Non-electrically isolated slugs?

[purduephotog]

I'm going to be purchasing some Avago dies- and they apparently don't offer the 3W units I want with electrically isolated slugs. Usually I drill and tap, but I was going to go hog wild and do this entire batch with epoxy.

My thought: Thermal Epoxy a piece of anodized aluminum to the slug, then thermal epoxy it to the heatsink. That sounds better than the drill/tap approach that I'm currently using with the ledengin dies using mylar to electrically isolate them.

Suggestions? It's 13 blue to a DX driver, then at 2.2V it would be ~20 reds. Repeat. I can get the Cree unmounted cheap enough but I can't find any small MCPCB that take more than one and don't require insane amounts of prep time.

 

[kethd]

All your LEDs will be sharing a common heatsink?

"People say" that anodized AL is electrically isolating. If this is true, you have four layers of elecrical isolation between each LED junction and the common heatsink: two layers of epoxy, two layers of anodizing (if you are very careful when glueing to ensure uniform layer with no direct metal-to-metal contact). That means you would have eight layers of protection between any two LED junctions. That should be enoough!

Guess you should carefully inspect surface condition of anodizing for scratches etc? I would want to take anodized sample and agressively attack with ohm-meter pointy probes in buzzer mode, find out for real just how electrically isolating your anodizing layer is. Better yet, make a simple 100-volt very sensitive buzzer circuit to probe for anodizing breaks, with one end bolted through into interior metal.

Are you going to too much trouble? Why not just very carefully thermal epoxy to common heat sink (preferably with an anodized surface)? Seems like you could put down some thin insulator (electrical tape?) leaving opening smaller than the slug surface, thermal epoxy to remaining window and be sure of no direct metal-to-metal contact?

Or, wouldn't it be a lot simpler to just use the commercial double-sided thermal pad material, and make sure your common heatsink is oversized and cool enough to live with that one not-so-great thermal interface?

Depending on the specifics of your application, there is alternative of direct bolting to individual heatsinks, just keep them separate -- you need about the same amount of total heatsink surface either way, and it is actually better in theory to just distribute the heat sources than clump them and then have to distribute the heat...

Whatever you do, if you have a way to measure your resulting die junction temps, would be interesting to hear about.

 

[purduephotog]

Well, that's the inherent issue- how to ensure that there is no metal contact. I looked at some mesh as a possible method of ensuring some form of standoff- but it is all fairly thick which would defeat the purpose.

edit:

I may end up taking some mylar bags (anti-stat bags) which have worked well for isolation in the past, hole punching a small hole in them and laying down a strip of them. Dab the epoxy in the middle of the hole, place LED on said hole, push down- should work.

I hope.

 

[kethd]

Like the idea of using a hole punch to make windows for epoxy, if you have slug bases a little bigger than the hole punch size. (I'd try punching the holes in electrical tape.) If you try this let us know how it works!

If you build your original design, you cannot measure the insulation resistance of each installed layer. But you can come close -- you can measure the megohms (or inverse) of each epoxy-anodizing pair. If they all measure good, you would have so much redundancy that the chance of ever having a slug-to-slug failure would be very slight.

Even if you epoxy directly to a common heatsink, if you measure the resulting electrical isolation, you still end up with one level of redundancy, enough for most purposes. But you would want to put fuses in your drive circuits, and generally be confident you had some protection from catastrophic failure.