The thing that kills leds most is the heat- all the way heat: ofc that is valid if you dont overdrive them great time.
Overcurrent and voltage spikes also kills them rapidly
Mousture is quite cunny too, it atacks solder joints
Usualy, leds dergade slowly and after some point they start blinking, either cob chips and single die emitters
Some of them die all in an instant yet most of them do the final disco dance :sunglasses:
I watched this video
And I am native german, so I did understand well why low currents are far worse than heat to kill LEDs
The current protects the very thin active light emitting zone from electrically defective spots to migrate into it from the surrounding material, this active zone is a lot thinner than the rest so migration is a huge issue
Of course heat also increases the migration of electrically defective spots
On our CC moonlight flashlight the LED can age over 10 times faster than on max. rated current
You see the plot at 16:00 of the video
And that plot shows the LED running on 3% of its normal current
Our moonlights and firefliey modes on CC lights get below 0.1%
The odd thing is if we get good cooling an overdriven LED ages slower than at rated current when they got the same junction temperature
Interesting.
If this holds up, it occurs to me that people who almost always use low modes might be better off carrying a light with a less powerful LED - think XPE, not XPL - for routine use of low modes, and a second light with a more powerful LED if they want higher output from time to time.
That way, the same low current for the low mode would represent a larger percentage of the less powerful LED’s optimum current, hopefully reducing the defect migration effect.
Still have to watch the complete video, but as a conclusion I’d say our often used FET+1 design is a good idea to realize moon modes since the LED is driven with at least 350 mA (with a very short PWM duty cycle).
Yeah but in flashlights you use an LED which can live >50000h
If the life cycle gets below 5000hours on CC Moon mode its not a big deal
as its not your living room light that burns on abeverage 4 hours a day
Gor people that use flashlights as joule thiefs for their bedroom lighning every day its not good news if those lights have CC driver
Same counts for dimmable lights in your household, if you always dim em to a very low level they age pretty fast, but the effect is not dramatically above 10% of current, but below its really noticeable aging of it
I was having my floor lightning every night on the lowerst dimming at approx. 2% on so my cats have light in the night, ot does not take much power, but now I know it wears down the lights very badly
The closest thing I have in terms of long running LEDs is a pair of LED Lenser blue 5mm pig tail LEDs wired in series. I am pretty sure they are Nichia re-branded and marketed by Lenser. I can’t for the life of me remember what kind of current and voltage I am driving them with. But they have been running constantly and continually for 8-9 years in a solid state headphone amp that never turns off (a modded Larocco PPA for those interested in a google search). They have dimmed slightly over that time. I am pretty sure I am over-driving them to some degree.
As the Osram guy explains in his video, an overdriven LED ages less if the temperature is the same as on lower currents because you cool em better
Two things let age LEDs
It sounds odd but current protects the active zone from aging by migration of faulty spots into it from the surrounding p-n substrate
At 16:15 you see the aging diagram from an 350mA LED driven at 10mA
and the impact is far worse than one driven at 100mA at 150dC junction temperature
Also what testers like Djozz see that at hard overdriven LEDs the radiaten decreases again does not come from heat
In fact its an effect calles Auger recombination
He explain it that there are 3 electrons in the active zone and instead of all 3 producing light two of em use their energy to kick the 3. out of the active
Okay, this is all interesting, but has this study been confirmed by an independent lab?
Since I don’t understand German, I’m not sure if they tested moonlight mode currents at low junction point temperatures, and compared it to high currents at high temperatures. That is what you will get with flashlights. It’s always been my understanding that heat is what ages an LED.
If moonlight mode only ages an LED rapidly when it’s also hot, who cares? I’m not sure if that’s what they were testing, but it isn’t what occurs in real life.
And, in my own anecdotal experience (which I know doesn’t make for any scientific evidence), I have used some flashlights on moonlight mode for thousands of hours (XML2 LEDs). They’re on almost 24/7, and have run that way for over 2 years. If moonlight mode was so damaging to them, you’d think I’d have noticed a significant dimming by now.
Moonlight ages the LED very quick even at low temperatures
But who cares if my LED ages a lot quicker if the base lifetime is with 50-100k hours so damn good
But in any case you wont notice until the light run 1000hours or more on moonlight
PWM moonlight dont age the LED
The low current aging is not a problem in most applications as PWM is used a lot
Its physics that let age the LED at low current and most manufactors dont speak about it just showing you lifetime plots at test and max. currents
So much to learn and absorb. I love it!!
Industrial led lighting suppliers generally give a 5 year warranty due to the drivers failing before that time. A 10 year warranty is being held back by the drivers, not the LEDs themselves.
Never personally seen an led returned having failed. The drivers have failed though. I think the caps are the weak spot
Well, I have used that on a couple of lights, and I don’t notice any difference when I compare their moonlights to other lights. They’re maybe 3000-5000 hours max, but if the damage is as bad as the video seems to claim, wouldn’t I see them a lot dimmer by now?
What about LEDs on things like electronic equipment (TVs, microwaves, radios, etc.). Some of those have been running for well over a decade, and I don’t see anything getting very dim. Does the video test different kids of LEDs, to see if some are more susceptible than others?
Actually, 50000 hour lifespan means a lot more for moonlight mode applications than it does for high brightness applications, at least for flashlights. Nobody is ever going to use their flashlight on high for 50000 hours, because battery changing would be insane. But I could see me using moonlight modes for a large fraction of that time, because they’re on almost all the time.
Did the video claim that? Just because you have a rapid on/off, doesn’t mean the same chemical processes aren’t taking place as a low constant current. At a high enough frequency, the chemicals don’t care.
The LED doesnt age by chemical process at low current, heat does mean age by chemical detioration as well
With like refrigator cooling or on antarktis station with temperatures always way below 0dC an LED lightning keeps even at 100% current in a junction Temperature below 25dC the LED may reach 200,000hours or more till it reaches 70%
Its migration of defects in dotation, if enough current flows there is not much migration possible because the electrons fill those defects
at room temperature the moonlight mode wears down the LED in 5000-10000hours to 50% depending which base lifetime the LED has
Only if junction temperature on moon would be 150dC its 2000h
Isn’t that just chemistry?
That doesn’t seem like a long time, though. It’s only a year of 24/7 use. It’s a short enough time that people should be noticing this effect. Well, okay, flashaholics should be noticing this effect.
What about solar LED yard lights? I’ve had some of those running for years. They’ve certainly put on 10000 hours by now. I don’t have a base-line to compare them against, but wouldn’t people be complaining about them getting too dim if the LEDs only lasted 5000-10000 hours? That’s just 2 years of “night” time. They’re dim to start with, but 50% dimmer should be noticeable.
The LEDs in electronic indicators, and clock radios: they’ve been running for over 100,000 hours. Shouldn’t they be almost dead by now?
If it is run at its normal current they can reach 100,000hours with 20-30% loss in brightness
In most lights you wont run them at very low currents
At 30% of its current the LED can live 50000h without problems as you can see in the 100/350mA picture as you have a lower junction temperature, than on full current
If you look at the 10/350mA chart at 55dC junction temperature you got 80% left at 10000h
Okay, but that chart only goes out to 10000 hrs. Are you extrapolating it to 50000 hrs?
Is the dotted line the lower current in the graphs, and the solid line the 350mA current?
Okay, I’m looking at the pictures you posted, and I still don’t think I understand them.
The 80% at 10000hrs in the 10/350 picture corresponds to the solid red line (for 55C). Is that the curve that represents the 350mA current, or the 10mA current? It sounds like you’re saying it’s the 10mA current. But what does the dotted red line represent? It gives 65% at 10000hrs. I thought that would be the 10mA curve, and the solid line the 350mA curve?
I think what’s confusing me is that the solid red line in each graph (which is the 55C curve), doesn’t give the same curve in both graphs. So, it can’t represent the same 350mA current at 55C. What is it? I think I must be slow today.
Also, this may just be a translation issue, but why are the graphs labelled with B10/B50? Isn’t that the nomenclature that denotes failure of a LED (i.e., B10 means that 10% are expected to fail by a given length of time). What we seem to be measuring here is the L70 value (or L whatever value, that corresponds to a brightness after a given amount of time). Typically, manufacturers use the L70 value when stating the lifetime of the LED.
350mA is the rated current of the diode
I am not sure which line represents the normal age rate he spoke of B10 and B50 curves but I dont know what this means, one of the is represented by solid the other by interrupted line
Okay, I guess I’ll just have to accept that something is lost in translation. B10 and B50 would normally apply to something that fails completely, like an incandescent bulb. LEDs use something like L70 to indicate lifespan.
I understand that 350mA is the rated current of the LED they’re testing. So the “worst case” 10mA that they’re using as an example of low current, would put out something like 3 lumens (assuming a forward voltage around 3v). Moonlight mode on flashlights would be far worse than that.
I’m tempted to test this hypothesis. I have a couple of Astrolux A01’s that I have hardly used at all. I could compare outputs, then leave one of them running in moonlight constantly, changing the battery every few days. But it would take so long! Likely over a year. I’m not sure I have the patience for that.
I have used a couple of other lights like that, but I have no baseline to compare them against. They don’t seem any dimmer than when they were new, but that’s strictly by memory, and an old memory. So, it’s not a reliable test example.
Why change the battery cant you power em wih an external power supply?
I don’t have one, and, it might be tricky to wire up on a twisty. That’s why I never do current measurements on them.
In any case, it’s just too long of a test. I don’t mind running tests for a month or two. But years?