XP-L Hi, and would-be-XP-L featuring in recent SRK-clones, two emitter tests

First the XP-L Hi, I'm a bit late to that party. I tested a U5 bin 5000K led from Mouser. there's better bins around since then, so the performance can be a bit better than in this test. Pflexpro did already some convincing testing of the XP-L Hi inside flashlights.

(stereo picure)

It is build a bit different than what we are used to from Cree, and in fact the same buid-up as what I saw with the Nichia 219C. It looks like they build the complete blue led, then added a phosfor layer over the complete led (you can see in the stereo picture that the edge of the yellow layer is visible on the side of the led, there's phosfor over the bond wires too) and after that a white mask is added around the die, covering the part of the phosfor that is not over the die itself. (and then a silicon layer is added)

Being 'SC5' technology, one could expect better thermal performance than the original XP-L, but I do not see quite that: the output maximum is at about the same current as the XP-L (the XP-L data in the graph are from my original XP-L test a year ago, in that thread the method is also explained)

* as already noted above, the thermal performance of the XP-L Hi (in red) seems not better than the XP-L (in green)

* the voltage of this Hi led is 0.2V lower than the XP-L that I tested a year ago. That is nice and that difference seems a bit higher than what you expect from normal variation between leds.

* the led blew between 8 and 8.5A: shattered bond wires!

*there's a weird dip at 6A, but with higher current the output went up again. That looks like a section of the led failed and a picture of the led after the test shows, apart from the incinerated bond wires at their base (and so it looks a blown ESD wire too), a bubbled-up section on the silicon layer, perhaps some dirt on the led caused the problem, and the debris absorbed some light. My hypothesis is that that is what indeed caused the dip at 6A, and that the max of this led is usually a bit higher than what the graph shows (it looks like you can extrapolate the curve from 5.5A on)

* the output of the led is, as usual, higher than Cree's own numbers. I still do not completely know what to make of that, of course the led is on a DTP copper board, but that should not matter much for the lower currents that Cree provides output numbers for (487lm@1500mA, 614mA@2000mA, 828lm@3000mA). There are some clues that my 'djozz-lumen' is about 9% too high, but that does not close the gap enough to exclude other causes. Let's just assume that Cree is wrong :bigsmile: (the numbers are at least very comparable to my other tests btw)


Now for the 'fake' XP-L that comes in Sky Ray King clones from various vendors at the moment (DX, FT), see this thread. At least it does not even look like a XP-L, unlike the upcoming chinese 'Latticebright' leds that are very convincing Cree counterfeits. L4M4 was so kind to send me his lot of fake XP-L's because I did not have them anymore (had to send the light back to dx undamaged).

First, it appears like the phosfor that converts the blue light to longer wavelengths is diffused in the entire dome, so the effective die-size that the optics of the flashlight sees is the entire yellow area. This is about the die size of the XP-L. But if we look at the performance, it is very close (only just a tad better) to the XP-E2:

So, although this is not a bad led, it does not come close to the performance of a real XP-L: the output at all currents is way less, it maxes out at a way lower current, and permanent damage is already occurring at 3.5A (data not shown in graph: at 1A, I measured 313 lumen, with the current at 3.5A I turned it down to 1A again, and got only 143lm leftover). It finally completely died at 7A.

So, its big die size considered, every modern Cree led is to be preferred over this chinese replacement.

The led looked a bit darker after the test than an untested one:

Some cold dedoming later it was clear why :evil:

Hmm, let's try that cold dedoming (just pushed the dome off with a forceps) on an untested led:

that looks pretty clean actually, does it work?

Jay! A pretty neat blue led that is, and easy to make by just prying off the dome. The die size is almost exact XP-G so it throws less than a royal blue XP-E2, but if die size is not an issue it is an ok performer in the colour led category.

That's all!

Thanks djozz. Excellent work as usual with invaluable information. That is some very high voltages to go along with the high amps.

I was getting worried my package never showed up.

Thanks for the work again

Oh yes, I failed to report that to you, it obviously did show up, sorry about that! (pd68 bought the XP-L Hi's for me from Mouser)

No worries!

Thanks you very much for the test and information djozz :beer:

Thank you for the excellent testing and reporting djozz. :)

Did the silicone layer for the XP-L Hi also become darker after the crash test? Or is it just different camera/settings/lighting/etc.?

Wonder if the phosphor packet from the fake XP-L could be stuck to the top of a dedomed MT-G2 that has lost it's phosphor layer. Could be 2 birds with one stone kind of thing.

EDIT: Thank you L4M4 for contributing your emitters to the cause. :)

djozz thank you very much for all tests :wink: always your tests are very useful to me :smiley:


It is the inferior lighting when I took the photograph, the silicon is still clear.

Apart from size problems, the dome should work on a dedomed MT-G2. It does not have the most advanced of phosfor mixes around though (heavy understatement).

After my dispute with DX was settled, I fired the light (with the fake XP-L's) up for a moment. Although it clearly wasn't MG-T2 quality by any means, it didn't seem like a half bad neutral tint. Now, I admit I am a lousy judge of tint and CRI.

But if someone has a ruined MT-G2, that fake xp-l packet seems to pop of cleanly and intact. Seems a drop of silicone or led seal could be used to stick it to the top of the dedomed mt-g2. Of course, it won't be a premium flashlight led, but it could maybe used for a lantern or whatever. If there is a better way to recover a ruined MT-G2 (excluding remote phosphor), I'm all ears. I'm sure I will eventually ruin one of mine.

Thanks again djozz! As always, excellent work.

On my 20 MT-G2 overhead light I had a mishap where I ripped two of the 3000K domes off, phosphor and all. They broke in a clean pattern, so I stuck them back on with some LED seal just to see. They lined up clean and are still running upside down a few weeks later, so it must work OK. Of course I don't know how much output was lost, but the tint is still good and I can't tell the difference by eye with it turned on.

Good review as usual djozz. :slight_smile:

Terrible fake “XP-L” though. :expressionless: It doesn’t even look like the real XP-L in appearance.