**Important** reality check on motor, voltage, current etc.

[rhitee05]

Sorry! That's unfortunate timing.

 

[John in CR]

Sorry! That's unfortunate timing.

If we learned something then great. I have ridden that stretch many times, though maybe not ever that slow, but that is the first time with a controller pushing it's limits. What could have made it blow at such very low power throughput? Could it be extra energy dissipated in the controller that never made it to noticeable motor output put from tiny pulses of the throttle on the bumpy road? The strange thing and as happened on the other occasions, the very slow speed stuff set it up to blow at the next normal load acceleration.

John

 

[TylerDurden]

What could have made it blow at such very low power throughput?

PWM is switching full power with each pulse, until BEMF suppresses motor current, yes?

 

[johnrobholmes]

High intermittent loads at low duty cycle, too high of phase current limits, and a lack of gearing to compensate for the speed.


This illustrates why we can't have a vehicle volted/ geared for 50mph that will also take technical off roading too. The controllers just can't handle the lowest throttle usage for very long. Having a switch to change voltage or motor wind would have been advantageous if this is a hub motor.

 

[swbluto]

I think it may have been a combination of the diode loss in the controller at the low duty cycle (Because it's free-wheeling during the mosfet OFF period of the PWM) and the inefficiency of a hub motor going 5 mph. At 5 mph, I'm guessing the majority of your power/current is going towards no-load losses, so only a little bit of the current goes towards actual propulsion. Basically this means that the controller sees that kind of free-wheeling current above that minimum current. But that shouldn't be that much current... So I'm thinking...

Also, from my experiences, I also wonder how much of it was truly "low throttle". It seems particularly hard to maintain an actual constant throttle, and I seem much more prone (When going slow) to going to 10%, then bursting to 40-70% to gain speed after avoiding a hole or whatever, then dropping to 10% and slowing down, bursting with 40-70% to speed up and so on. That would exacerbate the problem even more because it'd be seeing high phase currents, nearing the magnitude of full throttle stalling currents (Worst kind), while at partial throttle on a regular basis and not a "one-time" launch basis.

 

[liveforphysics]

Do'h! Sorry to hear about the controller loss my friend!

Light throttle at walking speed on the motor is just hell on everything. Even the poor motor is turning like 90% of the power you're feeding it into heat at those speeds with a direct drive setup.

As far as the idea 300w of controller heat is too high of an estimation, for that controller if it were at 100v 100amps out, then 300w of heat would mean its 97% efficient. It would be 150v 100amps out capable, which would make it
98.5% efficient. Unfortunately... that condition would never happen with low turn motors, (because it would mean like 150mph!)

 

[denito]

Evan,
He and LFP are both saying the power in stays the same, so the reduced average voltage out means current out must go up accordingly.

Homerun for John!


Power IN needs to match power OUT. PWM drops the average voltage across the coils, power needs to balance, so current goes UP by the same amount voltage dropped.


The system works just like a buck-type power supply. The motor winding is the inductor. Example of a bucking power supply would be starting at 100v and 20amps IN (2000w), and your supply switches at 50% duty cycle to buck the 100v IN into 50V out, but we've still got 2000w of power, so the current becomes 40amps to balance the equation.

You're mixing up independent variables with dependent variables. You're taking power to be an independent variable and making current dependent on it, and saying that if power is constant then reducing duty cycle means current must go up to maintain the same power. Well, yes, but only if you incorrectly assume power is constant or an independent variable. But it's not - reducing the PWM duty cycle also reduces power. (How the hell do you think you are able to ride the bike slower at part throttle and consume battery energy more slowly if reducing PWM cycle doesn't reduce power consumed?) The "POWER IN" doesn't magically maintain at some wattage as the controller throttles down the PWM.

A lot of the bullshit that flies around about all these motor power/torque/current/voltage/winding etc. discussions seems to stem from using (parts of) the right equations, but then holding the wrong variables constant or other incorrect assumptions about what the equations model. Usually the author takes the behavior of what a computer controlled or feedback controlled circuit would have to actively do to maintain the unquestioned false assumptions and then generalizing that to claim it models the behavior of a circuit that is not actively trying to maintain that variable. For instance in the bucking power supply example you give, the behavior you're describing is what a regulated power supply would have if it's trying to power a constant wattage load in the face of a changing input voltage. But this is an example of a microprocessor controlled or feedback loop controlled active circuit that is designed to maintain constant power output. That's not a model of an e-bike controller, quite the opposite, an e-bike controller is designed to vary output power given a constant (over the short term) input voltage.

Furthermore your (correct) statement that inductors tend to maintain a constant current is self-contradictory with your (incorrect) conclusion that current must go up when PWM goes down. Yes, inductors "want to" maintain a constant current. So if there were, e.g., 20 amps flowing into the motor coil when the MOSFETs were on, then there will still be approximately 20 amps flowing through the motor coil during the first milliseconds after the MOSFETs switch off. This current flows through the flyback diodes while the transistors are off - that's why the flyback diodes are there. But your conclusion that current must go up when PWM goes down, based on the unquestioned assumption that output power is constant, would require the current to spike during the off periods (low voltage) in order to maintain constant power - but a sudden unexplained increase in motor coil current is self-contradictory with what you started out saying that inductors do (maintain constant current)!

 

[deVries]

What could have made it blow at such very low power throughput?

PWM is switching full power with each pulse, until BEMF suppresses motor current, yes?

Why not design and add a surge/pulse protector for low throttle and slow speeds by only allowing voltage/current to pass from the battery in an on/off pulse every so many micro-seconds before entering the controller. This on/off pulse could be timed just right to prevent fet heat build-up and allow just enough time for the fets or whatever else needs extra cooling for these short periods of time to recover. Also, maybe have a fan turn on to blast air over the heat sinks or heated areas when at slow speeds and 1/2-1/3 throttle or less under load or at x temperature. Also, if the on/off pulses are timed just right to the controller, then one might be able to reduce the sensation of motor hesitation too.

Isn't it true if a motor is a 20 turn motor vs an 8 turn motor, then the fets pulse less current for the same or similar amount of work output that the motor does?

 

[rhitee05]

I have ridden that stretch many times, though maybe not ever that slow, but that is the first time with a controller pushing it's limits. What could have made it blow at such very low power throughput? Could it be extra energy dissipated in the controller that never made it to noticeable motor output put from tiny pulses of the throttle on the bumpy road? The strange thing and as happened on the other occasions, the very slow speed stuff set it up to blow at the next normal load acceleration.

My guess? Your controller probably had enough thermal mass to absorb the excess heat from a WOT launch when cold. You may have been operating beyond the continuous limits, but it takes some period of time for the temp to build to dangerous levels. If you're in a more efficient regime by that time, you'll probably be okay.

In this case, you would've warmed the controller up during the low-speed run. When you launched this time, the controller was closer to the "red zone" and couldn't absorb as much excess heat. Poof.

It's a shame these controllers don't do synchronous rectification, because that would make this quite a bit less of an issue. Your controller was a 15-FET, right? If we assume 100A, the diode loss would be 130W. If you did SR this would be replaced by I^2*R loss. Assuming 3x 4115 FETs @ 9.4 mohm each, that would give total Rdson about 3.1 mohm. Allowing for the increase with temp, we'll say 7 mohm. At 100A that would be 70W loss. Almost 50% less!

 

[Tiberius]

Evan,
He and LFP are both saying the power in stays the same, so the reduced average voltage out means current out must go up accordingly.

Homerun for John!


Power IN needs to match power OUT. PWM drops the average voltage across the coils, power needs to balance, so current goes UP by the same amount voltage dropped.


The system works just like a buck-type power supply. The motor winding is the inductor. Example of a bucking power supply would be starting at 100v and 20amps IN (2000w), and your supply switches at 50% duty cycle to buck the 100v IN into 50V out, but we've still got 2000w of power, so the current becomes 40amps to balance the equation.

You're mixing up independent variables with dependent variables. You're taking power to be an independent variable and making current dependent on it, and saying that if power is constant then reducing duty cycle means current must go up to maintain the same power. Well, yes, but only if you incorrectly assume power is constant or an independent variable. But it's not - reducing the PWM duty cycle also reduces power. (How the hell do you think you are able to ride the bike slower at part throttle and consume battery energy more slowly if reducing PWM cycle doesn't reduce power consumed?) The "POWER IN" doesn't magically maintain at some wattage as the controller throttles down the PWM.......

Clears throat.... Ahem

Denito, I'm guessing you haven't read every word in the thread, and who can blame you. But that is exactly what does happen. We are talking about the situation where the controller is acting to limit the battery current. In that case the "POWER IN" does magically maintain at some wattage.

Nick

 

[liveforphysics]

You're mixing up independent variables with dependent variables. You're taking power to be an independent variable and making current dependent on it, and saying that if power is constant then reducing duty cycle means current must go up to maintain the same power. Well, yes, but only if you incorrectly assume power is constant or an independent variable. But it's not - reducing the PWM duty cycle also reduces power. (How the hell do you think you are able to ride the bike slower at part throttle and consume battery energy more slowly if reducing PWM cycle doesn't reduce power consumed?) The "POWER IN" doesn't magically maintain at some wattage as the controller throttles down the PWM.

A lot of the bullshit that flies around about all these motor power/torque/current/voltage/winding etc. discussions seems to stem from using (parts of) the right equations, but then holding the wrong variables constant or other incorrect assumptions about what the equations model. Usually the author takes the behavior of what a computer controlled or feedback controlled circuit would have to actively do to maintain the unquestioned false assumptions and then generalizing that to claim it models the behavior of a circuit that is not actively trying to maintain that variable. For instance in the bucking power supply example you give, the behavior you're describing is what a regulated power supply would have if it's trying to power a constant wattage load in the face of a changing input voltage. But this is an example of a microprocessor controlled or feedback loop controlled active circuit that is designed to maintain constant power output. That's not a model of an e-bike controller, quite the opposite, an e-bike controller is designed to vary output power given a constant (over the short term) input voltage.

Furthermore your (correct) statement that inductors tend to maintain a constant current is self-contradictory with your (incorrect) conclusion that current must go up when PWM goes down. Yes, inductors "want to" maintain a constant current. So if there were, e.g., 20 amps flowing into the motor coil when the MOSFETs were on, then there will still be approximately 20 amps flowing through the motor coil during the first milliseconds after the MOSFETs switch off. This current flows through the flyback diodes while the transistors are off - that's why the flyback diodes are there. But your conclusion that current must go up when PWM goes down, based on the unquestioned assumption that output power is constant, would require the current to spike during the off periods (low voltage) in order to maintain constant power - but a sudden unexplained increase in motor coil current is self-contradictory with what you started out saying that inductors do (maintain constant current)!

Denito- You're badly confused here. I dont know how many damn times ive said it now, but this is for the time in which the controller is in battery current limiting mode. Aka, the whole area to the left side of the vertical red line on the example graphs. Up until this point, the power in is constant when the controller is full throttle. Example, a 20amp controller at full throttle will continue to draw 20amps from the battery until that red line. Last i checked, if battey amps and voltage are held constant, that means the power in is constant ( until that redline). What isnt constant? The voltage the motor sees because its chopped up to something lower by PWM until you reach that red line on the graph. Power in constant, voltage out lowered.... means current went up.

Makes sense?

 

[johnrobholmes]

Oh what a fun thread! I have to hand it to you guys for keeping it decently civil. 13 pages and going strong.

 

[tostino]

I agree! I've been reading this since I first saw LFP post it a few days ago, and just check in every few hours lol. Very informative, and yes, surprisingly kept pretty civil.

 

[donob08]

later

 

[Goethe]

Just an hour too late for my controller. Since I modded the shunt a few days ago, I've done numerous repetitive launches at both max acceleration and partial, all with no issue. What killed it was 200m of road that was so bumpy that I had to ride at about 5mph. Then when I pulled out on the smooth road and took off the controller failed almost immediately. That was the 3rd time controllers blew in exactly that manner on 3 different motors, even one brushed. Each time was on normal takeoff after someone road around for a while at near idle, probably including very small short pulses of the throttle from bumps. Each time the low speed usage made the controller scorching hot.

John

That is pretty much the same conditions that killed both of my CC ESC.

1:st. Spinning the wheel in the air shortly. Stopped the wheel quick. Opened throttle a little and pooof....
2:nd Going uphill at a bumpy track, 25% load. Wheel bounces some. Shuts the throttle after strange ticking sound. Open the throttle after a while and poof...

I think there is a issue with the start up logic in the RC ESC that kill´s the controllers.
When the wheel is going over bumps the wheel and motor stop/reverses shortly. This forces the ESC to toggle between start up sequence and BEMF commutation. During this rapid toggling something seems to go wrong in the logic and allows FET opening when it not should.

/Goethe

 

[northernmike]

(quietly wondering if any of this applies to my brushed eTek + Alltrax 4834 powered moped conversion... I have the option to change my motor-to-wheel gear ratio, and I haven't committed to a battery voltage. If I were to save my pennies for the new 90C capable LiPo, what would be an ideal end voltage, or, how would I figure that out?)

 

[amberwolf]

All of the controller/motor phase current issues apply to any type of PWM motor control, regardless of brushed or brushless or number of phases. It's actually even easier to see this type of problem with a meter on a single-phase brushed motor, since you can put two CAs in the circuit, one in series with the motor and one in series with the battery, and watch how your phase currents multiply during any type of current-limiting situation.

Now, if you have a battery that does not require any current limiting at any time, meaning it can supply any current the motor asks for, and you have a controller that can do this too, then you won't see the problem happen. The motor will get hot as it actually gets all the current it wants, and it might cook if not cooled properly, but the rest of the system probably won't get hot.

@deVries: Adding anything to actively control power in series with the battery/controller is just another PWM controller, and will have more voltage drop and more heat loss, although in a different way than the motor controller itself. It would act more like a half-bridge brushed motor controller, with the actual brushless motor controller taking the place of the brushed motor. There would still be inductive loading on it, and phase current surges, it's just that these loads would be passed thru the regular controller. I have too little math ability to figure out exactly how it would really behave, though it would definitely introduce more losses in the system, just in a different place.

@John; pretty similar conditions caused sufficient heating in my little 6FET to kill at least one capacitor, although when I heard the cap sizzling I stopped before the controller itself would've died. If I had been on a main street instead of a quiet back street, I wouldn't have heard that, and it'd've probably popped the FETs shortly. I was riding very slow trying to conserve battery power and not pull too much constant current from the NiMH (which can damage them) and not overheat myself (pedalling) or the motor, in the Phoenix heat (well over 100F, I think it was nearly or over 110F, but I can't remember now--it's in the DGA thread). I was also pulling a trailer, and had quite an extra load for the DGA bike for a longer ride, what would have been just under 20 miles or so round trip if I had been able to complete it; I didn't even get half of that.

That heavy load basically put the controller into the same situation as what's being discussed here, as it could never reach a point where current could drop from BEMF.

If I had been able to go WOT with no battery current limiting, the motor would have been the part I'd've worried more about instead, but the NiMH don't like that kind of current draw for any length of time (3C-4C, I think).

 

[rhitee05]

That is pretty much the same conditions that killed both of my CC ESC.

1:st. Spinning the wheel in the air shortly. Stopped the wheel quick. Opened throttle a little and pooof....
2:nd Going uphill at a bumpy track, 25% load. Wheel bounces some. Shuts the throttle after strange ticking sound. Open the throttle after a while and poof...

I think there is a issue with the start up logic in the RC ESC that kill´s the controllers.
When the wheel is going over bumps the wheel and motor stop/reverses shortly. This forces the ESC to toggle between start up sequence and BEMF commutation. During this rapid toggling something seems to go wrong in the logic and allows FET opening when it not should.

I have limited knowledge of the RC ESCs, but I believe they're even more susceptible to this problem than regular eBike controllers. I don't think RC ESCs make any attempt to limit phase current at all. I believe most will limit battery current, but as we've been discussing that doesn't help under take-off conditions or steady cruising at low throttle. They're designed for use mainly with propellers, which have low inertia and maintain high speed, so there is only a very short interval when phase currents will be high. Moving heavy people around at low speeds makes them very unhappy.

There may also be some issues with the sensorless controller maintaining lock under some conditions. A bumpy track sounds like it would give it a tough time. A sensored controller would probably be more robust under these conditions. I don't know if that would cause extra heating in the controller, though, or just be annoying for the user.

 

[rhitee05]

(quietly wondering if any of this applies to my brushed eTek + Alltrax 4834 powered moped conversion... I have the option to change my motor-to-wheel gear ratio, and I haven't committed to a battery voltage. If I were to save my pennies for the new 90C capable LiPo, what would be an ideal end voltage, or, how would I figure that out?)

Generally speaking, higher voltages are more efficient because you can get the same power with less current, and most losses (in both the controller and motor) are related to current or current^2. It also provides higher peak power capability, but you can always dial that back with current limiting. With 90C LiPo, you won't have to worry about drawing too much current out of the battery even if you have a single 5Ah string and it'll still have good energy capacity at higher voltages. I'd suggest choosing the highest voltage your wallet/controller/bravery can handle.

The biggest takeaway from this thread should be to choose your gearing appropriately. Don't gear for 50 MPH if you plan to spend a lot of time cruising at 20 MPH. Cruising at a reasonable fraction of full throttle should keep you out of the controller death zone. It'll usually be the most efficient region for your motor, too.

 

[northernmike]

Generally speaking, higher voltages are more efficient because you can get the same power with less current, and most losses (in both the controller and motor) are related to current or current^2. It also provides higher peak power capability, but you can always dial that back with current limiting. With 90C LiPo, you won't have to worry about drawing too much current out of the battery even if you have a single 5Ah string and it'll still have good energy capacity at higher voltages. I'd suggest choosing the highest voltage your wallet/controller/bravery can handle.

My worry here is that a lot of my riding is in city traffic - MANY stops and starts. Is it a mistake to look at the 18V, 100Ah Peugeot Scootelec as an example? It seems to be the antithesis of this discussion.... but I have no knowledge of it's controller dynamics whatsoever.

What do you guys think about that?

The biggest takeaway from this thread should be to choose your gearing appropriately. Don't gear for 50 MPH if you plan to spend a lot of time cruising at 20 MPH. Cruising at a reasonable fraction of full throttle should keep you out of the controller death zone. It'll usually be the most efficient region for your motor, too.

This is one of the main caveats I have with hubmotors - the inability to regear.

I do wish I could have kept the CVT + clutch of the ICE, sometimes...

 

[John in CR]

Again someone who knows far more about this stuff than I is having a problem (Denito in this case), but only because of the wording used, though I can't believe he's disagreeing with power in = power out + heat loss. The problem is the wording duty cycle going down and phase current going up. More accurate is "When duty cycle is less than full duty, phase current is higher than battery current, and low duty can result in very large phase currents." The situations we are talking about are during acceleration so the duty cycle is increasing but still low, so phase currents are multiples of battery current.


Regarding my controller failure, it's definitely something about the very low speed, possibly some unintentional throttle pulsing on such a rough road, but not speeding up and slowing down of significance. It was very low power riding, less than 100w. The workouts I gave the controller before were while the controller was hot too, not just a single launch. I'm talking about repetitive launches, stops, and repeat. I easily squeezed 30 rides worth of stop and go into a few rides, because I wanted to get comfortable that the controller would hold up before getting too far from the house, since my bike is impossible for me to pedal only for any significant distance. This is the 3rd time this exact thing happened, so it has to be some issue with very slow riding. The strange thing is I've ridden very slow many times with a less aggressive shunt modded identical controller.

John

 

[rhitee05]

My worry here is that a lot of my riding is in city traffic - MANY stops and starts.

So long as you have a controller which is either a) grossly over-rated or b) able to limit phase current, it'll be fine if the limits are set conservatively. Even the approximate phase current limiting the Infineons use is good enough under most situations if the limits are set appropriately.

 

[donob08]

Eric

I do understand that the phase current will change with the speed of the motor.

Iphase = (D*Vbatt – Vbemf)/ Rphase will try to reach infinity when the motor isn’t turning and Iphase will equal 0 when the motor reaches the speed set by D*Vbatt since Vbemf = D*Vbatt at that point.

I’m just looking for a way to hold some things constant and compare Iphase.

We know at 0 speed the phase currents would like to be the very big whatever we set the D to. The big current at zero speed which is limited only by Rphase is pretty hard to provide so the current is limited one way or the other. The battery will probably give up or the controller will pop.

To help me understand, I look at the situation as a function of Vbemf. For a given bike and motor the Vbemf defines the speed. The torque of the motor, weight of the rider and tons of other stuff determine the path to establishing that speed. But for given hardware knowing Vbemf tells exactly the speed.

For D =1 Vbatt = 60 volts the max speed is with Vbemf = 60. We can’t get there if D = ½. The effective voltage is only 30 volts.


So let’s compare at Vbemf = 30

For D =1 I = (60 – 30)/Rphase = 30/Rphase
For D = ½ I = (30- 30) /Rphase =0
The ratio Iphase(D=1/2)/Iphase(D=1)is 0
That’s not a very informative result.

For Vbemf =15
For D =1 I = (60 – 15)/Rphase = 45/Rphase
For D = ½ I = (30- 15) /Rphase =15/Rphase
The ratio is 1/3 = 33%

For Vbemf =10
For D =1 I = (60 – 10)/Rphase = 50/Rphase
For D = ½ I = (30- 10) /Rphase =20/Rphase
The ratio is 2/5 = 40%

For Vbemf =5
For D =1 I = (60 – 5)/Rphase = 55/Rphase
For D = ½ I = (30- 5) /Rphase =25/Rphase
The ratio is 5/11 = 45.5…..%

For Vbemf =1
For D =1 I = (60 – 1)/Rphase = 59/Rphase
For D = ½ I = (30- 1) /Rphase =29/Rphase
The ratio is 29/59 = 49.19%

For Vbemf = ½
For D =1 I = (60 – 1/2)/Rphase = 59.5/Rphase
For D = ½ I = (30- 1/2) /Rphase =29.5/Rphase
The ratio is 49.58%

I think this is saying that the current ratio approaches the ratio of the D’s in the limit as speed approaches 0. But the ratio probably won’t ever be measured as low as Current (with D = ½) = ½ Current (with D =1) because there is a good chance both or one of these currents can’t be supplied by the battery.

So making D small, maybe by an algorithm in the controller, can improve chances. If we go to WOT at a standstill we need current limiting. D can to be set to (Isafe*Rphase)/Vbatt. I’m still not sure what happens in the very first PWM cycle when the current still doesn’t know it’s going to see a voltage effectively reduced by PWM. Luckily that first current pulse is limited by the motor’s inductance. Maybe that’s what saves us.

And of course Ibatt = D*Iphase always as Justin's video so nicely shows.

But, and this is an Important but, Ibatt goes down to around D(squared)Ibatt original. Ibatt goes down that's why we like PWM. So for D = 1/2, Ibatt about = 1/4 Ibatt original

So
Pin = Vbatt*(1/4)Ibatt orig
Pout = (1/2)VBatt * 2 * (1/4)Vbatt original since Iphase = (1/D)* Vbatt
Both voltage and current are down by a factor of 2

 

[John in CR]

My worry here is that a lot of my riding is in city traffic - MANY stops and starts.

So long as you have a controller which is either a) grossly over-rated or b) able to limit phase current, it'll be fine if the limits are set conservatively. Even the approximate phase current limiting the Infineons use is good enough under most situations if the limits are set appropriately.

Yes, I abused my bike for a year and a half in stop and go riding in traffic, but that was a less aggressively modded controller, and that was still with the hard to drive low turn count motor that propelled me at speeds up to 60mph. Just to be sure your combo doesn't have the same issue try a couple of hundred yards of very slow speed on a flat road. Then stop and feel how hot your controller is, and no matter what leave the bike off for 10-15min before doing any normal riding, just to be on the safe side. Do it on the way home to make it more convenient. If the controller is hot, then stay away from that extremely low speed stuff except very short distances.

 

[donob08]

John

I think the first quote you gave is what bothered me. I think the statement 'low duty can result in very large phase currents" would be better if it pointed out that low duty cycle didn't cause "very large currents" it just passes along a reduced version of them if torque and speed requirements wanted them. It might also say "When duty cycle is less than full duty, phase current is higher than the significantly reduced battery current

The problem is the wording duty cycle going down and phase current going up. More accurate is "When duty cycle is less than full duty, phase current is higher than battery current, and low duty can result in very large phase currents."
John