I used your mounting technique to polish a 10511 today. Ended up using white jeweler’s rouge on a soft cloth towel, held by hand onto the top of the optic while the optic sat in a spinning cordless drill in a vise. Gives more throw and a tighter spot than the stock 10511, but I’ll have to wait till dark to see if I like it better. The stock 10511 has by far the smoothest transition from spot to spill in any of my lights.
Late to the conversation about conduction and capacitance of copper and other metals but I wanted to upset the apple cart a little. 
I subscribe to the theory of copper slug to copper layer for convection aided by a fan. For a flashlight I believe that flesh and blood do a better job of removing heat through conduction and fins although pretty interfere with the flesh to copper connection. Another plus is that copper kills MRSA, C. difficile viruses and more.
For the triple XML there is a free file at . Oshpark. I can send a link if you like. The center hole is small but ok for an anchoring screw, just have to poke holes for your wires.
The issue is that there aren't any 20mm optics for XM-L.
I have one of those boards, just haven’t set out to mount it yet. Was thinking today about how to mark a copper sink to make risers up into the thermal pads in that board so the emitter will actually be sitting on copper instead of g-10 laminate.
This.
My question would be how does the XML/XML2 dedomed compare to an xpg2 in terms of total lumen output and throw.
Good question. Anyone know the answer? I’m also interested.
de-domed XM-L2 vs domed XP-G2? The de-domed XM-L2 will beat it in output and throw, typically, generally... de-doming a XM-L2 loses about 20% in lumens and about doubles the kcd. I believe even with the 20% it will still have higher lumens. I'm thinking a XP-G2 max's out at around 950 lumens, while an XM-L2 can do 1,800 - 2,000 lumens, under ideal setups of high amps, large reflector, etc.
What! I thought the Oshpark was copper. Ahhh crap. Here we go again.
Richard, these were designed to use with the XPG Triple optic. De-doming is suggested. I dunno- I was going to try both after I did my MTG2 build that hit a road bump (me).
Now Dale is about to explain how I screwed this up with G10 Laminate… Dale what’s G10 laminate?
Like Tom said, and the XP-G footprint will have a narrower beam profile, especially if de-domed…less spill will make it seem brighter in the hot spot but it won’t be putting out near the light. The smaller footprint is good for isolating the hot spot, with little illumination to the sides. The bigger emitter will do a good job all around, showing you some side illumination while still putting considerable light down the middle.
That’s what I was guesstimating from what others have posted but no real world experiences yet.
Now Dale let’s get back to this G10 laminate… I thought the stars that I got were all coppery goodness, pumping out dark matter and channeling heat. What did I miss- you said there is G10 in the way…
G10 or FR4 Glass Epoxy is PC board Material.
Its FR-4 PCB. wikipedia: FR-4 (G-10 isn’t commonly used for PCBs anymore). Thin copper circuit traces bonded to fire resistant fiberglass.
Led stars are MCPCB, metal core PCB. Standard MCPCBs have thin copper circuit traces ontop of a dielectric layer bonded to an aluminum core. Noctigon removes the dielectric under the center thermal pad of the LED. Plus Noctigon has a copper core instead of aluminum.
How about at the same current? Say… 2 amps.
OSH Park, for those prices, will not/can't do true metal core. They have all sorts of restrictions on the boards, including only one width. Your job (order of boards) gets bundled together with many others jobs (differnet designs) on a panel, where the panel is one huge rectangle of all the same material, same thickness.
They look like bare copper underneath. That are a two layer copper and FR4 “sandwich”.
From the website…
Specifications
All 2 boards are FR4 170Tg/290Td which are suitable for lead-free processes and temperature.
4 Layer boards are now FR408 (180Tg). See
the support site for the stackup.
They have ENIG (gold) finish for superior solderability and environmental resistance.
They’re 1.6mm thick (0.063 inches) with 1 ounce copper on both sides. For four layer boards, the internal copper is 0.5 ounce.
The minimum specs for 2 layer orders are 6 mil traces with 6 mil spacing, and 13 mil drills with 7 mil annular rings.
That sounds good, right?
Here: http://flashlightwiki.com/Cree shows at 1.5A, an XM-L2 U2 will do about 680 lumens, while a XP-G2 R5 will do about 538 lumens. 20% is about the max you will lose on de-doming, so maybe dead even, maybe the XM-L2 is still ahead. The XM-L2 is a proven beast. Guys have done incredible things with them here. I did a 5,400 lumens Shocker (domed), and a 1,650 de-domed XM-L2 in a TN31 for example.
Look at Match’s results, both on copper and aluminum stars here
The triple XM-L board I have appears to be G10 with perforated metal pads, silver plated copper? But it doesn’t appear to have a solid metal thermal path to the pill or heat sink below it. That’s what I was referring to.
And Qlite’s appear to have G10 as their substrate, I’ve got lots of G10, as big as 2” thick x 12” round. I’ve made so much from the stuff I recognize the smell when filing off the attachment tips to fit it in the pill.
Even made custom grips for a .45 Ruger Vaquero out of G10. Almost indestructible stuff, especially at that kind of thickness.
So if I do a build using Richard’s and Comfychair’s FET driver with the Oshpark 3XML designed by Scaru the difference in time that I can run it on high before it’s too hot to hold versus a more ideal copper star with direct thermal path will most likely seconds? Unless Richard comes up with a copper spacer for the S6, I’m thinking of a home rolled P-60 in a Solarforce. Ala dorpmuller -if that helps visualize thermal capability. Thanks for any input or suggestions.
A poor heat path would result in a cooler light to the touch but more heat for the emitter. That means lumen/heat sag or worse at higher currents.