Luxeon emitter heatsink help

[MrAl]

Hi there,

Has anyone here been using the emitters only (1w) and
if so what did you use for a heatsink?
I recently ordered an emitter and i would like to
know what has worked in the past.

Thanks in advance...

Take care,
Al

 

[HarryN]

Hi - I have used 1 and 3 watt emitters - all I use. The hot lips is a popular item, and I think I remember someone making a self centering version.

My personal method is to use silver filled epoxy bonded to anodized aluminum. The anodizing provides an electrical isolation between the heat sink base of the emitter and the world, and the silver filled epoxy is a very nice thermal conductor. I also use the same epoxy to electrically bond the emitter wires to other connectors - this keeps it within the temp / time limits of the emitter.

Mcmaster carr carries it as well as others.

 

[Doug Owen]

Yes, I've used a few. In each case the heatsink followed the normal rules, good device to heatsink contact, lots of surface area, short thermal path. At least as best I could.

Cases that come to mind include a pair of the stamped aluminum wings used for TO-220 packages screwed together. The device is glued about where the die on the T0-220 would be. In another case, a small aluminum block is itself bolted to a heavy machine frame. In another I used a sort (say two inch) heavy (say 5/8 inch) stainless steel bolt, not the best choice for thermal conductivity, but it has to live in an acid environment. I also used a metal disk about the size of a quarter...much like the star which I should have used but didn't have handy.

Being that sort, I generally lapped the emitter and contact area and used a small weight on a ring around the emitter to get a good thin layer of the Arctic epoxy.

Just like power transistors.

Doug Owen

 

[robk]

If it's for a "D" mag, I have a few extra "slugs" with pedestals. You need to machine out reliefs for the leads and drill holes for the power wires. If you want a couple, let me know. You've been a tremendous help here with your circuit ideas, I'd be pleased to help you out with it no charge. If you need it drilled, I can do that for you. Photo attached.
Rob

 

[MrAl]

Hi again,

Thanks very much to HarryN and Doug Owen and Robk!

Let's say i dont have any Arctic epoxy or other thermal
epoxy...what would be second choice?

Also, is it possible to clamp the thing down somehow
using a method similar to CPU's (on motherboards)
heatsinks ? ... these use spring pressure...

Rob: it's for a minimag 2xAA type.

Thanks again,
Al

 

[robk]

Since it's a MiniMag, your best bet (cheap but seems to work) is to use a circular piece of copper clad PCB to transfer the heat to the mag body. I have some "emitter boards" that would work, the only issue is getting the emitter perfectly centered by eyeball, and using a good thermal epoxy like Arctic Alumina available at "The Shoppe".
Rob

 

[pbarrette]

Hi al,

I've been thinking about machining an aluminum pedestal which would have a "neck" that was only slightly larger than the emitter. I would then machine a tube with a thin lip. The lip would be inside the tube and have a hole just large enough for the emitter to fit through. The tube would be just large enough to slip over the pedestal base and would provide a friction fit. The idea was to hold the lux firmly in place, yet still allow it to be easily removed for replacement.

I'm guessing that you don't have a lathe or mill though.

In a pinch, I've just used a very thin layer of epoxy. Super-glue doesn't seem to work as well though. Another option might be fingernail polish. I've used that in a pinch before. It's really just an enamel and it seems to do well at higher temperatures. You just have to use a very thin coat and wait ~24 hours for it to dry completely.

I also drilled a hole in a plastic, screw-type bottlecap. By placing the cap upside down over the lux, I can set weights on top of the bottle cap while the adhesive is drying.

Another thing I've used before is a just small amount of thermal grease. I purposely clipped the wires from the driver slightly short. Then I pushed down on the lux as hard as felt safe while I soldered the lux in place. In that one I was just relying on the opposing pressures to hold it in place.

I guess in the end, the quality of heatsinkning would depend on the amount of heat you plan on generating. You could probably cheat some for a Lux-I at 350mA, but probably wouldn't want to cheat on a Lux-III at 1A. I've been able to get away with cheating at the lower end of the scale, anyway.

pb

 

[MrAl]

Hello again,

Rob & pb thanks for the info.

Rob:
The thermal epoxy 'Arctic Alum' seems like a good idea.
How much is that stuff for a small quantity?

pb:
Some interesting ideas there. I do have lots of thermal
compound but it's the standard stuff not epoxy so maybe
i can use your 'push down and solder' idea. Im wondering
if there is enough room on the emitter (it didnt arrive yet)
for a small 'sleeve' to fit over it where i can use
Rob's pc board idea combined with some spring steel wires
to hold the thing pressed firmly against the copper clad.

Im planning to run at 350ma. I've tried another Lux at
higher current but dont see an appreciable light increase
to warrent operation like that for any length of time.
In one light i went from 2 ohms series resistance to
0 ohms and the light increased, but not really 'that' much.
The extra heat just doesnt seem worth overdriving.

Thanks again,
Al

 

[KevinL]

MrAl, what I do with non-adhesive Thermalloy Thermalcote grease is that I apply enough to cover the metal slug of the emitter, put it down on the heatsink, then use a toothpick to apply cyanoacrylate-based superglue around the rim of the emitter. The rim is raised slightly, and the low-viscosity glue wicks inside perfectly. You might be surprised at the bond strength - I know I am.

 

[Doug Owen]

Al,

The AA is fairly reasonable (about $15 IIRC), the silver a mite more.

You can use standard 'goose grease' if you clamp it somehow. There's a back plastic housing you should be able to rig a bit of bent piano wire to press down on I bet.

Otherwise, I kinda like the 'glue it after the HS compound' idea, perhaps even with normal epoxy?

Doug Owen

 

[PEU]

Robk, on what diameter you settled for the recessed luxeon extruded cut? 8mm or a bit more?

I second your offer Rob, I can do the part for you too Al.


Pablo

 

[lambda]

MrAl, you can mix some of your thermal grease with regular epoxy and it will work almost as well as Artic Silver for your application.

 

[robk]

MrAL - I think I paid less than $8 for "AA" epoxy on some overclocking website. Try a search.

PEU - The dia. was .320", which my caliper shows as a little under 8mm, I think the next run I will slightly taper the depression, .320 on top, about .310 on the bottom, as the emitter is slightly loose and the taper will allow the epoxy to neatly flow up and around the black plastic part of the emitter. See the attached drawing (original design, no taper yet).
Rob

 

[robk]

MrAL,
It was Plycon:
http://www.plycon.com/grease/grease.htm
$8.99 - scroll down 3 or 4 products.
Rob

 

[MrAl]

Hi and thanks everyone for the great help with this issue...


KevinL:
That's a very interesting idea for sure, and i'd like to try it.
The only thing i would worry about (and i had this same worry
when using superglue for gluing cracked magnetic inductor cores
back together after cracking to form a small air gap) is what
is the temperature rating of the superglue? If the Lux gets
warm does the superglue soften up much, and if so what temperature?
Then again, i guess i'd have to ask this same question about
thermal epoxy too. I know regular epoxy melts at some temperature
as it acts like plastic.
Any ideas about the super glue temperature melt (or soften) point?

Doug Owen,
I guess 15 bucks isnt that bad. I should prob get some anyway for
future use too.
Yeah, i was hoping there's an 'edge' to catch the wire on. I was
even thinking of soldering the ends of the wire to the clad which
would make it easy to repair too.
I like Kevin's idea about glue after HS compound too, but i'll pose
the same question to you as to him...
What do you think the temperature rating of super glue (soften)
would be?
I like your idea of using 'normal' epoxy too, as i have some
PCII which has a fairly high temp rating. Would have to be clamped
very tightly overnight for cure. The only worry after that is
'thermal pumping', which i think a spring clamp handles a little
better.

Pablo:
Thanks very much Pablo, i really appreciate it. I'll try the pcb
idea first and see how it goes and maybe save you the time.
I could even pick up a piece of copper bar which would conduct like crazy /ubbthreads/images/graemlins/smile.gif
Havent some people even used a copper US penny?

lambda:
Wow, that sounds interesting. I'd be afraid that the grease would
interfere with the proper curing of the epoxy however. Have you ever
tried this on an actual LED and if so how well did it work at 350ma?
I've mixed paint with epoxy and even silicone rubber to keep it
a little flexible, but that was with very small amounts.

robk:
Oh, that sounds pretty good...i'll do a search...thanks.
Oh ok, i wont have to do a search /ubbthreads/images/graemlins/smile.gif Thanks!


Take care,
Al

 

[andrewwynn]

I was in an 'impatient' mood and didn't want to bother with mixing up thermal epoxy so i used the kevinL trick, but a little different.. i didn't want the heat conducting paste to inhibit the glue so.. i put the heat sink compound on about 80% of the area of the slug.. and used two drops of superglue on opposite sides and stuck it on.

In testing i got the heat sink up to about 170F and the emitter is still hanging tough.. i don't know hos much heat superglue is made for.

This was with a 3W emitter and i actually have two on the heatsink.. one is epoxied with arctic silver the otehr with this superglue trick and just RS generic 'heat sink compound'. I should do some temp measurements to see if one stays cooler.

I have run them for extended periods at 1A and no problems yet...so i think the trick works fine. The only caveat is that super glue can be impact sensitive.. it's conceivable that a drop could knock it loose.. when you work out the math.. even with light objects it's possible to put 100s of G forces on something with a fairly simply drop... say you drop a flashlight... but the tail hits something on the way down so it spins.. and it falls 4-5ft before hitting cement, but is spinning.. easily over 100gs on the head when it hits.

 

[KevinL]

Not sure on the breakdown or softening point of the cyanoacrylate, but this trick has been applied to virtually every emitter I've worked with, probably something like two dozen. No reports of field failure yet...... yet /ubbthreads/images/graemlins/wink.gif

andrewwyn: cool, it worked for you. I coat the whole slug with thermal grease, then put the cyanoacrylate on the plastic collar of the Luxeon. It's possible that a sufficiently strong shock may jolt it loose, and anything that compromises the integrity of the plastic collar will also break it loose.

However, this has saved me before on a severely botched mod. I was able to use a very thin knife to slice the glue around the collar, and gently pry the emitter loose. It turned out to be easier to cut it off than I expected, so that's a TV1K that lived to light another day /ubbthreads/images/graemlins/smile.gif

 

[MrAl]

Hello again,

Andrew:
Oh yes, i hadnt thought about the fall impact effects...
thanks for mentioning that.
I guess what i could try is a layer of HS grease followed
by two to four dabs of super glue, then let dry, then
apply a ring of high temperature epoxy (which i have already).
If that doesnt hold it i guess im beat /ubbthreads/images/graemlins/smile.gif

Thanks for all the ideas, which means i'll be able to use
the emitter as soon as it arrives rather than wait a bit
longer for another order of thermal epoxy.

Come to think of it, Andrew's notes are good to think
about for any flashlight construction with or without
special gluing techniques. One thing i hated about
bulbs (God rest their pitiful souls) is that one drop
and the filament would usually break rendering the
flashlight useless if you didnt carry a spare bulb.
In order for LED flashlights to truely overcome this
problem some care has to go into the construction.


Take care and thanks again,
Al

 

[snakebite]

cheap and quick mounting with just ca works fine even without compound if you force it down so the ca is mostly forced out of the joint.
same with jb weld(steel filled epoxy)

 

[MrAl]

Hello snakebite,

Oh yes, ok, i would imagine anything in a thin enough layer
would conduct heat fairly well. I just want to be SURE
nothing comes loose because i'll be running this little
thing at nominal current (350ma) which will heat it up
and burn the phospher (causing a blue sun spot) which
will ruin the flashlight good if something goes wrong.

Take care,
Al