Hi,
could somebody help me clear out something?
Am i understanding this correctly: with an appropriate PWM frequency and duty cycle, i can drive a led without any current limiting device in series, not even a resistor?
I'm therefore assuming the pulse current going trough it, must be within reasonable amplitude and length limits. Thus keeping within the limits of its thermal inertia.
Thanks
Hey, let me know how that turns out for you.
Me, I'd not risk enabling a LED from a MCU pin without a current limiting resistor.
"with an appropriate PWM frequency and duty cycle" - surely. For example, with duty 0% 
it's a high power white LED, so i'd use a high speed mosfet and probably a gate driver if the frequency is too high.
I'm reading substacks about PWM driving, but it's not clear to me if they are talking only about dimming or actually driving the thing. I'm confused.
The reason being better efficiency compared to a standard current sink driver or resistor.
Running from a lipo battery, i estimate the pulse current would be 2-4x the allowed constant current.
Let me know how it turns out by not using a current limiting resistor with a LED.
No problem. But i don't have any cheap HP white leds by me at the moment, only a $$$ OSRAM one, which i don't want to sacrifice to the electron gods just yet. 
Worst case scenario i'm gonna use a driver, but i wasn't sure if the PWM thing, is a thing.
No!!! The average current and power is lower and you might "get away with it" but you are still over-stressing the LED (and maybe over-stressing the MOSFET and power supply) for some percentage of the time. The nature of LEDs makes it hard to know the on-current so you probably don't know how much you are stressing it and the current can change with temperature.
Usually high-power LEDs use a constant current (or controlled current) supply instead of a resistor. The LED usually dissipates about as much heat & energy as the LED so it's inefficient and you need a power-resistor. High-power LEDs usually need a heatsink so the last thing you need is more heat generated by a resistor. And it's inefficient (not a big issue with regular little LEDs).
...Some people get-away without a resistor with regular little LEDs driven from the Arduino. But this is "bad engineering" and the LED and/or Arduino may have a shortened life.
Some LEDs may be rated for short-term higher current in something like a "strobe" application but the current should still be somehow known/limited/controlled.
Switchmode power constant-current LED supplies, are controlled by a kind of PWM to control the current, but the actual current out to the LED is filtered to DC with an inductor.
Not forgetting a flyback diode too.
@DVDoug
I see.
So it's one of those "in theory yes, but in practice no" kinda things.
Better not risk those expensive osram leds then.
Thanks for clearing it up.
I might as well ask, suggest any minimalist led driver IC for a li-po battery system?
Gosh darn it! The make smoke party has been ruined.
No.
It is one of those don't even think about it things.
One of those it's only done by idiots sort of things.
Now now. No need to be harsh. I'm not a complete novice.
I've dumped large value capacitors with twice the Vf into leds before and they survived just fine.
Some manufacturers even declare length and max current for pulsed applications of some HP leds. Usually in the microseconds and several times the forward current recommended. Cree has even an whole article on their xlamp series regarding pulsed overcurrent usage.
Yes, obviously, dumping 1000V into a 3,7Vf LED is not a good idea, but a li-po cell doesn't have 1000V. Extrapolating the Vf/If curve of a HP led i might do 2x the max allowable current. And in theory maybe it could work.
That is why i've even started this thread. I'm not a LED expert, but after browsing a bit, i was under the impression that people used this technique. I merely wanted a second opinion.
Thanks anyway and no hard feelings.
You can't extrapolate these curves in the data sheet. It simply doesn't work like that. There are other considerations apart from thermal ones. Depletion of charged carriers is one.
I doubt they survive well at all. The thing is that a lot of people think that if they don't fry immediately then it is fine. What happens is the life time of the component drops and the light output drops. You can't see any change because your eye is not that sensitive, but with the right equipment you can measure the loss of brightness after about a week. Then you can work out the brightness half life of an LED.
The lifetime also is not measured in terms of when it fails. But in the increase probity of failure.
I worked in a consumer electronics design company. And a product done by another group was designed to over pulse an LED because the client wanted the 20 LEDs on the product to be a bit brighter. The result was that after a year 10% of the boxes had one or more LED had failed.
This triggered an epidemic failure clause in the contract and cost the company a fortune. They tried to sue the company who supplied the LEDs, but they successfully argued in court that the pulsed rating could not be extrapolated.
I'm totally with you regarding the lifetime issue. I'm obviously aware that over-driving any component shortens its life. Hence why professional products do it properly so stay within specs.
But for non mission critical DIY stuff a 50% lifetime reduction might be well worth the lower BOM and space saving.
Assuming we both agree, the million dollar question is: what are the "just right" parameters to achieve this. Each LED has different properties, manufacturer doesn't provide the data. It's a trial and error thing then.
Ultimately I'll probably squeeze in a proper driver since those OSRAMs i have, cost a fortune.
But it's been nice chatting out of the box with you. I always like talking about corner cases.
You can't say how close you are to over driving the LEDs, basically because you are not in control of the current through them. This will change with current, temperature and age. So while you might start exactly with the right current over time the characteristics of the LED you have will change.
Unless you want to throttle back drastically on the white LED's brightness to the extent it is hardly worth having such a powerful LED, then your only other option is to supply a constant current drive.
Think of balancing a broom in the palm of your hand, you need constant adjustment to make it stay put. You stop providing that constant adjustment and the broom falls over. Unless you want to make the broom have a two inch handle, which is useless.
No because once it is right it doesn't stay right for very long. As the temperature rises because the LEDs are generating a considerable amount of heat, things can change over a few seconds.
Powering high power LEDs are almost a unique case in electronics, because you need constant adjustment just to keep things stable. Like a lot of modern fighter aircraft they are inherently unstable, and the only thing that imposes stability is constant control adjustment.
No, that will destroy the port pin, and possibly even the Arduino AND the LED.
Point taken.
Might be fun playing around with some cheap leds, but ultimately some kind of driver is the best and proper solution.
I would use a switching (with inductor) constant current LED driver, not a mosfet.
They work on frequencies between 0.1 and 1Mhz, and can be PWM-ed or voltage-controlled.
Don't use a too high PWM frequency (<1kHz) if you want to dim low enough.
Leo..
This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.