LED lifetime and output degredation with over current.

[StephenVB]

I have been designing LED taillights. I limit myself to the recommended current but have to compete with people who overdrive the LEDs significantly, from 2 to 5 times.

I can find no good information on how much over current or higher ambient temperature affects LEDs lifetime or light output. I saw one study but it didn't say what LED was tested, so the composition, recommended current and lifetime were unknown.

One problem with automotive applications is that while the lifetime of an LED is commonly from 10k to 100k hours the lifetime of a car is considered to be 4k hours (200k miles @ 50 mph) and the stop and turn are probably on for less than 100 hrs / yr. so you can overdrive LEDs a bunch and they'll still last past a year warrenty.

I don't plan to over drive the LEDs but I would like to know what the effects of overdriving are, so I can justify my position, or even change it.

 

[itsmoked]

I understand your problem. There is really very little information on temp/life. The ambient is also a large variable. What if someone sits on the brake pedal for an extended time in El Paso on a Hot Day?

One suggestion would be to add thermal feedback. Things get really hot, dial back the output a little. Then you could run higher up the curve without dumping the reliability.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[IRstuff]

Thermal acceleration of failures is pretty well known:

http://en.wikipedia.org/wiki/Arrhenius_equation

Most semiconductor failure mechanisms have about 1 eV of activation energy.

That said, however, overdriving can greatly increase one particular failure mechanism well beyond what the thermal acceleration would be. That particular failure mechanism is electromigration, where the electron momentum of the current actually causes displacement and movement of the metallization, causing voids in the conductors, as well as metallization shorts of the semiconductor junction itself.

I can tell you that we had run a laser diode at about 30% over its rated current, accidentally, and the lifetime (sample of 1, bear in mind) was reduced to less than 5%.

In some cases, the gradual degradation of LED output is supposedly due to smaller, less violent, shorts of the junction.

Your best bet would be to do a small experiment, driving say, 10 LEDs per different current drive, and see what the acceleration factor is. Then throw in the Arrhenious thermal acceleration. You've only got what appears to be a factor of 3 or 4 margin right now, though.

It's unclear whether you're pulsing the LEDs, but that's supposedly one option to crank the current up, but only for a 10% duty cycle, or thereabouts. It supposedly results in a perception of "brighter" output, even though the integrated output is actually lower than a continuous illumination. You'll need to run the repetition rate at over 50 Hz to minimize the flicker from the pulsing.

TTFN

FAQ731-376

 

[MacGyverS2000]

IR,

Pulsed outputs being perceptually brighter is only true for certain chemistries (usually those used to make reds/oranges, so in this case it will probably work).

I have a boatload of info to post, but it's 6:30 in the morning and I'm in no mood to post much of anything. Search old posts here (particularly ones with my handle in it) and there should be some more useful info hanging around.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

Pulsing LEDs in automotive taillights (even at 50 Hz) could result in a trail of dots to the viewer (depending on the diffuser). I'm not sure that would be acceptable.

Going beyond spec (this applies generally) works until the manufacturer changes their process or design; cutting the margin while still meeting spec. Then suddenly you have a warehouse full of LED modules with reliability issues. If you're going to go this route, you had better get into a relationship with the LED maufacturer.

 

[MikeHalloran]

Pulsing LEDs in auto taillights DOES result in a trail of images of dots on my retina. Worst offender is recent Cadillacs. I find it VERY tiring.

How tiring? On my planet, shooting out the lights, and the driver, would be a misdemeanor at most.





Mike Halloran
Pembroke Pines, FL, USA

 

[MacGyverS2000]

If you're doing a single color, there's no reason not to pulse at a very high rate (240Hz+) unless you're a complete cheap-ass (and car manufacturers count in that group).

Dan - Owner

http://www.Hi-TecDesigns.com

 

[lumenharold]

Another issue with over driving an LED with regard to temperature is encapsulant degradation. The encapsulant will discolor and lower the power output as well as change the coloration of the light output.

The gains as the LED is over driven are not linear so it would not seem to make much sense to me. Better optics for outcoupling and good lighting design would be time better spent in my opinion.

Harold
SW2010 SP1.0 OPW2010 SP1.0 Win XP Pro 2002 SP3
Dell 690, Xeon 5160 @3.00GHz, 3.25GB RAM
nVidia Quadro FX4600

http://www.lumenflow.com

 

[MacGyverS2000]

Harold,

Depends upon the encapsulant... if we're talking about typical T1-3/4 plastic, then yes, it will degrade. If we're talking about flex silicone typically covering the high-wattage stuff these days, then degradation is significantly reduced and could probably be ignored for such an application.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[StephenVB]

Thanks for your help. In some order:

The ambient temperature is an issue. Here in Oregon the inside of my car gets quite hot on that rare sunny day. I can just imagine what it's like in El Paso, inside the lens housing.

The problem with making my own test it that I have to test different colors and kinds of LEDs, at different currents and ambient temperatures and record brightness over a very long period of time. I was hoping someone else had done that for me.

I'm not sure pulsing LEDs works. I heard that pulsed LEDs looked brighter so I set up a test for my daughter's science project to test and record people's perception. They compared a red LED at 20 mA to a pulsed LED at 100 mA. With a 1 Khz PWM of 20% the average current was equal. The subjects started both with the adjustable LED dimmer and with it brighter. The results were surprisingly consistent. The apparent brightness was the same when the average current was the same.

I use lots of LEDs at their max rated current so there's little room for much more circuitry (Admittedly, I guess I'm also a cheap-ass). I already have active circuitry to control the current and brightness as the voltage changes. With a string of 4 LEDs and a resistor, the current about doubles from 12 to 15 volts, which doesn't seem to bother anybody else.

My issue is that I have to compete with people who over drive their LEDs significantly and don't regulate the current. I was hoping to get some documentation to help me.

The LED manufacturers have supposedly determined the LED life time and maximum current for their parts. I would think that in the process of establishing these values they'd have data on LED life and light degredation at different currents. Apparently not.

Thanks

 

[IRstuff]

They might, but they probably aren't here, and they probably consider it proprietary data. However, if you are a large customer, they may be more willing to share their data with you.

TTFN

FAQ731-376

 

[StephenVB]

Isn't that the way of it. I am a lowly consultant, designing boards for myself and my curstomer. If I or they ever became large customers, it would be too late to matter.