Do high power LEDs last long?

I've thought about spending big money on a really bright light, but when I hear about hundreds of lumens coming from a tiny little LED emitter, I have to wonder how long such a little thing like that can perform. It makes me think that they just can't last very long. But do they?
Only those of you who own these powerful lights would know, so I'm asking you instead of assuming the worst.
I've considered spending $60 on a flashlight, which is a big step for me. But I'd like to know that it would last quite awhile. I've still got a minimag that I bought 15 years ago (now with a nite ize LED). Would something like a Fenix LD20 last that long?

Yes, and no. If you drive them hard enough, you can kill them quite quick, and conversely, if used at very low drive levels they could last darn near forever.

You will see claims for life span ranging from 10,000 hours to 50,000 hours or more for quality power LED's like Cree XRE's, Seouls P4's etc, but if I recall, the spec. current for those LED's is typically 350 mA. For a really bright light, you will probably be driving the LED at around 1,000 mA, give or take a couple of hundred, so no way it will last those kind of hours at full blast. Keep in mind that 10,000 hours is well over a year, non-stop.

The key for me is using multi-level lights, since aside from biking, I rarely need a long run of max-brightness. 350 mA is a common "medium" level, and is DARN bright for most purposes, so most of my runtime comes from even much lower settings. In general, if you can keep the LED cool, it will last a long time, so low levels, short runs on high levels, good heat sinking and a way to remove heat from the outside of the light all help. That last one means try not to set the light down in still air on high for long. In your hand, heat will be transfered to your body rather effectively, and when attached to a bike helmet or handlebar, the stiff breeze is extremely effective at keeping a light running on high cool.

All of these currents get scaled up a fair bit if you get into the quad-die LED's like the Cree MCE and Seoul P7. I just got my first MCE, a PLI from Shiningbeam, for my bike. It runs at over 2,000 mA on high, and medium is almost 1,000 mA on a fresh charge, and is indeed about as bright as any of my single-die lights. Surprisingly, it does not seem to build heat any worse at higher power levels than the single die lights, but of course goes through a battery much faster.

To answer your specific question about the LD20, the way I use my original L2D model (it's predicessor, the first digital Cree model), absolutely, I will probably be able to give it to one of my grankids, if I ever have any. If you crank down the head to turbo, and run out full charges of NiMH's daily, I am not so sure how long it would last before you notice some loss of brightness, but if it were done on a bike for commuting, as many have done, I still think it would last a very long time. Of the many daily bike users of that light that I have read reports from here, I have never yet seen one about loss of brightness in the Cree, and they are the biggest users of long turbo runs, also with good cooling to match.

LEDrock said:

I've thought about spending big money on a really bright light, but when I hear about hundreds of lumens coming from a tiny little LED emitter, I have to wonder how long such a little thing like that can perform. It makes me think that they just can't last very long. But do they?

Click to expand...

They can last a very long time (50000 hours), but it depends on how hot they get.
Some lights are too small to handle the heat from the led and depends on being in a hand, i.e. if you run the light at full power and place it on a table, it might cook itself, but as long as you are holding the light in your hand it will stay cool enough.

HKJ said:

They can last a very long time (50000 hours), but it depends on how hot they get.

Click to expand...

That's why the light you use in a certain situation should be suitable for it. OK, if you only want to buy one single light, you have to make a compromise. A PD20 can be such a compromise, for me it would be a larger light however.

At my keychain I have an LD01ss for example. That one serves to find the way from the car to a nearby place, to have a look at the engine and so on. This means, I only use it shortly, I wouldn't go trecking at night with it. If I do, I take a TK10/11/20 sort of light, which first has a low level which is bright enough and second is able to take some heat from the emitter if I need it a longer time on high.

If you only have a PD20, use the modes in a wise way, don't use turbo unless you need it, you'll mostly be well off with medium. Unless it's cold outside, hold it completely in your hand to give it some blood cooling. Unless you have a bad LED, it should last longer than you'll need it. :huh:

I think that unless the LED is defective, if the flashlight as a whole can't dissipate heat well enough, or the LED is being overdriven, it would probably last long enough to the point where you'd feel like replacing the whole light for a brighter, more efficient one before the LED burns out. Also, just buying a reputable brand can possibly help insure the LED will last as long as it should, and may give you a warranty replacement if it fails.

Big_Ed said:

I think that unless the LED is defective, if the flashlight as a whole can't dissipate heat well enough, or the LED is being overdriven, it would probably last long enough to the point where you'd feel like replacing the whole light for a brighter, more efficient one before the LED burns out. Also, just buying a reputable brand can possibly help insure the LED will last as long as it should, and may give you a warranty replacement if it fails.

Click to expand...

Is there a way to know if a particular flashlight LED is being overdriven? I've seen lights by Eagletac that run 220 lumens, while the fenix L2D run up to 185 lumens. I think both use the same LED (Cree Q5). Does that mean the Eagletac's LED is being overdriven and will fail sooner? Is the L2D also being overdriven, but just not as much?

As long as the LED is heatsinked properly, it will last for years of continuous use. There's a member here who has a power LED running in his basement on a pretty high drive level for a couple years now. No dimming, still as powerfull as it once was. I forgot who, he posted in a thread with a similar question to this one once, maybe you can find it by searching.

Anyway, the point is, don't worry about cooking the LED. In an LD20, even if you run it on turbo for a couple hours daily, it will not be the LED that fails, unless you let it run in a very hot environment or something is wrong with the light (in which case you get to enjoy Fenix warranty). With smaller lights, however, this is not true. When running an LD10 on turbo, make sure the heat can get away (put it on a metal surface like a cold radiator or keep it in your hand or let a fan blow over it). If you run smaller, or badly heatsinked lights that hard, like an LD01 with a 10440 Li-ion cell (not recommended), you can cook the LED.

Don't worry about the lifetime of the LED while using your light. It will not be the first thing to fail. If your light gets warm while running on turbo, that's good, that means heath is being drawn away from the LED. If it gets really hot, however, either run it on a lower setting, or give it some kind of heatsink.

If you're new to high power LED's you might not know this, but LED's are nothing like incandescent lights. An incandescent bulb emits light because of the filament being heated until it glows brightly. So, the heath is the cause of the light and a good thing.
In LED's, however, the tiny chip you see in the "bulb" converts electricity directly into light. The heath is an unwanted byproduct and is wasted energy and if it can't escape it will destroy the chip. I think around 60°C is too hot for LED's, but I'm not sure. As long as the heath can escape, the LED willd last at least 8000-10000 hours (much longer actually, see the first paragraph), which is one year of continuous use. 24 years if you use your light one hour each day on turbo. Which you probably won't.

As for LED's being overdriven, there's not easy way to determine this just by looking at the specs. The Fenix LD20 is not overdriven. The Eagletac P10A2 is run harder, but not hard enough to fail after a couple hundred hours of use, as long as the heath can escape from the light.

Hi LEDrock, I am still new to high powered LEDs and did research as well before I purchased three fenix light (LD10, TA21, LD01). These guys here have probably said what you needed to know, especially on heatsinking and safe usage when using on high or turbo mode. I can't say much about the Eagletac, but I support JBorneu that the LD20 is not overdriven. To give you an example the LD10, TA21 and LD01 I own are using the same Cree Q5 LEDs but each of them has different max brightness output, this was because I think that Fenix designed them according to the heatsink capabilities and where it will most be applicable on, like if it'll be used as candle mode, hold it in your hand, hang it, mount it, etc. On turbo mode, LD10 is at 120 lumens, TA21 at 225 lumens and LD01 (high) at 80 lumens, same LEDs, but different heatsink size and batteries. That might also answer if an L2D is being overdriven, which I think it's not,, with the understanding of safe usage on high/turbo modes. There are CPFers here who did crazy run times on Fenix lights on turbo and had no problems with it because they kept the light cool.

JBorneu said:

There's a member here who has a power LED running in his basement on a pretty high drive level for a couple years now. No dimming, still as powerfull as it once was. I forgot who, he posted in a thread with a similar question to this one once, maybe you can find it by searching.

Click to expand...

I think it may have been me. I've been running a Q bin Luxeon at 350 mA (its rated current) since January 2004. That's over 48,000 hours. I didn't buy a light meter until roughly a year after I powered it up, but it hasn't dimmed appreciably since I've been checking it. It measured 10.3 lux @ 1 meter when I bought my light meter (early 2005 IIRC). This thread made me decide to check it now and it's 8.5 lux @ 1 meter. No idea what it was initially, although a slightly brighter Luxeon from the same batch measured 11.6 lux when I tested it for lumens. So this one may have started at roughly 11.0 lux, and now it's down to 8.5 after 48,000+ hours. That's 77% lumen maintenance. Even if we assume 11.6 lux like the one I tested, that's still 73% lumen maintenance. Keep in mind also that this was an old style Luxeon rated for a maximum current of 500 mA. The current crop of LEDs would probably do as well at 700 mA.

Bottom line, if properly heatsinked and not grossly overdriven, the LEDs in flashlights will last longer than their owners. General lighting is another story, but even here I have good lumen maintenance after the equivalent of 33 years of 4 hour per night running.

jtr1962 said:

I think it may have been me. I've been running a Q bin Luxeon at 350 mA (its rated current) since January 2004. That's over 48,000 hours. I didn't buy a light meter until roughly a year after I powered it up, but it hasn't dimmed appreciably since I've been checking it. It measured 10.3 lux @ 1 meter when I bought my light meter (early 2005 IIRC). This thread made me decide to check it now and it's 8.5 lux @ 1 meter. No idea what it was initially, although a slightly brighter Luxeon from the same batch measured 11.6 lux when I tested it for lumens. So this one may have started at roughly 11.0 lux, and now it's down to 8.5 after 48,000+ hours. That's 77% lumen maintenance. Even if we assume 11.6 lux like the one I tested, that's still 73% lumen maintenance. Keep in mind also that this was an old style Luxeon rated for a maximum current of 500 mA. The current crop of LEDs would probably do as well at 700 mA.

Bottom line, if properly heatsinked and not grossly overdriven, the LEDs in flashlights will last longer than their owners. General lighting is another story, but even here I have good lumen maintenance after the equivalent of 33 years of 4 hour per night running.

Click to expand...

Nice work.. :bow:

jtr1962 said:

I think it may have been me. I've been running a Q bin Luxeon at 350 mA (its rated current) since January 2004. That's over 48,000 hours. I didn't buy a light meter until roughly a year after I powered it up, but it hasn't dimmed appreciably since I've been checking it. It measured 10.3 lux @ 1 meter when I bought my light meter (early 2005 IIRC). This thread made me decide to check it now and it's 8.5 lux @ 1 meter. No idea what it was initially, although a slightly brighter Luxeon from the same batch measured 11.6 lux when I tested it for lumens. So this one may have started at roughly 11.0 lux, and now it's down to 8.5 after 48,000+ hours. That's 77% lumen maintenance. Even if we assume 11.6 lux like the one I tested, that's still 73% lumen maintenance. Keep in mind also that this was an old style Luxeon rated for a maximum current of 500 mA. The current crop of LEDs would probably do as well at 700 mA.

Bottom line, if properly heatsinked and not grossly overdriven, the LEDs in flashlights will last longer than their owners. General lighting is another story, but even here I have good lumen maintenance after the equivalent of 33 years of 4 hour per night running.

Click to expand...

So if I understand correctly, LED's enemy is heat, not being overdriven. So does this mean that an LED could be powered up to pretty much any level without failing as long as proper cooling was in place? I know computers have the same problem. They get more powerful all the time and use a small fan to keep the chip cool. But when that's not enough on the really powerful machines, they start using a water cooling system. Could an LED used for general household lighting use a system like that to stay cool? That would bring LEDs into a whole new realm of possibility!

But back to flashlights, it would seem that an LED could then be pushed to pretty much an infinite amount of power provided that it wasn't run long, correct?

I'm sure there's a limit. There is a phenomenon known as "instaflash", sort of "10,000 lumens for 1/10,000th of a second", or . Within reason, you can keep the junction temperature inside the LED dome under control if it can get heat through the base of the LED slug and into the heat sink fast enough, with a good path from there on out beyond the light body. But as the current single die lights start to push the LED much past 1,200 mA, I don't think that will continue to be possible for long.

I guess I think of the heat as water trying to get out through a pipe, and how good each one of the interfaces it has to cross being analogous to the diameter of the pipe. You have no control of the path within the LED itself, so it will eventually experience a "back-up" and accumulation of damaging, or fatal, heat when pushed hard enough.

Nothing can be pushed to an infinite amount of power, no matter how much you cool it.

Current density affects the lifetime too. IIRC, it's because current doesn't spread evenly across the die, creating inefficient hot spots. Search "droop" for more. We have been waiting for Lumileds' GenX for ages. It was supposed to practically eliminate droop, but it's been several years and the wait continues. :sigh:

There are companies that will provide a lifetime warranty. You might want to look into one of those companies.

Now, HDS says Life of Product, I think that is lifetime, but I'm not too sure. I know Surefire's warranty states that lamps will burn out. I don't know where they stand with LED's.

divine said:

There are companies that will provide a lifetime warranty. You might want to look into one of those companies.

Now, HDS says Life of Product, I think that is lifetime, but I'm not too sure. I know Surefire's warranty states that lamps will burn out. I don't know where they stand with LED's.

Click to expand...

My understanding is, from various threads I've read, that they'll replace the LED heads (such as those on the E-series), since they're an integral part of the light, as oppsed to an LED bulb/drop-in, such as the P60L....

divine said:

Now, HDS says Life of Product, I think that is lifetime, but I'm not too sure.

Click to expand...

That could be construed as until discontinued or the LED runs its rated 50,000hrs, but given HDS' handling of the EDC warranty issue I suspect they mean until violently killed.

LEDrock said:

So if I understand correctly, LED's enemy is heat, not being overdriven. So does this mean that an LED could be powered up to pretty much any level without failing as long as proper cooling was in place? I know computers have the same problem. They get more powerful all the time and use a small fan to keep the chip cool. But when that's not enough on the really powerful machines, they start using a water cooling system. Could an LED used for general household lighting use a system like that to stay cool? That would bring LEDs into a whole new realm of possibility!

But back to flashlights, it would seem that an LED could then be pushed to pretty much an infinite amount of power provided that it wasn't run long, correct?

Click to expand...

Heat is the main reason for degradation, but current density also has something to do with it. Even if you can keep the LED die at room temperature, at higher currents it will degrade faster. I guess a water cooling system for general lighting might help, but in the end it might work just as well to have more LEDs, each driven at lower current, to accomplish the same task.

Not true that an LED can be pushed to infinite power for short periods. Look at how thin the bond wires are. At a high enough current those will open up like fuses. Even if they don't, the LED die has a thermal resistance between it and the heat sink. Put enough current to the die, and it will get very hot, even if the heat sink is at room temperature. The interesting thing with temperature is that for roughly every 10°C it decreases, lifetime doubles, and vice versa. The die temperature of the LED in my test probably never gets much over 40°C, even in the summer, as it's in the basement in a room which reaches at most about 28°C, even on the hottest days. If it was a mere 10°C warmer, then the LED might have degraded this much 2 or 3 years ago. If it were 50°C warmer, it may have only lasted weeks. Conversely, if I could keep the die 10°C cooler, I might exceed 100,000 hours before hitting the 70% point.

Thanks for the input.

LEDrock said:

...
But back to flashlights, it would seem that an LED could then be pushed to pretty much an infinite amount of power provided that it wasn't run long, correct?

Click to expand...

Everything has a limit and are depended on many factors, not just by itself. Even if LED technology advances till liquid nitrogen cooled, it still has a limit, you know what i mean