Of LED drivers and PWM and current regulation

[swiftwing]

Through the course of customising an MC-E Mag from nailbender, i;ve come to realise that my knowledge of LED drivers is woefully inadequate, and possibly downright wrong. So i come to seek knowledge, for knowledge is power. And power, hopefully leads to brighter lights.

First, the basics. I always read that for example, Fenix lights have better runtime than others due to them having a current regulated driver intead of a PWM one. Now it seems that we are refering to current regulated dimming vs PWM dimming here, so am i right to say that PWM dimming drivers like those in say the NCD10 are also current regulated in a sense that they attempt to supply a constant current to the LED and dim it by PWM duty cycle?

So assuming we have 2 drivers, 1 PWM dimming @ 100Hz and 1 current regulated dimming, and assuming both drivers have physicsland 100% efficiency and have a max current of 1A, would it be right to say that at 50% power, the current regulated driver would be sending 0.5A to the LED while the PWM driver would be sending 50 1A pulses out of 100 per second to the LED. The current regulated driver would give a longer runtime due to the 0.5A draw being less hard on the batteries than the 50% duty cycle 1A draw of the PWM driver. Is this line of reasoning correct?

My next question concerns the term direct drive. I've always though that direct drive means no electronics between the batteries and the LED, just a resistor to limit current. Now that i know about the d2flex direct drive driver, my head just about exploded. I've always thought that driving an LED without a resistor would lead to it drawing more and more current and killing itself, so how does one prevent this from happening when direct driving say a P7 / MC-E with the d2flex without using a resistor?

And then there's the KD Super P7 driver, which claims to be a current regulated driver, yet uses PWM dimming, which makes my head go . How does that work?

Help me guys, help me...:mecry:

 

[VidPro]

PULSE:
Q1 , actually the batteries dont seem to mind the pulses so much, many batteries are even rated with nice high pulse numbers even though thier constant drain levels are lower.
the battery being a chemical reaction (as part of it) when the reaction is given a break it can recover pretty easily. (to a point of course).
but
Getting the most capacity, IS one possibility when driving things hard pulsed, it just isnt concidered very often, probably because with LED (vrses hotwire) the load doesnt reduce the capacity that much, specially when dimed.

IN the same sence the LED HEAT can recover a bit with pulses, as the heat is drawn off the led, through the stuff to take the heat away, so that leaves the biggest issue with it really being mostly the LED Drive current, and efficency from a clean low drive current.

Current , even pulsed:
at the Lowest end of drive , like >5ma for a high power led, they are then very efficent. As you raise the current up within normal ranges, they get slightly less efficent, then it starts getting really bad in the higher currents, where efficency drops off a lot. all this stuff is also effected by other elements, like heat for one.

because the GATE parts of the LED where the jumping is going on , trigger VERY fast, and stop triggering very fast, the led gate parts have a on/off time that is very fast, unlike a glowing Incadescent filiment which has to cool way down.
so even at very fast pulse speeds the LED is turned on and off fully.
so there is no residual free ride or "averging" going on there to work in a "digital" way. Without the averaging your just pounding it at the full current in pulses. then refer again to the above paragraph, the higher the current the less efficent it is for the quantity of light.

of course we would have to add in the "persistance of Phosphors" and the burning out or aging of phosphors to the equasion too, and any burning of gates that are imperfect, and phosphors dont last forever.
the jump points in the little molecular stuff that i know little about, are very irregular, i would bet that at high drive currents the weak points are what fails in TIME also. Depending on the power going in, (& heat and stress) would depend on what breaks over time, and what phosphors burn and what phosphors even trigger.

so not only does the full current PWM run in the lesser efficency state for the led, but over long and long times it will probably reduce its total output capability also.

oops, did i ramble again , i hate when that happens oo:

all PWM only curcuits can be cleaned up with a couple a parts, which would make them act more like current control, ANYWAY , and you could even call that then current control, because you could flatten the pulse, and it would in words control the "current".
even a 10c capacitor added on the led side you can see the brightness go up a bit, as it flatens out the pulses some.

 

[VidPro]

Direct Drive:
Q2 Direct drive, yup loose term for not controlling the current, or changing the voltage, in fancy ways. Indeed it should have been called 2 things "Direct drive", and "Resistered drive", but often there is enough resistance in play IN the switches and springs and stuff already even with "direct". Still the current/voltage , or total power going in is being controlled somehow, or the LED would pop, either the voltage is low enough, or the battery limitations, and all.
IF
you have a direct drive with a DIMMER, that doesnt mean it has a driver its just a "dimmer", but you could LABEL it as a driver, because it is used to drive an led. or call it a "controller". What its actual electrical things it does, is more important than the words used to describe them.

and its ok that your head can pop, because there isnt anything you didnt mostly understand, its just more of a labeling and communication thing, and the deeper you go the more your head will continue to pop anyways Best bet, is to analise the spec sheet, or try and figure the method used for real.

Current Control with PWM dimming:
you can control the current with many methods, then toss PWM ALSO in on the side to do the dimming. and then you can call it anything you want.

The LED is supposed to have some GREAT VAST NEED for the current to be controlled, basically you gotta control it somehow, even if you lower the voltage which reduces the current. Dont be fooled by some of the semantics of power, most of the stuff does have some method for controlling current in some way, although digital pulses dont directally "control current", they can USE pulsing TO actually control the current (by some definition).
so
PWM is digital (on-off-on-off) , and current control is concidered more analog (half on) , but you could control the current both with digital pulses, and with analog resistance, or with charge pump things, or with DC-DC converters, all depending on the curcuit.
PWM is pulses , and you can DIM with PWM or "power" control, depending on the circuit. You can actually even voltage control to dim or power the led if you do it correctally. There isnt ONE method, there is the BOX, and people doing cool things outside of it.

its just a matter of how much rambling they want to do when you go to sell the thing , each aspect of the curcuit that is designed to do certain things could be labeled, or not, they could say "microprocesser controlled Current regulated PWM , mosfet regulated Torrent coil capacitative resistive driver"

it might be easier to label them as "pulsing output" or not, boosting (raising voltage) or bucking (lowering voltage), DC-DC , Digital or analog.
once they get ahold of a microcontroller , analog gets tossed out the door, just like on your computer, they can still fake it though .

and you thought you were confused before, HEY where is my interpreter? and some SCOPE pictures.

 

[VidPro]

You dont have to overpower a LED into thermal run away, to burn it out, and you dont have to have huge power to get to a bad thermal situation.
So
You can burn them out just fine without seeing a VF droop, and the power were using HAS its own voltage Droop resistance abounds in things, the battery droops, the heat goes up.

the Bench is the bench, the flashlight is the flashlight, i can cause different situations in each of them completly different, it just aint the same.

on the bench, i can keep the voltage at exact specs, while the current ramps up, all while trying to cool the thing, it basically doesnt help that much to know that,
BECAUSE
in the light, the heat keeps going up, and the voltage and current of the battery go down, 2 worlds each can display different features of the hard driven led Tack in a little light with no heat removal capability and the Vf of the led changing is the least of the problems. then put in the little battery , and the battery cant even keep up.

so the scenario of the voltage lowering on the led (which it doesnt that much) you can see on the bench, doesnt translate to the Direct drive or resistered light very well, instead the BATTERY voltage goes down, the heat goes up and its freaking DIFFERENT, were just supposed to know these things so we work within how it acts. the bench teaches valuable lessons, then we still have to get out of the pitri dish.

Its more important to know that over TIME even without thermal runaway, and current psycosis, and heat frying stuff, you CAN burn out completely and/or age a led faster, by not keeping the total power down, and getting the heat off of it.
if
your going into current runaway situations in a portable power light, then it is WAY to far off to last TIME, and that is some hella battery or driver you got .

 

[emr]

:duh2:

 

[swiftwing]

Someone once said, "before i came to you, i was confused. Now, i am still confused, but at a higher level." lovecpf

Vidpro, for the PWM dimming vs current dimming runtime issue, what you're saying is basically that the extra runtime of the current dimmer comes from the fact that driving an LED at 0.5A will produce slightly more output than driving an LED at 1A at 50% duty cycle PWM? Hence, to get the same amount of light, the PWM dimmer has to run at a slightly higher duty cycle and thus consume more power? Which also means that one can match the runtimes of the current dimmers with PWM dimmers, just that you'd get less output?
As for the direct drive thing, what you're basically saying is that, if for example, i direct drive a 3.7V Vf P7 with 4AA Nimh which gives 4.8V, the voltage sag from the high current the LED is trying to draw + the internal resistances of the batteries and the connectors and stuff will combine to limit the current drawn to some equilibrium?

 

[Marduke]

The other part of the puzzle is many PWM circuits are more efficient than CC circuits. There is a rather through discussion on just that topic around here somewhere.

 

[VidPro]

an LED at 0.5A will produce slightly more output than driving an LED at 1A at 50% duty cycle PWM?

thats it, why dont you write it
Then add in, depending on the Method used to do EITHER of them, will depend on the WASTES in the curcuit itself. The curcuits themselves are often WAY less efficent than they even list
so
curcuit inefficency can often be more than the LED efficency problem.
And when it comes to TIME, they dont care , a flashlight isnt on enough, and doesnt last enough years for THEM to care. So driving the led at lower currents (actual not averaged) to get 20,000 more hours of time out of it isnt a concideration very often.

As for the direct drive thing, what you're basically saying is that, if for example, i direct drive a 3.7V Vf P7 with 4AA Nimh which gives 4.8V, the voltage sag from the high current the LED is trying to draw + the internal resistances of the batteries and the connectors and stuff will combine to limit the current drawn to some equilibrium?

You might be able to, first it isnt 3.7v if you ramp up the current going to the led, it can be higher even still, like 3.8-3.9v. and the battery voltage can hit higher (5.2v), right after a charge.
i am not going to say you SHOULD 4XAA Direct a P7 or a MC-E or whatever because it depends on all the things you listed, with the wrong combo of things you could burn it out.
Once you understand (though) what you just said, you could make a 4XAA that is MOSTLY direct, you would just have to check it and mabey add in tiny resistance.
and dont forget HEAT, if you cant remove the heat, it wont mater, you can burn out a led running well below "spec" if it gets hot enough.

the biggest problem with direct, is the battery. Manufactures of high ticket lights arent likly to use direct, because consumers will use differenet battery types and get completly different results. whereas the modder can easily toss something together that is MORE efficent, but THEY then know what its power source needs are and how that would effect that.
if your kids toss lithium cells in (for example) any previous calculations would go out the door.
manufactures of low priced lights HAVE made many direct drive lights, and i myself have fried a few of them by running lithium in it.
so if you make it, or if they make it you could still fry it.

you mentioned the resistance of the battery, EVEN a different rechargable battery used in the same scenario may sag more or less enough to change the total power enough to cause a problem. but if you set up something that works, and stick with that same battery, it usually works for , , , for a heck of a long time.

with the right driver, you have "Regulated" output, so the output level doesnt change as the battery voltages change,, they can also keep the led from burning out as those voltages change around. the complexity of the whole thing continues with the different types of drivers.

 

[swiftwing]

So, we can generally classify drivers as such,

direct drive vs current regulated (which includes buck, boost and buck/boost drivers) This would also include input and output voltages and drive currents.

then we have the dimming modes if its a multi mode driver,
PWM vs current regulation dimming (am i right to conclude that direct drive circuitry will only be able to give you PWM dimming since it has no way of controlling the current? It'd be a current regulated driver if it could, right?)

then we have the driver efficiency, which usually ranges from 85% to 90%+? ( is a direct drive PWM dimming microcontroller circuit always 100% efficient?)

Am i missing anything?

Also, i've got a specific question about my light. Its a 1D Mag which has been modded with an MC-E and the d2flex microcontroller, direct driven by either a 18650 li-ion or 3AA to 1D adaptor. If i wanted to increase my runtime by adding another AA and using a 4AA to 1D adaptor instead, would it result in a slightly brighter light with a longer runtime, or would the increased drive voltage result in a much brighter light consuming much more current, or something else?

 

[VidPro]

then we have the driver efficiency, which usually ranges from 85% to 90%+? ( is a direct drive PWM dimming microcontroller circuit always 100% efficient?)

you wish i find that the REALITY of driver efficency starts more around 85% and goes down from there , to more than 40% losses (in these low voltage apps) , heck i cant even figure out where its GOING on some of the stuff :thinking:
but its also "the way things work" lots of them drop voltages through parts , and therin are WAY more efficent when the input and output voltages are HIGH. Loosing .2v in a 3V setup isnt as bad as loosing .5v in a 12V setup. the high voltage setups where they are using 10+volts and good efficent drivers the efficency is very good compararably, but then it bring up the (minor) issues of many Series cells when discharging.
THis (of course) all varies widely with the various things.

PWM only dimming has Way less total resistance, if just a mosfet is the "path" then the resistance can be cut down to the miliohms (tiny), but there is still resistance through the boards paths and all that too. it is never a lot via measuring, but in application at the Low voltages and High currents its significant, or at least worth recognising it exists, it isnt a straight short copper wire, , , just almost.

PWM current controlling devices still have to have some way to "meter" (sence) the current, just like on your ampmeter, so it most often has (already) added resistances.

for your 4.8v direct drive, you probably will want a bit of added resistance, just depends LOTS on all the resistances you already got going, and added with any adapter thingeee. the resistance will give it some flexability, OR a good curcuit for that voltage is the AMC1735 thing (the glorified resister).
NOW
if we compare THAT initial voltage to the 3X initial voltage or the li-ion, then your starting (charged) more OVER in the voltage (and therin total power) needed, so you can set it up so it is brighter . as far as runtime that will always be relative to how the load your putting on the battery, how much current is flowing.

When the voltage drops on the input to the led , like say 3.5V (under the load) going in, the current to the led is very low, the output of the led is lower, but the runtime just keeps increasing , BECAUSE the current load keeps going down. it is also nice on the batteries, it can be a win win situation, unless you MUST have constant full brightness. if you must have constant brightness through the discharge range of the batts, then you must have a driver or current controller.

basically with any direct drive light (and similars) you get a bright light when your batteries are fully charged, and ALSO you get a light that virtually runs forever
if you increase the voltage , then tone it down a bit with resistance , the Variance of that will still exist, but it will be less variation. the light will stay brighter for longer, and it will not run as long before the batteries should be recharged.

also with the 3X or single li-ion , you can run it Direct (depending) there almost isnt enough voltage to begin with to hard drive (the quads specifically), it will be lower faster, as you might have noticed. With the 4x you can buffer a bit of the voltage down through resistance (or again mostly direct) and drive the quads higher for longer. you will also be draining the battery faster obviously.