges"; they're both apples with a fill factor and a kV. The big HXT motors seem to have half the fill factor of hub motors, so with a kV of approximately 20 times higher, you'd expect a continuous power output rating of roughly 10 times higher, which is within the predicted range I suggested.

It's possible I underestimated the difference in fill factor, however. The relationship still holds, though, that RC motor > hub motor for continuous power output. This only really makes a difference if you're trying to climb steep long hills at higher speeds or you're climbing hills with heavy loads or you're trying to get the maximum "sustainable" top speed. Just as a maybe not too fair example, luke says his motors climb hills at 60 mph while not getting too hot; try getting close to that with a hub motor that doesn't weigh more than 4 times the amount.

And, yes, I was directly referring to current appropriate e-bike applicable controllers with adequate heatsinking. RC controllers, in general, just don't have the heatsinking needed for sustained high power output.

 

[johnrobholmes]

There are many more factors involved too, such as the lamination quality, magnet quality, and heatsinking ability. From my experience, the lamination quality on hub motors is about as good as tin cans. Magnet quality is relatively unknown as well, so the that is an area that would need to be addressed to really get apples to apples.

The copper fill is a great place to start however, and it is interesting that just accounting for copper fill and rpm gets a predictable relationship between the HXT and whatever hub motor you are using. We do know that geared hubmotors have a higher power for less copper, so that upholds the assumption as well.


I have a feeling we will see real EV controllers that will run bigass outrunners soon. It is just a matter of time before the market fills the light EV need.

 

[swbluto]

There are many more factors involved too, such as the lamination quality, magnet quality, and heatsinking ability. From my experience, the lamination quality on hub motors is about as good as tin cans. Magnet quality is relatively unknown as well, so the that is an area that would need to be addressed to really get apples to apples.

The copper fill is a great place to start however, and it is interesting that just accounting for copper fill and rpm gets a predictable relationship between the HXT and whatever hub motor you are using. We do know that geared hubmotors have a higher power for less copper, so that upholds the assumption as well.


I have a feeling we will see real EV controllers that will run bigass outrunners soon. It is just a matter of time before the market fills the light EV need.

Lamination and magnet quality has a lot to do with the no-load. As towards heatsinking ability, that does have a practical affect on the continuous torque limit, but it still seems outrunner RC motors have better heatsinking than hub motors(They have less "thermal resistance" anyways, not necessarily less thermal mass; less thermal resistance is good for sustained power output like a long steep hill or high continuous speed, whereas more thermal mass is good for shorter durations of higher power, like smaller hills.).

Anyway, I was thinking back to what the practical limit of the motor RPM would be (And thus the continuous power output limit, which can be apparently increased to infinity with increasing kv, assuming no other limiting factor) and remembered/read from an earlier post that the no-load current increases linearly with RPM. That is, the no-load power loss increases linearly with motor RPM which takes into account the "iron losses" and the "mechanical losses". And, so, eventually the motor is spinning so fast, that there's so much "iron losses"(Losses in the stator and magnets?) that it's getting far too hot to run continuously. However, does the thermal dissipation rise faster with an outrunner's motor RPM than would its no-load power loss, since a faster spinning can would be able to shed more heat? If so, I guess that wouldn't be an issue. However, I don't think an inrunner's thermal dissipation increases much with higher motor RPMs, so that could be an issue. I'm thinking the main limiting factor with the motor itself is probably mechanical limits - how fast it can actually spin without breaking apart. Apparently, AC motors in cars can't be spun "no load" because they'll break apart from spinning too fast. What are the mechanical limits of typical e-bike applicable RC motors?

So, for "open" outrunners, I'm thinking the mechanical RPM limit is the limit on continuous power output of a motor, assuming thermal dissipation increases faster with increasing motor RPM than does the no-load power loss. For inrunners and hub motors, it may be the iron losses or mechanical RPM limit, whichever comes first.

Anyone have a good idea what RC motors are mechanically capable of spinning at?

 

[Jeremy Harris]

Sensored commutation doesn't have speed limits built in, it is the shenzhen controllers that are slow.

Out of curiosity, I did a test this afternoon, albeit not a particularly challenging one. I ran a sensored outrunner (a 12 slot, 14 magnet one) at just under 7000 rpm on a 6 FET Xiechang controller. It seemed happy enough, but I didn't have enough volts at hand to push it further.

Has anyone actually established what the limit is for these controllers?

Jeremy

 

[swbluto]

Sensored commutation doesn't have speed limits built in, it is the shenzhen controllers that are slow.

Out of curiosity, I did a test this afternoon, albeit not a particularly challenging one. I ran a sensored outrunner (a 12 slot, 14 magnet one) at just under 7000 rpm on a 6 FET Xiechang controller. It seemed happy enough, but I didn't have enough volts at hand to push it further.

Has anyone actually established what the limit is for these controllers?

Jeremy

I don't know, but Justin Le had this datum:

The limit on the infineon controllers we sell now is pretty high. I'm not sure just where it is since I haven't hit it yet, but they can at least drive the high rpm (20") geared eZee hub at 60V, which is around 550 rpm * 5:1 gearing * 16 poles = 44,000 electrical RPM.

Crystalyte sensored analog controllers had no issues with high RPM, but their digital units and sensorless units (at least from 2008) would totally choke under this circumstance.

Justin

Which would imply a lower bound of at least 44,000electricRPM/12poles = 3666 RPM. That's actually fairly easy for an RC motor to obtain, so that doesn't tell you much. As far as I know, LFP mentioned that his 130 kV HXT motors run excellently at 100V with his infineon controller, so that would imply a 13,000 RPM minimum or 13,000*12 = 156,000 electrical RPM (Assuming his motor has 12 poles).

 

[swbluto]

Okay, I'm doing some simulating with my simulator and I seem to running into the same truth that I extracted from this thread and I'm trying to find a way around it, but it seems I'm not having such luck.

The basic problem is that I want to slow down the motor RPM to slow down chain speed to minimize the chain noise. But, in doing so, I noticed that to attain the same hill climbing performance resulted in greater motor heating with "the best gearing". Why is this?

So, I thought back to this thread. If we apply a voltage to two different kv versions of motor, though their currents will be different, for a given amount of torque output, their heating is the same - that is, roughly, torque/heat = constant.

Since power is torque * RPM, the higher kV motor will then having a greater output power for the same amount of heating, resulting in a more efficient motor. The unfortunate converse seems to be that a lower kV will result in higher motor heating to attain the same power output.

Maybe I'm totally off my rocker here (And I'm hoping I am), but here's a few graphs of the 130 kV HXT motor applied with 24 volts at the bottom. The two closeups show the efficient "end part" of the motor (where you'd want to gear towards).

Since it might be too inconvenient to click on, I'll just cite a few value pairs.

130 kV...output power: 700 watts heat: 80ish watts
60 kV...output power: 700 watts heat: 260ish watts

Of course, since the 130kV motor will be spinning faster, it will have a higher no-load current which isn't taken into account. But, at 24 volts, the 130 kV motor is spinning at around 3000 RPM instead of around 1400 RPM for the 60 kV version and the difference in load current will be relatively small. Likely less than 50 watts, and very likely to be less than 80 (That'd be more than 3 more amps at 24 volts). So, even taking that into account, it seems like the 130 kV motor still "wins" by a large margin.

So the final conclusion - Less noise at lower RPM with a lower kV motor = more motor heating at a given output power level (Geared optimally, of course, to actually achieve that power level efficiently.). Does anyone know a way around this?

 

[Miles]

So the final conclusion - Less noise at lower RPM with a lower kV motor = more motor heating at a given output power level (Geared optimally, of course, to actually achieve that power level efficiently.). Does anyone know a way around this?

Variable gearing or a larger motor. That's the choice.

 

[swbluto]

So the final conclusion - Less noise at lower RPM with a lower kV motor = more motor heating at a given output power level (Geared optimally, of course, to actually achieve that power level efficiently.). Does anyone know a way around this?

Variable gearing or a larger motor. That's the choice.

How does variable gearing help get less heat at a certain power output? Yeah, I understand how it can get you to run closer to the no-load speed, but doing so just causes the power output to decrease and ultimately you go slower (Although your motor survives the heat.).

 

[Miles]

How does variable gearing help get less heat at a certain power output?

It trades speed for torque without increasing motor heat. Of course, you go slower to get more torque

Seriously........for a given level of technology (magnet strength, core material etc.), torque to heat ratio is proportionate to stator volume. I don't see any way around that.

 

[liveforphysics]

When it comes down to actually making choices to setup a non-hub drive, I see it as a fairly simple choice.


What top speed do you wish to travel, and what wheel size are you going to use? This answers the question of what rear wheel RPM you need to target. Multiply this RPM by about 1.25 to give yourself some speed range overhead.

Look at what the maximum reduction you're able to fit in your desired build.
I hate lots of extra complexity, so for me, the answer is a single stage drive. Look at your mechanical limits of a single stage, generally about 14:1 is really pushing things for a single stage (at least with high-torque capable chain sizes), but 10:1 is often fairly simple (like a 110T rear sprocket, and a 11T front for example).

Once you know the reduction you're capable of getting, then multiply it by your first RPM number you calculated, and now you know the KV range you're looking for with your motors.

If you're doing a dual-stage reduction, something like 50:1 perhaps, then by all means, run a motor that has a KV in the 500-600+range, and enjoy the increased power density. If I had a 50:1 reduction on my bike, I could remove one of my motors, and re-wind the other for a much higher KV, and get the same performance.

But... the question I had to ask myself was if I could fit a second-stage reduction in less space and for less weight and less losses, and then deal with the decreased reliability, and additional noise, and twice the current burden on a single controller, rather than just adding another 3-4lbs motor to my bike.

Sometimes it makes more sense to crank up the motor KV and add additional reduction stages. Sometimes it makes more sense to add another motor.

 

[recumpence]

Luke is absolutely correct.

One other issue with high KV and very low gearing is throttle response. I do not have any math to back this up, but, I know my trike is a freakin insane beast running 15k RPM. It is nearly undriveable. So, I went from Delta to Wye and upped my gearing to reduce the insane hit I was getting (that and helping my poor ESCs to survive).

It all depends what you are looking for.

Matt

 

[liveforphysics]

Luke is absolutely correct.

One other issue with high KV and very low gearing is throttle response. I do not have any math to back this up, but, I know my trike is a freakin insane beast running 15k RPM. It is nearly undriveable. So, I went from Delta to Wye and upped my gearing to reduce the insane hit I was getting (that and helping my poor ESCs to survive).

It all depends what you are looking for.

Matt

Going from Delta to Wye should be much much more friendly to your poor controllers! I think you made a good call on that one.

 

[rolf_w]

Interesting thread and I hope I can get some valuable feedback to my question: we have started a Swiss e-moto dirt bike, enduro and super motard championship this year.

Most of the bikes are from quantya, some from zero. All teams use the same Lynch type motors (Lem) or Agni). These are axial field brushed dc motors ('pan cake'). Due to production differences the same motor type vary in the no-load rpm by +/-5% , thus Kv (or Ke=Kt=1/Kv) varies. From the construction principle it is clear that the number of Windings (N) and geometry is always identical, thus the magn. Field B varies.
All teams obviously want to win and are eager to have the fastest machine. Everyone concludes: go for the highest no-load rpm motor. Is this the best strategy? (considering the racing conditions where one of the main failure is an overheated motor or battery)