Duty Cycle and the Controller

[eP]

First you want me to drop the discontinuous voltage effects because "you don't need so much precision" and now you want to add new formulas to simulate the effects of the wires?

Yes of course !!

Now you seem to be just fooling around.

No you are totally wrong.

Keep in mind your motor have 25 mohms only at this example.
So if you add next 20 mohms than it totally change your duty and duty average current.

You have to keep in mind that current must flow through controler's switch (FET/FETs ) so you have no magical current multiplication.

Try to do calculations for 10 mohms wires resistance, 2 mohms switch resistance, and 3 mohms capacitors bank Rs (serial resistance).
And dont forget apply internal resistance for 40s3p cells packet at yours simulations of course.

In other way it is completly useless !!

When you include the primary parameters as above it is still will be simple and not precise but far close to reality and valuable.

That's okay, a little fun at my expense is fine... and the "minimal" spreadsheet is an idea I had for a while anyway. It's nice to have a simple version that people can easily figure out.

The "minimal" version is very basic and is "good enough" for most people who just want a "ballpark" answer to how their motor works.
[/b]

Its good enough for people which doesn't know what such simulation do and what don't.
You cannot arbitrally choose (fundamental) parameters and skip other (the same important or even more).

Try to solve simply RL cirquit and skiping R or L.
If you do it you get a nonsense. The same story is above.

Apply all important parameters as i said above and show me example with average tork is 0.001 Nm greater in PWM mode than at continous mode when you pump the same amount of energy from the battery at both cases.

If you could show such fenomena then Nobel Prize is yours. :wink:

 

[safe]

Wiring is sort of a "yes or no" question... as in "yes or no are you using huge wires so that the resistance can be effectively ignored?"

I've actually went through that because I had some connectors that had a lot of resistance and it definitely effects the overall system, but now that I've upgraded to Dean Connectors and very, very thick wire and resistance has dropped to next to nothing.

The Peukerts Effect has been included as a parameter with the battery for those that are using SLA, but beyond that I have not taken it any further. Again, this is that "extra step" that if it makes enough sense to get into I might just do it. It's a bigger factor on bikes that have small battery packs that can barely handle the peak current. I've tended to use a large battery so that this problem is less significant.

There are so many factors that you do end up having to pick and choose which "add value" and which do not. My primary goal has been to simulate the multiple geared electric bicycle.... so my BIGGEST goal has been to be able to get the gearing set right and know what the top speed will be and how much of a hill I'll be able to climb. Accurate range numbers have been lower down in the priority queue...

 

[eP]

Wiring is sort of a "yes or no" question... as in "yes or no are you using huge wires so that the resistance can be effectively ignored?"

I've actually went through that because I had some connectors that had a lot of resistance and it definitely effects the overall system, but now that I've upgraded to Dean Connectors and very, very thick wire and resistance has dropped to next to nothing.

Dont forget that you need flow 429 Amps through these wires.
Its not a trivial task even for very very fat wires.

So wake up man, and go from the sky to the ground.

The Peukerts Effect has been included as a parameter with the battery for those that are using SLA, but beyond that I have not taken it any further. Again, this is that "extra step" that if it makes enough sense to get into I might just do it. It's a bigger factor on bikes that have small battery packs that can barely handle the peak current. I've tended to use a large battery so that this problem is less significant.

So answer me simple question what is greater 429 Amps of load or 150 Amps which can give that pack ?

What you think: what will happend when controler will try draw 429 Amps from that insufficient pack directly ?

Are you still sleeping ??

This problem is still less significant in your opinion ??

There are so many factors that you do end up having to pick and choose which "add value" and which do not. My primary goal has been to simulate the multiple geared electric bicycle.... so my BIGGEST goal has been to be able to get the gearing set right and know what the top speed will be and how much of a hill I'll be able to climb. Accurate range numbers have been lower down in the priority queue...
[/b][/color][/size]

Yes there are many factors ...
and if you want to do usefull simulations you must to know all about.

Your BIGGEST goal is worth NOTHING
completly nothing. Inaccurate is not right word at this place.
Your results are completly FALSE !!

The same way you could simulate 4290 Amps at 2.3% duty, or 42.9 kA at 0.23 %

Somebody must be insane (or must be complete ignorant) to take such simulations seriously.(to think they are valuable or usefull )

I have to ask you again: do you know that your simulated current 429 A must flow through controller ??
What switch could do that ?
Do you realize what size of radiator is needed for such task ?
Such current is hard to swiching for controler at 100 000$ sporty monster vehicle at few thousands cells on board.

And you simulate it for 40s3p pack

Many thanks for such "nice" simulations


Try do it again and you will see the difference and no magical effects.

 

[safe]

Dont forget that you need flow 429 Amps through these wires.
Its not a trivial task even for very very fat wires.

You know that particular simulation you are looking at is actually for a motorcycle motor. That's the "mighty" PMG 132 which a lot of the racing electric motorcycles use. For the stuff that I'm actually concerned with for electric bicycles the amps are much smaller.

It's true, for something as big as the PMG 132 that you have to think about the raw power that goes through the lines.

But that example was done more as a way to show a comparison of extremes...

 

[eP]

Dont forget that you need flow 429 Amps through these wires.
Its not a trivial task even for very very fat wires.

You know that particular simulation you are looking at is actually for a motorcycle motor. That's the "mighty" PMG 132 which a lot of the racing electric motorcycles use. For the stuff that I'm actually concerned with for electric bicycles the amps are much smaller.

Are you still kidding man ???

For bicycles amps are much smaller but motor's resistance are higher.

And primary weakness of your "simulation" is that you have no idea which current are small enough in particular case at particular duty cycle.

It's true, for something as big as the PMG 132 that you have to think about the raw power that goes through the lines.

But that example was done more as a way to show a comparison of extremes...
[/b][/color][/size]

That example show that this way you can do such insane simulation for any even the most insane and stupid combination of asumpions and still lying than yours results are reliable and valuable.

But in fact for the sake of your methodology yours "simmulation" is 0% believable.

When 150 Amps pack is good enough for 429 Amps current, so it is the same level of insanity as assuming 15 Amps pack good enough for 42.9 Amps current.
Your "simulation" don't give you any warning at such insane combinations.
Because it give you only complety unbelievable and useless numbers, nothing more than that.

Have i must remember you again:

But if you get bumps spread out then each bump will get the "full travel" possible. That's what happens in discontinuous mode... it's smooth "plush" voltage conversion. So the result is that you get all the energy transferred and the voltage of the pulse can hold it's full intial value. The value of this is that relatively higher voltage at low duty cycles can carry relatively higher current and therefore a little more torque. The "magic" is that a sort of "sub-band" powerband gets created at extremely low duty cycles.

... at extremely low duty cycles.

So simply show us one believable case with your "magic" effect.
Any smooth voltage conversion.

I dont see nothing magical in yours case above.

I see big big heat dissipation Only, and your very strong avoiding to sum such huge wastes.

Are you still thinking: this methodology is right way to go to any goal ???


I see it is way to get "magical" results only.
I can't belive you can't understand that.

 

[safe]

For several months I actually argued YOUR position and thought the idea of "current multiplication" was insane. The logic goes like this...

On the battery side you see something like: (for example)

The battery can supply a peak of 300 Amps.

48 Volts @ 100 Amps (the controller limit is 100 Amps)

...now when you "chop" up that voltage you end up ALLOWING (since the controller is in the "business" of deciding "how much to allow") enough current to pass through to the motor so as to generate a current on the battery side to equal 100 Amps. But on the motor side it "sees" a lower voltage since it's being "chopped".

Okay...

Now unless you want to create a "perpetual motion machine" the energy on the battery side MUST EQUAL the energy going out the other side. (right?... you have to balance both sides)

So now we get:

BatteryVoltage * BatteryAmps = MotorVoltage * MotorAmps

48 Volts * 100 Amps = 24 Volts * 200 Amps (@ 50% duty cycle)

I know it SOUNDS INSANE, but this is the logic I've been "indoctrinated" into and there are others on this messageboard that will swear that this is in fact how things work. The 5304 Hub Motor simulation creates a chart that follows these principles.

http://www.ebikes.ca/simulator

 

[Matt Gruber]

u both are right!
u won't get more torque than if u bypass it(no magic)
u will get more amps, much like a step down transformer

 

[safe]

http://www.4qd.co.uk/faq/bmnc1.html#curr

Remember motor and battery current are not the same: because our controllers use high frequency chopping, the motor's inductance sustains and smoothes the current so that it is pure d.c. with very little ripple. However the battery current is chopped on and off, only flowing when the motor is connected to the battery. So at 50% modulation (i.e. at half full speed) battery current will flow 50% of the time, so you will measure a battery current equal to half the motor current.

Battery current X Battery Voltage = Motor Current X Motor speed.

So this is an online "reference" that points to this weird "effect" of "current multiplication". It's really more a matter of the battery side "appearing" less. The idea of a "motor side current limiting controller" has been talked about and would be a way to make a more linear torque curve. It really all depends on WHERE you are reading the current. Do you measure BEFORE the controller or AFTER?

 

[eP]

u both are right!
u won't get more torque than if u bypass it(no magic)
u will get more amps, much like a step down transformer

Hi Matt

We dont talk DC-DC step down converter effect.
It is well known.

Fechter wrote:

I'm not really sure what you mean by "PWM effect". To me, it's anytime the controller is at less than 100% duty cycle, the motor current will be higher than the battery current.

As soon as you reach the current limit, the controller will start backing off the duty cycle to keep the input amps below the limit. Output amps begin to rise when this happens.

The ratios of amps / volts / duty cycle, stay nice and linear until the duty cycle gets low enough for discontinuous operation.

Whether continuous or discontinous, the power input is very nearly equal to the power output. There's nothing magic or desirable that happens when you go discontinuous. Losses increase with current and with decreasing duty cycle. Controller losses will be minimum at 100% duty cycle.

And safe answer was:

"Output amps begin to rise when this happens."

This doesn't always happen right away though, the "distortion" that begins to separate the input amps from the output amps is dependant on the discontinuous mode to happen. My point was that the "boundary" upon which the "effect" takes place depends on the configuration of your machine. If your machine has a very large gap between the allowable current (current limit) and the current that the discontinuous state takes place then you DON'T get the effect at the higher duty cycles. Looking at the chart, you just have to think in terms of:

"Where does my current limit intersect with the PWM 'effect'"?

It's possible that with some configurations you could even get some discontinuous operation while in high throttle usage... looking below at the chart... imagine if your "normalized current" matched an equivalent value that corresponded to your controllers "current limit" and it's normalized value equaled 0.10. At a value of 0.10 your duty cycle needs to only be 70% for there to be some intersection. This can happen even at full throttle when your controller limits the current and lowers the duty cycle to resolve the rpm based voltage that the motor is bound by.



As to the "magical" quality of the "effect", there is actual distortion of the voltage when in discontinuous mode. When continuous the voltage remains at a level that is proportional to the duty cycle. But when the gaps between the pulses get large enough the inductance seems to have enough time to "rebound" sort of like a shock absorber. If you hit bumps really fast with a shock absorber the wheel never has a chance to fully extend after each bump. But if you get bumps spread out then each bump will get the "full travel" possible. That's what happens in discontinuous mode... it's smooth "plush" voltage conversion.

So we mostly discus safe's "magic" effect.

Cheers

 

[eP]

http://www.4qd.co.uk/faq/bmnc1.html#curr

Remember motor and battery current are not the same: because our controllers use high frequency chopping, the motor's inductance sustains and smoothes the current so that it is pure d.c. with very little ripple. However the battery current is chopped on and off, only flowing when the motor is connected to the battery. So at 50% modulation (i.e. at half full speed) battery current will flow 50% of the time, so you will measure a battery current equal to half the motor current.

Battery current X Battery Voltage = Motor Current X Motor speed.

It is undenyable.
But you keep in mind that ripple is dependent from average current at fixed frequency and dependent from duty.
And keep also that this ripple is the extra source of heat waste.
Lower duty at the same current and frequency the higer ripple loses.
So the the best is case is 200A @24V at 100% duty (zero ripple), worse is 100A @ 48V at 50% duty
and even more worse is 20A @240V at 10% duty (highest ripple)

So you cant get any extra magical tork when the duty is falling.

So this is an online "reference" that points to this weird "effect" of "current multiplication". It's really more a matter of the battery side "appearing" less. The idea of a "motor side current limiting controller" has been talked about and would be a way to make a more linear torque curve. It really all depends on WHERE you are reading the current. Do you measure BEFORE the controller or AFTER?

Before and after.
When current is huge and you dont use huge smooth capacitors you take the same current from the battery for a while what will kill them.
That also could kill the controller at very low duty.

So instead you get your magic effect you can frying your controller.

 

[safe]

It's been a looooooooong evolutionary process to understand the various things that are going on. At first I just wanted to disregard anything having to do with varations of the current and my first spreadsheets were so simplified that they simply used a constant current limit value. So many people complained about that being not "good enough" so I started to look into the continuous verses discontinuous situation and went through all the education to solve those and realized that their "magic" is mostly non-existent. (it's there, but very small) Later it was brought to my attention that you really need to solve for the motor side current and not just the battery side current and so that different aspect of the controller was included. Again, this makes some effect, but there is not all that much "magic" to it either. There does seem to be a slight increase in torque that occurs at least in parts of the powerband as a result of "current multiplication", but it's probably mostly lost to other efficiency issues.

The "bottom line" as they say is that the really simple charts (the first ones I made) were good "ballpark" estimates for figuring out what gearing to use. The charts have matched my actual bike pretty well and the results are usually within a mph or two of being right on. These newer charts make slight changes that seem to increase the accuracy slightly. The differences are not "huge" no matter what you do.

What was learned was that if your bike can handle full throttle and first gear across the entire powerband that you will never need fear any condition that might take place that might break your hub. (partial throttle never means torque above what full throttle might give)

 

[eP]

I see we are aproaching to agreement now

It's been a looooooooong evolutionary process to understand the various things that are going on. At first I just wanted to disregard anything having to do with varations of the current and my first spreadsheets were so simplified that they simply used a constant current limit value. So many people complained about that being not "good enough" so I started to look into the continuous verses discontinuous situation and went through all the education to solve those and realized that their "magic" is mostly non-existent. (it's there, but very small)

All magic is non existent.
All magic (small or very small - it is non important) is only at quadratic equations when you apply them limitless (for every possible combination of data).
In fact they are good but only for limited data. Some data are out of range when you can apply these equations succesfully (with true resulsts).

So when you found any magic results you should check:
if you skip something important or use something out of usability range - first of all.

When you have done precise and sometimes very complicated checking and it is still teling you have done everything all right, so you can start to telling about magic.
Not before.

Later it was brought to my attention that you really need to solve for the motor side current and not just the battery side current and so that different aspect of the controller was included. Again, this makes some effect, but there is not all that much "magic" to it either. There does seem to be a slight increase in torque that occurs at least in parts of the powerband as a result of "current multiplication", but it's probably mostly lost to other efficiency issues.

Current multiplication occurs, so the increase of tork occurs the same.
But for the cost of efficiency of course. Loses are proportional to current square.

So you can get big multiplication but for the cost of huge loses (big square).

But still you need keep in mind:if you want get resuts ( how far multiplication is at particular case) close to reality you cannot afford for skiping as much important parameters as pack's internal resistance or switch resistance.
Even if you could sucesfully skip them at 99% cases.

That is the base of right methodology.

You can do it for simplicity as hidden fixed parameters. Fixed close to reality is much better then fixed to zero (or skipped).

The "bottom line" as they say is that the really simple charts (the first ones I made) were good "ballpark" estimates for figuring out what gearing to use. The charts have matched my actual bike pretty well and the results are usually within a mph or two of being right on. These newer charts make slight changes that seem to increase the accuracy slightly. The differences are not "huge" no matter what you do.

At 99% of cases you are right, but what about the rest 1% ?
When your sheet dont generate warnings for suspicious data combination, you simply dont know if your actual case is in the that 1% or in others 99%.

What was learned was that if your bike can handle full throttle and first gear across the entire powerband that you will never need fear any condition that might take place that might break your hub. (partial throttle never means torque above what full throttle might give)

If you can handle full throttle at stalling hub when the current goes to maximum.
Right done hub should work with far great margin of safety then controller itself.

Regards

 

[safe]

It's near the "stall torque" that is the greatest potential threat to an internally geared multispeed hub. So my goal had been to produce the "worst case scenario" for torque (most idealistically optimistic) so that I could look at that and compare it to what the hub could handle. The Rohloff hub publishes their "upper limit" as 100 Newton Meters of torque as a maximum to the hub for first gear. The Sturmey Archer hub makes it's first gear a 1 to 1 gear ratio which "should" mean that it can handle high torque loads pretty well. The lower the first gear ratio (when first gear is lower than the chain/sprocket setup) the more extreme the low end torque can get.

So the more inefficient things get in the "real world" the more safety the hub will have against damage...