Partial throttle and motor efficiency

J

John in CR

100 TW
Joined
May 19, 2008
Messages
14,954
Location
Paradise
Ok I'm spoiled with more performance than I have the balls to use, but there are aspects that come with it that I refuse to give up, such as still having scary acceleration at any point in the speed range I typically ride. The unavoidable shape of the power and torque curves means running a voltage high enough to make top speed faster than you use, but is there a cost beyond what comes with using the extra performance?

There seems to be a cost in terms of controller efficiency, since riding at high power but partial duty has always made my controllers run hotter than they do running similar performance at lower voltage. It doesn't seem to make a difference to the motors I use, because I don't see them make more heat unless I up the current or increase the load for more time at similar current. Is this my imagination or is it real?

The distinction is important as I build some long-range ebikes with big battery packs of 45-60ah or more. I don't have a problem running more capable controllers than are necessary to keep their resistance lower and reduce losses, but there's nothing I can do about the motor(s) and I much prefer having the extra performance on tap. Plus, in the mountains a small difference in heat generated can add up quickly.

It boils down to one question. For the same torque and rpm, does it make a difference in the motor at what duty cycle the controller is running, or is the difference insignificant? I think the answer is no, though it may depend somewhat on the motor. Does it depend on controller type too?
 
FluxZoom said:
https://endless-sphere.com/forums/viewtopic.php?f=31&t=65031&start=50#p981766
Click to expand...

Interesting info, and once prices become reasonable of course I'll try FOC, but it doesn't address the primary question that boils down to whether there's an efficiency cost at the motor for running a higher voltage than the speeds typically required. Giving up the much greater acceleration in the mid to high speed range will require a significant difference to give up.
 
I can't have a set up with unlimited speed as living in So Cal 40mph is fine. It is very efficient after 32 mph being a 5t 3,000. Going faster would just be trouble in the states. I'm buying a 4t turbo and fine out how I like it and what speed and efficiency at what speed.
I would imagine your setup is most efficient at a very high speed, what speed ? Your ca will tell. Now go do a 95mph wh/m
Test.
 
999zip999 said:
I can't have a set up with unlimited speed as living in So Cal 40mph is fine. It is very efficient after 32 mph being a 5t 3,000. Going faster would just be trouble in the states. I'm buying a 4t turbo and fine out how I like it and what speed and efficiency at what speed.
I would imagine your setup is most efficient at a very high speed, what speed ? Your ca will tell. Now go do a 95mph wh/m
Test.
Click to expand...

That's not my point. If your bike's top speed is 40 and you ride at 40, then acceleration before you get to 40 sucks. Imagine a car with an absolute top speed equal to the speed limit. No one would buy it. It's not about going fast and the videos I've done and with much better to come are only to demonstrate my motor's potential. That's not how I ride. In fact, I ride slower with my 100mph+ capable bike than I did when it was only capable of just under 70. What I love and use every ride is the far greater mid range acceleration. The question is whether there is a difference in motor efficiency with an identical system cruising at say your 40mph if it's volted for 40 vs volted for 70? Even if you stick strictly to a max of 40 the ride experience is drastically different.
 
Hi John.

The answer is, the motor doesn't care. It only feels the current. It doesn't feel the duty cycle.

The controller has to deal with issues. Like higher voltage FETs have much higher forward resistance. Higher voltages require more time to switch. So more time in the loss region. Etc.

It isn't the duty cycle per se, it is the higher voltage and the effects that has. Whether you are running 5% or 90% duty cycle there are exactly the same number of switching transitions. The average current is different, and that of course increases loss through the I squared R in the FETs.

But the motor doesn't really care. As you have established empirically already.
 
Alan B said:
Hi John.

The answer is, the motor doesn't care. It only feels the current. It doesn't feel the duty cycle.

The controller has to deal with issues. Like higher voltage FETs have much higher forward resistance. Higher voltages require more time to switch. So more time in the loss region. Etc.

It isn't the duty cycle per se, it is the higher voltage and the effects that has. Whether you are running 5% or 90% duty cycle there are exactly the same number of switching transitions. The average current is different, and that of course increases loss through the I squared R in the FETs.

But the motor doesn't really care. As you have established empirically already.
Click to expand...

I pretty much understand right up until you bring up average current. With the higher pack voltage, wouldn't average current (at least thru the input side fets) be lower, so if I run equal total mosfet resistance, wouldn't the higher voltage controllers run cooler...or is it something along the line of "bucking down" to the lower effective voltage for the motor generally become more lossy in the controller? Now I'm trying to understand how a pair of 24fet controllers running 111V nominal can run quite warm even just riding around leisurely at neighborhood speeds of 30mph or so.

Now that I'm changing the subject to the controller side, I just don't get how my current irfb4115 based controllers run so warm even at riding near legal ebike performance levels, and we're talking about of pair of 24 fet controllers with decent air flow driving the motor. I'm soon going to a pair with 24 irfp4568 mosfets in each with great airflow over a big heat sink, and massive copper on the traces, so if the overwhelming majority of controller losses are I2R losses in the mosfets, I should see significant reduction in controller losses.

John
 
John in CR said:
Alan B said:
Hi John.

The answer is, the motor doesn't care. It only feels the current. It doesn't feel the duty cycle.

The controller has to deal with issues. Like higher voltage FETs have much higher forward resistance. Higher voltages require more time to switch. So more time in the loss region. Etc.

It isn't the duty cycle per se, it is the higher voltage and the effects that has. Whether you are running 5% or 90% duty cycle there are exactly the same number of switching transitions. The average current is different, and that of course increases loss through the I squared R in the FETs.

But the motor doesn't really care. As you have established empirically already.
Click to expand...

I pretty much understand right up until you bring up average current. With the higher pack voltage, wouldn't average current (at least thru the input side fets be lower, so if I run equal total mosfet resistance, wouldn't the higher voltage controllers run cooler...or is it something along the line of "bucking down" to the lower effective voltage for the motor generally become more lossy in the controller? Now I'm trying to understand how a pair of 24fet controllers running 111V nominal can run quite warm even just riding around leisurely at neighborhood speeds of 30mph or so.

Now that I'm changing the subject to the controller side, I just don't get how my current irfb4115 based controllers run so warm even at riding near legal ebike performance levels, and we're talking about of pair of 24 fet controllers with decent air flow driving the motor. I'm soon going to a pair with 24 irfp4568 mosfets in each with great airflow over a big heat sink, and massive copper on the traces, so if the overwhelming majority of controller losses are I2R losses in the mosfets, I should see significant reduction in controller losses.

John
Click to expand...

Hi John.

The SAME Motor current flows through the FETs REGARDLESS of the supply voltage. Current is torque. Same torque, same current.

The peak current flowing from the battery (when the FETs are ON) (really supplied mostly by the caps in the short term) is the SAME as the motor current. it flows for less time (lower PWM duty cycle), just long enough to transfer the power that's being used.

The battery sees lower current, feeding the caps. The only place the current is lower due to high voltage is between the caps and the battery. From the caps on downstream, current is motor current, determined by torque and motor physics.

But the FETs carry this average motor current ALL the time. (Really battery current is nearly irrelevant). Only motor current matters. Even when the FETs are switched off (not supplying battery power during the OFF portion of the PWM cycle). It just flows through different FETs at that time (or their body diodes, which is even lossier than when the FETs are on).

The I squared R losses when the FETs are on is higher as R is higher for these high voltage FETs. The I is the same.

During switching time windows the E squared / R losses in the FETs are higher as R in the FETs is higher, the switching time is longer, and E is higher, and it is squared so it is much higher.

Just every way you look it is lossier. Higher voltage is MUCH harder on the controller. You've empirically established this already.
 
Remember John,

You picked a motor with low voltage and high current.

You are stuck with this high current to make torque. Anytime you are making torque you have this current flowing through the controller.

Regardless of the battery voltage.

Controller current makes losses, and battery voltage makes losses.

Sinewave controllers are generally better, are you using them? But they have the same battle, just it takes more engineering to make a Sine controller so they do better on a lot of things, and get a few percent less losses.

Edit - PS - thanks to ES for helping me study this. Years ago when I was doing my EE I didn't think motors were interesting, I was more interested in circuits, data acquisition, control, computing and radio. So I skipped all the three phase motor and power stuff. A few years ago I decided to build an ebike and found ES. So now I've had to go back and study what I skipped. A lesson for us all - just because it seems boring now doesn't mean it won't be exciting and useful later on! Study and learn everything you can, it will serve you well in unexpected ways later!!
 
999zip999 said:
So John what's your setup ? My guess. Hubmonster, two 24fet contollers a 120v 85ah battery. Controllers set to 150 amps each.
Click to expand...

The current setup is 2 bikes each with a ventilated HubMonster in-wheel, both with 30s packs, 111V nominal, though I may go to 31 or 32s on one for some top speed videos. I've lost almost 70lb since my ridiculous peak a year ago, and my SuperV's HubMonster had a 19.25" OD tire back then and it never got hot, only pretty warm (under 105°C) when ridden at max aggressiveness or up a 20% grade accelerating hard out of each low speed curve. With my big load reduction the old bike is getting a 20.5" OD wheel, and the new one has a 16" moto alloy mag wheel instead of the 13" scooter rim and the tire I chose has an OD of almost 22". The new one (a build nearing completion) has a shorter wheelbase with a higher saddle and CG, so it remains to be seen how high I can set the current, but those controllers have a switch on the fly 3 current limit switch with a max range of 50%. That means I can have a lean over the bars current setting for use in embarrassing motos and sportscars at red lights, while retaining lower chance of flipping for typical riding. The pack on that bike is nowhere near 85ah, but the 14+ah pack is capable of burst current of well over 1200A, something I'll never get close to...it's for low sag for the top speed runs.

When I go to an emoto build with 60-100ah of capacity, the added weight means it will get dual HubMonsters, though a big pack is foreseeable on a long-range cargo hauler, but depending on speed and power requirements, that would be a single HubMonster or MidMonster running as a mid-drive to get the desired lower speed gearing capable of big loads.

John
 
Alan,

Thanks for the explanations down to or near my level of understanding. I filled in some gaps to take it that the higher voltage converting down to signal needed by the motor below at partial speed is more lossy in the controller. I take it the motor doesn't care, because its inductance causes it to see more or less constant current at cruise regardless of the duty cycle of the pulses.

Back to the controllers though. The positive and negative battery leads are connected directly to the input side fets and output side fets respectively, so I don't get the some of what you said. While yes there are capacitors, they are in parallel with the primary power wiring, so I thought they just cushioned the blow of what the controller asks from the battery and what comes back from the motor...ie aren't they just helpers? There's no way total current is flowing through the wires of the capacitors.

That gets me back to needing to ask in the case of the higher than necessary pack voltage, where do the extra losses show up in the controller? If it's primarily I2R losses, then I can make them go away with better (or more) mosfets and thicker internal wiring. If significant, I can't imagine it being somewhere else, since nothing else is attached attached to a heatsink.

No sine wave controllers for me so far, at least not FOC type. I refuse to pay more than the motor costs when quite economical trap controllers give me nearly silent operation with these 6 phase motors, not to mention that I can't find any FOC controllers remotely reasonable in price that can handle over 96V nominal.

John
 
Hi guys, this is some very interesting stuff with tech details well beyond me. Perfect opportunity to learn something. Is my thinking here correct?

Naturally to get the most from any motor/controller we want the best combo of volts and amps.

Volts should be looked at essentially as gearing as well as the other half of the power.

So when you have a motor with high kV, it's better to focus equation of volts x amps towards the amps side, and utilize mechanical (gears or wheel diameter) means to achieve desired top speed beyond a certain voltage.
That top desired speed is affected by preference of watts VS rpm and how well motor and controller deal with losses associated within their matchup, and ride design/ use preference.

So maybe John's controllers are working harder (showing more loss) because the motor would prefer better mechanical gearing for speed, with the balance of the power equation shifted towards amps?

. . . . .
Yes, v<a = v>a for the motor. Watts are watts. But, overall efficiency/loss is affected by your choice of gearing VS where (what motor rpm range) you are applying power (ride design vs use), and how capable the motor is at handling loss, right?

So it's best to choose a top motor rpm (via volts and other 'gearing') about 20% max above your desired vehicle mph for best efficiency (at top mph), but when motor is too stressed with the amount of watts you're throwing at it in it's lower rpms, you can either go easier, or drop amps, or sacrifice the +20% 'gearing' vs speed ratio.
If you tend to run @ 0-40% rpm far more often than 100%, and motor or controller are seeing too much loss, it might be better to just drop volts till your max desired top mph is max rpm- trading top-end efficiency for low-end efficiency.

. . . . .
My bike: low kV mid bht (10-11lbs) with 18fet. The motor gets hot @ 6kw, but get's hotter quicker if I play with 6kw more in the lower portion of it's gearing. (0-25mph play w/ 40mph top speed).
The controller doesn't get hot whether the controller is using 80v x 75a or 60v x 105a. And the motor get's hot either way. So the motor is the weak point. I can make the same top speed with wheel diameter or reduction for the lower voltage, and the motor can't tell the difference.

It's my job to determine if I want low speed draw vs top speed. Thus, I may benefit from not over gearing it past max desired speed (or even drop max speed or look for variable gears).
Because the motor is low kV, I should focus as much as possible of my speed gearing with voltage (to some degree unknown to me- so I'll just say realistically like 100-120v), and then match amps with what the controller and motor handle in they're pairing and with my riding style/attributes/use.

Hopefully that hits the points of discussion. If it's all correct, there must be a simpler way to say it.
 
Would it not be possible to switch a couple of, say, 36v battery packs that could be paralleled or seriesed? i.e 72v for high speed and then 36v with twice the amps for low speed or would a controller not like this? I know nothing about controllers.
 
Do controllers actually go into block commutation at full throttle, which results in lower switching losses? Or is this a myth?
 
Punx0r said:
Do controllers actually go into block commutation at full throttle, which results in lower switching losses? Or is this a myth?
Click to expand...

Zombiess scoped it and found it to be myth. Corrected by Teklektic's post below.
 
John in CR said:
Alan,

Thanks for the explanations down to or near my level of understanding. I filled in some gaps to take it that the higher voltage converting down to signal needed by the motor below at partial speed is more lossy in the controller. I take it the motor doesn't care, because its inductance causes it to see more or less constant current at cruise regardless of the duty cycle of the pulses.

Back to the controllers though. The positive and negative battery leads are connected directly to the input side fets and output side fets respectively, so I don't get the some of what you said. While yes there are capacitors, they are in parallel with the primary power wiring, so I thought they just cushioned the blow of what the controller asks from the battery and what comes back from the motor...ie aren't they just helpers? There's no way total current is flowing through the wires of the capacitors.

That gets me back to needing to ask in the case of the higher than necessary pack voltage, where do the extra losses show up in the controller? If it's primarily I2R losses, then I can make them go away with better (or more) mosfets and thicker internal wiring. If significant, I can't imagine it being somewhere else, since nothing else is attached attached to a heatsink.

No sine wave controllers for me so far, at least not FOC type. I refuse to pay more than the motor costs when quite economical trap controllers give me nearly silent operation with these 6 phase motors, not to mention that I can't find any FOC controllers remotely reasonable in price that can handle over 96V nominal.

John
Click to expand...

Hi John.

I squared R losses in the FETs are mainly from the high motor current required for torque. This is determined by motor selection, and by FET forward resistance (R). High voltage FETs have poor forward resistance hence large losses. Motor current is constant (with some ripple) so FETs carry this current all the time.

Voltage losses in the controller are inherent in the switching losses. Switching takes time, and during this brief window the losses are huge. The only thing that keeps the FETs from exploding is the fact that this event is very brief. The higher the voltage, the longer it takes to switch and the higher the switching losses are. Higher voltage FETs don't want to switch as fast which adds to the problem. The faster you want to switch, the more engineering it takes to do it well. The more noise it generates. The more problems it causes. The high voltage losses show up in the FETs as switching losses. Adding more FETs helps I squared R losses, but makes switching more difficult and doesn't inherently reduce the switching loss. For example, at the midpoint of the switching the losses are E squared over R where R is the halfway resistance where the voltage across the FET is half the supply. This R doesn't depend on the FETs, the only thing to reduce this loss is to minimize the time spent in this region. To reduce switching losses requires switching faster. Which is difficult, and even harder with more FETs. Switching faster requires more engineering, more drive power, faster drivers, better capacitors, more expensive parts and better, more costly design. At the end of the day lower voltage is more efficient for the controller.

An old rule of thumb in low to midrange power conversion was a dollar per watt for solid reliable gear. It has come down, but you are a couple orders of magnitude below that. By constraining yourself to cheap electronics you doom yourself to higher losses. Yesterday's designs. Mediocre parts. And you avoid rewarding the companies that are pushing the state of the art forward. When we spend money we are choosing who to reward, not just getting product.

So what you're asking for is a very low loss high voltage high power controller, but you don't want to pay for the engineering and components required to make it.
 
Thanks Alan and for correcting me Teklektic,

I can't believe I missed that one at 100% in Zombiess work. How can I quantify switching losses, because it looks like I need some kind of series parallel switching for at least a portion of my pack on long distance cruising. Around town regardless of voltage I'd never be at WOT to make a meaningful difference.

Is achieving block commutation independent of load and rpm, ie simply throttle at 100%? If not then how significant a portion of the time can it be.
 
John in CR said:
Thanks Alan and for correcting me Teklektic,

I can't believe I missed that one at 100% in Zombiess work. How can I quantify switching losses, because it looks like I need some kind of series parallel switching for at least a portion of my pack on long distance cruising. Around town regardless of voltage I'd never be at WOT to make a meaningful difference.

Is achieving block commutation independent of load and rpm, ie simply throttle at 100%? If not then how significant a portion of the time can it be.
Click to expand...

FOC controllers switch all the time, and they don't get hot. They just invest the engineering into doing switching more efficiently.

If you want to do 100% block commutation, then run a DC-DC converter in front of the commutating FETs. Then the motor and commutation doesn't see PWM at all. The DC-DC converter uses PWM but has a lower loss inductor and the high frequency loop is contained inside the circuit so it can be done more efficiently. It also doesn't have polarity reversal to deal with.

Like you, I've experienced many (actually all) controller failures at essentially zero speed. This is because the controller did not keep PWM pulses very short. If a controller allows too large a pulsewidth the current peaks will destroy the FETs. Controllers MUST not allow long PWM pulses or block commutation until motor currents are limited to a safe level by back EMF. Uncontrolled excessive currents cause a lot of low speed controller failures. Excessive voltage spikes probably account for the rest.

Block commutation with standard controllers means you have no control, no speed reserve, nothing left. It isn't where you want to operate normally. You want a controller that gives great control and maintains PWM without getting hot and without being lossy. FOC controllers do that. The Sabvoton on my CroBorg gives the most wonderful throttle control and is very efficient and stays cool. For those bursts of speed the field weakening feature can be used - it is like a voltage boost. It doesn't cost you anything till you use it. Justin's measurements show that it doesn't cost all that much when you do use it either. There is a lot of engineering in the good FOC controllers.

Switching voltages is a very interesting idea, and one that I think could work particularly well for us. It is easy to switch two half packs in series/parallel with some diodes and FETs. There are a few problems to solve, but it could be a good way of only paying the higher voltage losses when necessary. A 30 mph / 60 mph setup might make a lot of sense. BUT you are going to have to use the high voltage controller, and the I squared R losses are going to be paid all the time. So you will reduce switching losses, but suffer the increased I squared R losses of the high voltage FETs. Field weakening doesn't have this problem.

But switching voltages adds a lot of complexity to the system. Perhaps the better approach is to work harder on minimizing switching losses. Use lots of FETs and excellent engineering and drivers to switch really efficiently. Keep the System Simple. That's what Jeremy and Arlo are trying to do.

As consumers we should find folks who are building better controllers and buy their stuff, vote with your wallet. The cheap products tend to hurt these companies, if we don't support them they will leave the market and we won't have improvement. One example of theseiare the BAC series from Ebikes.ca. Perhaps the Kelley sinewave units. Sabvoton. Adaptto. Seek them out. They are trying to help.
 
In the topic: Looking for a perfect motor controller? gather here

Bluefang said:
Just put a pair on my motorbike driving a Hub-monster. Soooooo much smoother and quieter. I am in love with sine wave controllers.

[youtube]1VjRGx9yN3k[/youtube]
Click to expand...

Was this one of John's Hub-monsters with two sabvoton controllers?
 
Martin A said:
In the topic: Looking for a perfect motor controller? gather here

Bluefang said:
Just put a pair on my motorbike driving a Hub-monster. Soooooo much smoother and quieter. I am in love with sine wave controllers.

[youtube]1VjRGx9yN3k[/youtube]
Click to expand...

Was this one of John's Hub-monsters with two sabvoton controllers?
Click to expand...

Yep. We're both hoping that his brother can pull off the twin 200V Lebowski based controllers to explore the true limits of properly ventilated HubMonsters and MidMonsters.
 
John in CR said:
Martin A said:
In the topic: Looking for a perfect motor controller? gather here

Bluefang said:
Just put a pair on my motorbike driving a Hub-monster. Soooooo much smoother and quieter. I am in love with sine wave controllers.

[youtube]1VjRGx9yN3k[/youtube]
Click to expand...

Was this one of John's Hub-monsters with two sabvoton controllers?
Click to expand...

Yep, and Bluefang modded those Sabvotons for much higher power than stock. We're both hoping that his brother can pull off the twin 200V Lebowski based controllers to explore the true limits of properly ventilated HubMonsters and MidMonsters.
Click to expand...
 
Yes I got a new 18fet Sinewave controller 100v 80amp controller 200.00 it's smooth and deadly quiet. Love seeing how that hubmonster quite down. Like magical.
 
Looks like you want an FOC controller, John. Or maybe a beast of a 60V trap controller and a 1-turn motor ;)
 
You must log in or register to reply here.

Similar threads

D
Motor Smoothness: FOC vs. Six-Step BLDC
Replies
1
Views
120
amberwolf
amberwolf
M
Parallel multiple brushed controllers on one big brushed motor
Replies
8
Views
249
matt912836
M
B
    • Like
Arduino Virtual Flywheel Throttle controller
Replies
10
Views
498
DingusMcGee
D
C
Optimizing DD motor and moped rim for 60km/h and 2.5kW (and other questions)
Replies
5
Views
951
chadez
C
S
Actual max power of th48ve TSDZ2B with OSF/
Replies
3
Views
230
dgc
dgc
Back
Top