Running a 1W Luxeon with no Heat Sink!

[dat2zip]

Well kinda.

No Slug heat sink that is. Since the introduction of the 3W and pictures of it showing a fatter lead I've speculated that the lead frame being on the same plane as the die is where the LED die is initially attached to. If this is the case, there is a good thermal path from the leads to suck off some heat directly to somewhere else. Problem is most of the mods cut the lead off and attach a small wire to it providing a poor thermal path to the flashlight housing.

Thus, the experiment. Is there a thermal path here or not? Can I run a 1W without heatsinking the slug?

At first you might think this is crazy. Well, I know I'm crazy, but, that's not the question here...

Here is a picture of my setup.

I can tell you this that the heat sinks do get warm and the voltage on the LED stabilizes from 3.28V to 3.138V which if I can believe the -2.5mV/C yields a die temperature rise of 56.8C above ambient which was 81F (you convert that to C).

The heat sinks stabilized at around 95F and were rather warm.

Next time I do a mod, I am seriously considering heat sinking the leads to the body using copper tape.

Wayne

 

[dat2zip]

O, I forgot to add, that's my ESD generic carpeted workbench that those heat sinks are on and why I can't find the LEDs when I drop them.

One of these days I need to upgrade to a real workbench...

 

[Darell]

Man, you are seriously turning into Mr. Science these days, Wayne. Neat stuff. Did you spit on it or anything? /ubbthreads/images/graemlins/grin.gif

 

[McGizmo]

Cool Wayne! BTW, that looks like a Luxeon torture device! Do I see the heat sinks slowly moving apart from eachother? "The Rack"!

If we can come up with some good thermal bog, enveloping the LED in it up to its ears could aid as well if the leds are covered too.

- Don

PS. were both heat sinks the same temperature? What current level did you have going through the LED?

 

[shiftd]

Cool Wayne.
you always come up with something new. /ubbthreads/images/graemlins/thumbsup.gif
then, will the slug no longer become the main thing for heat transfer?

 

[dat2zip]

I started the test at 200mA. After that stabilized I cranked it up to the spec'd 350mA current.

shiftd,

The Slug is still probably the primary thermal path, but, adding in the thermal transfer from the leads will only help in aiding thermal transfer. I would speculate the the lead frame has a better lower thermal resistance but, being a thinner material the few millimeters of length degrade the efficiency of sucking off too much heat.

It would be an interesting experiment to attach a thermal probe to one of the leads and the heatsink in a regular setup to see if there is a significant temperatute difference from the slug and heatsink to the leadframe.

If there is adding thermal bog from the lead to the base and then re-measuring the lead to see if some of the heat has been transfered.

Wayne

 

[hotbeam]

Daring Wayne. Care to take a pic of the emitter by itself... close up? /ubbthreads/images/graemlins/naughty.gif

***EDIT: Somehow I read you were using no heatsinks with a 3W!! No need for pics of the 1-watter

 

[Slick]

Interesting! It would appear that you've found a new way to "push the envelope" that extra margin.. /ubbthreads/images/graemlins/grin.gif

 

[dat2zip]

Here is a picture of the LED close up???

BTW: No spit was used in performing this experiment, taking measurements or gathering data except for the drooling anticipation /ubbthreads/images/graemlins/drool.gif

 

[AilSnail]

Hey I recently read that Doug_S had measured some luxeons' thermal quantum mechanix, they were found to have -3.5mV/C not -2.5mV/C as stated in the spec sheets. I'll try and dig up the thread... maybe I'll even find out what they call the relationship.

Edit: ok, i'll just call it Vf tempco. here is the link http://www.candlepowerforums.com/ubbthreads/showflat.php?Cat=&Board=UBB14&Number=121408&page=0&view=collapsed&sb=5&o=&fpart=2
and here is the relevant quote from Doug_S [ QUOTE ]
Not figured out by me. Done long ago by others smarter than me. I do cringe at your terminology "lost resistance" as this does not reflect what is happening. The falling Vf with increasing temperature is a quantum physics phenomenon. As a personal practical matter, I have not yet developed confidence in my ability to measure absolute die temperature via Vf. Every time I have attempted to measure the Vf tempco of Luxeon LEDs I come up with a value around -3.5mV/C. This is greatly different from the datasheet typical value of -2mV/C. I have not had time to devote to resolving the discrepancy.

[/ QUOTE ]

 

[FalconFX]

Very nice...
Maybe a setup, where the 2 leads are enveloped in copper slug (the LS sunken into a copper holder, and the leads grounded outside the slug) could get the most heat drawn out of the LS...

 

[MR Bulk]

I think I'm the one a little slow on the uptake here, but...is that a 3W? I swear it looks just like a 1W. Hey, very clever marketing by Lumileds!

 

[Rothrandir]

if the leads are being used as a thermal path from the luxeon die, then wouldn't soldering wires onto the leads be more problematic?

are the 3w leads the same as the 5w leads? i can't tell from you picture, but from the picture peter took, they looked identical.

where these samples provided to you? are they available for purchase? would you consider disecting one to find out exactly what the differences are?

 

[dat2zip]

Guys,

I don't have a 3W. I was refering to a picture posted by someone else and discussions with Future and others.

The image as shown is a generic 1W LD. Probably a lonely misplaced Q3 lying on the bench. Since it had been isolated I can't tell you what bin it is. I guess I could look at the beam. White-Q3. Purplish-P4, Yellowish-N3... I don't have too many bins.

 

[Rothrandir]

doh! /ubbthreads/images/graemlins/ohgeez.gif

 

[AilSnail]

Which adds up to 40,6C above ambient. (spec sheet says -2.0mV/C, not -2.5 like I said above)

Since you put just over a watt to it, it means it has a total thermal resistance junction to ambient of ~37C/W right?

On the black column beside fig.5c it says "Maximum Forward Current vs. Ambient Temperature. Derating based on Tjmax=120C and Tambient max=75C..."
This could mean that junction to ambient thermal resistance on the star board is something like (120-75)/1.1=40.1C/W, which means you may have bested the factory heatsink setup!

What kind of voltage drop do you get @350mA with a decent heatsink on the slug, or with a star?

 

[dat2zip]

Yup, -2.5 is my brain memory which most of the grey matter has now gone and died. /ubbthreads/images/graemlins/grin.gif

Voltage drop? I suppose if I carried this experiment to the next level would be to put the same LED on a heat sink and do similar tests. I haven't done that yet, so, I don't have any data for you at this time.

I have visually watched the negative voltage slope many times and have a general feel of the slope and thermal characteristics. In watching this I saw it leveled off and a rough guess was 300mV before destructions. I was going to abort if the Vf dropped more than 300mV. Since it leveled off before that I knew it was actually settling in at some safe level. I verifed with my calibrated lip after running the test to measure the temperature of the LED top surface and the two heat sinks.

I haven't thought farther than this test and I doubled checked your calculations and it seems you might be correct on that.

I see two more tests to round off the experiment.

1) Mount the Luxeon on a heat sink. Apply 350mA and see where the Luxeon settles in at and the Vf drop.

2) thermally bond the lead wires to the heatsink in some form. Copper tape? AA epoxy, thermal slurry epoxy. Then retest 1) above and see if the Vf drop changes.

If the number is ~37C/W and the die to slug is ??15C/W then the two combined should be the sum of the two or:

1/37 + 1/15 = 1/Rt_total = ~10.6C/W

In a 1W environment, this might not add up to much of any savings, but, in when overdriving this oould play a significant factor in increasing Lux output efficiency.

NOTE: Any number, formula or english is the residual grey matter left in my brain.

Course if it's -2mV/C then the numbers are much worse and the leads would not contribute much in lowering the overall thermal resistance.

Wayne

 

[AilSnail]

[ QUOTE ]
NOTE: Any number, formula or english is the residual grey matter left in my brain.

[/ QUOTE ]

You have probably dropped too many LEDs /ubbthreads/images/graemlins/sssh.gif

 

[koala]

Hey wayne great info there! I must say the luxeon volunteer is one brave LED! /ubbthreads/images/graemlins/crazy.gif /ubbthreads/images/graemlins/popcorn.gif