LED Enclosed Bulbs

thermal runnaway

thermal runnaway? So, you are saying they are getting too HOT ....

http://ledsmagazine.com/

Definitely due to heating. The sealed enclosures (if provided) trap heat and gradually this builds up, leading to thermal degradtion of the LED's. Remember that each diode would likely generate 0.3 mW each which adds up over 50 bulbs and over time too. Solutions? Heatsink the whole lot by filling the bulb with a thermally conductive fluid (messy if it breaks!) or pulse the system at a high frequency so that the light output remains the same and average heating is reduced.

refer to old issues of LEDS magazine. Lots of methods to heat sink and other ideas. Track doen to advertiser web sites for the actual LEDS, they give you free data to use them.

Indeed they get hot, but it can be managed,

pulse the system at a high frequency so that the light output remains the same and average heating is reduced.

Sounds like the easiest way, but how would I do that?

Donald

pulsing is a method of using the thermal mass of the LED and the substrate to reduce the average operating temperature to one that the LED can tolerate. If you take an LED rated at 20 ma you might be able to give it short duty cycle pulses at 200 ma as long as you averaged that same 20 ma. Why do this? Often you get more light than you would get with the continuous 20 ma. Of course, if the temperature get a little too high, your 100,000 hour LED turns into a 1000 hour LED or a 1 second LED.

ADhere closely to the drive voltage/current andheat sink specs of the maker, as the maker knows all about these LEDs and how far they can be pushed. Push too far = toast.

Develop the skill set to measure the heat sink you are using in terms of junction temperature of the LED, as this tells the tale. Bad heat sinking will give you a cool heat sink and a burned out LED. A good heat sink cools the junction and get warmer.

The primary cause is poor design, bad engineering. LED's aren't perfect and their output declines over time depending on the operating temperature. They do not experience "thermal runaway" it is simply electro-migration and perhaps other factors that are indeed accelerated by excessive temperature rise that is caused by either bad engineering or overzealous marketing that compel the engineers to push design limits beyond what is wise in order to be first to market. Sometimes that backfires. Before long we will have good LED lighting. Next year select vehicles will be available with LED headlights but chances are they will be well engineered.

http://en.wikipedia.org/wiki/LED

Good article, I stand corrected, thermal runaway is a failure mode though I doubt that it applies to a gradual degradation in output. More likely, it would result in catastrophic failure and either open circuit (=DED Dark Emitting Diode) or short circuit which would temporarily increase the brightness until the current got high enough to fuse the bonding wires.

LEDs and most semiconductors depend on precisely trace alloyed N and P junctions. They call this trace alloying "doping" and the "dopants" give the ability to conduct electrons or "holes" ( missing electron from an atom).

As you know atoms can diffuse within solids in the same way ink diffuses in still water. This diffusion proceeds slowly at room temperature for LEDs etc, but this diffusion is a function of temperature. Higher temperature = faster diffusion.

A led than would last 100,000 years at room temp will last a lot less time as the temperature goes up. By burning out leds at high temperature first at very low currents in baking ovens and measuring the time to destruct at various temperatures. This destruction at a few microamperes is not caused by the current, but by the oven temperature. The high temperature causes the dopants to diffuse to where they should not be and the material stops beind an LED. You then plot it on a log scale, hours versus temperature and you get an estimate of how long it will last at room and other temperatures. They call this accelerated aging under voltage. You can repeat this with larger currents and eventually characterize the LED with a set of curves you then give away to the users. Competent engineers will know that they must design within those curves to give the life they want, be it 500 hours for a high brightness short lived use(surgical lamps etc, where they can tolerate the cost in trade for the brightness) to long lived lights for trucks and other uses that want 100,000 hours of 24/7 burning (12 years) or more.

Those of you who wish can search for the details of these procedures, I wrote it in generalist language for wide understanding

Thanks for that input, makes sense.. But you know, the market drives the sales force, so things have to start somewhere. It's just a shame that the customers are the ones that get the worse deal from poor initial quality.

Donald

Donald writes:

"But you know, the market drives the sales force, so things have to start somewhere. It's just a shame that the customers are the ones that get the worse deal from poor initial quality."

You're absolutely right, Donald.

And if this weren't bad enough, some customers continue getting a bad deal when the product's initial quality never moves beyond the 'poor' stage. One product immediately springs to mind, and that is a well-known, popular computer operating system manufactured by a well-known, less-than-popular company headquartered in Redmond, Washington.

If the truth be known, I suspect the continued popularity of this product derives not so much from its own technical merits - superior alternatives exist - than simply from the sheer momentum of customers having had made considerable investment in its use. Like junkies, once 'hooked,' this company's customers have but little hope to continue throwing their precious pearls before the swine.

(My apologies for getting off-topic.)

--Europium

Don,

Heat management! They over drive LEDs.

kcj

Thanks Guys, for all those GREAT comments, and Technical Information.

I am compiling all of your comments, and forwarding on to the people that SHOULD already know these things!

Just another example of getting the "cart" ahead of the Horse.

Sincerely,

Donald

To truly understand the answer, you must understand that it was the wrong question...a question to which there is no definitive answer and can be none. The correct question is: Why are the LED bulbs advertised as performing better than they actually do perform, or are likely to perform? This is a question (and implicit answer) that will always apply in such situations. The solution for one feeling disappointed/cheated is to attempt to assert a warranty and obtain a replacement. Don't laugh; it is as likely to work as not. Then, if a reasonably satisfactory portion of the claimed performance (the puffed performance) can be had with the original plus one or two replacements, in that case you can feel satisfied that the product has reasonably lived up to its promise. This would be a reply that smart engineers could provide depending largely on the company for which they worked and their position in that company. But in most circumstances it something better addressed by management/marketing types.

It is easy to make a cheap design with a short life that will be bright.

Good design costs more and uses more costly materials and those lights last longer.

They may not be as bright, being designed to last longer but not brighter.

Marketing: Necessary evil; Spawn of Satin; See sociopathology. In industry is called "Public Relations" In politics, Propaganda but the military has the most succinct term "Perception Management"

Your comments are right on the mark, as usual!

BTW, is "Spawn of Satin" by any chance Velour?

(Yes, I know you actually meant to say "Spawn of Satan," but the temptation to have some fun with this was just too much to resist.)

--Europium

Thanks. One thing for sure is that with so many eyes not many errors get by unnoticed. I hope they hurry up and finish working on that meaning checker because the spelling checker is only a partial solution.

One problem so many answer.

You have not indicatied that at what voltage you are operationg each LED.

LED is a diode and is very sensitive to voltage if you are operating it at excat and optimum voltage?

about above 0.6 volts any increase in voltage is not going to converted in to light energy, means any increase in voltage above otimum value is going to incraese the haet of LED, This incraese in junction temp is going to shorten LED life

If you want to measure this voltage just disconnet LED from CKT.dont measure with LED connected.

You may be not able to measure juction temp by external method due to thick casing.

if you dont have a exact voltage source then you should use searies resistance with your ckt. this resiator will abosorb any extra voltage and life of LED will not lost.

although any series resistor is going to reduce the energy efficiency of your system.

also all the LED should have reasonably close optimum voltages.spacialy if you are using them in parallel.

Different colors of LED emit light at different voltages. When conducting they are forward biased with a sharp knee, so they need current limiting if the voltage can spike upwards.

Think of an LED like an electron waterfall.

When the water is below the top you can add to the level and get zero flow(save leakage through the stones). Once you reach the lip of the dam and water goes over it is now hard to raise the level as it flows over.

In the same way a green LED that starts to emit light at 1.8 volts at 20 ma will draw a few amps at 2.0 volts and fail in milliseconds.

Variance in LEDS means you must have both current limiting resistance and load sharing resistsnce if you want them in parallel.

Series is no problem, they will not conduct at all until the first one reaches it's knee and will start to emit. You will still need current limiting for the tim when they are all conduction as any increase will cause huge current draws = failure.

http://en.wikipedia.org/wiki/LED

When did I said that all LED glows at the same voltage.I have just given example.

You have posted so many times in between these days but you never reached on this conclusions.

just by changing words you cant change the meaning.whatever you stated is the same what i said.

cut and paste answers from net dont work all the times.

You said this:

"about above 0.6 volts any increase in voltage is not going to converted in to light energy, means any increase in voltage above otimum value is going to incraese the haet of LED, This incraese in junction temp is going to shorten LED life"

what is that supposed to mean? No LED emits at 0.6 volts.

The initial LEDs emitted in the range of 1.5 to 2 volts with variance from manufacturers and colors. More recently the voltage has moved to 1.5 volts to 5 or so volts, dues to different processes, dopants, colours etc.

In general a LED will not conduct until you reach a threshold voltage and it emits and starts to conduct. At all times it abeys basic thermodynamics. If you have a 2 volt drop and 10 ma you are dissipating 20 Milliwatts. 3 volts at 1 amp = 3 watts. and so forth.

This means the plastic encapsulant and the metals in the LED will reach a temperature that is a function of this wattage dissipated. The 10 milliwatt lamp may well work well with no heat sink by just the dissipation of the leads. The 3 watt one will need to be attached to a heat management substrate that will remove the heat and not allow the LED to cook itself into a blob of useless metal. Most of the problems with the new brighter leds having short lives are due to bad heat management. You must not ignore the makers data, or your LED may shine brightly, but not very longly....

Why do they do this? With LEDs being only 5-10% efficient there is a need to 'puch the envelope" and try to get more light for the same current. Since few people can properly test an item and cannot tell if it will last 100,000 hours or merely 1000 hours we get a lot of LED light that die young.

A low cost spot pyrometer will help you measure the junction temperature. If that is too high, the lamp will die very soon. This device must be able to read the small LED junction and give you a reading in 3-4 seconds or less. An OEM can advise you on this.

aurizon I just said hello to you on e-bay, and you said you didn't know who I was.

This is Donald the originator of this thread.

Donald

ps thanks for the input!!

Donald, I do now. The Ebay system is designed to stop good cummunications lest we make side deals.

Try these forums. these guys ferret out a lot of good information and you can also sel them items.

http://flashlight-forums.com/index.php

http://candlepowerforums.com/vb/forumdisplay.php?s=547a9ed65e30ea3fb71f7eba6c713815&f=66

Another data reference publication.

http://lighting.sandia.gov/Xlightingbizlit18.htm

http://www.freepatentsonline.com/7052180.html

http://ledsmagazine.com/press/14289

more on LED temperature measurment

http://www.google.ca/search?hl=en&safe=off&q=measure+%2B%22led+junction+temperature%22&btnG=Search

yes, the key problem is the heating generation problem