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

[rhitee05]

Yes, it sounds like you understand my point now. PI control is generally very flexible, and a well-tuned set of parameters should provide reasonable performance over a broad range of loads. A little tuning would be required for optimum performance with any given system. If the controller is smart enough, self-tuning is possible without any user intervention.

There are a lot of app notes out there which explain different methods. I was able to find one with this helpful diagram:


What most controllers implement is mode 2 - direct control of duty cycle with no feedback (except limiting for current). What I was talking about is mode 3, where the input is a current demand. The system shown in mode 1 would be good to implement for cruise control. The user supplies the desired speed, the outer loop demands the current necessary to maintain that speed, and then the inner loop sets the duty cycle.

Perhaps the optimal controller design would implement this double-loop system. My default the throttle would bypass the outer loop and control the inner loop directly, providing a torque-based throttle. The user would then have the ability to switch so the throttle inputs to the outer loop for speed-based control, or use the outer loop for cruise control.

For reference, the diagram came from this app note:
http://ww1.microchip.com/downloads/en/appnotes/Sensorless BLDC 00901a.pdf

 

[donob08]

Eric

I guess the BIG Question is when will someone make a control with feedback so that Current is a function of Throttle. I would think the big gun motor guys would happily pay.

I have no horse in this race. My little system gets me around on my errands with an acceptable level of sweat.

My interest in this was only tweaked by what seemed like the bad idea of current clipping.

I was also kind of bothered by the fact that for two currents which have the same peak level but where was is intermittent, the continuous one is said to be a multiple of the intermittent one rather than saying the intermittent one is a divided down version of the continuous.

I have the feeling you are an EE. Did the school you attended use this language?

Last thought on this. We seemed to overlook the fact that the controller can figure KV pretty easily. It sees Hall counts for speed and it pretty well knows Voltage from D and Vbatt.

have fun, Don

 

[rhitee05]

From what I understand, some of the big-boy controllers (perhaps even most) do implement current-based control. I'm talking about car- and motorcycle-class controllers mainly. The cheap Xie Chang controllers seem to be practially the de-facto standard on bikes and they do not implement it. It's not very difficult and doesn't require a lot of extra processing power, but there might not be any incentive for them to put in the effort. The controllers I eventually plan to build will definitely use this method.

As far as the current multiplication thing, all I can say is that it's an analogy and analogies are inherently imprecise. It sounds like you've gained a good understanding of what goes on, so in the end that's more important than terminology. FWIW, I believe current multiplication is the usual term of art. It really only makes sense in terms of averages, so that was probably the source of your confusion. There's a lot to be found on the 'net about buck converters and it's worth reading - everything you read about a buck converter is applicable to motor controllers with a few minor exceptions.

Yes, I'm an EE. I took a power electronics class, but I don't really what (if any) terminology was used there. As I said, the concept is more important than the path you take to get there. Although, understanding terminology is important to be conversant with other people!

And yes, in theory you're correct that the controller can figure out the motor's Kv, although there are a couple of practical difficulties. First, I don't know of any controllers that are actually configured to read this (obviously not an issue with a custom controller). Sensorless controllers do need to monitor the phase voltages for control, but they only need to look for a zero-crossing event so I expect most use a comparator rather than measuring the voltage. Second, it's a very noisy environment to get good measurements. You have to time sampling to work around the PWM pulses, and you might miss the BEMF peak. I'm sure it's doable, just not quite trivially.

 

[donob08]

Eric

I'm an EE too. I don't think I've developed a lot of new understandings of concepts. But I do have an enlarged appreciation for the value a Glossary could have in situations like this. I think terminology was right up there with the biggest problems in this discussion. I guess new understandings of how words are used is a new understanding.

I think terminology hid from some people that Iphase with PWM is something like D*Iphase (with D a fraction less than 1)what it would be without PWM and it is only multiplied compared to Ibatt because Ibatt with PWM is something like Ibatt*Dsquared (so Dsquared is a smaller fraction) it's value without PWM. In this case terminology misleads.

enough, We can't fix it.

Don

 

[liveforphysics]

So if you draw say 20amps from the battery, and you're sending 60amps to the motor you think it's improper to call that current multiplication? If you have better nomenclature for what is happening, please share it with us.

 

[Toorbough ULL-Zeveigh]

If you have better nomenclature for what is happening, please share it with us.

won't call it better, but perfesser emeritus had his own nomenclature to go along with his 'theories' when searching the arkive.

 

[swbluto]

If you have better nomenclature for what is happening, please share it with us.

won't call it better, but perfesser emeritus had his own nomenclature to go along with his 'theories' when searching the arkive.

Well, it arguably isn't the best terminology because if PWM weren't occurring (i.e., full throttle or absolutely no current limits), the phase currents would be even higher. So, to make it "multiply" seems like it could mislead a few as it's decreasing the phase currents. Indeed, this is the impression I've got so far from recumpence and a few others.

I don't have a better suggestion though. Maybe, "battery multiplication" rather than simply "current multiplication" to make it more obvious which currents are being multiplied...

 

[TylerDurden]

:doubleFacepalm:

 

[donob08]

Luke

I don't have a one word description. But what people should understand is that when you start PWM the phase current becomes a fraction of it's former value and since the battery current is intermittent it is divided by the same factor again. So if you look at the causal path, not the result, PWM divides current in one case by 1/D in the other by 1/Dsquared. I think the term multiply hides the real fact. I don't offer a replacement, I'd just get rid of the misleading term. No current anywhere is multiplied by PWM. Both currents are divided down.

Don

So if you draw say 20amps from the battery, and you're sending 60amps to the motor you think it's improper to call that current multiplication? If you have better nomenclature for what is happening, please share it with us.

No, Luke you never start out without PWM drawing 20 from the Batt and after PWM begins send 60 to the motor . If you use PWM you will be sending a fraction of the Original batt current to the motor. Your words are misleading to me. Because they don't talk about the change in current that PWM caused. They just talk about the relationship between two currents.

I guess to me it's all about the change PWM makes. Not the relationship between Iphase and Ibatt.

 

[donob08]

Maybe "Power Division" would be a good term.

Before PWM

Power Batt out = Batt Voltage * Batt Current Original
Power motor = same

After PWM simple case D =1/2
In this case we'll look at Power motor since it is determining things.

Power Motor = Batt Voltage/2 * Batt Current Original /2
Current Batt out = (1/2) * Motor current

Power Batt out = Batt Voltage * 1/2*Motor Current
Power Batt out = Batt Voltage * 1/2*Batt Current original / 2
Power Batt out = Batt Voltage * Batt Current original / 4
And rewriting Power Motor = Batt Voltage * Batt Current original / 4

So the Important thing is that we have cut power to 1/4 its original value, forget about the ratio of Iphase to Ibatt. That only tells half the story since the voltage is being cut. We could have just as easily called the process "Voltage Division".

 

[rhitee05]

No, Luke you never start out without PWM drawing 20 from the Batt and after PWM begins send 60 to the motor . If you use PWM you will be sending a fraction of the Original batt current to the motor. Your words are misleading to me. Because they don't talk about the change in current that PWM caused. They just talk about the relationship between two currents.

Besides the fact that Luke never talked about before/after PWM, that comparison isn't relevant. No one here has been comparing what happens under PWM to what happens without. Except for the trivial case of 100% duty cycle, PWM is always happening. Any comparison which ignores PWM is simply not applicable because this discussion is about what happens during PWM.

I guess to me it's all about the change PWM makes. Not the relationship between Iphase and Ibatt.

I guess you've missed the entire point of this discussion, then. The relationship between Iphase and Ibatt is exactly what we're talking about.

 

[donob08]

Eric

My point is the question"Should the ratio of Iphase to I batt be what we are talking about"? We might as well talk about the ratio of Vphase to Vbatt. Then we would call it Voltage division. The "current multiplication' only matters to the wires connecting things to the controller. Everything else cares about what we have done to power.

Don

 

[rhitee05]

The phrase "current multiplication" refers to the ratio of phase and battery currents, so yes that is what we're talking about. The original context here is that phase current is the important parameter for torque output and heat generation. Most controllers only measure battery current, so the discussion was about how the two were related and how current limiting works under various conditions.

Going back to Luke's original post:

It explains a critical element of putting together a battery/motor/controller system.

The phase current and battery current relationship, and how and why phase currents are multiplied.

Why does this matter to you? Because battery current is never seen by the motor, only phase current, and phase current is what determines torque output of a motor, as well as the heating the FETs see.

 

[donob08]

Eric

I agree. Phase current is important. I'm just concerned that all this talk about "current multiplication" can cause people to miss the point that Iphase goes done when PWM is used. If the issue is totally aboout measuring Ibatt to infer an Iphase then, FOR SURE we need to consider "Current multiplication". I think we need to avoid confusing people about Iphase changes. Doing both, sounds great.

Don

 

[liveforphysics]

So, as far as I'm aware, nobody on here sees any measurement other than battery current displayed on their bikes.

The purpose of this thread is to show people that during current limiting, the current value they see displayed is not representative of the current load on the motor or controller. It is instead some higher value (some multiple >1).

The secondary purpose was to give an idea of the factors that impact how much the current gets multiplied. This enables people to make better choices of posts when chooseing the parts for a drive system.

 

[ZapPat]

I think the Don's issue isn't so much of technical nature, but more of one of human nature. I guess we might call it ego or whatever, but it can be hard to admit that we were confused after debating a point at length. When we finally see we were not understanding the issue, it can be very hard to admit it and thank the other people for teaching us. It's just one of the many

Imagine that Don started this thread thinking that it was impossible for the current output of a controller to be higher than the input current, and even further that all the PWM "off time" energy was wasted as heat in the controller! He has come a long way in his understanding of how motor controllers work... so now have the humility, Don, to thank the people who helped you understand something new instead of nit-picking the terms used.

Cheers,
Pat

 

[donob08]

Guys

I'm totally happy saying current multiplication happens in PWM. I would just like everyone to be clear that is a multiple of a much REDUCED current so that the Iphase is equal to Ibatt original *D where D is a fraction less than one. I was concerned that the multiplication made people think the current got BIGGER. If we are all clear on that. I'm happy.

I think saying Current Multiplication without mentioning the other things that happen can lead people astray. It just doesn't describe all that happened. Using Current Multiplication applied to measured Battery current to get Iphase is good. The measurement will show that Ibatt has fallen to Dsquared*Ibatt original. Then we can see, using Current Multiplication, that Iphase is equal D*Iphase original. This will make it all clear to everyone.

Don

 

[ZapPat]

Don't ber a troll, Don.

We do get the point that PWM also limits battery current too, and that it is even smaller than the phase current... Come to think of it, that's the whole reason this thread exists, right? It's because LFP was trying to make people understand that there was more current flowing out of the controller than we might think when looking at only the controller's input (battery current), and why this is important to understand.

From Luke's opening post on this thread, transcribed from PWM explanation image:

Since power in roughly equals power out of the controller, and power in equals voltage*current, then anytime PWM is used to regulate battery current, average phase voltage drops and average phase current increases to balance the power equation.

Let it rest.

 

[donob08]

Your quote says just what we need to correct. It says phase current increases. It does NOT. Phase current is divided down. But because battery current is divided down more, phase current becomes a multiple of battery current. This misunderstanding or misstatement by Luke is just what I think we need to clear up.
The fact is the power into the contoller is reduced dramatically by the much reduced battery current so BOTH the voltage AND the CURRENT of the phase are reduced. For D = 1/2, Ibatt becomes Ibatt original/4. Vphase is divided by 2 and Iphase is divided by 2. So we have power in/out balance but at 1/4 the original level.

Don't ber a troll, Don.

We do get the point that PWM also limits battery current too, and that it is even smaller than the phase current... Come to think of it, that's the whole reason this thread exists, right? It's because LFP was trying to make people understand that there was more current flowing out of the controller than we might think when looking at only the controller's input (battery current), and why this is important to understand.

From Luke's opening post on this thread, transcribed from PWM explanation image:

Since power in roughly equals power out of the controller, and power in equals voltage*current, then anytime PWM is used to regulate battery current, average phase voltage drops and average phase current increases to balance the power equation.

Let it rest.

 

[donob08]

I know I don't explain things in a way that many of you find understandable. Let's look at data.

Set your bike up with the wheel off the ground

Using CA or whatever measure/record the battery current at full throttle

Repeat the process but with 1/2 throttle.

Look at the ratio of Ibatt(half throttle) to Ibatt(full throttle)

I found at about half throttle Ibatt was about 1/4 its value at full throttle. That let me accept that both Vphase and Iphase were cut in half. I could apply the "current multiplier" concept and say Iphase must be at 1/2 its WOT level.

The argument that when Vphase goes down Iphase has to go up would only be true if Ibatt didn't change. But it does and the power in changes.

 

[liveforphysics]

The argument that when Vphase goes down Iphase has to go up would only be true if Ibatt didn't change.

Hence the entire f*cking point of this thread about what happens DURING CURRENT LIMITING!

Can I make it any more clear? It does not apply to being out of current limiting. There are loads of other effects that can happen when not in current limiting, hell, you can be sending current from the motor back into the batteries, or be totally off-throttle, or be running at no-load speed down a hill, or in regen braking, or plug braking, or hundreds of other conditions, but this thread isn't about those conditions, it's about people learning that when in current limiting ( like 90-100% of time spent riding for some people, maybe 30-70% for a normal ebike), the phase currents are some greater multiple than battery current. Is this impossible to accept?

 

[Tiberius]

The argument that when Vphase goes down Iphase has to go up would only be true if Ibatt didn't change. But it does and the power in changes.

Except in the control regime where Ibatt is constant.

This is what happens when the the motor is on load and controller decides to limit the battery current. The wheel off the ground experiment is different.

I refer the honourable reader to page 5, where Eric (rhitee05) posted the math; all the pages where Luke (lfp) posted the graphs and page 6, where I listed the control regimes.

Nick

 

[donob08]

Luke, Tiberius

I agree the Current Limit situation is different. And more importantly, I need to work on remembering that we are talking a specific instance here. There, I think Tiberius' words that Iphase = Power In /Vemf at that time make things clear. I don't think it is because Vphase is getting lower that I phase gets higher as Luke's quoted comment suggests. It is because Vbemf is too low. I accept that. I understand that. I'm not sure everyone understands the concept of Power in and Vbemf taking control of Iphase. So we/they don't know seat of the pants on bike seat, when when the normal PWM rules no longer apply. I understand the graphs I just can't get it to be intuitive.

I continue to be concerned that we are misleading people about the way PWM works for them when it's fulfilling its ideal role. Granted the time it is in that role varies.

Don

 

[donob08]

I think I’m on the right page now and I figured a measure of what I phase will be as a function of Ibatt limit, the limited battery current.

If I’m correct then when Vbatt = 100 volts and Rp = 0.1 ohms and we set Ibatt limit to 50 amps, the phase current at low or no speed will be 223.6 amps. I got this number from multiplying SQRT (Ibatt limit /( Vbatt / Rp ) ) times Vbatt / Rp . The logic behind that is below. Basically it uses the known relationship between Ip , DVbatt , Vemf and Rp . Added to that is the idea that as lack of Vemf drives the Ip up, the current limit logic will drive D to a smaller fraction so that D times Ip the battery current will stay at the battery current limit.

Seeing how bad this is, as Luke has been pointing out, I see one way to make things better. If the battery pack were set up with multiple taps: one for full voltage, one for half voltage and one for quarter voltage and if the controller could be made intelligent enough to fire the relays to switch between the different taps as a function of either Ip or Ibatt, since they have a known relationship, then Ip could be controlled. The controller would have to change LVC as a function of which relay it fired.

The other thing this says to me is that high power systems should have variable mechanical gearing. If the system were put in low gear at low speeds the Vemf would be higher since the motor speed would be higher. Not only would this save the battery from the big current draw, it would save the motor from heating due to the high current.


Ip = (DVbatt - Vemf )/Rp

D = Ibatt limit/Iphase

I p = ((Ibatt limit /Iphase) * Vbatt - Vemf )/Rp

Multiplying by I p

I p squared = (Ibatt limit / * Vbatt - Vemf * I p )/Rp

I p squared + Vemf * I p /Rp - Ibatt limit* Vbatt / Rp = 0

Solving the quadratic

I p = - Vemf / Rp +/- (sqrt(Vemf/ Rp)squared - 4 Ibatt limit * Vbatt / Rp)/2

If we express Ibatt limit as a fraction of Vbatt / Rp say:

Ibatt limit = C * Vbatt / Rp

I p = - Vemf / Rp +/- (SQRT ((Vbemf/ Rp)squared - 4 C * (Vbatt / Rp)squared )) / 2


For Vemf = 0, I p = SQRT(C) * Vbatt / Rp

 

[rhitee05]

I have to admit I have no idea where you're going with this.

Seeing how bad this is, as Luke has been pointing out, I see one way to make things better. If the battery pack were set up with multiple taps: one for full voltage, one for half voltage and one for quarter voltage and if the controller could be made intelligent enough to fire the relays to switch between the different taps as a function of either Ip or Ibatt, since they have a known relationship, then Ip could be controlled. The controller would have to change LVC as a function of which relay it fired.

This is pointless. The controller is a buck converter. It's designed to convert a higher DC voltage into a lower DC voltage. Why do we need relays to do this when it's part of the controller's intended function?

Phase current limiting works. We don't have to re-invent anything. The main purpose of this thread was to emphasize that it's important to set the limit to an appropriate level.