What kills Luxeons? I need to know

[ufokillerz]

So i killed 6 1 watt luxeons in several days, I'm using DD mag 3D's, 5 blew after a little while in the first Mag 3D, then i took it apart, kept the heatsink, and put it into a new Mag 3d, this too blew after turning on and off several times. Is it grounding improperly or something? I've had it on the Star PCB, and off the PCB, same thing, haven't figured out a resistor to use yet either, i've put Luxeons into 2 3C flashlights and i think they are still alive, i will be making a new heatsink by hand tomorrow and hope that helps. I've read that many people DD luxeons off 3D and it works fine. Every time after a while, the light dims off, then dies. I dont think its overheating either, being that its never on for more then several minutes, and half of it was on the Star. I also had a 6d 4 luxeon DD, worked fine, and my 5W in 2C and 6D work find also.

 

[Slick]

What are you using for heatsink material? And what are you using to mount the emitter (what kind of adhesive)?

 

[ufokillerz]

I'm using a 1" x 1" x .5" aluminum graphics card heatsink, i was using artic alumina epoxy at first, then on my last 3 i used artic silver epoxy. Run time was never above 2 mins. I'm guessing it might be my heatsink somehow.

 

[Slick]

Well, that sounds OK.. You're not using so much epoxy on the emitter that it squeezes out enough to touch the positive lead does it? Arctic Alumina is non-conductive and is my choice for thermal epoxy for emitter mounting. Are you frying emitters no matter what kind of epoxy you use or are the ones that were built using the Artic Alumina still alive?

The way you describe your emitter's failure, I'm inclined to thing that your positive lead is grounding out somewhere through a high resistance short (like low conductivity epoxy). When this happens with the emitter powered, it's "good-night Irene"...

 

[ufokillerz]

i make sure the epoxy doesnt touch anything other then what its supposed to. It burns out regardless of epoxy.

 

[Rothrandir]

maybe you have very low vf luxeons...if couple with less-that-ideal heatsinking, this could result in death...

were they from the same batch? maybe they are just crappers? i've never heard of this happening but...

they are stars right? what is the bin number on the back?

 

[ufokillerz]

i've tried M4JW's and N1KW's

 

[MR Bulk]

I have not heard of 6 in a row. Maybe one or two, driven off SIX volts (Yes, I've tried it during one of my Mad Scientist rampages). Must be the particular way the positive lead is soldered in this one light you used.

Also, I have rarely used any 3 "D" DD 1w Luxeon setups. Only "C" or "AA"...in fact I have never personally used "D" setups, bought a couple -- no, in fact I bought some TWO "D" lights with three "C"'s in them. That is all...

 

[Silviron]

I had the exact same thing happen to me a few months ago; I lost five white batwing LSs in a row, although the two cyans and the one 1W lambertian white I built at the same time are still working perfectly in the same configuration.

Before that, when I blew an LS, it was because I was careless and hit em with too much voltage. These were actually slightly under-driven, and didn't heat up at all before they died, and even my old dead LS resurrection trick wouldn't work on these last 5.

That's the main reason I quit making mods and lost interest in LED lighting in general.

 

[ufokillerz]

How do you resurrect a LS? What guage wire is good to use? I'm using the kind of wire you would find on a powersupply molex(16guage-18guage?), i'm pretty sure that its too thick. Very weird how i'm not the only guy suffering from this. Atleast non of my 5W luxeon whites have blown yet. Btw, Silviron, were yours driven in a 3D or something else?

 

[Silviron]

Sometimes you can resurrect a dead emitter by hooking your negative lead directly to the heatsink slug at the bottom of the emitter (Not the actual heatsink/PC board, but the little ~1/4 inch oval on the bottom of the emitter itself.

I did a big post here, including photos a year or so a go on this if you want to take the time to search for it.

I was using a somewhat discharged home made 3 X (4/5A size) NiMH battery pack on these guys when they blew. No load voltage was about 3.24v. Never checked what load voltage was, but it had to be less.

Oh, here is that old thread:

CLICK HERE

I'm going to tear myself away from CPF now for a few days, so I hope that answers your questions.

 

[gdan66]

I've been using a 3D DD setup on my bike helmet light for about 6 months now with no problems...so its unlikely that's the main cause of your failures.

Dan

 

[gdan66]

Maybe if you were using NiMH cells it would cause that problem. I've heard they don't have the internal resistance of Alkalines and even thought the voltage is lower, the current is higher and fries the LS's.

Dan

 

[ufokillerz]

I'm using alkalines

Edit: actually i've used panasonic super heavy duty crap batteries, can't remember if those were in when it fried each time, but i also have 6 of these in a Mag 6D that used to have 4x 1W LS, now runs 1 5W LS.

Can anyone confirm this battery thing?

 

[Slick]

One thing you need to do is measure current consumption in ANY configuration. Have you ever checked?? I do this to every mod that I build just to make sure that I can catch and correct a thermal runaway condition before the emitter is toast..

 

[ufokillerz]

[ QUOTE ]
Dr_Slick said:
One thing you need to do is measure current consumption in ANY configuration. Have you ever checked?? I do this to every mod that I build just to make sure that I can catch and correct a thermal runaway condition before the emitter is toast..

[/ QUOTE ]

please explain /ubbthreads/images/graemlins/confused.gif

 

[Slick]

You'll need a DVOM. You use it to measure how much current is flowing through the Luxeon. After powering up, it is normal for the comsumption number to climb a bit as the LS heats up, but it needs to stop and level off. If it doesn't, the hotter it gets, the lower the resistance gets - then more current flows through...... a vicious cycle until heat induced burnout occurs. I don't have any mods powered by D's yet, only C's and I do know they will drive harder than a smaller battery like an AA due to lower resistance. So checking your current comsumption is the only was you're going to rule that out as a cause.

 

[Doug Owen]

Abuse kills......

"What kills Luxeons?"

This one's simple, too much current.

Does it one of several ways. Too many amps times volts (watts) for too long a time without enough heatsink cooks 'em. Higher currents will eventually cause failure, no mater what sort of heatsink. Think of that thin bond wire you can see if you squint hard and look sideways through the lens as a bitty fuse.

So, how much current are we talking about? Voltage bins J and K (at 350 mA) are 3.27 to 3.51 and 3.51 to 3.75, let's say 3.5 volts. This is to say a 'direct drive' of 3.5 volts will give us (full rated) current of 350 mA, the maximum the maker intends. Typically the one ohm internal resistance accounts for .35 Volts ( .35 Amps times one ohm) of that total, meaning 3.15 Volts 'across the diode itself'. Now we can add the internal resistance of our three D cells (.136 ohms each from <http://www.duracell.com/oem/Primary/Alkaline/mn1300.asp> ), for a total of 1.408. Assuming 1.5 Volt cells, the total drop across this resistance is three times that less the 3.15 Volts, 1.35 Volts. This 1.35 Volts divided by the 1.408 ohms (since it 'appears across it') gives us a current of nearly an amp (.96). Nearly three times the 'absolute maximum' number specified by the maker, the point he guarantees the bond wire not to fail for instance. Over 3 Watts.

Interesting things can happen with heat. For instance the -2 mV per degree C number. A 100 degree rise (not itself fatal) will cause our 3.15 Volts to drop to under 3 Volts (2.95). This will further raise the current as the voltage 'left across the resistance' goes up. It's now 1.55 divided by 1.408 for well over an Amp (1.1 in fact), which of course heats the diode still further......

This is a slug temperature of 58 degrees (18 degrees per Watt times 3.245 Watts) lower for a heatsink that's 'warm not hot'. Yet the junction temperature is at the "absolute maximum" of 120 (assumeing a 20 degree C start). This assumes perfect slug to heatsink bond, of course not possible, our real world results will be even worse.

There's also the matter of saturation. There are only so many electon/hole pairs there, and the electrons can only move so fast, try to force more through and the voltage drop skyrockets (and with it heat).

Just the extra .3 ohms from using C cells probably kept you alive then? It represents what, a 20% reduction?

You're basically fooling around with the 'direct drive' condition CMG *didn't* use in the Reactor III. This was with AAA cells with even higher (.181 ohm) internal resistance. You really should consider adding some resistance. Using an additional two ohms or so (like they did) will hold your 4.5 Volt current to (1.35 divided by 3.408) about 400 mA. Full rated output (plus a bit, at least at the start) and *increasing your run time* more than proportionally.

I know there are lots of folks out there hotrodding stars into this range, but then again they're also having some "accidents" as well.

At least that's how I see it.

Cheers.

Doug Owen

 

[ufokillerz]

its probably the amperage as i had thought, but never cared to use resistors, i will be driving one of the same luxeons at 5v with 225ma only, and see if that blows, if it doesnt, i know my luxeons are good to go heh. Its probably the thermal runway thing as slick had said.

 

[Doug Owen]

[ QUOTE ]
Dr_Slick said:
I don't have any mods powered by D's yet, only C's and I do know they will drive harder than a smaller battery like an AA due to lower resistance. So checking your current comsumption is the only was you're going to rule that out as a cause.

[/ QUOTE ]

I know the above statement makes sense. As happen so often, the truth is counterintuitive. The *facts* are that AAs can (and do generally?) 'direct drive' harder than C cells, at least initially. Check the specs. Mallory lists the AA (MX 1500) at 81 milihoms, the C (MX 1400) at 210 Miliohms, .13 ohms per cell more. A very big difference, more than double. Even D cells are higher than AAs.

The lesson here is direct drive is skating on thin ice. And sacrifices run time. And LED life to a largely unknown degree (although often it shortens that '100,000 hours to *right now*!).

On the plus side it's bright and simple.

The advice still stands, 'measure your current'. Or expect surprises.......

Or so I see it.

Cheers.

Doug Owen