So no sinkpad, just wondering how you dissipate heat from the Led? The sink pad adds more area to do so and also helps with with shorting out the led.
you would go through hell soldering that by hand and if you go for no solder the LED won’t stay in place long enough for you to turn on the power
how did you create the image?
It would work. You might try using solders with different melting points for the central and side contacts. It might even be possible to use adhesive rather than solder for the central thermal pad. I mean, it'd work if the thermal limits of the adhesive aren't exceeded, but the big question is if it would conduct enough heat. Some here say that there's no difference in thermal conductivity between some adhesives and solder, so it might just be good enough.
I don’t see how it wouldn’t short out?
That's already been done long ago, though mostly on LEDs with larger footprints (MTG2, XR-E). It does work if you can assemble the parts without shorting anything.
edit: But it doesn't offer any performance benefit over using a normal direct-thermal MCPCB. Even a poor join between MCPCB and pill has more surface area contact than the LED's thermal pad.
Not a bad idea if it’s done like a pill shaped sink pad. Would need reflow soldering and every pill unique.
JB Weld will tolerate the heat needed for reflowing (at least with 63/37 paste, dunno about higher temp lead free, as if anybody uses that). You could embed the wires in the epoxy, would give you a lot more time to get it all properly arranged, then sand it down to a level surface across the center pad & wires, and then once it's all fixed in place reflow the LED like normal.
I have done it before on a SST-50 a while back, it was ok but a royal PITA to do.
Some great comments.
Indeed I was thinking reflow on the stub that is part of a machined P60 pill. Use a “pillar” as an adapter for heating the pill just below the emitter and tin the mounting surface. The XM-L will float into place so shorting is not an issue.
The wires can easily be attached after the fact. 22AWG solid copper wire will act as an additional heatsink.
I use the 22AWG solid wire in all my lights. Add these to the driver and trim and bend once the driver is installed in the pill.
The question is how hard you can drive this. The XM-L has a nice large center pad already (~3x5).
I got the idea from an XM-L that popped off its MCPCB due to a poor solder joint. I run this XM-L at 350ma with only 2 22AWG wires attached. Running an XM-L at 1400ma in this configuration would be a non-issue. A full 3 amps? probably okay… 5-6amps? Yea, that’s the real question. But it is a direct heat sink like a Sinkpad with the mass of a P60 pill.
Thanks guys for the info, sorry for my lack of knowledge
I first did this many years ago on a pair of AKoRay AAA's in order to use an XPE led. One neutral white, the other warm white. It worked quite well, but I'm not pushing more than 400mA. The NW is my EDC and works flawlessly.
Just last month I did this with an XPG:
I'm sure this mounting method will take a fair amount of power, but I can't help but think that a sinkpad will work better.
-Crux
It won't be better or worse comparing this way to using a Sinkpad/Noctigon (or any other MCPCB without a dielectric layer under the center pad). After all, there's minimal difference between aluminum & copper Sinkpads, they work because of the lack of dielectric layer not because of the base metal.
I think it would work but I think you'd be better off soldering a copper mcpcb directly to the brass pill using a high-temp solder then flowing on the emitter and connection wires with a lower temp solder. Performance would be much better and no shorting out. I also think there would be far less chance of the die coming loose if dropped.
You’d have to go digging, but I think Lambda does (or did) this to all his copper pill mag mods. I remember him complaining about what a huge effort it took to re-flow directly to a 2-3 lb copper pill and that it was testing his patients even after he had designed a solder stencil. I think there was also a compassion between lux performance of the direct bond versus a MCPCB mount, but I dont recall if the MCPCB was copper, al or a direct thermal design. Either way, there’s some good reading and innovation demonstrated amongst those threads.
It does bypass the thermal grease barrier. This is only useful until you hit thermal saturation. Doing this with a finned pill that extends outside the housing is when this could have an advantage.
It would be fun to do some thermal imaging with all the great options available today.
Again, just an idea.
I don't think we know for sure that there would be no improvement at very high currents. I know djozz has done testing comparing different bases, but I haven't seen any such testing of a set up like you mention in the OP. I'm no expert. So there is probably testing out there that I'm not aware of. One might infer from djozz's testing that there wouldn't be an improvement. But without good testing, I don't think we know for sure.
Maybe that work that Flash linked to has an answer.
The thermal grease barrier is not an impediment, even compared to the same parts with the same prep work attached with solder, and even when the thermal grease used is the cheapest white silicon-based stuff you can find (Fujik is not thermal grease, so please don't jump to the 'I guess you think we should just use Fujik to stick aluminum stars to hollow pills, huh?' garbage). If it made a difference I'd be all for it.
Do it if you like, but there are no gains to be found. Originally it was done because there were no direct-thermal stars available. Now there are.
For most general purposes I agree because the degree is indeed minimal. However, if this was not an issue, over-clockers wouldn’t have argued over this for decades now. I do a little work in this arena. It matters.
I use the Arctic Silver epoxy for all my projects. Never had a failure from heat, but a teardown is nearly impossible. I always machine the pill before attaching the star and sometimes I face the star as well depending on the performance I am aiming for. I did finally opt for a Noctigon XM-L2 in a recent rebuild. But to date, I have not over-driven an emitter. The idea of reflowing a Noctigon in the pill is worth considering.
For the short term, I can easily machine a simple pill with a stub… and I have a bare XM-L. And the mass I’m, considering isn’t enough to fret about floating an emitter onto it using a hotplate.
I did find some of the other articles using similar techniques. I didn’t fully appreciate the technology of the Noctigon and the SinkPad until recently. Not that I’ve been looking into this all that long. I do build my own lights and I am all for developing alternative techniques.
Am I reading this right… people pushing XM-L2 to 12amps?
I like it just for the simplicity of design and asthetic quality even if it would be a PITA. I say cool, do it. Convention and logic be damned. Slick.