LEDs wearing out?

Katherine Alicia

Enlightened
Joined
May 15, 2020
Messages
836
Location
Central UK.
We often see a LEDs lifetime quoted as about 50,000 hours for instance, but what actually happens after this time, since it`s only about 6 years ON time.
I have a clock radio that`s never off and uses the red 7 segment displays and it`s nearly 30 years old and perfectly fine!

So I`m a little confused here?
 
They gradually get dimmer. The term often used is "lumen maintenance". So the LEDs in a light bulb might be expected to be at least 80% as bright (maintain 80% of their lumens) after 50,000 hours (by which time some of the other electronics in the bulb may have failed).
 


Thanks! this article explained it perfectly for me: https://www.digikey.com/en/articles/understanding-the-cause-of-fading-in-high-brightness-leds

and in a nutshell: "The temperature to which the LED is subjected is a primary contributor to the rate of dislocation production. A hotter chip will form dislocations faster and hence fade quicker. This is a key reason why manufacturers advise keeping the chip's junction temperature down."

it`s encouraging to know that just by running my lights at a lower lumen count (and thus cooler) will extend their life and that it`s little to do with just "Age" per se, and probably explains why my clock is still working perfectly, the leds aren`t really driven hard, and the turning On/Off causing expansion/shrinkage in the die substrate allowing for more of these dislocations wouldn`t really be an issue for a clock that`s never off.

it makes perfect sense now, Thanks! :D
 
Overheating will kill them fast, they get dim and may start to flicker.
 
That seems to explain why although HDS lights are considered underpowered compared to the rest of the pack, they'll still be going when the others fail.

What was that line from Blade Runner? The light that glows twice as bright.....
Something like that.
P
 
The lumen wars are contributing to this, LEDs get too hot because they are over driven to put out more lumens. I doubt the LEDs are gonna last any where near 50,000 hrs, but since most of these lights have components that tend to fail before the LEDs or die the first time you drop them (despite their phony 1.5M ANSI specs) and people that buy them tend not to use them that much, it might be too big of a problem.
 
Ah, thank you Alicia.
I went to see that at the movies when it came out and had a strage feeling that I knew the plot. Do Androids Dream of Electric Sheep? was one one of my favourite books. When I used to commute to London I got through about 2 books a week.
P
 
One data point:

I homebrew'ed some outdoor lighting in the final week of 2009 that's been running at nearly 100% duty cycle with marginal heatsinking (4x Rebel triples @ ~3.5W/unit sporting perhaps 12 in² of effective heatsink). Assuming a 95% duty cycle, that's just shy of 90,000 hours of operation. I do not at all doubt that the system has dimmed over the years - likely close to half with LED performance degradation and shmutz working its way into the housings - but it remains adequately bright for the task of lighting the driveway adjacent to the garage.

I will however very likely replace it this year - I'd like more coverage, better-focused light, and would rather not be replacing it after it has failed.
 
Another data point: In 2010, I bought some of the cheapest LED strips I could find on ebay. I put some short segments on a 12V wall wart in my workshop and never switched them off again. They are now at ~88,000h. Compared to the control samples, they have not become dimmer.
 
that`s a good point actually, I have colored LED fairy lights in the bedroom that have never been off in the last 6+ years, and a LED rope light running down the stairs that`s been there for nearly 5 years and never off, both are still fine.
though I don`t recall there ever being a lifetime rating on the (long gone) box of either, so maybe it`s just the high powered stuff?
 
Unless using calibrated measurements, even a 20% or 30% decrease in output is not necessarily going to "appear" especially obvious.

And, yes, I would expect higher power / heat / output emitters would likely suffer more from "burn time" losses.
 
Unless using calibrated measurements, even a 20% or 30% decrease in output is not necessarily going to "appear" especially obvious.

And, yes, I would expect higher power / heat / output emitters would likely suffer more from "burn time" losses.

Correct - a 50% difference is generally perceived as incremental; thus I merely claim that my installation's performance remains 'sufficient' since I've not so much as a cheap light meter to throw at the problem.
 
I am surprised to not be hearing/reading anything about deterioration of the white LED phosphor over time with
high temperatures, contributing to fading. This appears to be the drive behind "remote phosphor" by Philips and
I'm sure other companies; getting it away from the hot die. Not doubting contribution of the die itself, just wondering
how this all fits together.

Dave
 
Top