New approach to brushed motor controller

[Leeps]

I was thinking about the way a motor controller works. Fixed pwm period with an input that adjusts the duty cycle. We all agreed that its the easy way to do it but a current mode throttle would be better, this requires added circuitry or does it. What if we did away with the pwm generator (the triangle wave and comparator) and had instead two comparators one that got the throttle signal the other that got a divided throttle signal as the reference. The shunt voltage gets amplified and goes to the other inputs on the comparator. The comparators serve as trigger points, the mosfet turns on when the lower signal comparator turns off and turns off when the upper comparator turns on. This would automatically adjust the pwm period to just what is required for the specific motor based on ripple current and motor inductance, the ripple current also is adjustable by the difference in reference voltage between the two comparators.
What do you guys think, am i just reinventing the wheel, i really like the idea of a controller that is purposely designed to be current mode from the ground up.
Joe

 

[fechter]

Seems like it should work. I don't see why they don't do something like that anyway, it wouldn't add much to the cost. It could automatically compensate for different motor inductances too. I think I've seen some switching power supplies that work like that.

 

[safe]

What if the throttle was the battery pack?

What if you eliminated the PWM altogether and simply designed a battery pack that had relays on every cell that could either be on (allow the cell to be included in the pack) or off. (exclude the cell)

You would have a degree of "gradularity" to your throttle because each increment would equal the cells voltage. So your "digital" throttle settings would be for Lithium"

3.6
7.2
.
.
.
...whatever number of cells you have.

Maybe you could even solve balancing and charging problems while you're at it?

 

[safe]

How about keeping the Series - Parallel pack design with the (large) parallel "subpacks" that gave you certainty that you never exceeded the current limits of the cells and then simply had a relay for each "subpack".

At 72 Volts that gives you a "granularity" of 20 settings which should be "good enough".

You would probably need some sort of randomness on which "subpack" to bring "online" because otherwise you would always be using the first "subpack" and only sometimes use the others.

How about a selection algorithm that selects the "subpack" with the highest energy "first". That way they "rotate" evenly. When the throttle is reduced the "subpack" with the lowest energy is released first. So it all balances dynamically.

Basically you always draw from the strongest "subpack" and release quickest the weakest "subpack".

You would always get 100% performance... which is nice...

Current limiting would still be needed however (to safeguard the motor).... hmmmmmm...

 

[fechter]

That's pretty much what they did back in the '20's before they had transistors. Resistors were also used to avoid a huge inrush current when you switch. Gives sort of a jerky ride, but it does work and it is efficient if you don't use the resistor much.

Such a system would also be EMP resistant :wink:

 

[safe]

That's pretty much what they did back in the '20's before they had transistors. Resistors were also used to avoid a huge inrush current when you switch. Gives sort of a jerky ride, but it does work and it is efficient if you don't use the resistor much.

Such a system would also be EMP resistant :wink:

Back then the cell voltages were probably more "granular" (bigger voltage jumps) than the small 3.6 Volts we have today.

With so many little increments you would barely notice it.

Is there any way to prevent the current from going overboard other than a resistor?

Isn't this a little bit like "Boost Control" in that you could "sense" the current and have an algorithm that would take a "subpack" offline if that's what was needed to get the voltage down. Lower the voltage and the motor cannot draw as much current.

 

[fechter]

We're getting pretty far off Joe's original topic here,

but if the granularity is small enough, the current surges will be absorbed by the motor. You'd want to be able to use up the juice from all the sub-packs evenly, so I'm not sure exactly how you would that.

 

[Lessss]

Cell scome online 1,2,3,4,5 go offline 1,2,3,4,5 That way it isn't always the same cell coming on first and going off last getting the max drain.

 

[jackatfsi]

This is an interesting idea...even w/ only a few cells.......I would envision the cells connected in a series circle (but not all at once !) such that the number of "active" cells would be determined by the voltage required (ie vs RPM) and the PWM would be done by the "rotary" switching of the string of "active" cells around the circle (ie the string would increment its position each time the current is pulsed so that the cells always discharge evenly) Should be a straightforward design w/ a DSP and cheap mosfets....

I'd been thinking about a similar scheme to charge all the cells evenly w/ a charger voltage equal to one cell (or one battery etc etc)

 

[fechter]

I've seen that approach used before to charge a bunch of 12 SLA batteries with a single charger. Sort of a rotary switch done with transistors.

One problem is it makes it hard for the charger to know when to switch to float or off, depending on chemistry

 

[Toorbough ULL-Zeveigh]

Back then the cell voltages were probably more "granular" (bigger voltage jumps) than the small 3.6 Volts we have today.

Other way round actually.
Generally speaking, the larger the per cell standoff voltage the more desireable it is since you need fewer cells to reach a given voltage. The electrochemical half-reactions have pretty much all been worked out for a few centuries, with lithium having been known for a long time to exhibit about the highest emf of any of them. It's simply taken this long to practically implement.

 

[jackatfsi]

I've seen that approach used before to charge a bunch of 12 SLA batteries with a single charger. Sort of a rotary switch done with transistors.

One problem is it makes it hard for the charger to know when to switch to float or off, depending on chemistry

For charging the switching speed can be slowed down compared to a motor. The problem at the moment seems to be the charger not being able to switch back at the next rotation if it needs to.......it doesn't look like I can use the "smart" charger as is.........back to diodes I guess.....(forward voltage drop)

 

[fechter]

I'm not sure what kind of batteries you want to charge, but if you slowed it waaay down so it completely charges each cell, then moves on to the next (which could be triggered by the charger detecting end of charge), then it would behave OK. This should charge the entire pack just as fast as switching quickly, since the charger output and battery capacity will determine the time to fully recharge.

You could make something like this with an old-school stepping relay.

You could have a LED bar graph that shows which cell it's on and the charging progress.

You just wouldn't want to take off with a half charged battery.

 

[jackatfsi]

I'm not sure what kind of batteries you want to charge, but if you slowed it waaay down so it completely charges each cell, then moves on to the next (which could be triggered by the charger detecting end of charge), then it would behave OK. This should charge the entire pack just as fast as switching quickly, since the charger output and battery capacity will determine the time to fully recharge.

You could make something like this with an old-school stepping relay.

You could have a LED bar graph that shows which cell it's on and the charging progress.

You just wouldn't want to take off with a half charged battery.

Right !

The stepper might work anyway cuz' it can charge at convienient pace using an RC circuit to step and reset the charger.........I must have some steppers around here somewhere......

 

[bobmcree]

discharging batteries unevenly and swapping their connections around might have been necessary once, but with cheap high current fets the only thing that makes any sense is to use pwm to convert the battery voltage to the level needed to drive the motor the required speed. there are some exotic systems that swap motor windings and batteries around, but they are very complicated and not really worth our consideration, IMO.

you can do the pwm in a torroid and then commutate the adjustable DC voltage to the motor, and this is done in the stealth drones for the ultimate in efficiency. when you do the pwm in the motor like an ebike controller, there is some loss in the motor core that might be reduced by making the conversion in a high Q inductor in the controller.

a current mode throttle is a great idea, and does not really require getting rid of the pwm. all you need to do is measure the voltage across a shunt in the negative battery lead, scale it appropriately, and compare it with the throttle setting. when the current is too high it reduces the pwm and the converse. this actually works fairly well with the crystalyte controllers. Fechter's recently posted version is pretty good, and i am in the final debug stages of my own version which does not require pots for zero or gain adjustment. this is done by using a precision shunt, auto-zero op amp, and 1% components that should provide a smooth 0-40A output over the throttle adjustment range.

the main deficiency of most throttles is of course that they are open loop control systems, and when the load changes the speed will change. current feedback can make it easier to control, the next step is of course cruise control with speed feedback to the throttle system.

-bob

 

[jackatfsi]

I was just noticing that some of the LED drivers use an up-converter IC from (9V to18V) to get out 54V for the LED string........the IC has switching current feedback, voltage control and frequency inputs............the major external components were an inductor, MOSFET switch, and a capacitor........claiming 95% efficiency into a (resistive ??) load......

The idea would be to keep my 12V batteries in parallel for charging and up-convert the DC to the optimum voltage/current ratio for each driving condition.......

So the question is.....what is the optimum ???????

 

[safe]

Variable Voltage and Variable Amps

Lets say you have 120 NiMh cells. You might assemble them as 36V 40A or you could do 48V 30A or even 72V 20A. Now with a PWM controller you can get very good efficiency for a particular selection of battery configuration, but what if you wanted to get the best of everything?

If you could create some sort of controller that could dynamically shift from a configuration that was low voltage and high amps and then high voltage and low amps then you would be able to place less stress on your batteries. (you could stay below 1C more easily)

So the idea would be to dynamically reconfigure the battery connections to completely change the battery voltage and amps. This is going waaaaay past the idea of simply loading parallel "subpacks" in series to build up the voltage. This idea goes all the way down to the individual cell and not only loads and unloads it as needed, but even changes the series or parallel choice of how it gets used.

If every cell were individually wired into a sort of "grand central station" you could then use some sort of computer like algorithm to combine them. Maybe even use an actual computer? So all you really need is the "plug in board" with the built in switching mechanisms. The computer could then supply all the logic. You then program the controller with software. (my area of expertise)

Imagine how few Lithium cells you could get away with for a given power capability?

This is what is theoretically possible with:

<<< 156 - 3.7V Lithium Cells at ONLY 1C >>>

Volts -- Amps

3.7 -- 229
7.2 -- 114
14.8 -- 57
37.0 -- 22
48.1 -- 17
74.0 -- 12

 

[Matt Gruber]

use an inverter
and a 120vdc treadmill motor w/ speed control

 

[safe]

use an inverter

How does inverting from DC to AC help any?

The "big idea" is to change the series and parallel relationships of the battery configuration.

Lowest voltage translates to all parallel and high amps.

Highest voltage might mean all series and low amps. (probably never need to go all the way to full series though)

If on the PMG 132 you need massive amps at low rpms and high voltage at high rpms the battery reconfigures itself to satisify the needs of the motor.

 

[Matt Gruber]

use a $20 treadmill motor
and a $42 60-80ah LA.
helps u go cheap.
fun to experiment!
can be hooked up today, no special parts

 

[safe]

use a $20 treadmill motor
and a $42 60-80ah LA.
helps u go cheap.

How do you go from series to parallel?

I'm not sure if you understand the concept I'm suggesting. When a battery is in series it has high voltage, but low amps. When a battery is in parallel it is high amps, but low voltage. Any time you convert from one to the other with circuitry you lose a lot of energy.

So the idea would be to DYNAMICALLY configure the battery connections and as an added feature you do this as a sort of throttle...

 

[safe]

Choppers Won't Work

If you take a high voltage battery with low amps and then use a "chopper" (buck converter using PWM) what you are doing is REDUCING what you have. It's always a voltage reducer. The result of lowering the voltage is that the duty cycle goes down and less gets through. At best you could get current equal to what you expect for the high voltage series connections.

So using the numbers from above:

74.0 -- 12 amps

Using the PWM "buck converter" the best you could get:

37.0 -- 12 amps

But what you WANT is:

37.0 -- 22 amps

So you've LOST the amps...

 

[xyster]

So using the numbers from above:

74.0 -- 12 amps

Using the PWM "buck converter" the best you could get:

37.0 -- 12 amps

No, the best you get at 100% efficiency is 37.0 volts and 24 amps. You just don't seem to believe in conservation of energy. Watts In = Watts Out. V_in X A_in = V_out X A_out.
If efficiency is ~90%, the best you get is 22 amps.

 

[safe]

So using the numbers from above:

74.0 -- 12 amps

Using the PWM "buck converter" the best you could get:

37.0 -- 12 amps

No, the best you get at 100% efficiency is 37.0 volts and 24 amps. You just don't seem to believe in conservation of energy. Watts In = Watts Out. V_in X A_in = V_out X A_out.
If efficiency is ~90%, the best you get is 22 amps.

Well, we just went through that with the PWM "effect". Yes, it's true that at low duty cycles if the inductance is low enough and the frequency is perfectly tuned that in discontinuous mode you get what appears to be higher voltage (to the motor) and therefore an increase in amps. But due to the ripple effect you don't get all of it through. At best (if perfectly tuned) you get about half of the amps back.

To get the FULL amps at low voltage you need to be running predominately in parallel.

Think of the REALLY low voltage:

3.7 Volts -- 229 amps

Do you REALLY think that:

Duty cycle - 5% - 74V becomes 3.7V

...out of this you think you can convert 12 amps to 229 amps?

(the PWM "effect" isn't that extreme)

 

[xyster]

Arrow Do you REALLY think that:

Duty cycle - 5% - 74V becomes 3.7V

...out of this you think you can convert 12 amps to 229 amps?

(the PWM "effect" isn't that extreme)

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?

http://endless-sphere.com/forums/viewtopic.php?t=500&highlight=amps+pwm

I borrowed the clamp on AC/DC amp probe from work and did a quick test. Not super scientific, since I had a hard time loading the motor enought to get full amps.

I have my trusty analog meter on the battery wire, and the clamp on over one of the three phase wires. For maximum current draw, I was starting from a dead stop on an 18% grade. With the brakes on in some cases.

The maximum "multiplication effect" seemed to be around 25 amps input, where I was measuring around 100 amps output.

As the motor picked up speed and the duty cycle increased, the output/input ratio dropped.

For a 35 amp limited controller, the motor will see a little over 100 amps at low throttle settings.