Thank you very much for such useful information RaceR86. 
IMO measuring the actual lumens output is necessary to find out if CREE has changed these emitters for good or bad. Although the acceptable peak current of these emitters have reduced but on the other hand the Vf has increased, this could mean that the total power the emitter see can still be about the same? For example the emitter in your video died at 5.70A but the Vf at that moment was 4.04V, that is still a whopping 23W already.
Indeed. I wish I had a weekend to just sit and test.
And I also wish I had the equipment.
Thank you guys for the information. For those of you with the icharger 106b (and maybe other models and clones or originals), who are willing to test LEDs, the motor drive function can be used as an approximate power supply.
Thanks to the OP for all the information.
To me, if Cree did cut the production costs somewhere along the lines, be it cheaper gold bond wires, or in some other area, I can't fault that, since producing in the US is getting higher priced every day. Maybe they have gone the same way of many manufacturers, where they count on process control being their quality control, in an effort to cut overall personnel costs. No matter what, it does go to show that modding anything beyond the stock specifications is a risk and can have bad side effects.
What we do should never be seen as poor quality of Cree emitters, as what we do is abuse pure and simple. When they start blowing at 3 amps consistently, then Cree will have a problem.
I haven't touched an XM-L2 in a long time. Don't seem to care for them any more. I am still stuck on MT-G2 for most everything right now. I hope they don't go South too, but I have gotten to where I don't bother with the super high amps any more though. Just isn't that much more impressive or useful to be worth it. Like they say, once you've seen two LEDs, you've seen 'em all.
ImA4wheelr, If there there is a need I could start a spreadsheet. I dont think it will be necessary though. Might also be higher chance that people report more data if its kept simple. That way people can post what they know. At the end of the day, I don't think there are too many people who have that much data when it comes to these things. Especially not beyond the capabilities of direct drive. Im not going to encourage people to push their emitters as far as Im often willing to test. I like to know that I got some headroom. When that is said. 5A seems to be a nice place to measure Vf.
[quote=rjorge] Just measured a XM-L2 U3 1A and 2A. 3A: 1A, 3.34v. 2A, 3.64v 4A: 1A, 3.50v. 2A, 3.85v 5A: 1A, 3.68v. 2A, 4.10v Both on 20mm Noctogons, 18 gauge wires, sinked to a 4x4x5 chunk of Aluminum. Power supply: Kenwood PD18-20D. [/quote]
WOW. Looks like you won the U3 1A lottery.
That looks like a U3 binned emitter that could potentially handle 7A+ (not saying you should try it). Are you able to do a light output comparison between the two emitters at similar amps (3,4,5A) ? That would be very interesting. Any more info? Vendor? Date ordered? :D
@18sixfify. Keep is updated on your adventures. :)
[quote=bibihang] Thank you very much for such useful information RaceR86. :beer: IMO measuring the actual lumens output is necessary to find out if CREE has changed these emitters for good or bad. Although the acceptable peak current of these emitters have reduced but on the other hand the Vf has increased, this could mean that the total power the emitter see can still be about the same? For example the emitter in your video died at 5.70A but the Vf at that moment was 4.04V, that is still a whopping 23W already. [/quote]
:beer:
I agree. More output data would help. There are limits to how much time I can put into testing. I was mainly trying to achieve 6,6A with the best possible emitters. Hopefully others will share what data they have as well. I find it strange that the U2s Ive tested either die around 5,6A, or they are good for much more. So far I have not seen anything in between.
Yes, the U2 died at 23W (5,7A). Its a lot of energy for a 10W emitter, but at the same time, its not that high for an emitter on copper. In comparison the U3 1A I tested died at 27,33W (6,4A). And the "China T6" emitter was still going strong at 30,2W (7,6A). I would not be surprised if that emitter could be pushed beyond 33W. 10W more compared to a bad U2 emitter.
Needless to say, Id like to get my hands on well performing top binned emitters that are capable of 25W+.
Id pay extra to get emitters that was verified to do high amps for the most hardcore builds. Right now it seems to me that emitter performance can be a lottery.
No doubt that everything beyond 3A is at our responsibility. But its its CREE`s responsibility to produce emitters that are within spec. (Although I dont know what tolerances they have. Anyone knows? Look at the differences in the emitters rjorge posted.
One U3 emitter had a Vf of 3,34V. Which is where they ideally should be. Problem is when another similar binned emitter have Vf of 3,64V then something is off. That is a massive difference. They should be at 3,33V. Yes, our copper mcpcb`s can make that a bit higher, but 3,64V...
If its not something Cree have done then I wonder if there could be something wrong with the re-flow process. Maybe I should stick to only buying bare emitters. I dont know...
I wonder if OL would comment the same way if we were discussing MTG-2 leds and not XML-2? 
Just thinking... if the U3's have such a high Vf but do have U3 output, they are still very suitable for <2amp single li-ion flashlights. Driving the XM-L2 lower than 2A sounds a bit odd in general BLF-thinking, but it makes a very efficient flashlight. The higher Vf is not very relevant because overhead voltage is burned off anyway by the driver and at these low amps the voltage is still low enough for the flashlight to stay into regulation for most of the battery drain.
Yes, yes it does.
Sounds very unnatural. Who would do such a hideous thing? Think about that poor helpless little emitter. It would be freezing. I take good care of them, put them on copper and make sure they run close to rated temperature. I help my emitters live up to their true max potential. Once they enter my house they have a bright future ahead of them. 0:)
My most efficient flashlight is a Uniquefire S10 with 2x7135 driver and a XM-L2 T6 0D, at 90 OTF-lm/W.
:evil: emitter torture! :evil:
EDIT: pointless question (deleted)
Not at the moment, but hopefully soon. I got them from RMM a couple of weeks ago. Not sure what to do with the 2A, I had planned to put it in a SupFire L5 and LD-1. Not anymore!
I appreciate your time and effort on testing these LEDs. You should have already spent a fortune in buying lots of emitters.
So I ordered a 5 amp adjustable power supply.
From the posted spreadsheet, it looks like between 3 and 4 amps is where the discernable vF change becomes clear.
From here on out I will be testing in that range to see vF and relative output before I use an LED…………
Ive done a new "test session". I`ve added more data to the spreadsheet. All new data in bold. Im not going detailed through everything in a post like last time. At least not now. But for me, several interesting finds on several of the latest and popular emitters. Feel free to check it out and leave a comment.
On top of that I can add a bit more data. Tailcap measurements from a bunch of different UF-T20 lights I put together for a local. (For the record, I mostly just make lights for myself). Cell voltage when I started doing the tailcap readings was 3,87V. I did not use the absolute lowest resistance DMM either, but its good. All results comparable and all purposely out of regulation to easy see the difference.
Flashlight specs: 10x7135 Qlite driver. Spring "bypass". New switches in all lights, all emitters reflowed by me. De-domed XP-G2 R5 1A in all lights except 3. Those had de-domed S2 2B (which as expected does not de-dome that well). All emitters ordered not too long ago (new).
| LG HE2 3,87V | |
| amp | |
| S2 | 2,73 |
| S2? | 2,65 |
| 3,4 | |
| 2,54 | |
| 2,98 | |
| 2,9 | |
| 3,03 | |
| 2,2 | |
| 3,18 | |
| 2,65 | |
| 2,66 | |
| 3,15 | |
| 2,9 | |
| 3,2 | |
| 3,14 | |
| 2,93 | |
| LG HE2 3,81V |
I knew which light I had one of the S2 emitters in. It was also clear due to the worse tint. I easily located another light with a bad tint. I was 97% sure it was another S2. Very similar amp reading too. The third S2 I could not say for certain I was able to locate. I suspect it was the one at 2,2A due to combination of fairly bad tint and low amp. I later swapped out that emitter to another S2, which performed better.
I suspect de-doming of these emitters could help generate a larger difference in performance compared to normal. Either way.. It is what it is.
Anyone else got anything to share?
nice info,thanks.
Thanks RaceR86 again for the effort put in here. It appears the trend here is that the XPL is the new go to emitter for very hi current draw applications. It would be interesting to compare light output but with the smaller die of the XPL the same reflector could not be used. This is where a sphere would come in handy.
Thank you for the testing.
It is nice to see that it is highly unlikely to blow a led up with a single battery setup.
Thanks you RaceR86 for the tests and datas.
May be multiple emitter in parallel will become more popular, hoping battery technology will make big improvement too.
Thanks for more and more insightful data, RaceR! Who would have thought that led voltages, so relevant for single-cell flashlights, would vary so much. And there is no precise specification of Vf by any seller including Cree itself!
With a really powerful buck driver I would put my money on the XP-L based on what I have seen so far. Buying new XM-L2`s (from recent batches) and trying to push them to the more extreme currents seems a bit like gambling with bad odds. Unless you have verified how far the emitter can be pushed, which I now seem capable of testing quite well without frying any emitters. :) Sadly, most XM-L2s from recent batches does not seem that capable.
Not sure I understood everything you said. Smaller die on the XP-L? XM-L and XP-L have the same die size. Suitable XP locating/centering ring can make the XP-L`s fit perfectly in any reflector that you use XM-L if you need one.
I disagree.
If DD and low resistance cells could push XM-L2`s and XP-L`s to death at 5,5-8A+ that would be a good thing IMO. If the emitters were like that, many lights could maintain 3A much longer. Good for everyone who wanted a bright light.
Hot rodders could make single cell hotrods like say an HD2010 with spring mod with a 5200mAh cell push 6A to to an XP-L and maintain that for say 30% of the runtime before it started to decrease.
The" Vf safety net" we have are holding us back IMO. This is quite evident in the increasing increasing amount of "low resistanse mods" we are seeing. People just cant get "high enough" with regular cells. Not even with low resistance cells...
Id much rather see some lower Vf emitters. If the XP-L had a Vf of 3,2V instead of 3,43V @ 3A (25C) then I don`t think the XM-L2 would be used much.
When that is said, there is a place and use for every emitter/driver combo depending on what people want. Just need to find the best. :)
I was a bit surprised when looking at the highest and lowest amp readings. Still not sure if de-doming can change things a bit though.
I have seen and reported Vf issues before though. Check out OP in the L4 mod thread and scroll down to "Why did I use a bunch of time modding this light for probably zero visible gain?".
Cree have lots of precise data. You can see Vf here for any given emitter at any given current and temperature. My spreadsheet data is actually very close to Cree`s data. At least if you average out the numbers.... Lots of the numbers are within 1%. Problem is, +/- 5% can give large differences in flashlights.
Earlier in this thread I assumed the XM-L2s had Vf of 3,33V at 3A. And it does, at 85C. My testing is basically at 25C based on mcpcb temperature.
To give a few examples at 25C Cree says Vf are:
XP-G2 @ 1A : 3,16V
XM-L2 @ 3A : 3,43V (3,02V @ 1A)
XP-L @ 3A : 3,48V (3,06V @1A)