Soldering an LED directly to the pill?

Hey guys, I've been lurking for a while now. I'm looking to build myself an XM-L drop in, and I was wondering if it would be better for heat transfer to solder the emitter directly to the pill? Has anyone tried this before?

These two pages suggest that solder has much higher thermal conductivity than even high performing thermal grease like AS5.

Sn63Pb37 solder is about 50 W/mK while AS5 is 8.7

Also, I wanted to experiment with thermal potting; has anything come of this?


I haven't done it myself but as long as the pill isn't aluminum you can solder to it. Since brass isn't a good heatsink material it feels like you might lose the benefit of soldering the LED unless your pill is copper. Don't ask me why they often seem to use brass anyway as aluminum is certainly better. I do understand copper is a pain to machine tho.

You do need more power to heat the larger mass of a pill.

You don’t plan to solder the bare emitter, do you? The bare emitter has electrical contacts under it and the pill would short them all. Maybe you can somewhat solder a copper PCB to the pill and add the emitter on it at the end. I don’t know if this can be done without damaging the PCB first.

I mean soldering the star that the LED is mounted on. Reading the datasheet, it doesn't seem like heat damaging the LED should be too much of an issue. I'm just more interested in better heat transfer from the emitter to the pill.

Brass is the easiest to machine; aluminium's a lot harder than brass.

Soldering an aluminum star to a pill of any kind would be a royal PITA. Add in the fact that getting the star hot enough to properly solder it to the pill would most likely ruin the emitter and suddenly it's not a worthwhile idea.

Kokopelli had about the only practical alternative and that is to use an LED with a copper star (rare but they do exist). You'd have to remove the emitter first, then, after soldering the star to the pill, reflow the emitter back onto the star. Reflowing it back in itself would be extremely difficult without a reflow oven.

Given the available surface area, and the likely thickness of the solder if not soldered perfectly and under pressure, you would probably end up with a thicker layer of solder, and likely see better results from lapping the pill->star contact and using a good thermal paste with a positive pressure design (i.e. pressing down on the star, like in torches such as the skyray, the D10/EX10). I'd also say that the thermal conductivity from the pill to the body is also one that could likely be a more significant bottleneck if you have done a good job on the star.

The loss of output that is likely to result from extended high temperatures will not help either.

IMO, its not worth the effort, and it would be better to simply lap the meeting surfaces better. I find all stars I have installed have suffered from a poor fitting surface due to the metal drawn during the shearing process. Minimal work significantly improves the contact area.

Well thats what I think, which is probably more because I really don't think I have the means to solder a star to the pill. I have seen people affixing emitters directly to copper posts, bypassing the pill completely, but I don't recall them soldering them together, id say that it would take too long to heat up, and too long to cool down.

YES, and here someone did some experimentation with that - note that simply bolting the heatsink was superior to soldering. This is obviously too small a sample size, w/ too many variables to be considered scientific evidence, but imo, it still has some value.

another good description of lapping — making as near as possible perfectly flat surface contact between heat source and heat sink:

Remember, thermal compound or tape, no matter how good, is much less effective than actual contact between the metal surfaces of source and sink; the more metal to metal, the better heat transfer.

Manufacturers who ‘goop on’ a blob of thermal stuff insulate their emitters from the heat sinks — not quite as poor heat transfer as a big air gap, but darn close.

Any Chinese manufacturer who starts selling properly lapped emitters on copper pills to go into flashlights is going to be widely recommended.
Show us your heat sink effectiveness.

Welcome to the forum. You could remove the emitter from the the star, solder the star to the pill with high heat and acid flux( clean both surfaces thoroughly), clean again to remove and neutralize the acid, and then reflow the emitter back onto the star/pill combo. You are likely to ruin an emitter or two learning this process. Another thermal dam is between the dropin and the host but all of the thermal interfaces count for something. Lapping and using a thin layer of grease is likely the most improvement/difficulty for most people.

Soldering the LED to the pill (copper) is possible (not easy), the heat transfer must be the best.