So we are talking of much lower currents for some hours in total. This could therefore mean that the dark spots or damage to the silicone may well occur at lower currents than 9-12 A I used here for testing (which were only used for 1-2 min in total), which then again depends on the previous runtime at high operating current.
At which current this damage occurs, I can’t tell. Maybe in general it is a good idea not to operate this emitter with more than 6-7 Amps.
Yeah, personally I wouldn’t use it at more than 6A due to the efficacy and heat, and it’s already double the rated current.
A low CRI version might handle higher current without damage since it heats up less.
Could be possible, this was also an idea from the folks of the German flashlight forum.
I hope to get my hands on a highest-bin low CRI version in cool white soon, so I can compare the light flux and maybe the damage with the HighCRI one. If I have somethin new, I will post it here of course.
Between the offset die shown in the OP and the brown spots shown by thefreeman, it looks like Cree does have some defects going on.
Somebody get Mila Jovovich on the phone and find out what’s going on with Cree!
Hmm…
I’m guessing that she doesn’t work with CREE or even cares about flashlights.
She is pretty photogenic, though! 
I just saw this in my pic - is it just me or is the LED chip top left smaller than the other ones?
The problem probably doesn’t just affect overcurrent, it might have other effects in parts we don’t notice much, like longevity, but manufacturers for other devices probably will.
Having bubbles or contaminants in the silicone isn’t exactly a desirable trait on an LED and Cree does “hidden” changes relatively often so it wouldn’t be the first time something gets changed without a new revision either.
Upon closer look, BOTH left-side dies are noticeably smaller than the right-side ones, so it’s not just you! And it appears that the top-left die is slightly smaller than the bottom-left one.
I wonder what LED configuration the SR32 uses, like how many LEDs are wired in series and parallel, and if some of the LEDs see higher amps than the others.
About the design of the SR32: according to a review, the light has a turbo output of 100k lumens which instantly nosedives towards ~17,500 lumens. It hits ~17.5k at ~40 seconds and maintains this output thereafter for ~47 minutes.
It’s a 17.5K lumen light if measured at ~45 seconds or in terms of continuous output. It maintains this level from “high” mode as well.
17,500 / 32 = 547 lumens per emitter.
The XHP50.3 HI (J4 flux bin) is rated for 2016 lumens minimum output at 85C at maximum drive current.
2016 x 32 = 64,512 emitter lumens * .85 for optical losses = 54,635.
If they limited the turbo to the max spec of the LEDs, it should still put out ~55k lumens. This is about what the “high” mode of the light puts out for a brief moment.
If Immolate designed the light without the 100k turbo mode, probably the type of people that buy their stupid lights wouldn’t buy it because it wouldn’t be hyped by reviewers in the same manner because then it isn’t “the worlds brightest” anymore, but that aside, there might be no issues of LEDs burning up.
I love it! 
Great analysis.
Somewhat unrelated, but @koef3 have you tested the XHP70.3 HI? I’m trying to find a flux/current graph for it and it can’t be found.
Naturally I’ll probably find it 10 minutes after making this post, as the laws of the internet work.
In fact I haven’t tested the XHP70.3 HI yet, I don’t even have a light flux chart for it. 
Testing these is also planned as well for other LEDs… Problematic will be, that I only have one piece of XHP70.3 HI, if it is also damaged because of burned silicone like XHP50.3 HI, I am in trouble…
I’m pretty confident it’ll be safe to go to 10A at least, seeing where the XHP50.3HI burned, the danger zone probably being over 15A though it probably could get brighter, but it will at least give a good reference point for current/output.
I have them in my MT09R, which is a FET driver so I can’t give an exact current but it is incredibly bright with 2S2P Samsung 20S3 cells, and hasn’t had any damage yet, so they should be safe to drive to whatever current matches the Vf of 2S li-ion with a bit of voltage sag, at least for the R70 versions.
If I had bought a fourth emitter, I’d happily send it your way for testing. I’m definitely looking forward to the results, hopefully in both R70 and R9050.
BTW, I would love to know what the Nichia 219C-V2 is like, if you can get your hands on some.
It’s a newer domeless emitter, same release date as the 719A but the 219C has a lot of overdrive potential.
Someone is supposed to send me some, not sure exactly when though.
When they do I’ll send a couple to @koef3 for testing if he’s interested.
Of course I am interested. Dedomed 519A with guaranteed color binning as it seems. It is similar to dedomed 519A but the LES is a little bit smaller as for 519A (hopefully higher luminance). The thermal resistance is slightly higher, the typcial Vf is more or less the same.
It should be using the same die as the 219C, I would assume, which has a long history of use in flashlights, but without dedoming it being common. Hopefully a bit more throw than dedomed 519a too
Hopefully better than the XP-G4 HI that has just released in the same die size class.
Even if this is not the case, the design of the XP-G4 HI smells for Cree rainbow™
The 219C-V2 (weird name tbh since it is a completely different LED compared to earlier 219C) should be much better in this regard.
I’m not so sure, seeing how a lot of newer domeless emitters have the entire surface covered in phosphor, like the CSP2323 of the TS10 and the new FFL351A that Fireflylite got made. Fingers crossed it’ll be nice, and in the worst case that it’s easy to remove the extra phosphor.
Please do test it! I will buy you another one if it gets damaged.
