Back in the day, at least in my nabe, when your car starts coking up with carbon deposits, etc., from stop’n’go driving, an IT would be taking it on an open highway and running it wfo for as long as possible to burn off all those deposits.
Kind of instinctive, and it usually worked!
Mainly to prevent color-shift from high/low currents. Dunno if designers even knew about this phenomenon. I’ve played around with LEDs since dinosaurs walked the earth, and this is the first time I ever heard of it!
But it makes plenty of sense for the lowest mode. My Zebralights typically use less than 5mA in moonlight, and I’ve run some of them in moonlight mode for many months (cumulative). Often I just use them as a nightlight, and sometimes don’t even bother to shut them off (since they’ll last a couple of weeks on a charged cell).
Granted, I’ve probably only used about 10,000 hours in moonlight mode, at most. But I still see no sign of dimming (on the high modes). If low mode was really more damaging than high mode, I should probably start to notice it.
That’s only anecdotal, so I’m not saying it’s proof. But until I see proof that moonlight mode really is damaging, it’s difficult to accept it as anything more than the speculation of one person. And since I don’t understand German, I’m not even sure if that’s what he said.
I also have a Sunwayman D40A n/w that has a ultra-ultra moonlight mode “feature” (actually it’s a bug). If I shut it off from moonlight, it goes into a very very low moonlight mode, that stays on indefinitely (until I use it in high mode or I change the battery). It stays this way for months sometimes, if I’m not using the light. Again, probably at least 10,000 hours in that mode. Since it uses just microamps, surely I’d see some damage by now.
- Dr. Berthold Hahn is the Senior Director of Chip R&D at OSRAM OS
- here the pdf with the slides he shows during the presentation
- the important slides are 15 and 16
- they show the reduction in output of an LED depending on temperature and current
- B50 = L70
- B10 = L80
- the dottet lines are for the lower noted current and the solid lines for the higher current
- first slide shows 350mA and 100mA
- second slide shows 350mA and 10mA
- 10mA reduces the lifespan compared to 100mA and 350mA
- reason according to him: at low currents LED ages faster because of diffusion of interference currents in the “active zone” (he doesn’t go into great depths here…)
- so it tells you that an LED (probably 1mm^2 or 2mm^2) from Osram will lose 50% of it’s output in 9000hours when driven @10mA and 85°C
- at 100mA and the same conditions it takes maybe 30,000h (the graph doesn’t go that far)
for flashlights it absolutely doesn’t matter! Why you ask? Because at low currents like 10mA the LED will never become hot. Low currents are only bad in combination with very high temperatures,.
The presentation was held in 2014 at a big lighting event from a university in Germany where experts from different companies held presentations on current lighting topics.
I am only mildly interested in each and every detail so I just watched parts of the video to understand the problem.
Here is what I learned:
- LEDs age from impurities (defects) in the semiconductor that move into the active zone (where the light is ‘produced’). More defects - less light. The spectrum also shifts.
- The first mechanism that drives the motion of the impurities: High temperatures. The hotter the more motion.
The second mechanism is low current density. The slide (in English) from 11:00 on shows a red area of low current density. A lot of the electrons go into the impurities without emitting light. They are feeding the impurities with energy and make them move more.
When you drive a 350 mA LED with 10 mA only it will fade faster that when driven at 100 mA (see 14:30 vs 15:40).
The low current ageing is so bad that it can be better to drive LEDs harder to get a better life span (13:30 green curve, different LED, under 1 vs 20 mA).
This is a bit unexpected and most suppliers do not tell you.
However with low current you will always have a much much lower LED temperature (around ambient), so we are talking about the upper right corner of diagrams 14:30 and 15:40. Maybe at 30°C the life span of the LED will not be dramatically low.
.
So good news for all modders: It is the temperature, not the current.
Bad news for firefly mode fans: Your light may not survive mankind.
I’m not sure I understand the slides. The first slide indicates that the 350mA 55C graph reaches about 94% after 10,000 hours. But the second slide it looks like this same graph reaches about 94% after only 1,000 hours, and hits 80% after 10,000 hours.
Are they showing the same thing?
Okay, that is a key point. Temperature is still doing the killing. A typical 10mA current in an LED will generally result in room-temperature (or close to it) at the junction point.
I’d still like to see some further testing of this, especially at room-temperature, but this is an interesting start.
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:
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
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
low current
heat
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
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
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?
