Great works! I didn't expect such results! Thanks again for sharing
that's interesting - I always wondered why it was such a git to solder to an LED already mounted vs. sitting on the bench. I lap my LED stars and where they mount to with 400 then 2000grit wet'n'dry then use Mother's aluminium polish and a dremel polishing wheel. Then I use a thin layer of thermal paste and screw the LED down. The housings of my lights always warm up, which is a good sign of thermal transfer.
Match, amazing results and well done for all the effort with testing it. If I ever find a piece of copper big enough to turn down into a MagD heat sink, I might give this a go :)
Thank's for using my spreadsheet (also to Slim Pickens). I have no objection against your tweaking the spreadsheet. It's there to be used. In fact with my little provocation I had in mind that more people should benefit from it. I have published and mentioned it a few times on the forum but only a few have responded.
I agree that Match's numbers can be obtained with an optimistic T6 from the "LED cold [lm]" result, i.e. the numbers for the LED at 25'C.
To cover overdrive from max current to double that (from 3A to 6A for XM-L) I made an graphical extrapolation of Cree's curves with an unlinear ruler (called a "bone" in my country) so the accuracy in this range could be a little less than in the "certified" range where it should be within 2 pct. for a "nominal" led.
I hope that Match will continue these interesting investigations and as I found his first measurements plausible I have no reason to distrust his lumens measurements in the integrated sphere and I will leave this subject with the known words: "the truth is out there..".
Seriously.. how much would it cost to build and distribute these, whether it be exact specs or scaled down a bit? Would you even be interested if it wasn't too much trouble?
That was kinda my feeling to, but I can’t quantify it. 1000 real OTF lumens at 3.2A just sounds too good to be true, but I want one if they put it in a torch.
It would be cool (!) if you could play with this thing in an Absolute Zero environment and see what kind of light it gave off....
In an absolute zero environment it would give off no light as at absolute zero all motion stops, electrons wouldn’t flow, now light could be made.
None.
Edit: Bose beat me to it. :P
Check out youtube. Some vids show 5mm LEDs immersed in liquid nitrogen and they undergo crazy color shifts (orange to yellow, then green)! Makes you wonder just how much current can be pumped into an XM-L under similar conditions.
A classic broadside, shipmate......
U2 1C @ 3A ca 40.000 LUX on copper star
U2 1C @ 6A ca 60.000 LUX on cooper star 
(LED Tech Germany)
Reflector: V2 Catapult with UCLp
Regards Xandre
Copper stars are used in the Jetbeam BC20/40 line, this could go someway to explaining the high readings I get for the BC40 on high; 948 lumens.
Many of them are not stars :) A similar shape is present on Niteye flashlights.
Ha ha yeah sorry, I meant PCB not 'star' literally ;)
any places to buy xml on Copper 14/16mm ?
CW: http://www.led-tech.de/en/High-Power-LEDs-Cree/CREE-XM-Serie/CREE-XM-L-U2-on-Copper-PCB-LT-1790_120_170.html
NW: http://www.led-tech.de/en/High-Power-LEDs-Cree/CREE-XM-Serie/CREE-XM-L-T5-on-Copper-PCB-LT-1810_120_170.html
For a little extra money you can also buy the PCB alone or have it equipped with other emitters, such as the NW T6.
I did with hot-rodding an electric scooter. Now that’s a fun hobby too and addicting. Here’s my thread on that particular subject and I can say that without a doubt combining copper with aluminum fins plus dynamic cooling works like the dickens. That scooter motor dropped temp by a bunch which released hidden power and run time. It got some views I’ll say that. Probably somebody’s selling something like it now. lol
Very interesting stuff you craft up. I like how you approach things and solve them as simple as possible.
Lumatic
http://forum.modifiedelectricscooters.com/viewtopic.php?f=9&t=103
Nice data gathering, thanks!
I been asking myself the same question about the lumens increase at high current with optimal cooling.
I built a very special light:
Direct reflow on copper tubes with a flat spot that match the pads. Coolant flow into the tubes.
It is in the nose of an FPV airplane:
Based on your results, the light output might be around 19000 lumens (16 XM-L driven 4A) I just rebuild the light with XM-L2 U2 driven 5A wonder what is the increase?
-Hugo
What kind of power source do you have on that plane? Something plutonium based?
4S 5000mAh lithium-polymer, good for 20mins flight.
-Hugo