Explosive failure of LED

Steelwolf

Flashlight Enthusiast
Joined
Feb 6, 2001
Messages
1,208
Location
Perth, Western Australia
Wow.... I'd heard mentioned once or twice about the way LEDs die, but had never seen one for myself. I had the chance recently when one of my white (non-Nichia) LEDs was damaged during a circuit test. It had been used in a step-up circuit, which because it had be wired wrongly, was delivering full reversing AC pulses instead of DC. Since it no longer lit under nominal power (3.6V), I hooked it up to my 12V battery pack (8 x 1.5V AA) just to see what would happen.
grin.gif


It glowed orange and quickly started to smoke, crackle and pop. It emitted a weird smell. The die turned black. Then finally it blew up!
shocked.gif
The plastic around the die was split neatly down the centre, exposing the die.

As Donkey from the movie Shrek, said: "Wow! Let's do that again."
grin.gif
And I might, if I damage any more LEDs during circuit testing. I have the carcass of the blown up LED and should be able to post pictures if anyone is interested.
===================================

No stone left unturned in the pursuit of truth, knowledge and the brightest LED light on the planet.

To infinity and beyond!

Warning:
Due to explosive nature of failure, protective goggles and long-sleeved lab coat should be worn for personal safety.
Do not point LED that is about to die at others.

Ave Caeser! Avis morturi te salutant!

Mr. LED has requested, in the event of his death, that his body be donated to science. As such, only a memorial service will be held.
 

Chris M.

Flashlight Enthusiast
Joined
Jan 17, 2001
Messages
2,564
Location
South Wales, UK
Yeah they do that sometimes. I most often find cheaper (usually red) ones just flicker and go out as the gold anode bond wire gets instantly fries, but higher power ones can be quite fun, presumably as the wirebond is thicker and can pass more current. Not that I`ve ever deliberately tried to cook an expensive GaN LED, I couldn`t afford to do that....but now you mention it I do have an old damaged white one somewhere....nah better not. Don`t want to risk damaging my power supplies too....


shocked.gif
 

The_LED_Museum

*Retired*
Joined
Aug 12, 2000
Messages
19,414
Location
Federal Way WA. USA
That's funny...
smile.gif


One of the disclaimers I post on my website reads as follows:


4: LEDs which are severely abused have been known to strike back at their tormenters by exploding rather violently, shooting pieces of broken epoxy at high speeds which can cause eye injury.
 

Silviron

Flashlight Enthusiast
Joined
Jun 24, 2001
Messages
2,477
Location
New Mexico, USA
Hi Gang-

I searched the archives here and found a few posts that touched on this question, but couldn't find any exact answers ::

For all practical purposes, how hot can you run a Nichia 5.6C LED without serious heat sinking or other cooling, without degrading the longevity seriously?

Now,I realize that these leds haven't been out long enough to have a lot of "real life" data on this kind of thing, but I'm hoping someone has done more experimentation on this subject.

The Nichia website lists ~85C (185F) as maximum operating temperature, but I could find no indication if this was in "free - air" or heatsunk (or does it matter if heatsunk or not?)

And realizing that manufacturers generally are pretty conservative in the operating factors, how hot can we run these things for practical purposes and still get at least -~-~- (say) 25,000 hours MTBF?

Last night I was testing a couple of my newest designs and found both running at temps too hot to touch for long:

On fresh charge /fresh batteries :=

The 4 series LED (no resistor)@ 14.4 V (300MA) rechargable light was running at about 100C on the LED lens (It is "potted" so I can't measure lead temps). It is "sort of" heatsunk in the metal Radio Shack reflective led holders.

The 12V 9LED array (3 parallel sets of {3 series LEDs with 47ohm resistors}) is running at about 95C on the negative lead, 75C on the positive lead and about 92C in the center of the plastic lens array. This is in "free-air"- Not PCB mounted or heat sunk at all.

On the 14.4 V rechargable light, I'm getting about 16V for the first half hour or so of operation, and 13.5V off of the 8AA Alkaline cells powering the 12V array which falls to about 12.8V in half an hour.

After an hour or so, the rechargable pack is down to about 15V and in two hours it is down to about it's nominal voltage of 14.4 and around 80-82C.

After a couple of hours the 12V is running at about 12V & 80C, and after 18 hours is down to ~ 11.2 V and 40C.

I'm designing lights primarily for caving/mining use, (trying to keep costs down so want to stay away from using any steppers or regulators if possible) so reliability is just as important as long battery life.-

I'm planning to run the 12V (3X3 LED array on SLA (gel cell lead/acid) batteries in the future, but the arrays will be moderately well heat-sunk when they are "Ready for market", so they will have more amp-hours pushing them and a much slower voltage / temperature decay rate- I don't know if the heat-sinking would be enough to compensate for the temps I'm getting in free air, given the flatter curve.

So, am I going to burn these things out quickly or not?

Is this is overdriving them too much for too long? Should I maybe put a "high-low switch" with an extra resistor for the first couple of hours?

Thanks!
 

The_LED_Museum

*Retired*
Joined
Aug 12, 2000
Messages
19,414
Location
Federal Way WA. USA
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by Silviron:
Hi Gang-

For all practical purposes, how hot can you run a Nichia 5.6C LED without serious heat sinking or other cooling, without degrading the longevity seriously?
<HR></BLOCKQUOTE>

With only minor heatsinking of the cathode, Nichia recommends 55mA continuous as the absolute maximum before serious degredation begins to occur.

When run hotter than this, or when run at this current without adequate cathode heatsinking, a rapid degradation can occur which manifests itself as a stronger than normal bluish/violetish tint to the light and a noticeable brightness decrease at every battery change.

At 85mA, a typical specimen of a Nichia NSPW500BS white LED will fail within 4 days; at 90mA they last about 24 hours, and at 100mA they generally last a number of minutes - as few as 3 and as many as 90.
Much higher than that, and they usually fail immediately, and sometimes violently.

Flashlights can sometimes get away with running them higher than 55mA because they're FLASHLIGHTS. That is, they aren't turned on and left burning for half a day at a time like household lamps.

The 85°C maximum temperature refers to the actual die temperature, not the case or lead temperature.
 

Silviron

Flashlight Enthusiast
Joined
Jun 24, 2001
Messages
2,477
Location
New Mexico, USA
OK, thanks for that info! Now, because of my always lousy math skills and my long forgotten electronics knowledge, I have another question, based on your input:

On series connected LEDS I THINK that basicly all the LEDS will be drawing more or less the same current at a given voltage. Correct? (I realize that there will be some variations between each LED , but this is a pretty safe "rule of thumb" isn't it?)

So If I have 25mA total running into an array of 4 series LEDS, then each LED should be drawing about 25MA each at about 1/4th the total voltage??? Right?

But with my 9 LED array (which is 3 parallel sets of 3 series LEDs}, with fresh batteries, according to my meter it is drawing about 100 mA total on a fresh charge- (the draw seems to increase as the array warms up - initial draw on a cool array is around 70 mA, which goes up about .1 degree per second or so)

So, is each LED drawing about 33.3 mA since there are three parallel sets (100 MA / 3) or is it more complicated than that?

And since each series of 3 LEDS has a 47 ohm resistor, does the resistor make any significant effect on the current draw to each LED or series of LEDs?

Thanks again.

Oh, by the way, since these are for caving / mining lights, they will be burning continuously for many hours at a time, so they won't have the intermittant cooling / use periods like normal flashlight use will have (unfortunately).
 

Steelwolf

Flashlight Enthusiast
Joined
Feb 6, 2001
Messages
1,208
Location
Perth, Western Australia
Silviron:
In series, each LED draws the same amount of current, but the voltage across each individual LED may vary.

In parallel, the voltage across the LEDs will be equal, but the current may vary in each branch. In this case, where similar components are being used in each branch, you can safely assume that the current divides itself evenly (ie. 1/3)

You can think of the entire concept as a pipe with water flowing through it. Given only one pathway, all the water has to flow through that way. Given several pathways, the water splits and flows down each pathway depending on which allows it to flow the easiest.

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>And since each series of 3 LEDS has a 47 ohm resistor, does the resistor make any significant effect on the current draw to each LED or series of LEDs? <HR></BLOCKQUOTE>

Yes, the resistor does make a significant impact. You have a 9 LED array with 3 branches of 3 LEDs each. I assume you are using a 12V source.

With a 12V source, the voltage drop across each branch is nominally 12V. Some differences may occur in the actual circuit due to variations in each individual component, but they will all show the same voltage drop across each branch.

Looking at the individual branches now, you have 3 components with 3.6V dropped across each of them. That is a total of 10.8V, which means you still have another 1.2V left over. This is dropped across the resistor.

Using the formula I=V/R, you find that the resistor would cause the current to flow at 25.5mA If you didn't have that resistor in place, you will probably find that the current flowing through the branch would be around 40mA, but this is hard to calculate for sure as the internal resistance of the battery is unknown.

As well, the resistance of the LEDs will increase as they heat up, thus increasing their own resistance, which, because of I=V/R, causes more current to be drawn, which heats up the LED further. This is known as a thermal runaway and leads to a spectacular failure.

If the 47ohm resistor is there, thermal runaway cannot occur because the resistor sets the value of the current flowing through that branch. You will probably not have to worry about thermal runaway because the internal resistance of your batteries will provide sufficient resistance to limit the current flowing through the circuit.

Since you have the 47ohm resistor in each branch, each branch will have 25.5mA flowing through it (nominally). This all adds up to 76.5mA being drawn from the battery. You can actually power a LED as much as 40mA without much worry of over-heating, but I generally try and keep to 30mA as I don't see much improved light output vs run time from there on.

My latest LED light was for my car and I used a 33ohm resistor with 3 of the blue 3.6V LEDs. That gives me a nominal current of 36mA. The LEDs do warm up a little, but not hot.

Hope all that answers your questions.
 

Robl

Newly Enlightened
Joined
Jun 20, 2001
Messages
23
Location
Green Bay, WI
I created three strings of three white LED's for a dome light and found that a car battery under charge is actually 13.8 to 14.2 volts. The LED's got so hot that the solder melted off the LED leads and the middle ones actually fell out of the fixture!

Since I am more interested in brightness than longevity, I run them as hot as I can. Of course sometimes I overdo it!
grin.gif


Robl


BTW, the LED's still work; with a resistor in place now.
tongue.gif
 

RonM

Flashlight Enthusiast
Joined
Nov 10, 2000
Messages
1,164
Location
NJ, USA
Robl - What size resistor did you use? Im interested in making some white LED fixtures for boat use. And yes, the voltage is 14.2V when a charger or the outboard's alternator is charging the battery. The only time you ever see 12V is when the battery is getting pretty run down.
 

Robl

Newly Enlightened
Joined
Jun 20, 2001
Messages
23
Location
Green Bay, WI
About two years ago when I first started playing with White LED's I found this formula on the internet:

E-3.5/.04

so 14-10.5 (for three LED's in a string)=3.5
and 3.5/.04 = 87.5 ohms. Then, because I like them bright, I grab the closest resistor below this value. In my case I used a 67 ohm 1/2 watt resistor from Radio Shack.


Rob
 
Top