reinventing the wheel - DIY sinkpad idea

Its not that hard. Getting the right tools is also not that hard. But I think I paid about 1€ for each drill in 1.2mm (XPG) and 2€ for 2.6mm (XML). And its not that easy to make a hole of 2cm into a massive block of copper. Also, I used a 600€ power drill because at least for the small drills because most electric drills dont spin round enough.

I only bother doing that for multi-emitter boards. For single emitters, Sinkpads are just so comfortable. :D

A round hole and copper pedestal was my first thought on this as it was much easier. I wanted to be difficult.
As much time as I spent making mine, I will not be doing it again. Nitro is taking preorders now so I will be going that route.
I had considered making the whole pill out of copper. But the time it takes to make the whole pill with such a small host just was not justified for me. The head and body of the light get so hot now after 10min, it worries me with the body wrapped around a 18650. The host can’t get rid of the amount of heat that’s already being transferred.
A person could possible use a drill bit with the correct size and a rod. If you do that, I would make the rod as close and flat as possible to the actual thickness of the pcb before press it in.
The bottom of the copper rod could easily be matched to the pcb by filing the excess then sanding to a uniform finish. But the top can not be done this way because of the printed circuit on top. You could (true up) flatten the rod and press to just the right height on the top of the pcb and then work on removing the excess from the bottom.
But nitro’s sinkpad will be here soon. :slight_smile:

Thats actually exactly what I did. You would have to do a lot of sanding to remove the traces.

If you look at this picture:

you will see that all the traces are free in the air but thats no problem. It would take a lot of time until you harm them.. and by then you would also start removing the solder pads for the LED. ;)

Well you proved me wrong. :stuck_out_tongue: It can be done. I would have figured you would have destroyed the pcb by the time the pedestal was actual flat. Have you ever done this with a single led pcb? Where there would not be as many points to keep the sanding process flat.

I've been kind of out of the game, but it's obvious that there's been a wave of reflow fever here on BLF and I'm about to catch it. But then again, with everyone doing it, maybe I should just wait until people start selling LED's already on a copper :) Great ideas you guys are having though!

There's been NightCrawl's but also 1 or 2 other variations posted. I was thinking more of a direct connect through an existing pill top down into a copper heat sink inside the pill, or even just a couple of copper discs. So basically you could adapt it to any existing pill pretty easily - though I'm not saying it would be easy...

The other posted variations were more involved with cutting out a rectangle in the star to match the full pad, but it would be a lot easier just doing a simple hole to match a copper rod's thinkness. I'm just not sure about the reflow when you have that copper rod prutruding well below the star - my frying pan reflow method won't quite cut it. What Matt is proposing is simpler in a way, but you don't get a direct bond down to a heat sink.

The only thing I'm curious about is what we are un-doing by having a direct thermal connection, including the SinkPAD's -- there's no electrical isolation, those substrate layers in the original PCMCB's are there to electrically isolate the thermal pad of the emitter from the host's ground, actually battery negative I believe. Has there been any discussions here about that?

No, not yet. I'm an all or nothing dude. ;)

But I will do it with XML and XRE once my order from LCK-LED arrives. I'll post pictures on that. :)

Dont have an XPG light that would need this kind of treatment.. but I might do it anyways if I'm really bored and find a spare PCB. ;)

@Tom E: It actually sounds really easy. If the pill is brass or aluminium, it even is easy. Copper is hard to work with, even if you "simply" want to drill a 1cm deep hole with a 1.2mm drill..

Your point about reflowing with a large copper pill is indeed something to consider. But on the other hand, it doesnt have to be complicated at all. If you put it all together and heat up the whole pill (will work in your pan) you can reflow the LED. If you want to fill the pill with copper pieces and solder afterwards, that should also work if you take a little care not to heat it all too much. Trial and error. :D

I think matts idea would work just as well. See this graph here:

Testing performed by German TLF member Basti. It shows that the material after the star is relatively irrelevant. Of course copper performs best, but the difference is negligible.

On what we are "un-doing".. there have been discussions about that. For most LEDs, there is no negative effect because the center heat pad is electrically neutral, except for some weird LEDs.

Regular PCBs are built the way they are built because its the cheapest way. And you need some of the layers on aluminium PCBs because its not easy to solder to aluminium. Another thing I want to try.. reflow an LED directly to aluminium.. a lot of work to do. :D

Sort of off topic, but my mod'ed ZY-T08 w/XM-L2 U2 SinkPAD, Nanjg @4.2A lost a 7135 somehow - maybe it never worked day one I'm think'n. Anyway, added another 7135, with the 2 Pana PD batts, measured 4.27A, and measured output: lumens - 1,395 @start, 1,366 @30 secs, throw: 100 kcd (broke the 100k barrier). Those Panasonic PD's from FT are outstanding for single cell XM-L2/SinkPAD's. A little better are the Samsung INR's - only got 2 from FT dirt cheap (pulled from a battery pack) @ 1500 mah. Ordered 3 more Samsung INR's at higher capacity off of eBay, and think I'll order more Pana PD's from FT -- excellent in the 7G9 and T08, plus my single cell mod'ed XinTD, UF C8, and UF-V3. The 7G9 did 125 kcd w/the 2 Samsung INR's and 1 Pana PD -- best #'s ever. The AW IMR does well but these batteries beat it.

I've done this, best way I found is to put a chamfer/bevel on the top & bottom of the hole you drill thru the MCPCB. Then cut the copper rod (14ga copper solid core house wiring works great... it's .064/.065" OD EDIT: BRAIN FART - .080") so it sticks out about 1/16 above each side, then squeeze it with vise grips so that it expands into the hole, and fills in the top/bottom chamfers, like a ye-olde-timey hot rivet. Trimming the ends down flush with hand files is A REAL PAIN, small drill press with end mills and a depth limiter would make it a breeze.

On MCPCBs with the center pad connected to bigger areas outside the LED footprint you can add the copper pins all over the board, as it's all the same copper top layer. Depends on how much free time you have.

that’s good info comfychair, thanks for the feedback. I especially like the tip about squeezing the wire to make sure it fits, although I was hoping to find something that I could press in with an interference fit for the best heat transfer. My plan was to push it through with a vise so that the LED end was flush with the star, then trim it off the other end with a dremel.

I just got some parts from a friend to build a milling attachment for my ancient mini lathe, so hopefully I’ll be able to use that to do a more “pro” job :slight_smile:

Ha, I was looking for this thread to make a correction in that post when the notification email popped up, I got the wire size wrong. xD

Try this: cut the copper pegs to length so you have around 1/16" sticking out both sides. Insert ALL the pegs, then file the tops flat while they're still clear of the solder pad, then push then down flush from the top, then do the squeeze/rivet expanding thing. MUCH less filing needed on the topside this way. Trimming the backside is no problem as it's got to be sanded flat anyway after all that abuse.

Oh, and doing it rivet-style is what gives you good contact, even with the initial clearance loose enough that it's easy to assemble. When squeezed the copper rivet gets shorter in the direction you squeeze it, and expands radially to create an interference fit in the hole I.D. The chamfer on top/bottom really isn't needed as the fit will be tight enough as-is, but what it does do is increase the surface area of the copper in the two spots where its needed most.

really sweet comfychair and thanks for the correction, I’ll definitely be giving this a go. I need to head to Home Depot this weekend for some other stuff, so I’ll pick some wire up while I’m there. What size drill bit do you use? I’m not very good with Imperial/ fractions :slight_smile:

great looking star by the way, looks very pro!

Drill is .079", 14ga copper wire is about .080". It's what I would call a 'light press fit', meaning you may need pliers, but only if you're afraid you might chip a thumbnail. Even if it's sloppy, if you have the copper peg long enough it will expand enough to fill it in. I just use 6" vise grips (with the jaws filed smooth, no nasty knurly grippers), squeeze a little, tighten the adjuster a bit, squeeze again, tighten adjustment, etc.

Cut up a bunch of little 1/8" long pieces of copper wire and squash 'em by themselves, get a feel for how much they will expand (quite a bit). Then practice doing rivets on some plain aluminum sheet, once you get it figured out you'll be able to feel when it's no longer squashing and has filled in the hole.

Oh and for the countersink/chamfer I use a Dremel round ball carbide burr, .100" dia. Turning it by hand, not in the Dremel! It's basically not doing much more than breaking the sharp edges around the holes.

Very clever. Great idea and thank you for the picture too.

I guess you could also extend this through the pill and chamfer the back of the pill instead of the MCPCB. Then press them as unit. Reflow the MCPCB to the pill first. This combines Nightcrawl’s and Comfychair’s great ideas.

EDIT: Sorry, just noticed the OP said the above already.
EDIT2: Well, kind of. Press into pill, not press together. Sorry, again guys. I am multitasking to much stuff at the moment.

thanks comfychair, I’ll check to see if I have a 5/64in drill at home (I do miss metric!). I’ll have to have a think about how to do the pressing in as my vise grips are large’n’nasty (mostly used on the car), so I may do better to rig something up to use in the vise instead.

I was thinking something similar and you wouldn’t have to worry so much about getting the “rivet” on the back of the pill as flush. You’d have to be very precise with the drilling so that the LED was centred properly, but this and a thin layer of thermal great under the star would make a fantastic thermal joint. Then all you’d have to do is reflow the LED to the star once it was all together. I may try this with the C20 host I have coming, although my drilling is notoriously crap (no drill press).

Matt, wouldn't you just use your lathe to drill the hole? Unless its a multi-emitter you're going for.