The noctigon is attached to the heatsink. I just tested it in my xp-g2 s3 3c on noctigon from intl-outdoor. I got 4.65A, 4.08v. I just connected the battery for about 2 seconds then disconnected it to get the readings in my watt meter. Then I discharged the battery to around 3.8v using load resistor, then connected again the battery to watt meter to get the reading. I do this at a decrement of 0.05v.
Here is what i got, I don’t know if that’s the correct way of getting the vf.
3.56A 3.81v
3.21A 3.75v
3.01A 3.7v
2.73A 3.65v
2.55A 3.6v
2.30A 3.55v
2.05A 3.5v
1.8A 3.45v
1.7A 3.4v
1.5A 3.35v
1.4A 3.3v
that looks like a workable way to do the readings without need of a powersupply :-) Do I understand correctly that the voltages were under load? (namely the readings from the watt-meter).
It is indeed the correct way to measure the Vf. The only measurement error is that there is a slight voltage loss in the wires to the ledboard, but if these wires are not overly thin that is not significant.
why i get so late the important threads ???
Race big thank you for this voluminous and time consuming job. i think this is really big help to us that like to play and experiment with flashlights.
that said, i feel very sad about the directions cree is going (assuming the last batches are not defective ones or just isolated batches )
as i see it they are going in the opposite directions of what i would like to see.
first no effort in new dies with increased luminance, but instead they go easy path adding 4 existing dies and present it like a new one 
not only this they take an existing die again cut here and there and again present like a new thing….
and now this, probably they are in saving cost revisions and the old gold cree we like so much can no longer be overdrive so much :_(
really hope this is not at all true…
Yes the voltages were under load. I am also thinking that there is a slight voltage loss because of the wire. The wire from watt meter is 14 awg.
I did another test this time xp-g2 s2 1d from intl oudoor. Very disappointed by this, i can only get maximum of 2.5A 4.1v. Do you think it’s faulty?
I have another xp-g2 s2 1d and i get 4.5A 4.07v.
Any plans to crash test some XM-L U3-1As? Specifically from the $3.45 deal @ LEDDNA. I have an S6 build that I decided to use an XM-L2 U3-1A instead of the XM-L U3-1A, but I cannot get it going beyond 3.4A presumably because of the higher vF. Was thinking if I'd be able to drive it harder with the older gen emitter...
I tried searching, but I couldn't find anything to save my life :(
One thing I like about using nimh AA is that you don’t have to worry about high Vf beyond figureing out how to fit more cells into the light.
And to think I was trying to figure out a way to run a 3S XML light off of 8S eneloops because the sum of the vFs of the LEDs at 3A was around 9.1V
Does anyone has any update on your overdriven XM-L2 U2/U3 emitter? Are these emitters still incapable of handling >5.5A?
I havent heard otherwise. if you want to drive this size led harder you have to use the XPL.
I see, so the XP-L is fine to be driven >6.5A so far?
I have killed XP-L's at unexpected low currents (<7A) as well, it is not just XM-L2 that shows the problem.
If the XP-L can survive >6A but <7A then it is still good to me. 
Looks sort of like the left hand bond wire lifted from the die in pic 4 of Post 5 . Poor die connections could be one source, another could be the use of weaker bond wires or a manufacturing process that has changed. Cree might not care about over current failures but mechanical failures like bond wires coming off the die could happen in normal use and should raise QC concerns. If this is the weakest link then maybe they will develope more LEDs that do no use bond wires.