New approach to brushed motor controller

[safe]

Take another look at the chart again:

If a battery is fixed with a particular AH rating (based on the configuration) then the amp line is essentially flat. You can deviate slightly when you go into discontinuous mode (PWM "effect") but that only happens at low duty cycles.

If you want to MIRROR the motor.... you need to have a dynamic amp battery... (the virtual AH rating needs to vary based on the voltage requirement)

 

[safe]

Fechter measured a peak 4X effect. I don't know, though ,if the output components were beefier that a 20X effect is out of the question. Fechter?

The equations make it clear that the PWM "effect" only occurs in certain situations:

If the inductance is low enough the effect is larger than if the inductance is higher. And the frequency is also a factor. You can play around with it on the spreadsheets... you can get some interesting results.

My point is that the PWM "effect" is limited. Also, it's operating in the part of the motors powerband that is extremely inefficient. If you were to design to actually be able to get the full amps without the costs of efficiency losses (which can be 50% down at low rpms) then you could have a big advantage.

You might be able to pull 15 hp from a relatively small Lithium power supply. (of course your range might only be a few miles at that rate)

 

[safe]

A Flexible Battery (a new max amp potential)

By being able to rearrange the configuration of the battery while riding you can get some incredible amp capacity that one would normally never even dream of with a 1C battery pack.

I did some calculations using the PMG 132 and set the voltage to 36 Volts and then went about comparing a "static" battery with the usual PWM controller verses the "flexible" battery that one could theoretically create.

The results are stunning... this is a chart that compares what we are used to with a 1C battery verses a battery that could be reconfigured:


Note: At 1C this battery can do 40 amps. The PMG 132 can handle a maximum of 200 amps. Since 120 cells of 1.2V NiMh can in parallel produce 10 AH * 120 cells you end up with potential of 1200 amps at peak for 1C at 1.2 Volts. On the other extreme you have 36V and a peak at 1C of 40 amps.

 

[Leeps]

Safe for the love of god please trust us, i promise we are not attempting to mislead you. The pwm effect, if you are talking about a difference between input and output current occurs ANYTIME the duty cycle is less than 100%.

Let me explain it in non mathematical laymens terms. Think of inductance as inertia to electricity. Switch closes current builds up like your in a truck and floored the accelerator, think of the current building like your truck speeding up, think of the force accelerating your truck like voltage. Now the duty cycle time is up switch opens, the current must continue (like a tractor trailor must continue unless something really bad happens) so theres a flyback diode which acts as a switch and allows current to travel around the load and inductor, in a motor the inductor is the load. But the interesting thing is that the battery is out of the loop but current is still flowing in the load. Theres many ways to explain this, Lets use conversation of energy. if you have 10 amps average going in fifty percent of the time you have 5 average amps going in, at 10 volts you get 50 watts. Now on the output you have the same 50 watts but you have those 10 amps all the time, well you dont get something for nothing so you have 50 watts/10 amps = oh boy 5 volts. voltage is divided in two amps go up by two. If the power supply was in continuous conduction your duty cycle is 50 percent. discontinous conduction means its a pain in the ass to figure out the relationship between output voltage and duty cycle.

that is all that the pwm effect is, all it ever did and all it will ever do is give you an easy way to adjust voltage.

Joe

 

[safe]

I've completely mastered the PWM "effect". The equations are included in my spreadsheets and you can verify them if you like. (they are right out of wikipedia)

My idea was not about modifying PWM in any way.

Let's look at some PWM examples:

The PMG 132 has an inductance value of 19 microhenry's. In order to get ANY of the discontinuous mode you need to get a controller that operates in the range of about 1 kHz... which don't exist to my knowledge.

The PMG 080 has an inductance of 2.6 microhenry's. Now all of a sudden our normal controllers that work somewhere between 15 kHz and 20 kHz have an effect almost immediately on the motors powerband. (this is what you see as "normal", but it's really more of a special case)

So ONE way get amps it with a very low inductance motor.

The OTHER way would be to make a battery throttle system that provides all the amps you might need without needing a very low inductance value. (it's a way to get enormous amps without the need for inductance tuning)

Many motors have high inductance values and therefore don't see any PWM "effect" at all. (it's effectively impossible)

Acknowledge this point at least... the PWM "effect" is not a universal for all motor / controller combinations.

 

[xyster]

The equations make it clear that the PWM "effect" only occurs in certain situations:

Safe for the love of god please trust us, i promise we are not attempting to mislead you. The pwm effect, if you are talking about a difference between input and output current occurs ANYTIME the duty cycle is less than 100%.

I've completely mastered the PWM "effect". The equations are included in my spreadsheets and you can verify them if you like. (they are right out of wikipedia)

Safe, all you've mastered is sticking your keyboard in your mouth. Leeps is right -- obviously (conservation of energy) the PWM current effect is in effect at all points and times when duty cycle is less than 100%, not just when the current is discontinuous. If it wasn't, where's all that extra power going? I suppose you'll answer "heat" from the motor and controller. But that's the heat loss due to motor/controller inefficiency already fully accounted for. Step back from your misunderstood, misapplied, cobbled-together equations and you might actually begin to understand what Leeps, Fechter, and everyone else here with a much bigger brain, and loads more experience, is saying.

 

[safe]

In continuous mode the voltage is linear so the amps are also linear. It's only in discontinuous mode that the "effect" of current amplification occurs. Read this: (again, since I've posted it before)

Motor speed control of DC motor is nothing new. A simplest method to control the rotation speed of a DC motor is to control it's driving voltage. The higher the voltage is,the higher speed the motor tries to reach. In many applicationsa simple votlage regulation would cause lots of power losson control circuit, so a pulse width modulation method (PWM)is used in many DC motor controlling applications. In the basic Pulse Width Modulation (PWM) method, the operating power to the motors is turned on and off to modulate the current to the motor. The ratio of "on" time to "off" time is what determines the speed of the motor. When doing PWM controlling, keep in mind that a motor is a low pass device. The reason is that a motor is mainly a large inductor. It is not capable of passing high frequency energy, and hence will not perform well using high frequencies. Reasonably low frequencies are required, and then PWM techniques will work. Lower frequencies are generally better than higher frequencies, but PWM stops being effective at too low a frequency. The idea that a lower frequency PWM works better simply reflects that the "on" cycle needs to be pretty wide before the motor will draw any current (because of moter inductance). A higher PWM frequency will work fine if you hang a large capacitor across the motor or short the motor out on the "off" cycle (e.g. power/brake pwm) The reason for this is that short pulses will not allow much current to flow before being cut off. Then the current that did flow is dissipated as an inductive kick - probably as heat through the flyback diodes. The capacitor integrates the pulse and provides a longer, but lower, current flow through the motor after the driver is cut off. There is not inductive kick either, since the current flow isn't being cut off. Knowing the low pass roll-off frequency of the motor helps to determine an optimum frequency for operating PWM. Try testing your motor with a square duty cycle using a variable frequency, and then observe the drop in torque as the frequency is increased. This technique can help determine the roll off point as far as power efficiency is concerned.

There are also high frequency PWM systems that work. The low versus high frequency for PWM of dc motors describes two totally different approaches. Both are valid. At low frequencies you get a mechanical averaging. When the drive is turned off, there is a momentary spike of voltage that the catch diodes clamp but after that transient dies out the motor is left to freewheel. You will typically hear the motor buzzing. But this is usually pretty simple to implement with small motors and low voltages (remember the motor must freewheel when the drive is off). At high frequencies the inductance of the motor (armature) does the current averaging. This is similar to a switching power supply (or a chopper drive). The catch diodes are more critical here because they carry full motor current a substantial amount of time (not so if you are driving the motor locked anti-phase). High frequency PWM is quite sensitive to the motor properties (inductance). For medium size motors usually 20 kHz or higher frequency works.

http://www.epanorama.net/links/motorcontrol.html

If you got LAZY and didn't read it the point was that if you don't get a "perfect tune" of motor to controller there will be little or no PWM "effect".

These sentences sum up the idea:

"Try testing your motor with a square duty cycle using a variable frequency, and then observe the drop in torque as the frequency is increased. This technique can help determine the roll off point as far as power efficiency is concerned."

 

[xyster]

If you got LAZY and didn't read it the point was that if you don't get a "perfect tune" of motor to controller there will be little or no PWM "effect".

I read that passage the first time, I just read it again, it says zip about there being 'little or no PWM "effect"'. That passage relates to high frequency versus low frequency PWM. That's it. Nowhere does the word "amp" or "amperage" even appear. And furthermore, obfuscating as usual not so adroitly, you failed to answer my question from above:
"Leeps is right -- obviously (conservation of energy) the PWM current effect is in effect at all points and times when duty cycle is less than 100%, not just when the current is discontinuous. If it wasn't, where's all that extra power going? "

 

[safe]

...the PWM current effect is in effect at all points and times when duty cycle is less than 100%

This "can" in some cases be true, but it's not "always" true.

It depends on the inductance of the motor and the frequency upon which your controller is running. It's not that complicated and the formulas (once understood) make perfect sense. Basically the 5304 is designed to work with it's controller so that you do in fact get what you get (the "effect") all the time. But for "the rest of us" the PWM "effect" is sometimes small or non-existent. My motor does not have large amounts of torque at less than peak power. Seriously, I've tried it in practice and there's nothing down there. My inductance value (for the MY1020) must be pretty high. Any value above about 6 Microhenry's and the PWM "effect" mostly disappears.

Here's an example I'm working on right now. The PMG 080 if run with a controller that runs at 20 kHz can barely even climb a hill. But if you could get a low frequency controller matched up with it you could (CHANGING NOTHING ELSE) be able to climb just about anything.

Controller tuning is everything!

 

[xyster]

This "can" in some cases be true, but it's not "always" true.

Well, at least we're heading in the right direction now. Perhaps I'll chalk up our schism to a mild case of autism and find some way to move on with my life...

 

[safe]

You might be confusing "controller frequency" verses "pulse width modulation duty cycle". It's the "controller frequency" that I'm talking about here. The duty cycle is the duty cycle and that's just a percentage of the full throttle. But the frequency has everything to do with the inductance. It's the frequency / inductance relationship that defines behavior.

It's a ratio.

There's another metric out there called that I forget the name of now that defines the time constant for the motor. (is it time constant?) Anyway every motor has a certain "speed" that it likes to swallow up it's pulses. If you match the "speed" well then you get a sort of "turbo" action that we call the PWM "effect". If you miss it with a frequency that is too high then the current reverts to continuous mode and the "effect" of discontinuous mode goes away.

Read that paragraph AGAIN and have the two concepts separate in your mind. (I suspect you have the ideas mingled together)

 

[safe]

Well, at least we're heading in the right direction now.

I figured the moment you realized that the 5304 is superior to many motor /controller combinations because it's in perfect tune you would suddenly like the idea.

Yep, the 5304 seems well thought out and optimized for peak performance... :wink:

 

[Leeps]

Okay first thing. By pwm effect do you mean the fact that input and output current dont match, or are you talking about the fact that your torque curve isnt flat. These are two very very different things, and neither have much to do with discontinous vs continuous operation. continuous conduction and discontinuous conduction have to do with maintaning inductor current or having inductor current hitting 0.Having inductor current hit 0 makes it hard to have good load regulation, thats the reason a lot of switch mode power supplies arent rated to be stable when your drawing less than 10% of the rated load of the supply, the voltage will have a tendancy to go up, In the same breath the impedance of the supply goes up also. think about what that implies for your supercharged discontinuous current mode turbo torque boost. you know ohms law.

As fechter said the only times your going to run into discontinous operation your going downhill or are barely moving, so well leave that as something that really isnt going to apply for serious torque. Torque is proportional to current high torque requires high current and some voltage to push said current into the motor, so usually (exception would be low inductance low resistance, and low switch frequency) your not going discontinuous. On the flip side that combination has a tendancy to blow your controller. The combination makes for very high ripple current, peak current can get out of hand quickly with average current staying modest, if the current limiter isnt fast then the mosfets get stressed. the combination isnt good from an engineering standpoint. Real switch mode power supplies switch in the hundreds of khz, and they have a very real difference between input current and output current, thats the name of the game trying to get the same power out as your putting in, or as close as possible (95% aint that bad my friend).

Anyhow back to the point, i remember how all of this started with you being interested in a non flat torque curve, i suppose somehow along the lines you choked it up to 'pwm effect'.
First realize the torque a motor produces is based on the current no more no less. The current that the motor can draw is based on (voltage in - backemf)/resistance no more no less. remember this we will get back to it later.

Okay now for the current limit. The controller checks current input not current output. current input*voltage input = current output *voltage output because of conservation of energy. Well the two inputs are constants but the outputs are variables. its obvious what happens as we drop the voltage.

Okay so for the situation were the torque is rising. As the motor is slowed it takes less and less voltage to maintain the same current across it due to a lowered back emf, thus the duty cycle is dropping. Well now we are reducing the vout of the equation current in*vin=cout*vout , so cout can go up, this is what is giving you the torque rise. The moment you drop vout something gives, most of the time the result is that the current input drops unless you drive like evil knevil always running into the current limiter.

somehow i think i wrote all of this in vain

Joe

 

[xyster]

somehow i think i wrote all of this in vain

As far as Safe, probably. But your explanations do help me. I start with conservation of energy, and work my understanding down from there. The rest follows, but I'm still weak on the details.
In Safe's incomprehensible universe, energy is not usually conserved (Energy In <or> Energy Out). Hence, he starts from a point far afield of the rest of us, and apparently proves his theories as far as internal consistency, but not external consistency (validity). So Safe claims mastery of this subject; but his mastery fails to extend beyond the realm of his spreadsheet.

 

[jackatfsi]

I'm trying hard to follow this as a newbie.........but the the energy in vs energy out by using the rule E1 = V1 x I1 = V2 x I2 = E2 is not correct when E2 is straight DC and E1 is chopped....E1 is equal to E2 plus the energy in the AC component of the chopped waveform.........this "extra" energy in has to be accounted for........

 

[fechter]

most of the time the result is that the current input drops unless you drive like evil knevil always running into the current limiter.

somehow i think i wrote all of this in vain

Joe

Uh... that should be Stevil Knevil

Thanks Joe.
Sometimes I think my explainations are somehow defective when the point gets missed. It can be challenging. Most of the simulations and graphs only apply to 100% throttle, and there's very little data on system efficiency at partial throttle.

You always have to look at the entire overall system, not just the pieces.

 

[Stevil_Knevil]

most of the time the result is that the current input drops unless you drive like evil knevil always running into the current limiter.

somehow i think i wrote all of this in vain

Joe

Uh... that should be Stevil Knevil

Twist and shout!!

:wink:

-S

 

[Malcolm]

Sometimes I think my explainations are somehow defective when the point gets missed. It can be challenging. Most of the simulations and graphs only apply to 100% throttle

As a newbie I'm learning a hell of a lot from the explanations that you, Joe and others give in your attempts to persuade Safe. I'm sure that counts for many others like me who aren't too keen to display their ignorance or don't know where to begin.

 

[Leeps]

I'm trying hard to follow this as a newbie.........but the the energy in vs energy out by using the rule E1 = V1 x I1 = V2 x I2 = E2 is not correct when E2 is straight DC and E1 is chopped....E1 is equal to E2 plus the energy in the AC component of the chopped waveform.........this "extra" energy in has to be accounted for........

I dont really follow on this. E1 is chopped. E2 ideally should be straight dc but there is some ripple voltage due to ripple current. The output filter capacitor ( left out on motor controllers) filters out the ripple voltage.

I dont really see unaccounted for energy in an ac component. use my example I1 can be 10 amps for half the time, this would go into the equation as if it was 5 amps.

i cant reply until friday but im interested to see where this is going.

Joe

 

[safe]

This is how discussions tend to go on this messageboard:

1. Step One - Assume anything Safe writes is wrong.

2. Step Two - Realize that what I thought Safe was saying was something different than what he was actually saying.

3. Step Three - Accept that what Safe actually said had some merit and is worth thinking about.

(or something along those lines)

There are two separate sets of equations. There are equations that relate to the motor. There are equations that relate to the controller. VERY FEW people have even attempted to integrate BOTH like I have and the result is now I see things (weird truths that are not obvious at first) because once you have the tools to play around you can see things that you can't without the tools. (simulations that use fixed variables for the inductance and controller frequency don't give you enough flexibility to really learn about the subtle aspects of behavior)

What I've learned...

The "big concept" I've realized is the importance of inductance and controller frequency in how it effects the motors torque behavior. This is due to the discontinuous mode current that can effect heavily current limited motor / controller combinations. If your current limit is high you will get very little discontinuous mode effect... sometimes. (laugh :lol: )

The bottom line is the answer to motor behavior is COMPLEX and you really can't make blanket statements about any one variable. It's the entire system that defines the behavior.

So having the combined equations in one place (like the spreadsheets) gives you a unique window into the complexity...

But getting back to where I started here...

It's possible to design a high voltage / low current battery pack and then combine it with a controller / motor combination that is in "perfect tune" so that you can exploit the current multiplication PWM "effect" (discontinuous mode) and convert that high voltage into high current while at low duty cycles. (3 months ago I wouldn't have understood my own sentences!)

But...

You could SKIP the "inductance tuning" altogether if you simply created a variable voltage / amperage battery pack. That was my POINT about all this. You CAN go down the "inductance tuning" route (sort of like choosing to use a turbo charger) OR you could go down a flexible battery option. The real question is about 1C battery drain. If you can create a battery pack that was in parallel when at low voltage you could get far more amps than you would ever need.

The central idea was about the battery.

When you look at the "natural" curves for a motor they suggest a straight line for current going from zero rpms all the way to the no load rpms. All I was doing is pointing out that a BATTERY that behaved in a similiar manner (linear from parallel to series going from bottom to top) would go a long way towards mirroring the motors natural behavior.

The chart below compares the potential torque a motor COULD produce verses the actual torque that a well tuned typical PWM solution would provide.

Finally I have to add that due to the fact that the motor can only hold so much current before it burns up, the NEED for so much current is in real doubt. So in the end the usefulness of having so many amps at your disposal is a real question mark. But it's INTERESTING to think about the extremes of possibilities. For smaller battery packs however, this sort of thinking makes sense.

Get the idea?

 

[Leeps]

Safe your typical torque curve could also be a motor controller with no current limit and somebody holding full throttle.

At this point your just not listening, i guess its nice to know that v=di/dt but it also requires visualizing what all those letters mean to understand the system. plugging them into a spreadsheet helps but you have to start from the ground up, if you dont understand the equation that goes into a graph your not going to understand the graph. I dont think you ever fully comprehended the graph of discontinous current mode, i know you liked the normalized voltage going up the implication that it only happened when the current load was down never fully dawned upon you. Voltage changing rapidly with load, sounds like a high impedance no?

I give up, you have the right to your opinions it just pains me to see misinformation being spread to people that might not know better.

Or i could go down the route of adolescent games and just say, i built a switch mode supply and you havent so HAH! but were more mature than that.

Joe

 

[safe]

I dont think you ever fully comprehended the graph of discontinous current mode, i know you liked the normalized voltage going up the implication that it only happened when the current load was down never fully dawned upon you.

I'm fully aware of the low current load requirement for discontinuous mode. (I made a big point of that a while back) But in some situations when the controller has a very low current limit you can indeed begin to see discontinuous mode pretty early. That's what the "tuned inductance" machine like the 5304 seems to demonstrate.

I really do understand the equations... it's really not that complicated and I'm at a loss for what troubles you.

What's the problem?

What exactly are you unhappy about?

So far I haven't heard any actual factual error that you have uncovered in what I've presented. I'd really like to have ANY ERRORS pointed out so by all means if an error is taking place please correct me!

 

[safe]

Okay now for the current limit. The controller checks current input not current output. current input*voltage input = current output *voltage output because of conservation of energy. Well the two inputs are constants but the outputs are variables. its obvious what happens as we drop the voltage.

Is this the area that you are focusing on?

So let's try some numbers here:

40 amps * 36 volts seen as Input

...

80 amps * 18 volts seen as Output (50% duty cycle)

...so is this what you might call the PWM "effect" from your perspective?

 

[Toorbough ULL-Zeveigh]

VERY FEW people have even attempted to integrate BOTH like I have

I hope you don't seriously believe that.
Do you really think you know something the boyz over at Teslamotors don't? I appreciate your exploring the issue, & while I'm not sure if I fully grasp what your on about, I know enuf that this is all well trodden ground.

You're terminology for some things is 'colorful' to say the least. Different from standard engineering nomenclature. This makes it a little difficult to grasp your meaning sometimes, but applaud your effort & ability to think this stuf through.

Although it was BL, what you're describing sounds suspiciously similar to 'adaptive motor control' that WCL spent 80 MEG$ to develop. Hard to say for sure since they never released any hard data on their system. All I can say is that any benefit to their adaptive whatever-it-is seems to be happening all at the bottom end at times of high current demand.

.Performance Optimization: What Makes the WaveCrest Hybrid System “unique?”

There are three main components to WaveCrest motor controls: the interface, the Digital Signal Processor (DSP), and the power electronics.

The interface converts operator inputs, vehicle or application-specific electrical inputs, or data from an application data link into signals that can be directly read by the DSP. The DSP monitors inputs from the interface and computes appropriate response to those signal based on proprietary control algorithms. Responses are converted into a specific drive sequence for each coil or series of coils in the motor through WaveCrest-designed power electronics. Depending on the application, several drive profiles are used to enable the motor provide optimized, efficient torque and power. The power electronics, at the command of the DSP outputs, adjust the provided current and voltage every three milliseconds. The motor hardware includes sensors that provide relevant and critical feedback signals to the DSP through electronics and control systems that form a feedback loop.

http://www.wavecrestlabs.com/technology/controls.html



No doubt there have been others in the past that have sunk a ton of money trying to follow this rainbow to the pot of gold.
Nothing any of us are doing here is breaking any new ground. We are simply the beneficiaries of recent advances in materials technology, so please don't kid yourself.

 

[Leeps]

well yeah from my perspective that is what im talking about. pwm effect isnt a real term that your going to find in a glossary, i assumed that this is what you were talking about, being that it is the cause for the curved torque curve that upset you oh so many months ago.

And the thing thats upsetting me is that your claiming that were claiming pwm has magical effects that we never ever claimed. ive said it in bold ive said in italics maybe i should say it in a new funky font, all the motor control does is manipulates the voltage. thats all it does and it does it well.

second thing that upsets me is this harping on about discontinous mode, your right bad combinations might give you early discontinous mode operation(perm running on a cheap chinese controller would do it) but your not going to get some magical boost in torque, remember you end up with a high impedance supply, voltage swings rapidly with a change in load. high impedance power supplies arent conducive to performance in a motor. basically your never going to climb up the hill in your discontinuous mode graph.

And tuning the motor inductance to the switching frequency. Theres merit to that. now now dont get too excited, easy puppy. The effects wont be too great in a motor controller system, The major effects are going to be on your ripple current. we can ignore switching loss and core loss these only add up to enough watts to count on your fingers and toes. Sure you might run into discontinuous early, see above. If the switching is too fast your ripple current goes down and you have cleaner power going through, which nobody will appreciate surely not the motor, and your switching loss and core loss goes up, once again not serious a handful of watts lost.

In a smps where your looking for clean power, theres a balance between size, efficiency, and clean power output. Inductor selection and switching frequency play a huge role in this.

thats pretty much it, my two problems pwm isnt magic it manipulates voltage with damn good efficiency (and all that this implies ). And that discontinuous mode conduction isnt going to give you a magical torque
boost.
thats bout it

Joe