Overdriving 5mm LED bulbs...how much shorter life?

Hello everyone, this is my first post here - I've had an idea for a project involving LEDs milling around in my head for a while, and a friend of mine said this would be a good place to come to ask some questions.

What I'm wondering at the moment is how much overdriving an LED will shorten its life? I'm looking into using fairly standard-seeming high intensity 5mm LEDs that are spec'd to run at 20 mA (30 mA "max"). Now, I know heat accumulation and how well heat can be sinked away will make a huge difference. At the moment, I'm not planning on doing anything special in that regard - in fact, currently I'm planning to have the LEDs in a tight cluster of three in a closed plastic housing.

Given those sort of circumstances, what could I expect to find in general in regard to the impact feeding these LEDs more current will have on their lifespans? Could I run them for a few hours straight at 25, 30, 40 mA, or what, without them burning out or going below, say, 2000 hours of life instead of the general claim of 100,000? In such an enclosed environment, could I even safely run them at the stated 20 mA typical current for a few hours?

I haven't been able to find any information online regarding this beyond that overdriving LEDs will shorten their life (or possibly burn them out entirely, if run continuously for an extended period of time instead of given periods to cool off) and that temperature plays a big role.

This is unfortunately a fairly specific question, and I can't provide more specific information on what precise LEDs I will be using. I've been looking at bright reds from ultraleds.co.uk or eled.com, for example. Is there any sort of general rule? Any sense or guidelines you guys could give me?

If I haven't annoyed you all too much with all the questions, one more: what about 3mm LEDs with similar specs?

Thanks in advance; I really hope you can help me out even a little on this matter.

a general rule for 5mm, they fry easy, they fry over time, and they dont like to be overdriven Lots, over driven a bit is easy, lots of the little lights are basically doing that.
the whites go more blue (or other non-white colors), either as soon as they are overdriven, or after time of being overdriven.

the - negative Leg, the one if you look inside the led is the bigger hunk of metal, is the one most of the heat goes down. sink that, and its good for more juice.

they also get brown burned epoxy around the base when overdriven pulsed after a VERY long time.
they seem to be putting out some of thier heat out front, because if you put a insulator on the face of a bunch of them, they go crasy from heat.

when ANY of the colors or types, starts changing colors (yellow going red) (red dimming) (white going blue) its dying really bad (minutes to live), and the efficiency is terrible.

actual specs , sombody else knows, not me.
i just know the symptoms and the cause, not the facts /ubbthreads/images/graemlins/smile.gif

I made the following chart based on the equations from some LED lifetime studies:



"Lifetime" is defined as the time to 50% brightness. Note that this chart is for 5mm white LEDs. 3mm white should be similar. Colored LEDs will also have the same drop off of luminosity with time, but the time constant will be longer. For example, red would be about 100,000 hours to 50% brightness at 20 mA.

Check out page 10 of this research paper . It requires Adobe Acrobat to read.

so from all that stuff, would you guys say 30ma is a max logical overdrive for a 20ma specked unit, when run for more than an hour? would last about as long as spec, in thousands of hours too.
it LOOKed like on the chart it was the one overdrive that does not overheat and reduce efficency. yet still has slight higher output.
SPEC looks like the best for efficiency and longevity.

about 55 (mentioned in there) looks good for intermittant use for a light that will ever be used for 1000 hours in its total life.
because 60 looks MAX before it starts to erode quickly and even reduce in output (meltdown) if a person backed off from 60 to 50-40, a short intermittant use flashlight would be hard driven without wasting to much extra power for its output or life.

Not necessarily. Many designs overdrive their 5mm LEDs to get more light and accept the resulting loss of longevity because most flashlights won't be used more than a few hundred hours in their lifetime, so 50 - 60 ma or higher currents are quite common, e.g. in the popular Dorcy 1xAAA. Think about what the total lifetime usage of your light will be and design accordingly.

(i had edited the above After you posted)

ok so as much as 70 will make it through 10 battery changes of a flashlight that ran for 30 hours. 300 hours, still running at 80%.

or about 1/2 as long as a incadescent /ubbthreads/images/graemlins/smile.gif
just so they can put 100,000 hours in thier advertizing.

hmm run 2 at 30, have the same output and some real longevity, then you only have to lie about 15x over when you list 100,000 hour specs

Wow! Thanks for the great info, everyone!

jtr1962: That graph is promising. /ubbthreads/images/graemlins/smile.gif Thank you for showing it to me.

nerdgineer: Wow, surprising results. My father's been an industry EE for a long time (but not terribly familiar with LEDs) and indicated that even a little overdriving could cause an LED to burn out if run for more than a few minutes at a time, not even considering overall lifetime reduction. Looks like LEDs have improved in that regard?

With the Dorcy AAA and other overdriven flashlights, will they hold up if left on for hours at a time, not used intermittently? That is...if used for 2 hours in one straight block, would they survive, as opposed to 2 cumulative hours in spaced-out 30 second bursts? How much extra heatsinking do they do, as opposed to just having the LEDs sitting there without much extra to help heat dissipation?

Assuming everything works like it shows here, 30-50 mA should be plenty for me. Unless anyone has some personal experience to add, I guess I'll just have to experiment once I have my plans more thoroughly laid out. Thanks again for the info!

I've run the Dorcy for hours and hours. I fell asleep and when I woke up it was still running fine.

Re: Overdriving 5mm LED bulbs...how much shorter l

I don't know where the "100,000 hour" number originally came from, but it's pretty obvious that manufacturers started copying it from each other since it sounds good. It's kind of like the "30 lumens" that gets advertised for Luxeon flashlights, even those that are greatly underdriven (like the UK 4AA eLED).

But I agree that it's almost pointless to get too worried about the exaggerations, because most flashlights won't be used for even 1000 hours. That's especially true for us on CPF, since we tend to abandon, sell, or trade our older lights once something newer and better comes along.

The real weakness of an incandescent lamp isn't its relatively short lifetime, but rather its manner of failing suddenly and without warning. Unless an LED is being insanely overdriven, it's just going to get dimmer and dimmer.

Re: Overdriving 5mm LED bulbs...how much shorter l

If you can easily replace the LEDs overdriving them is not that big of deal if they only last 500-1000 hours by the time they go out you can put better ones in the light, especially if they are 5mm.. those you can get fairly cheap.

Re: Overdriving 5mm LED bulbs...how much shorter l

which is fine for flashlights, that rarely get used.
but i am making Lights (lamps& lighting) :-( and flashlights.

but i had always figured that if you use Phosphors it would be relative to other phoshor using items.

my main flashlight gets used for about 250hours a year
and my lights get about 6000 hours a year of on time, but are underdriven most of that time, and mostly in LUXEON type leds.
if my Lamp doesnt run for 10Years, then i lose money on it.

Re: Overdriving 5mm LED bulbs...how much shorter l

Here is a good read for you, take a look a figure 3, and take note of the logrithmic scale...
http://www.lrc.rpi.edu/programs/solidstate/pdf/guSPIE.pdf

Take a look at Figure 5:
http://www.lrc.rpi.edu/programs/solidstate/pdf/narendran2004_jcrystalgrowth.pdf

Nichia 5mm:



Keep in mind, this test started in the time frame when the 350mA Luxeon I- hit the market, and is not an indicator of the Luxeon III or Luxeon IV.


If you go here:
http://www.lrc.rpi.edu/programs/solidstate/ongoingProjects.asp?ID=57

You will see how LED life, take a truely massive hit in life for heat (15C warmer), basically hits you with one half the life to 70% output (degrades 2X faster). And then holding the same 35C, but just slightly increasing the current from 350mA to 450mA, you get additional degredation.

Also notice:
-Blue and multi-chip white LEDs showed the highest rates of degradation.

"In general, the degradation rate increased with higher junction temperatures. However, the relationship between junction temperature and degradation rate is still unknown."

Junction Temperature increases with drive current.

Re: Overdriving 5mm LED bulbs...how much shorter l

they test pretty hot, what idiot is gonna put thier luxeon in a suitcase with a halogen light on in side

when they TEST, they often do like the mythbusters, and try to make sure something blows up before they are done
i mean they make sure they have stuff in there that is going to REACT, as opposed to work non-stop forever.

100*c is aufully hot for ambient temps mabey what they would be in a theatre lighting array.

i see the Logrythmatic scale, ends up being harder to THINK of its interpretation, than if they just streched it out for another 1600 pixels

so the high powered stuff, even its phosphors should hang for much longer.

Re: Overdriving 5mm LED bulbs...how much shorter l

Anybody here know if there have been any changes in the epoxy formulations of 5mm LEDs lately to slow the degradation? While what is being done with power LEDs regarding lifetime is great, for some uses a $0.25 5mm LED does the trick. It would be nice to know that they last longer than 6000 to 10000 hours.

Re: Overdriving 5mm LED bulbs...how much shorter l

[ QUOTE ]
VidPro said:
100*c is aufully hot for ambient temps mabey what they would be in a theatre lighting array.


[/ QUOTE ]


Are you talking about the 35C and 50C test I mentioned?

How do you get to the 100C comment?



jtr1962

I've asked vendors several times, nobody has said there were any breakthroughs in epoxy at all. What I did get was alot of stuff on silicone, and it does hold up alot better.

Thinking of your taxi project, and also the mention of the competitor doing 30mA, as well as the resistor method which raises ambient temps, the customer sounds like they won't need to wait for too long before they get a bad taste for LEDs.

I can tell you, that using short leads on substantial copper planes to spread and pull the heat out of the LED does extend the life alot. Been there, done that, tested that, seen the results. So does underpowering them.

[ QUOTE ]
VidPro said:
ok so as much as 70 will make it through 10 battery changes of a flashlight that ran for 30 hours. 300 hours, still running at 80%.

or about 1/2 as long as a incadescent /ubbthreads/images/graemlins/smile.gif

[/ QUOTE ]

I'm not sure where you got the 600 hour figure for incandescent bulbs. The average incandescent flashlight bulb is rated to last just a few dozen hours. Essentially, they they are being "overdriven" too, in order to get maximum brightness at the expense of long life.

it was more comment, than fact.
but there are bi-pin halogen, and many incadescents other than flashlights.
my interest in the life of the 5mms is for the (totally stupid now) lighting project i was working on.
which i should now scrap the stupid 5mm that i didnt want to wire anyways, and use LuxIIs instead.

average soft white bulb in here has been lasing about thier stated specs at 750hours
average halogen is going beyond 1500hrs, and is still doign great.
if a car headlight only lasted for 300 hours to 1/2 brightness, half the country would be OEMs (one eyed monsters) /ubbthreads/images/graemlins/smile.gif

300-600 hours is total Garbage, and 100% unaceptable for lighting, and its the LIES not the life that should be corrected.

Re: Overdriving 5mm LED bulbs...how much shorter l

i would also keep in mind that led technology is still significantly improving
so in some years you might want to swap them for brighter and more efficient ones anyway

Re: Overdriving 5mm LED bulbs...how much shorter l

Another good paper for lifetime of LEDs is found here:
http://www.lrc.rpi.edu/programs/solidstate/pdf/LedPerformance.pdf