Direct bonding to copper: p60 triple nichia build + p60 XML build

Flashlight Modding and DIY Parts
1

After recently reading some threads about the benefits of direct bonding leds, I have decided I want to do this in a p60 host. However, I have very little knowledge in electronics and stuff of this nature so I wanted to talk out my ideas first. I have decided on 2 builds, a triple nichia and a single XML drop-in. I have some ideas and was just wondering if they sound plausible and if you guys have any recommendations etc.

triple nichia:
So with this build, I plan on using a copper rod cut to size with a hole drilled down the middle. On one end, I would solder the filed down brass pill so I can easily mount a driver. Instead of sticking a triple XPG star to the top, I was thinking of making this whole copper chunk a “star” using the method OL talked about in his thread (I want to eliminate the thermal grease junction so would rather not make a separate copper star). I will solder 3 very small pieces of rectangular pads on top for bonding the LEDs. I will probably have to use a torch to heat the whole pill to reflow the LEDs.

XML:
Similar to the triple nichia, I want to use direct bonding to copper. Initially I was planning to just make a copper star (as mentioned in the OL thread) and reflow an XML-2 to it. I would then install this star as a typical p60 build. Then I would add a copper round inside the driver side of the pill for extra heat sinking. This I hope would help with heat management at the LED.
However, I was also thinking of drilling out the middle of the pill to get rid of the brass and to just stick a whole rod of copper in it instead. This copper chunk would serve as the star and an XML-2 would be reflowed to the top with a torch (I want to get rid of the star/thermal grase/brass pill junction.)

Does this sound crazy or would it work out?

2

Might be easier to solder a copper star to the copper rod and the reflow the led onto that. Also better with a copper star which places more copper than those little pads before the next solder joint.

3

I would do that for the XML drop in but I don’t think there are any 16mm sinkpads made yet. Same goes for the triple XPG boards. The goal is to get a direct bond/path from the LED to the heatsink with no dielectric in between.

Instead of making the whole copper pill the star though, I might just make separate stars then attach it to the heat sinks as it might be easier to reflow.

4

You might want to take a look at this thread:

Especially the first picture and the few lines directly under it.

5

Wow that is basically the ultimate multi-emitter p60 drop in build. Amazing.

I don’t have the tools to do that but if I did I would. Once I find some time (which might not be for awhile), I’ll start working on the drop in and post some pictures. I think I’ll first try out the OL method of direct bonding as that seems most feasible. If that does not end up working, I may have to just stick with a typical star and not do any direct bond to copper.

6

Good one! I was thinking of that thread. You could epoxy an aluminum star to the rod and drill through the center pad into the rod and solder in a copper link.

7

same here - that’s the first thread that came to mind :slight_smile:

personally, if I were going to the effort to make a funky pill for awesome thermal transfer, I would go with a light that has a screw in pill (C8? Bunch of EDC hosts on Fasttech) and work with that. The pill to housing thermal junction will be far superior to that of a P60 pill to housing. Another advantage (depending on the host) might be that you can adjust the height of the pill for the triple Nichia to account for the much shorter height of the triple optic (although you might then have problems with battery fit).

Adding extra copper to the pill is just going to increase the thermal mass of the pill, it won’t make any difference to LED temperature once it’s at a steady state (say after a few minutes on high). I’d focus more on getting the heat out of the pill first. Still, if you want to swap these pills between multiple lights, I can see the appeal of a P60 module, but that’s about it for me.

8

Wont replacing the dielectric and aluminum in the center pad with copper make a significant improvement on thermal transfer at the critical point, right under the led junction?

9

@matt: Thightly wrapped P60 dropins transfer heat very well. I'd say even better than the C8. Why? There is air between the threads. I think the myth about bad thermal design comes from people who dont wrap their dropins. My P60 lights get warm very fast after turn-on if I give them enough amps.

Also, the copper pill in triple/quad lights is not about thermal mass only but to get heat away from the LED more quickly.

Adding mass to the pill wont help with heat transfer, replacing the dielectric with solder or copper does.

10

IMHO, the poor thermal design of the p60 centers around the brass pill, a thin, poorly conductive material with no direct contact to the host. Any improvement here wil help.

11

I find these discussions very interesting, and as a newborn flashaholic am trying to implement some of what I’ve been reading. Now, this work is done by Fred Pilon and done quite beautifully, on a design idea by me. Fred implemented it better than I could have envisioned as he’s a perfectionist extraordinaire!

This light is small. 3” long running a 10440 IMR to power a Cree XP-G S2 emitter. It’s not complete yet, but very close. The body and head are Surgical Grade Titanium (Grade 23) so the copper heatsink fills pretty much the entire head, with exposed ribs to outlet the heat.

Your thoughts? Skyrider, you might look at this as Matt has said, threading the pill into the heatsink for maximum contact. The pill here will be copper, threaded into copper.

!!

And here it’s fitted into the Ti head with bezel mounted.

!!

Nightcrawl, I don’t think you’re going to find much air between these exquisitely machined .5mm pitch threads! :wink:

But as I’ve said, it’s not complete so the proof in concept is still waiting to shine.

12

A little thermal paste on those threads eliminate any air gaps. I wouldn’t ever use any sort of silicon or adhesives on it…

13

Yea, these posts/discussions really help me figure out what to do and what works best etc.

I really would love to be able to thread the inside of a p60 host and get a nice chunky copper pill threaded to fit but I just don’t have the tools or the money to get someone to do it for me. I am sticking to truly budget as possible.

The direct bond is to get the heat away from the LED as quick as possible. The extra thermal mass is for it to be able to hold more heat. Adding copper in the p60 should also help heat move faster to the host as a lot of it is just air initially. I will also use copper foil to wrap these drop-ins to a tight fit too to get as good as possible heat transfer from the drop-in to the host. I’m currently have 2 debates about which way to go for each drop in.

For the nichia I am going to first try and just direct bond the LEDs directly to the copper chunk. If this becomes too hard to do I will make a separate star. If this is still too hard, I’ll just stick with using a typical aluminum star.

For the XML, I’ll try boring out the brass in the center and just sticking a whole copper rod in it. If that is too hard, I’ll go with a copper star route or get a 16mm sinkpad if those are out by then.

I most likely will be running these relatively “conservatively” at 3amps max each. This will probably net 700 lumens OTF for the XML2 and 500 lumens OTF for a triple nichia, hopefully.

Once these build are done, I’ll probably also build one of those custom IS so I can get some relatively accurate lumen measurements.

14

that really is beautiful, Fred is quite the machinist! Personally I would have tried to make it one piece, with no pill, so that there’s a direct path to the outside air. Even so, the fine pitch of the threads should mitigate a lot of that and make assembling the light much easier. It’s certainly going to be a stunner when it’s done.

As for the whole thermal mass thing, I’m starting to sound like a scratched record here, but it’s pretty simple physics (simple enough for me to understand at least). The more thermal mass, the more hysteresis in the system. In other words, there will be more lag in heating up (as it can “absorb” more heat) and more lag in cooling down (as it has more heat to dissipate). For short bursts, lots of thermal mass is fine, provided you don’t saturate it. This will help keep the LED cooler for those short bursts. More thermal mass = more time before it’s saturated/ at equilibrium for a given thermal input. A pound of copper would most likely take longer to heat up (and it still has a decent surface area anyway) than most torch batteries would run for.

For sustained use (ie. once the LED and housing reach a steady state temperature), the ability of the housing to dissipate the heat generated (which is the result of the thermal resistance of the heat path and the surface area of the housing, plus the difference between housing and ambient temperatures) is more important. The fact that most of your high power torches cannot be left on for an hour or more without overheating (ie. housing temperature does not stabilise or will only stabilise at a temperature way higher than is safe) demonstrates clearly that the housing is unable to dissipate the heat generated. Adding more mass to the pill isn’t going to make the blindest bit of difference, it’ll just delay slightly how long it takes for the housing temperature to start rising. Even directly bonding the LED to the pill isn’t going to help if the housing can’t dissipate the heat.

As for P60 drop ins vs. threaded pills, the argument that you can create a thermal path between a P60 dropin and the housing isn’t evidence that the thermal resistance of that junction is lower than a threaded pill installed with the same care (ie. using thermal paste), it just means you’ve done a good job working around the disadvantages of the P60 format. As long as the pitch is fine enough and the threads fit well, you’ll have far more metal to metal contact (excluding the thermal past filling any gaps) than with a P60 pill - otherwise your pill will just rattle around and come loose, right?

15

Yea, I think a threaded pill will be better but a well fitted p60 drop-in should work relatively well.

The reason I want to add thermal mass is because it would also speed up moving heat to the host. I think all of this will allow for a longer period of time before the host to air heat movement is saturated. Although it still might not allow me to use high for a full battery, it would allow me to use it for longer and for the LEDs to not drop in lumen output as much (thermal sag). And if I use these lights on the lower level, the extra heatsinking will help a lot with keeping the LED output up.

16

I think, in your position, I’d get some of the copper sink-pads at 20mm made for the XM-L and reduce it’s dimension to 16mm, then reflow the emitter onto it which should be fairly easy as it’s designed for this emitter.

I am planning the same thing myself and have never done it before. My plan is to mount the sink-pad on a 1/2” aluminum bar using JBWeld, then use my cordless drill to turn the pad slowly while cutting it down with a grinding disc on my dremel tool. This should keep it perfectly round (or very near so) while keeping the emitter pad centered. Then cut the JBWeld off the backside of the pad and clean it up a bit with sandpaper. Laying the sandpaper on a glass surface, or formica, it’s easy to keep the part neat and flat when sanding. I also usually use a rotating pattern when doing this sanding, to ensure I don’t lean to one side…but still check it often during the process.

It’s my plan to use a 6” SS disc I have and put it on the burner element of a glass top stove. This will keep the element working properly and give me a good hot base to bring heat up in the copper sink-pad to reflow. I’ve seen some that recommend simply killing the heat when the reflow is accomplished and leaving it to cool naturally so as not to interfere with the joint or the emitter by moving it. I plan to relocate it to the glass stove-top though, to cool somewhat faster than that.

Like your plan to put copper in the core of the pill. Never have been able to understand why they use brass instead. Cost in that small of a part is offset by machining costs as far as I’m concerned, so add the extra buck or two to the finished light and call it efficient.

Will definately watch to see how you go about this build, the copper core in the pill really interests me.

17

I probably won’t be able to start this build for awhile as I am away for college and my tools are at home. I have to wait until May (2.5 months) to try this out. I’ll make a new thread once I start though to share what I build (and what I can’t if I fail :_( )

Will filing down those 20mm Stars to 16mm work? As in are the contacts close in enough that it would be ok? I might end up using that instead of building my own star. This is if my initial plan to direct bond the LED to a large chunk of copper fails. I’m also hopign the 16mm sinkpads are out by the time I start

18

I’m very impressed with Fred’s work, to say the least. Those threads, man, just incredible! He made a pill fob for my wife from the remaining part of the Ti bar I sent him. I’ve got that in hand and I can testify you will not find a smoother operating thread anywhere, and this is the fabled hard-to-thread Titanium! Same .5mm thread pitch, and smooth as hot butter on oiled Teflon.

This light will be worn on a Ti curb chain around my neck. :slight_smile: The piece-de-resistance of my flashaholism period. :slight_smile:

19

I'll try to make this short.

DBC, thats a nice little thingy. But you cant compare that to the usual C8 threads.

Pulsar, metal to metal is far better than thermal paste. But I'm sure you know that. It helps of course, but not much.

skyrider, if you get a direct to copper or brass bond of the center heat pad and wrap the dropin tightly, you are good to go. 3A is easy if you always hold the light with your whole hand. Human blood works very well as a heatsink.

matt, you have some good points but some are not so good. First of all: more thermal mass does not mean that the light will cool down slower when powered for the same time with the same power as one with less thermal mass. Thats just wrong. Adding mass does not change the ability of the body to dissipate heat.

Of course you need thermal mass for short bursts of high power. But not mainly because the light will get too hot to hold (you could use gloves) but because you move the point where thermal sag happens. Thats the main reason for copper bonding: more light. Who cares about heat, turbo mode has the name for a reason.

Now for your argument about threads. The pill doesnt rattle when screwed in, thats right. But not because the threads are so tight (you wouldnt be able to screw the pill in by hand so easily if that was true) but because once you fully screwed in the pill and screw it tightly, there is pressure on the threads. A little less than half the area of the threads will be in contact at that point. If you dont believe me, take a threaded rod and two nuts. You can screw them on very easy, now screw them tightly against each other. You wont be able to move them by without tools anymore now.

Ok, that wasnt short. Sue me. ;)

/edit: DBC, to make Sinkpads round, you could solder some thick copper wire to the back. Manufacturers use brass because copper is a PITA to work with, that would boost machining costs. Also, if you reflow, let the emitter cool down slowly. They dont like thermal shocks.

skyrider, its no problem to sand down stars. If the electrical pads come too close to the edge, file them off (I use a dremel for that).

20

Keep an eye on Nitro’s thread, he’s trying to get Sink-Pad 16mm’s made.

Vestureofblood is selling the 20mm copper stars now, in XP-G and XM-L formats.

Fred turned a 20mm star down to 10mm for my build, so anything is possible!! :slight_smile: