BLDC Efficiency at different voltages

[John in CR]

I'm looking at a lot of different Chinese made brushless hub motors and in trying to compare the supplied test results, it's often difficult because the tests are run at different voltages. I know a lot of us run our brushless hubbies at much higher than the design voltage, and I'd like to extrapolate these different tests to a common voltage of 60v or 72v, but I wonder if that approach is valid for comparison purposes. Does the efficiency of these motors change much with increased voltage?

John

 

[John in CR]

If you have say a 72V setup for short haul performance, but occasionally need maximum distance for which the speed at 36V is sufficient, is the efficiency the same at the lower speed using 72V at part throttle vs 36V at full throttle? Does the motor respond differently to full width pulses of 36V vs narrower pulses of 72V? Do the losses in the controller change in this scenario?

Does a motor's maximum efficiency change with voltage? eg a 48V brushless motor is tested to have an 85% maximum efficiency at 48V under X load. Is it likely to have the same peak efficiency with the right loads at both 36V and 60V albeit at different loads?

John

 

[billvon]

Controller losses at a given power output are _slightly_ higher at 72 volts than at 36 volts, because the pulse width for a given power has to be considerably narrower, and thus peak switch currents are higher. However, that's a pretty minor effect.

 

[John in CR]

So the minor difference in controller losses is the only difference in efficiency at two different voltages but equal power input???

John

 

[swbluto]

At a given power input, the higher voltage would have less current going in. Since this would decrease the energy loss for the battery and controller's resistance due to the lower current flow(P = I^2*R), the system would be overall more efficient at a given power output assuming every other effect is negligible. Since this would indicate a greater power output, you'd be going at a greater speed which also has the effect increasing cogging torque, so this would pose an efficiency reducing effect.

I should just bring up the formulas so everyone can eyeball it and spot things I'm not thinking of.

Now that I'm looking at the formulas, I'm realizing it's not as simple as looking at one formula. The existence of two separate regions(current-limited and throttle-limited) impairs the discussion a bit as there'd be three separate cases at a given speed with everything else equal:

where v2>v1(both different voltages)
1)
v2-current
v1-current
2)
v1-throttle
v2-current
3)
v1-throttle
v2-throttle

For the sake of discussion, the controller limited region suppresses efficiency. You'll notice the "bump" in the efficiency graphs at ebikes.ca's simulator(and my vehicle simulator) when passing from the controller limited region to the throttle limited region.

Here's the code for the simulator's core which contains the fundamental equations if anyone cares enough to deduce the needingly complex answer to this question.

http://svn2.assembla.com/svn/ebikecalculator2/EbikeCalc2/src/ebikecalc2/Motor.java

 

[ZapPat]

At a given power input, the higher voltage would have less current going in. Since this would decrease the energy loss for the battery and controller's resistance due to the lower current flow(P = I^2*R), the system would be overall more efficient at a given power output assuming every other effect is negligible. Since this would indicate a greater power output, you'd be going at a greater speed which also has the effect increasing cogging torque, so this would pose an efficiency reducing effect.

I should just bring up the formulas so everyone can eyeball it and spot things I'm not thinking of.

Now that I'm looking at the formulas, I'm realizing it's not as simple as looking at one formula. The existence of two separate regions(current-limited and throttle-limited) impairs the discussion a bit as there'd be three separate cases at a given speed with everything else equal:

where v2>v1(both different voltages)
1)
v2-current
v1-current
2)
v1-throttle
v2-current
3)
v1-throttle
v2-throttle

For the sake of discussion, the controller limited region suppresses efficiency. You'll notice the "bump" in the efficiency graphs at ebikes.ca's simulator(and my vehicle simulator) when passing from the controller limited region to the throttle limited region.

Here's the code for the simulator's core which contains the fundamental equations if anyone cares enough to deduce the needingly complex answer to this question.

http://svn2.assembla.com/svn/ebikecalculator2/EbikeCalc2/src/ebikecalc2/Motor.java

Hi swbluto - Your I^2*R losses are only the controller's conduction losses, but you need to add in the switching losses too. Conduction losses are proportionnal to duty cycle, current and Rds in a given controller, but switching losses are not directly affected by duty cycle, and are instead proportionnal to voltage * current * switching time * switching frequency. When using low resistance FETs in a controller, switching losses can become a large part of the controller's losses, larger even than conduction losses it seems to me from the math.

I am going to be probing a controller or two I have here to actually mesure switching times of typical ebike controllers (when I have time), because this would be very important to know for calculating the overall efficiency of these controllers.

 

[swbluto]

Hmmmmm... true. I know that's one of the main determinants of efficiency in SMPS and I think I remember quotes of 75%-98% efficiency from those, with higher efficiency at lower, audible frequencies(Like the high-pitched squeal I hear from the controller.). So might it be crudely guessed the switching losses impact it by ~5-10%?

(Wow, that actually seems fairly high.)

 

[John in CR]

All the motor performance curves I've seen have fixed voltage, but don't our controllers make the motors act like they're getting different voltages based on the throttle position? I ask because I have a marine application where I will need to run at 2 different speeds 95% of the time. I'll need WOT for maximum speed for about an hour out and back, with 6-8hrs at very slow cruise using 1.5-2kw. I'd like to do both efficiently, so I'm wondering how best to set it up.

The high power runs out and back will be relatively easy to predict performance, but that will use generator power, not the batteries. I don't have a clue regarding predicting the low speed efficiency, which will be important for determining the battery requirements. Due to the parallel connection with the generators, I'd like to avoid some kind of parallel/series switching of the battery pack to lower the voltage for efficient slow speed cruising, but doesn't the controller handle that for me via the throttle?

John

 

[swbluto]

but doesn't the controller handle that for me via the throttle?

John

The controller does lower the output voltage, but it's a basically a buck converter and a buck converter isn't 100% efficient. Since the typical BLDC controller is a synchronous buck converter, you might expect conversion efficiency anywhere in the 80-93% range depending on how much voltage reduction it undergoes. More voltage reduction = more inefficiency. If you were to able to change the voltage by rearranging the series/parallel connection, by comparison, you'd be looking at near "100% efficiency" save the losses you'd incur when you switched.

This inefficiency mainly comes from body diode losses and switching losses.

The peak efficiency of the motor itself seems to decrease as the motor voltage decreases, but this applies to both controllers and series/parallel arrangements so no real advantage here (Maybe. I'd have to think that through).

 

[ZapPat]

If you are demanding lots of torque at low speeds your motor will run pretty inefficiently, and it's losses will outweigh the controller's own losses in this situation.

Since most ebike controllers *don't* operate as a synchronous buck converter usually, your controller will heat up quite a bit when outputing a low voltage to the motor if you are demanding much torque(amps) from it. This is because the freewheel diodes (ie: the low side FETs working passively) will be conducting most of the time, and not doing a good job of it either.

Your motor must be really big if you're going to demand ~2kW from it at low speeds, or you have really good cooling. You might want to consider a chain drive or some other speed reduction setup if you want really low RPM's at those power levels (just how slow anyways?).

Pat

 

[Miles]

Also, how significant are the effects of PWM on losses within the motor? Eddy currents, hysteresis and copper losses....

 

[swbluto]

Since most ebike controllers *don't* operate as a synchronous buck converter usually, your controller will heat up quite a bit when outputing a low voltage to the motor if you are demanding much torque(amps) from it. This is because the freewheel diodes (ie: the low side FETs working passively) will be conducting most of the time, and not doing a good job of it either.

I don't think he'll be using a non-synchronous bike controller for a 10 kW motor. I get the feeling that the controller couldn't handle operating in the 70% - 85% efficiency range and dissipating at least 1.5 kW of heat.

But, as far as bike controllers go, you may be right about the non-synchronicity . Does that generality apply to high power bike controllers, too? (Like Meth's 10 kW ones) If so.... how does it dissipate so much heat? I thought heat sinks were impractical, size wise, above 10-100 watts or so and I didn't see any forced convection.

 

[rhitee05]

You'd probably get the best efficiency if you were able to set up mechanical gearing with at least two different ratios for WOT and loitering. Hard switching between two different voltages might also help. You could probably do that by using diodes to parallel the generator and batteries. That's assuming you'd choose a lower battery voltage than the generator voltage, something optimized for your low-speed operation. The diodes would prevent the generator from overcharging the batteries, then the batteries would take over when the generators shut down.

The issue of PWM and duty cycle is another good point. Eliminating PWM/throttle altogether would definitely improve efficiency to near 100%, but you'd have only as many speeds as you do mechanical ratios.

 

[John in CR]

No not high torque ever, except during acceleration to get the props up to speed. Dual motors, so 750-1000W each for cruising at 6-7 knots. Motors are unknown at this point. Those 5-7kw Golden Motors are cheap enough to carry spares, though even with a pair, I doubt they can get the top end I want above 15 knots. A 4 stroke outboard to help with the runs out and back may end up a requirement, plus I like that extra backup despite the reliability of electric motors.

What I don't like though, is the idea of giving up 15-30% in the controller. Is that correct, you can lose so much just in the controller when running at partial throttle? If so, then with brushless motors I'd probably have to go with a pack switching between 24V for trolling and 72V for maximum speed. Yes, I'm thinking of something like the Methods controllers, again priced right for spares.

John

 

[rhitee05]

Yeah, the problem with partial throttle is the lower duty cycle. When the transistors are off, the current's flowing through the body diode built into the FETs. Unfortunately, those diodes aren't very good - forward drop 0.7-1.0V, usually. A synchronous controller will turn those FETs on during this time, so the freewheeling current sees the Rdson of the FET instead. There's no hardware difference, just in the control of which FETs are turned on at what times. I don't know which controllers do this, but as swbluto pointed out it'd make a big difference in power dissipation.

 

[John in CR]

I'm looking to reduce the stress on my drive train without giving up any power by increasing the voltage and decreasing the current, and I'd change the gearing for the same top speed. My thinking is that the increase in iron losses may offset the decrease in copper losses in the motor, but the motor will dissipate heat better at higher rpm and my copper losses (quite warm phase wires) outside the motor will decrease. Also, the controller should be less stressed and cooler as a result of both lower current and lower gearing of the motor, so I should enjoy at least some increase in overall efficiency. My battery will thank me, because I'd keep the same ah, and just tack on some more in series.

Assuming the copper losses are greater than the iron losses, won't the motor also gain some efficiency at higher voltage for the same power input?

John

 

[Drunkskunk]

In my experiance, yes. On the Ebike with hub motors I only see 5-10% better efficancy.
But I learned the trick from RC planes. My tigermoth went from 20 minute flights to 30 minute flights by going from 2 cell to 3 cell lipo. it used the same wattage, but a lower KV motor. I had similar results in 6 or 7 other planes, though that was the most dramatic jump.

 

[John in CR]

In my experiance, yes. On the Ebike with hub motors I only see 5-10% better efficancy.
But I learned the trick from RC planes. My tigermoth went from 20 minute flights to 30 minute flights by going from 2 cell to 3 cell lipo. it used the same wattage, but a lower KV motor. I had similar results in 6 or 7 other planes, though that was the most dramatic jump.

Wow, 5-10% better efficiency would be huge. If I'm running say 75% efficiency, increasing that to 80% would be a 20% reduction in heat losses. That would make a huge difference the way I was running the rig right at the edge... phase wires beginning to show distortion in their insulation and the motor and controller too hot to touch for more than a second or 2.

 

[olaf-lampe]

Iron losses are lower than copper losses. And a huge chunk of iron can collect more heat than the tiny hub wires. You'll gain efficiency and safety as well. ( Despite the potential lethal voltage above 60VDC anyways )
-Olaf

 

[John in CR]

Olaf,

I'm at 74v nominal and going to 133v nominal, so I'm lethal already. How is it that safety increases? Do you mean by reducing the stress on wiring and connections with lower current?

 

[olaf-lampe]

Olaf,

I'm at 74v nominal and going to 133v nominal, so I'm lethal already. How is it that safety increases? Do you mean by reducing the stress on wiring and connections with lower current?

Yes John, no more melting cables in the shaft. The hard thing is to keep the current low. It's too seducing to use both: high voltage and high current
-Olaf

 

[Desertprep]

Given 2 motors with the same wattage and voltage, (say 48v and 1000 watts) what is going to be the difference in either speed or range if the efficiency of one is 80% and the efficiency of another is 90%?

 

[liveforphysics]

In speed or range, it would be minor, like 10%.

In power handling, the 90% motor could handle twice the continuous power (if they were both operating at peak efficiency of 80% and 90%.)

Likewise, a 95% motor could handle 4x the continous power of the 80% motor.

 

[Desertprep]

Wow! So an outrunner that is rated at about 95% efficiency is going to perform substantially better than a common bldc motor rated at 80%! It might be worth the extra effort and $$ I have found a number of average quality motors here in China that will do what I want, but most of the brushless ones are advertised as >= 80% efficiency. It really sucks that a lot of the good stuff is made here but never makes it out into the market places here. I can't even get to their website, or even a site that was referred to me to help me calculate the size motor that I need for my bike Hobby city in HK will not ship to mainland The best option I have before me is the cyclone, which advertises a 94% efficiency for their most powerful motor.

If I took a motor that was rated at >80% efficiency and rewound it and replaced the magnets, is it possible that I could raise the efficiency of the motor that way?

 

[Desertprep]

Will a motor's efficiency increase as its voltage increases? (within reason)