Mounting an XR-E without the star module

[Myself]

So, I squished some Cree XR-E emitters the other day, mentioned here: http://www.candlepowerforums.com/vb/showthread.php?t=223988

Today, I decided to do some experimenting. I desoldered them from their star modules, and attacked one with some serious scissors. The ceramic substrate on those suckers is HARD! But I was able to lop the corners off, which was my goal.

Why remove the corners of the substrate? Because that's where the vias are! I want to eliminate the thermal resistance of the star module by soldering my emitter directly to the heatsink. This requires either milling channels in the heatsink to run wires into, or somehow disconnecting the electrical contacts on the underside of the substrate. The latter is the course I took.

The vias are clearly visible here:

After desoldering the emitter and trimming the corners off, I verified with a meter that the bottom contacts no longer connected to anything. Then I soldered wires to the topside contacts and tested it:

My theory thus confirmed, I then scuffed a nice clean area onto the end of some copper stock, added a bit of flux, and applied plenty of heat:

Success! The emitter is soldered directly to the copper bar but both contacts are electrically isolated from it. I can now buy bare emitters rather than star modules, mount things in ways stars wouldn't allow, and enjoy a more direct thermal path. I can use pre-1982 pennies instead of star modules if I need convenient slugs for more elaborate mounting.

Full-res photos and mouseover annotations at http://flickr.com/photos/myself248/

 

[Scattergun]

Why don´t you just buy bare emitters instead?:thinking:

 

[Myself]

I didn't know until just now that this was possible!

I plan to buy a mix of bare emitters and assembled stars in the future, because stars are convenient for a lot of things. Aluminum, for instance. This trick only works on copper heatinks, unless there's some insane flux I'm not aware of that makes it possible to solder to aluminum.

I might start making my own "stars" out of copper slugs, to enjoy the better thermal conductivity in the first stage, before screwing said slugs down to larger aluminum heatsinks. (The way I've been doing it, attaching an aluminum star to a copper heatspreader, seems suboptimal.)

 

[bretti_kivi]

I don't get why you don't use heat transfer compound (arctic alumina) or glue? Why bother wanting to solder to the heatsink?

Bret

 

[Myself]

Because the thermal performance of epoxy or glue is utterly dismal compared to solder. Also, in order to maintain electrical isolation, you need a certain minimum thickness, whereas with this technique I can get direct substrate-to-heatsink contact and just let the solder fill the gaps, again decreasing thermal resistance.

 

[bretti_kivi]

good reasons...

So, you're using a similar technique as others use for Rebels. If you were to connect the copper to anodised aluminium and were to mill / saw some upright fins into it, it would probably make a *very* nice heatsink.

Bret

 

[Myself]

Yup. I figure all the heatsink in the world is useless if it isn't efficiently coupled to the LED. And any gains I can make in that path will make even a modest heatsink more effective.

Since my current interest is plastic-body retrofits, I'm obviously contstrained on that end of things. So improving the heatspreader's effectiveness depends on coupling it more closely to the LED itself.

Whether the trick detailed in this thread will prove useful in the long run, time will tell. (I already foresee problems with mounting optics, as they expect to rest against the top of the substrate where I'm now soldering wires.)