Soldered a Noctigon to a Heat Sink

Just as the title says. I put one of my hybrid Copper/Aluminum heat sinks in the oven and soldered a Noctigon onto the copper center goodness.


This is the heat sink before soldering.

After soldering and machining to fit into a mag reflector.

I'm pretty excited. I'm hoping to push some amps through this baby.

Different angle

Did you reflow the led on at the same time?
I just did it once and found it not so perfect because it takes forever until the pill heated up and then forever until it is cooled down again…
And you have a heavy chunk there which should be even worse…

You have the copper core to get the heat out and spread to the aluminum, and the aluminum to sink it. You can expect to be able to push some serious amps through that LED.

That looks awesome !
I hope you’ll provide more photos and details of this build.

Nicely done. Looking forward to the finish of this build.

Nice work. I'd be pretty exited too if that was my heatsink.

IMO don’t see the bennies of adding an inherent thermal-block through solder itself which isn’t a great conductor of heat to begin with especially if it’s 63/37 lead biased or to that effect. Plus it just puts added stress on an already reflowed led. OTOH disregard this comment if you used a silver-based solder mix that would thermally rock as would a copper solder like this….for that.

Anyway FWIW and if I didn’t want to mess around with soldering if it were me I’d screw down that Noctigon with a high-grade silver based grease and very little of it. Well first I would lap everything over a plate of glass then screw it down to that copper core nice and snug. The screw heads themselves if they can be elevated off the Noctigon somewhat can act as an additional heat wick off the core. Copper screws would be great as would aluminum.

But yeah either way it’s gonna crank out real good ampedlumes. Looks great!

Edit: BTW, I take it you’re gonna machine some fins too?

60W/mK of regular solder is good if you solder the Noctigon with some pressure. Way better than any thermal paste.

Grinding on a glass surface is not called lapping, its still grinding. Lapping is much more complex but gives great results.

surface or real lapping is better but does takes some work! I don’t have the patience for it. lol

I used ‘lapping’ as a term very loosely here when used through very very fine wet sandpaper, silver thermal grease, which for some folks lessens the valleys and peaks enough to work good enough to see improvement. But you’re right it’s more grinding than lapping but still a finer form of grinding. The trick here is applying even perpendicular pressure to the glass plate with each piece. I find the best way to achieve that it is with a mill drill chuck and a 3/8” piece of glass and some slight pressure. I guess you could achieve that with even a regular drill press if the speed is controllable. The less lead you have in solder here the better off you’re going to be. Lead is not a good transfer conductor of heat at all regardless of how much pressure you apply in the soldering process. Is it probably better than a very small amount of silver thermal paste clamped under pressure with screws? I kinda doubt it. Would silver or copper solder under pressure be better? I kinda think so.

But hey you may be more right especially if you have proof. :bigsmile: So I defer.

More pics coming as build progresses. I may just wire it up as is for some testing if I can figure out how to detune an sst-90 driver to below 7 amps.

Errrr Pb solder is far, far better than the best thermal grease? Sounds a bit wikipedia to me.

Cant wait to see the results. Got my xml2 up to 7 amps last night, whoops :slight_smile: but its still alive!


Tin: 67W/mK

Lead: 35W/mK

67*0,67 + 35*0,37 = ~58W/mK for 63/37 solder. The best easy available thermal paste I found is still below 10W/mK, so solder is about 6 times better. And if you solder with pressure, you can spread the solder as thin as thermal paste, possibly even thinner.

What proof do you want? If you grind both surfaces flat, to a mirror finish (2500 grid) and then screw them together you might as well not add anything inbetween. If you are more lazy, solder will do better.. and you dont need screws.


It seems wikipedia lists the same thermal conductivities as almost any other source, so whats your point?

Sorry, internet sarcasm NC :slight_smile:


This chart doesn’t show how silver thermal paste would compare but I’ll have to take your word for it.

Copper and silver solder IMO is still the best way to go so I agree with you at least in the soldering under pressure principle. The bottomline no matter what is that the less air gaps you have between materials the better. That’s why lapping works.

Diamond paste (zirconia substituted too maybe?) under pressure OTOH could be very interesting perhaps?

The limitation of using screws is that if you tighten them enough to get good enough contact, the board warps and you end up with either not enough paste to fill the resulting gap or needing a thicker layer of paste, neither of which are issues when using solder. So even if solder is on paper not as good as some pastes, in actual use it can be much much better. If we used our thermal pastes on a test rig those specs might matter, if the design of the actual real-world object doesn't allow the material to perform the same as it does on the test rig then the test rig results are no longer relevant.

Just search for any silver thermal paste and see if the manufacturer provides thermal conductivity. If that stuff would be that good, they would all advertise it. Even Arctic Silver is specced at only 8.5W/mK.

There are some glues that work very well. The DEFT uses some special stuff.. very expensive. Vinz had his own glue manufactured, I think it containts a lot of diamond powder. Main advantage is that you can glue it to almost anything and its still at around 60W/mK.

I have yet to see any thermal paste with >20W/mK that does not cost a fortune..

What I find most interesting is that even with cheap white generic silicon-based paste, and tiny little M2-M2.5 screws that can't provide enough clamping force to get a proper interface, and MCPCBs too thin to stay flat enough for good contact across a majority of the surface, it still works just as well as if all those problems didn't exist.

What's the light output change between cheap paste, expensive exotic paste, and solder? Is it even measurable? If it's not measurable is it worth spending money on?

The most important thing by far is that there is SOMETHING to fill the void, after that what it actually is, is of lesser importance. Just because lead is less of a conductor than copper does not mean that it will totally insulate the 2 surfaces. What is a good insulator is dead air. That is the primary purpose of thermal paste, to rid the dead air and replace it with something that is a better conductor. Here is a roundup of different materials used as thermal paste, including some common household materials. Of course some are better than others, but anything is better than nothing.

Soldering the PCB to the pill will make the light get hot faster, so it's worth to spend some money on. I use solder or Arctic Silver, although AS is considered being rather expensive. I don't think so, because 7g of AS are good for 20, maybe 30 lights.