NICE!
How did you wire the LEDs up?
I guess the coils cool the coolant and a little pump is used, I would like to see some details of this too.
Very brilliant craftsmanship!
I made a separate thread not to clutter this one:
I’ll reiterate my question to the OP:
Based on your results what would the light output be with XM-L2 U2 driven at 5A?
I guess the cooling effect is same or better than the copper pill.
-Hugo
I can’t understand how was the contacts soldered to the led…
Didn’t soldering the led to copper directly shorted out the contact on the bottom of the led?? :~
Thanks for the graph!
I have been playing with these LEDs for months while testing my Thermal Connectors. I was wondering what the graph looks like.
Here is what I use. TCAP-4325 Thermal Connector and an FR4 circuit board.
The thermal connector solders into the PCB, then a heatsink can be bolted on after. All that is required is a 0.1” diameter hole, and a minimum of 0.1” thickness. Anything can be used as a heatsink.
The parts are available from www.tem-products.com
Dont take this as being condescending but in what way is this a better option that a direct thermal path board [sinkPAD/noctigon]?It would seem to me like a standard direct thermal path board with its larger surface ares in contact with a heatsink could actually shed more heat than a small screw in connection like this.
Again not trying to jump all over you, I welcome any and all new ideas/products, I’m just not sure why this is needed over the already widely available products we all already use. What applications is this better for?
Also can the emitter be reflowed directly to the pedistal without the FR4 board and leads soldered directly to the contacts on the underside?
Oh and welcome to BLF!
I can see this being handy for leds with odd footprints, there are not Sinkpads for every led, this enables you to design a board (Oshpark) and have a direct-to-metal thermal path.
Thank you.
Its true. Nothing is better than covering the thermal pad with a FAT BRICK of copper. 
If you have a CNC machine, some serious skills, and a big budget, that is definately #1 for performance.
PowerPeg Thermal Connectors are designed for streamline design, and not necessarily for LEDs in specific.
I designed PowerPeg back in 2011 to address the overheating issues of semiconductors such as transistors on circuit boards. I needed a cost-effective way to manage the thermal buildup of a motor control chip.
I realized that heatsinks naturally COOL DOWN, and solder needs to HEAT UP during manufacturing, it would be hard to mass produce my product this way. This is how i invented a two-part system. Thermal connectors attach to the PCB, then heatsinks can be bolted on later.
PowerPeg is designed to take up as little PCB area as possible, and provide the lowest thermal resistance possible. The TCAP-4325 provides 2.0 °C per watt of resistance. More data is available in the Datasheet.
PowerPeg comes in handy any time an LED is to be mounted on a PCB.
Check out these pictures of one of my first projects with this system.
I’ve wondered about this, and you went and did it. Your results make sense. Thanks
Rusty
For the most part I think you’ve missed the boat for our hobby. Most of our applications are much better served by direct-copper MCPCBs. That term refers to a copper MCPCB that has had it’s substrate cut away, directly exposing the copper to the LED’s thermal pad. Take a look at Sinkpad and Noctigon to get more info on that subject.
That said, a product like that still has potential in some of our more niche applications. I’m thinking of Projects like tterev3’s MELD series. Nobody is going to make a direct-copper MCPCB for that anytime soon.
I noticed that your datasheet does not discuss the material beyond saying that it’s a “copper alloy”. It looks like brass to me, is that what we are dealing with here? Can you provide thermal conductivity numbers for your alloy? Thanks!
The alloy is C145 Tellurium Copper, which is 99.5% pure copper. It is plated with 24 kt gold. Lead Free and made in USA.
I am familiar with MCPCBs. PowerPeg has a few benefits.
The one you pointed out was the obstacle/price hurdle of making a custom MCPCB. Powerpeg offers comparable performance with an FR4 substrate. FR4 is nice because they are low cost, and easier to make fancy circuitry with! OSH Park makes some great boards, a PowerPeg is a go-to resource for thermal issues.
Just to be clear, PowerPeg DOES solder directly to the LEDs thermal pad through an aperture in the PCB. The top surface is 2.5 MM in diameter, and PCB layers also help to fully cover the thermal pad.
Another advantage of PowerPeg is the durable, definite connection. I have seen MCPCBs which use junky spring clips for mounting. I have seen others which use nuts and bolts. Either way, the MCPCB could flex when it is screwed down, especially when the quality is poor (Sinkpad and Noctigon boards look beautiful).
PowerPeg bolts-on with a single screw located in the center of the thermal interface. When the screw is tightened the surfaces MASH together to form a perfect flat connection every time.
I dont think the boat is missed!
PowerPeg provides 2.0°C/W. This means for every watt the LED outputs, the temperature rises 2 degrees Celcius. Thats not bad! MCPCBs may provide less, but certainly not less than 0!!
LEDs generate alot of heat, and transferring it away from the die is very important, but the “weakest link in the chain” is the heatsink.
In the case of a flashlight handle, I would (roughly) estimate the thermal resistance is 30°C/Watt. 30 + 2 = 32. If your LED is making 3 watts the temperature rise will be 96°C.
If thats the case, there is no reason to drop PowerPeg over a 3% difference. You need to re-design your heatsink!
Here is a picture showing the direct connection without FR4.
Ah, fooled with gold! It’s rare (I hope) to think that an item is brass and then find out that it is gold. 
Thanks for the info. Your product is much more attractive when we know details like that.
Eliminating the dielectric layer under the LED is the primary thing here. That dielectric layer on a traditional MCPCB is what kneecaps performance and we take a bad junction temperature hit because of it. That’s why direct copper MCPCBs are a big deal as well as the newer direct aluminum boards from Sinkpad. SOME kind of solution for where the LED solders on is definitely worth it for us.
Thanks wight. Im proud to share them.
Let me know if you need any custom sizes or special heatsinks.
Also, take a look at the GP-43-A precision heatsink. Shes a beauty!
In jewelry class, we used solder that was largely silver and zinc, more zinc as you worked on a piece to lower the melting point. The zinc also evaporated or oxidized out, so older solder didn’t melt as easily as new. We didn’t need oxygen or arc to melt it. We used acetylene and air, which is apparently hotter than propane and air. That might be too hot for an led, but perhaps with some indium or tin or something it would work.
Only issue I see is the extra metal on the backside of the PCB…mounting it into a pill will be “interesting” unless it is ground down
Djozz and comfychair did something like this with a star and riveted in a piece of 10ga copper from romex, drilled a hole, dropped in the wire, hammered it flat in then ground it off, then epoxied the entire star down to the pill
The benefit I do see this would be the multi emitter aluminum stars w/o heatsinks in SRK’s, drill out the aluminum pads, solder in the stand, then get a copper flat and screw to the stands on the back
I guarantee though the OP and his build to that copper shelf would definitely benefit from the design, he would have to cut the threaded section off, lap the little nub, braise to the copper slug, then attach the pin to the emitter…done
Much less machine work
I can see a bit more coming with these 'pegs'.
I am building a flashlight with 'scaru's scaryboard' (7x xml on 20mm diameter), the board could be made scarier with 2.5mm via's in the middle of the thermal pads and using these bits, an easy alteration of the board (but with some BLF-concern: a bit pricy at 7x $1.65 for the bits).
My main concern with the DQG-flashlights is that the led is badly heatsinked on the driver board. There are some gifted board-designers at the moment at BLF, I see versions of the new driver boards with on-board leds that can be quite easily heatsinked with these (fixing the board to the pill/shelf at the same time), making very thin pills possible. I am not sure if the the gap that is created by the pegs on the backside of a board is wide enough for back-side components.
I would have liked to play a bit with a few of these but for $31 dollars shipping to the Netherlands 
, I'm afraid it is not for me to do that :-(
At the very least…really nice + battery standoffs to clear battery side components 
hi TEMdude1, it is perhaps an idea to open a thread in the commercial sellers spot of BLF about your pegs (threads started there are also listed in the 'recent posts' page), it could catch a bit more attention than responding to this quite ancient thread. I can imagine a few more people who are into building flashlights would be interested in your products.
Interesting stuff, never seen it. Normally there are through holes filled with solder on such applicAtions. I have somewhere some links/PDFs from cree in which they compare through holes with mcpcbs…the conclusion was through holes are not much worse. I played with that on a 3XML fr4 board…
I also see the problem with the metal dish on the back, because it benefits from a perfect fit in the heatsink and for use with a fr4. To machine this hole in a existing pill is a challenge…but on the other side it is just a hole with a smaller hole in it.
They are made for use with your heatsinks but there is only a small one availabe, the gp34 has 58 or 17k/W which is with a single XML not overdriven at 10W to hot…
Have I overseen something…?
Another question I am wondering about…?
The height of the pin is 0.062 inch(maybe you should add some metric values in the datasheets…) so the fr4 board should be ordered in which thickness???never seen an option for 0.062 inch
And just one more
You only have one size peg? The 4325 has a input surface diameter of 2.5mm which is only suitable a small range of LEDs. XML should fit but the xpg not and for huge pads from the mtg2 it should be double the diameter…
I created a new thread in commercial sellers spot for TEM-Products.com here:
Lets continue the conversation there.
Thanks for all the replies.