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

[rhitee05]

I'm not trying to be stubborn but I think Vbemf will change because the speed changes.

Absolutely true. But don't forget about inertia. If you're sitting still and pin the throttle to 100%, the speed/BEMF will take seconds or even minutes to reach that 100% level. That's many, many, many PWM cycles. The current will increase in milliseconds.

Do we all also agree that with a non unity duty factor the average battery current is the 100% duty cycle battery current current reduced by the (duty factor)squared.

Again, no. As I said above, speed and throttle are not directly linked. Your car doesn't accelerate instantly when you push on the throttle and a bike doesn't either. As far as calculating phase/battery current are concerned, throttle/duty cycle and speed/BEMF are two completely independent variables.

 

[rhitee05]

In principle, a controller with a properly implemented phase current limit should be totally robust against ever blowing a mosfet. If the controller phase current is intelligently implemented, then it would start off near the maxiumum of what the mosfets could take, and as the controller heatsink temperature went up, then the phase current limit would automatically roll back to keep the FETs in a safe operating area. You would always be able to get the max power that the controller could support without ever worrying about cooking it.

A very, very excellent point, Justin. Does anybody know if the standard Infineon/Xiechang controllers have any sort of temp sensor? I know my Headline/Cyclone controller does have a sensor, although I don't know what it does with it. Probably just a shutdown, not a gradual rollback.

 

[liveforphysics]

Also back to a real world example - cruising along at top speed WOT I'm drawing say 30 amps @ 60km/hr. I come to a hill and the current rises to 60-70 amps. If I back off the throttle just a little I drop a little speed by the current drops significantly down to ~40 amps. But in doing so I'm making the controller work harder to multiply the current right ? So I'd be better for the controller to stay at WOT and draw more battery current and minimise the current multiplication ?

I don't know your loaded battery voltage or motor wind or controller battery current limit setting. So if you want to make it easy and a clear example that we can work out here on the forum, go to ebike.ca, click the simulator tab at the top, plug in your setup, put the graph up here for us to see, and we can work the numbers out on what is happening. If you have a substantial drop in motor speed, while only dropping from 60amps battery current to 40amps battery current, then it's entirely possible you are making things harder on the motor and controller by letting up on the throttle while climbing the hill. If the bike only drops maybe 10km/hr speed, but reduces battery current draw by 20amps, then it's possible you're breaking-even, or even lessening the heating of the motor and controller. But it's all just guess work until you go make your graph for us. Chop chop!


How do the % speed limiting modes come into the equation ?[/quote]

If it does it through lowering PWM (rather than a commutation frequency rate lock, or timing retarding), and the controller never gets out of battery current limiting mode because of it, then it's F'ing you. For high voltage setups with low speed limiter setting, they could be getting f'ed really really hard.

 

[donob08]

Point taken, but I thought we were discussing the steady state in our discussions, I phase at steady state etc.

I'm not trying to be stubborn but I think Vbemf will change because the speed changes.

Absolutely true. But don't forget about inertia. If you're sitting still and pin the throttle to 100%, the speed/BEMF will take seconds or even minutes to reach that 100% level. That's many, many, many PWM cycles. The current will increase in milliseconds.

Do we all also agree that with a non unity duty factor the average battery current is the 100% duty cycle battery current current reduced by the (duty factor)squared.

Again, no. As I said above, speed and throttle are not directly linked. Your car doesn't accelerate instantly when you push on the throttle and a bike doesn't either. As far as calculating phase/battery current are concerned, throttle/duty cycle and speed/BEMF are two completely independent variables.

 

[liveforphysics]

Do we all also agree that with a non unity duty factor the average battery current is the 100% duty cycle battery current current reduced by the (duty factor)squared. That is the average battery current is down by a factor of duty cycle because the current only flows during the duty cycle and it is down by another factor of duty cycle because the voltage driving it (VBatt*duty cycle - Vbemf), where the Vbemf is the reduced value caused by the reduced speed, is also down by a factor of duty cycle.

No. A motor could never accelerate if this were true, only maintain. You need to view these as dynamic rather than static systems.

We all agree that duty factor may be pushed below 1 at full throttle because of restraints on Phase current. And I think we all agree that the throttle is less effective, in this circumstance, it can only cut duty cycle from say 1/2 down to 0? In every other way, the system acts the same as when not current limited, right? There is NO amplification of phase current as duty cycle is decreased, phase current decreases, just as it always does. There is a greater decrease in IBatt, as always.

No... This does not happen until you are past the red-line on the graph.... Your example is STILL not for a constant-power-in system, as it is until you pass to the right of the redline on the graph.

 

[liveforphysics]

If you input a power level that would be capable of pushing the motor to 100mph, but you're only going 10mph, you get the resultant current of that power level at 10mph. It doesn't matter if in 5 seconds it will be 20mph, and the BEMF will be higher.

Also, a motor doesn't know a hill is coming, a downhill is coming, or if the hand of God is about to spin it to 100krpm in a split second, or if you're about to slam on the brake and lock the rotor solid. A motor only knows the NOW. All it knows is the current rate that the magnet is passing by that stator tooth at this instant. It does not prepare for an impending increase or decrease in BEMF, it only knows the situation it is in with that instant of flux in it.


A motor is not a historian, nor a fortune teller.

 

[rhitee05]

Point taken, but I thought we were discussing the steady state in our discussions, I phase at steady state etc.

I don't think I ever said I was limiting this to steady-state conditions, but I may not have been clear. In any case, I think the dynamic condition is more important since this is when a controller is more likely to be stressed.

It's also worth pointing out here that BEMF exactly equal to Vbatt*%duty is something of a fiction. This would imply that the motor is producing zero power, which in the real world means that you will not be maintaining constant speed. Motor losses, friction, air resistance, and especially hills all affect the actual steady-state point. You might be maintaining constant speed under full throttle when BEMF is only 25% of battery voltage when going up a steep hill, for example.

 

[Knuckles]

This thread just blew my controller!

 

[donob08]

So, Luke I think you are saying Vbemf will be smaller than when it reaches steady state, I agree. So Iavg = (Vbatt*D-Vbemf)/Rphase will be larger than its steady state value. That is what gives acceleration.
Say the rider changes the throttle hoping to double speed, maybe have D go from 0.4 to 0.8 in his mind, but because there was current limiting D is really going to go from 0.3 to 0.6. There could still be the increase in speed after a delay. Oh, I think I get it. If he is deeply into current limit, D won't increase at all. It is stuck at 0.3. The speed won't increase at all until he gets to a speed where he is not current limited. That make sense. I don't think I ever disagreed with that.
The part of your discussion that said, as I heard it, the phase current will increase was what bothered me. In fact, in this current limit case, the phase current will neither decrease nor increase. D won't change. It will stay just the same until he gets the speed to make Vbemf close enough to V*D to get out of current limit. This, I totally agree with.
This is consistent with your argument to design a system intelligently so Current Limits won't do funny things. In this case make the throttle response un-undewrstandable.
I kept hearing the words about Phase current increasing to keep Power out consistent as V*D changed.

I hope now we are together.

Do we all also agree that with a non unity duty factor the average battery current is the 100% duty cycle battery current current reduced by the (duty factor)squared. That is the average battery current is down by a factor of duty cycle because the current only flows during the duty cycle and it is down by another factor of duty cycle because the voltage driving it (VBatt*duty cycle - Vbemf), where the Vbemf is the reduced value caused by the reduced speed, is also down by a factor of duty cycle.

No. A motor could never accelerate if this were true, only maintain. You need to view these as dynamic rather than static systems.

We all agree that duty factor may be pushed below 1 at full throttle because of restraints on Phase current. And I think we all agree that the throttle is less effective, in this circumstance, it can only cut duty cycle from say 1/2 down to 0? In every other way, the system acts the same as when not current limited, right? There is NO amplification of phase current as duty cycle is decreased, phase current decreases, just as it always does. There is a greater decrease in IBatt, as always.

No... This does not happen until you are past the red-line on the graph.... Your example is STILL not for a constant-power-in system, as it is until you pass to the right of the redline on the graph.

 

[ZapPat]

There's been a suggestion or two here that the Infineon/Xiechang controllers measure and control phase current. This isn't strictly true, they only actually measure battery current, they guess at phase current from other factors, like the pulse width and battery current for a given throttle demand and then limit accordingly.

This is imprecise, although it does provide fairly good steady-state phase current control. Where it fails is when the load conditions suddenly change, hard acceleration, for example. To help protect the FETs under this condition, the Xiechang 116 controllers use a crude shut-down sensor. This just looks for a sudden rise in battery current (caused by the sudden increase in load) and sends an interrupt to the controller, shutting it down. As some have already discovered, this can result in unwanted shut downs if you mod the controller for a higher current and don't mod the emergency shut down limiter circuit.

The emergency shut down limiter is in hardware, so can't be programmed, but it can be adjusted by component changes. By changing the phase current limit in software all your doing is adjusting the ratio between phase current and battery current, under fairly steady-state conditions, at which the controller starts to apply pulse shortening to restrict phase current. We discussed this some time ago on a thread about modded controllers shutting down with sudden throttle application.

Jeremy

It's funny how on very active threads like this one there is often much lost/buried information (I had posted this same point a while back). It reminds me of the Police tune "Too much information"! :lol:

I agree with most of your statement above, Jeremy, yet I wouldn't describe the current override signal as being truely hardware based. As you mention, it most likely uses an interrupt in the MCU's firmware to achieve a fast reaction time, much faster than doing an AD conversion. But I believe there's something else going on inside the MCU other than just an interrupt and instant shut-down. Some people have observed that by just changing the controller model type during programming, we can make this current override signal react more slowly without changing anything at all on the hardware involved. For example, by programming an EB206 as an EB212 (and re-adjusting the current limits by 2X to compensate), it seems that the MCU doesn't cutout as much as before. So it looks like a mix of hardware and firmware involved, but we'll never really know because we'll never get the source code. :|

Pat

 

[ZapPat]

This thread just blew my controller!

You mean it blew your own personnal controller (your mind)?!

 

[liveforphysics]

So, Luke I think you are saying Vbemf will be smaller than when it reaches steady state, I agree. So Iavg = (Vbatt*D-Vbemf)/Rphase will be larger than its steady state value. That is what gives acceleration.
Say the rider changes the throttle hoping to double speed, maybe have D go from 0.4 to 0.8 in his mind, but because there was current limiting D is really going to go from 0.3 to 0.6. There could still be the increase in speed after a delay. Oh, I think I get it. If he is deeply into current limit, D won't increase at all. It is stuck at 0.3. The speed won't increase at all until he gets to a speed where he is not current limited. That make sense. I don't think I ever disagreed with that.
The part of your discussion that said, as I heard it, the phase current will increase was what bothered me. In fact, in this current limit case, the phase current will neither decrease nor increase. D won't change. It will stay just the same until he gets the speed to make Vbemf close enough to V*D to get out of current limit. This, I totally agree with.
This is consistent with your argument to design a system intelligently so Current Limits won't do funny things. In this case make the throttle response un-undewrstandable.
I kept hearing the words about Phase current increasing to keep Power out consistent as V*D changed.

I hope now we are together.

No! lol!
I'm way past bedtime here, so maybe I'm not making sense, but the most simple way for you to see what is happening here to phase current is to look at the graph. Any of the graphs will be fine. The height of the blue line represents the torque the motor is making. This torque is exactly representative of the amps in the phase current. Though the throttle setting is WOT the whole time, the controller doesn't hit 100% duty cycle on the FETs until that line drawn on the graph, at this point, battery current = phase current, to the left of the line, battery current is still the same, but torque is higher, this shows the higher phase currents, to the right of the line, battery current decreases and torque decreases together, because BEMF is now limiting the power the controller can draw, aka, you're out of current limiting, which is NOT the area of operation this thread is about, though everyone and their dog like to make examples of how it behaves to the right of that line, which is fine, but it doesn't help anyone learn about current multiplication (obviously).


EDIT*** here is a graph you can look at while you're reading this last post I'm making before bedtime.

 

[liveforphysics]

Here is the real-world proof/test station for all this motor calculator/current battery/current motor/ torque/ voltage etc etc jazz.

This is me with Justin le and his motor dyno. I'm the sleepy half-retarded sounding guy holding the camera sideways. lol

Now I'm going to sleep, which I haven't done for a while, hopefully for 8hrs, which would be like a dream come true (maybe literally, because sometimes I do have dreams about getting the opportunity to sleep).

[youtube]epIz7P3viio[/youtube]

 

[Doctorbass]

Nice test with Justin!! :wink:

I thought it could be nice to share something real too about my real experience using two opposite motor winding ( 5305 and 5302) and similar test conditions.

Setup no 1:

Speed record Killabicycle project ( summer 2009)
5302 on 20" wheel
18 fets controller 4110 that burst at 172A fluke shunt calibrated with kelvin 4 wire method
100A battery current and 160A phase current
Block time : 1
100V battery
Acceleration: 0-70 kmh in 5 sec

-Result after 4 or 6 accelerations to 70 kmh for test: The controller get very HOT !!!! but not blow


Setup no 2:

Mongoose with new 18 fets controller ( summer 2010)
5305 on 20" wheel
18 fets controller 4110 that burst at 221A fluke shunt calibrated with kelvin 4 wire method
100A battery current and 250A phase current ( 2.5 ratio)
Block time : 1
100V battery
Acceleration 0-60 kmh in 5 sec

-Result after 4 or 6 accelerations to 70 kmh : The controller still cold or warm.. no real feel of any danger or heat on the case


my conclusion: in very close conditions, the 5302 offer similar performances as the 5305 but the 5302 make the controller become very HOT !!! and the 5305 dont.

My understanding is that higher turn winding motor are better but if we want to have both torque and speed and not overheat something.. we must use something more.. ex: like dual winding setup like relay switched stator.

5305 or 5306 make the controller happy but have limited speed... 5302 and 5303 make the controller to overheat easy.. that's why i think we mest be able to switch from one winding to another and make it as fast as a switch!

Any other suggestions?

 

[John in CR]

Ok, now that we understand it, for those of us lucky enough to have programmable Infineon based controllers, how do we go about setting them properly?

Jeremy mentioned a hardward protection from over current. Does modifying the shunt fool that current limiter into a different effective setting? I've only experienced it once, and that was with a 9fet that had 4110's. I got carried away with the shunt mod, and it made the controller push the motor darn near the same launch as with the 18fet 100a controller Luke and I ran. At least for 2-3 seconds it did, before the limiter cut off. I pushed it for a handful of launches easing off the throttle a bit, and promptly let the magic smoke out of the controller. I assume the shunt mod effectively changed the current limiter, because that little controller had no business pushing those kinds of currents even for 2 or 3 seconds, and easing off the throttle a bit for launch still gave a pretty stiff launch without tripping the interrupt sensor. We need to find out what the interrupt sensor limits are for each board.

How do we go about setting the battery side and phase current limits? Is the battery side limit just to protect the battery, and the phase limit is to protect the controller? Assuming a battery that can handle extreme current, is there any reason not to set both limits the same? I realize it won't stop the current limiting and partial duty cycles during hard acceleration, but that's a function of the phase limits. Where does the battery side current limit really come into play?

At what limit should we set the phase limit? It seems that going about it as just a multiple of the current limit as I've seen people talk about is incorrect. Jeremy says it's imprecise, so it's not like the voltage limits were we seem to be able to go right up to a finite voltage limit without worry. That means we can't just multiply the current limit of the particular fet by the number of high side fets for a phase, but what kind of margin do we need to maintain to play it reasonably safe?

John

 

[swbluto]

Nice test with Justin!! :wink:

I thought it could be nice to share something real too about my real experience using two opposite motor winding ( 5305 and 5302) and similar test conditions.

Setup no 1:

Speed record Killabicycle project ( summer 2009)
5302 on 20" wheel
18 fets controller 4110 that burst at 172A fluke shunt calibrated with kelvin 4 wire method
100A battery current and 160A phase current
Block time : 1
100V battery
Acceleration: 0-70 kmh in 5 sec

-Result after 4 or 6 accelerations to 70 kmh for test: The controller get very HOT !!!! but not blow


Setup no 2:

Mongoose with new 18 fets controller ( summer 2010)
5305 on 20" wheel
18 fets controller 4110 that burst at 221A fluke shunt calibrated with kelvin 4 wire method
100A battery current and 250A phase current ( 2.5 ratio)
Block time : 1
100V battery
Acceleration 0-60 kmh in 5 sec

-Result after 4 or 6 accelerations to 70 kmh : The controller still cold or warm.. no real feel of any danger or heat on the case


my conclusion: in very close conditions, the 5302 offer similar performances as the 5305 but the 5302 make the controller become very HOT !!! and the 5305 dont.

My understanding is that higher turn winding motor are better but if we want to have both torque and speed and not overheat something.. we must use something more.. ex: like dual winding setup like relay switched stator.

5305 or 5306 make the controller happy but have limited speed... 5302 and 5303 make the controller to overheat easy.. that's why i think we mest be able to switch from one winding to another and make it as fast as a switch!

Any other suggestions?

My guess is that even though the phase current limit for the 5302 was lower, it was phase limiting the motor for a longer period of time since the motor has a far lower resistance, so your 5302's controller was PWMing for a much longer period of time which means it was handling switching losses at 160 amps for far longer. At 160 phase amps, you should ensure your controller is properly heatsinked (Add a few cooling fans or CPU units if you have to - the closer to the mosfets, the better) if you plan to do it "long enough". However... if this was like a 10 second hard launch, a CPU heatsink may have limited effectiveness.

 

[John in CR]

....Any other suggestions?

X5302- actively ventilated and iced down before the run.
200V600A HVHC controller
Just enough of those new 40-90C Lipos to get the voltage and current and distance.

If that's not enough, put another 5302 in the front wheel.

I'm trying to work up the balls to try option 2, but I don't have to go to 200V, because the no load speed of my motors at 140V in a 23.5" wheel (17" moto tires I have) is 156mph (250kph). I commuted by moto for almost a decade, but rarely went above 50-60mph because I just didn't care for it. I need to stop being a chickenshit and just go for it to set a number out there for you guys to break before you guys set one I don't want to try topping.

John

 

[Hyena]

make your graph for us. Chop chop!

On it boss!

torque is exactly representative of the amps in the phase current.

aahhhhh.... now the graphs and your comments make more sense!

My motor is a GM1000 which isn't in the simulator but it's the same wind as the 9x7 so I've used that.
Controller limit is 80 amps, 18S lipo so 75v off the charge but I've used 70v as it hovers around that for most of the discharge.

 

[liveforphysics]

This is running a lipo pack right?
Change that 0.2 battery resistance to something like 0.04, then regraph. It should extend the current limiting point out another 5kmh or so perhaps from batt voltage staying up better.
The knee of the graph comes soon for this combo after that knee where you see the blue line kink? Letting off the throttle will decrease the heating of the controller/motor after that point. So, keeping it past that speed minimum for hill climbing will keep the motor and controller most happy

You have well selected battery/controller/voltage/motor choices here. Your setup will only be in current multiplication for the times you're under 25kmh(maybe 30 with correct batt resistance), and at speeds above that it behaves intutitively. Great real world example for us, thank you Hyena!

 

[Hyena]

Thanks man
Yep 18S lipo that comes off the charge at 75v. I used a voltage of 70v as I wasn't sure exactly how the simulator allowed for nominal voltages and sag but I guess now with the battery resistance of 0.04v I can recalculate it more accurately using the nominal voltage of 67v.

So I guess from the graph I can now assume I'm going to be drawing phase currents of ~150a as I move off the line. Is this too much for a 12 fet with 4110s ? In theory should it be able to handle spikes of 200 amps ? Or does the 4x battery current thing come in to play ? I guess at 1/4 throttle it'd only be drawing 30-40 amps max which even at 4x still brings it back to ~150a max phase current.

Now to include another variable, what about block time ? I guess you probably should allow for that when specifying your controller current ?
Eg. My 80 amp controller has a block time of 3 seconds which allows short spikes of up to 100 amps. Off the line this is where phase current is the highest so it could actually be 165a according the the simulator regraphed for 100a...

 

[liveforphysics]

With any TO220 fet, even if it was magic and the RdsOn was zero mOhm and the silicon could handle a zillion amps, your still up against the continous current package limits of something like 70-75amps per FET. So, in this case, it looks like you're right about at the max you want to be to stay reliable if your block time was set to 0. Set to 3 seconds, you could definately toast your controller doing repeated low speed acceleration bursts. Otherwise, it looks like a well selected and setup system, and should work out well IMO.

 

[John in CR]

What's wrong with multiplication as long as you're not blowing the controller? I love it. You're saying Hyena has a good match, and I say having acceleration falling off like that above 27kph sucks wind. I say keep that knee at or above your typical max cruising speed if you want good performance by going to higher voltage. Who cares if a controller runs a little warm? They can take it if the settings are proper.

Still no takers on why not set the batt side current limit the same as the phase current limit as long as the battery can take it?

and

What kind of cushion should we include in our phase current limit setting? Is the starting point the max current for that type of mosfet X the number of low side fets per phase, and then reduce that by what kind of factor since Jeremy stated that is not a very precise limit?

 

[rhitee05]

Still no takers on why not set the batt side current limit the same as the phase current limit as long as the battery can take it?

I don't see any reason why that would be a problem. The phase limit protects the controller and motor, the battery limit protects the battery. It can be as high as your battery can handle, but it obviously won't do any good to make it higher than the phase limit.

What kind of cushion should we include in our phase current limit setting? Is the starting point the max current for that type of mosfet X the number of low side fets per phase, and then reduce that by what kind of factor since Jeremy stated that is not a very precise limit?

As LFP already said, IR rates the TO-220 and TO-247 packages at 75A per FET. That's a continuous limit based on heating, so short and widely spaced bursts a little higher than this would be okay (say, maybe 100A for a few seconds). It should be safe to pull 75A per FET on a continuous basis, but it never hurts to de-rate a bit for more reliability. For odd-FET controllers, a safe limit is probably halfway between (so, for a 9-FET, around 1.5x75=107.5A). The upper FETs do see the full phase current, but only for a partial duty cycle so they can be driven a little harder.

 

[John in CR]

I don't know what he meant by TO-220 and TO-247 packages, apparently EE lingo, though I do see you guys throw the term "packages" around pretty often. I seem to remember looking at the data sheet on the 4110 mosfets and saw something like 115A. Also, MWKeefer mentioned to me once that the common mosfet in controllers that has a voltage limit of around 83v was good for 100A, though I don't remember the part number.

John

 

[Hyena]

With any TO220 fet, even if it was magic and the RdsOn was zero mOhm and the silicon could handle a zillion amps, your still up against the continous current package limits of something like 70-75amps per FET

Cool, I might back it off to 60 amps battery current and mod the shunt to allow bursts of 80a with a 3 second block time. That should still yield good performance but hopefully keep the magic smoke in. Jeremy says the phase current isn't accurate as such in these infineon controllers, but what would you set it to in this situation? If I set it to 100 amps is the controller still going to try and multiply it up to 150 odd based on the simulator ?

You're saying Hyena has a good match, and I say having acceleration falling off like that above 27kph sucks wind

Don't be fooled by the graph, it pulls like a freight train up to ~60km/hr then tapers off as it gets to 70. I have alot of confidence overtaking slow moving traffic at 40km/hr by moving into the middle lane. I wouldn't want to do that on a lesser powered bike.