Dedomed LED Tints Compared: XP-G3, XP-G2, XP-L

XP-G3 has the hotspot/corona effect of the old Seoul S42180 High CRI emitters, wonder if that’s consistent across all tint bins?

Thanks MEM, great work.

Maybe you don’t know but even XP-G2 S2 1A are affected by new production process. Confirmed by several members.

By judging your pictures you may have old production process XP-G2 S2 1A.

Your pictures explains a lot and thanks to Djozz and your test here we can see that XP-G3 is not improvement at all… Well at least in de domed version.

The ones that are able to get old series of XP-G2 emitters are lucky persons… It seems that you are one of them.

OK now I’m at a keyboard instead of my phone, so I can elaborate a little more. :wink:

Yes. I was able to get old production XP-G2 S2 1As. Most likely from a spare bin in China—which is probably 1000 different locations we could only hope to locate. :stuck_out_tongue: (It’s also why I provided the high-res source images, showing the emitters head on to confirm emitter details.)

Things to keep in mind here;

The LEDs are being driven at a (bucked) constant current of a little over 4 amps (about 4.2 amps). What does that mean? First off, the XP-L V6s, and XP-G3 are not being driven anywhere close to max intensity. The XP-G2 however is being driven at about 4/5ths to 5/6ths of the intensity you will ever see from one in a test like this. The XP-G2 is already making most of the lumens that it can make at over 4 amps current, in other words.

With that being said, the XP-L V6s and XP-G3 are really just getting “warmed up” at a current of ~4A. Remember, with Djozz’s test, the XP-G3 hits max output at a whopping 8 amps! So if you are drawing the conclusion that it’s simply not as intense as the XP-G2, just remember that its volume knob is only at a 5 of 10 here. :wink:

My thoughts? I think the XP-L V6s dedomed are incredible performers for a 2mm² die size. Look at the intensity the V6 has in both tints over the unknown factory bin DD XP-L with greenish tint. I too find the 3D tint of the DD V6 to be remarkably good. So the common thoughts that only a low number tint bin (0D, 1A, 1C, 1D etc) are the tints that are most acceptable for dedoming has sort of been dismissed by this very data, I would say. It was more of a stereotype than anything.

On with the XP-G3. Although this test shows a very unlikable beam profile, in nature I find it to actually be quite usable! Yes, I mean it. The yellow ghost tint around the hotspot is warm, but there’s really some usable light being made there. It gives the hotspot sort of a “blending” ability, where it’s not as pronounced, but a gradient of color range is featured and I find that the whole gradient can actually be useful. Anyways, the moral here is that I’m not even close to unlocking the full potential of the XP-G3 in these tests, but this just shows at alike modest-amperage values, what you can expect from them.

I will continue to add tints to this thread, also. I have plenty more willing to shine! :smiley:

If you have a dedomed tint request, post it. :wink:

Great thread. Thanks MEM.

Thanks MEM for adding your work to the mix here. It is at the least very very helpful. I appreciate it!!!

So XPL V6 DD ? Although I said I will not I’am on the way to order them. But it does not look intense as old XP-G2 but I don’t even doubt it can outperform new production process XP-G2’s.

If you are able to get old production types XP-G2’s maybe you are willing to sell some? I mean your own selling thread and you can ship internationally since it is very light package?

And for the end my opinion is that new G3 will never reach performance of old G2’s… Well at least with FET drivers.
However you may improve it in buck lights configurations but I am type of guy who likes maximum performance in as small light package as possible.

And even then if it will really sag 8A to reach maximum performance than I don’t really want it even for free :slight_smile:

So for now I really don’t see the way how this G3 could reach performance of top throwing led (old XP-G2 S4 2B) but I wish you very best luck with your tests. :+1:

Oh… I just figured out hi res pictures.

I really notice something different in your own production process:

It seems you use 2 kinds of lacquer treatments on your led. One ceramic white one and other transparent.

Well on G3 you can only use one kind of lacquer(transparent).

So this is interesting…

By the way, you’re welcome everyone! Thanks for reading. I will try to help as much as I can around here when I have the time for this stuff.

Luminarium, the LED in that photo is stock. It’s the XP-L HI V2 1A stock Cree emitter. It’s the same starting tint as the XP-G2 (1A), but look at the difference in tints after dedoming the G2! (Blue vs White) The V6 emitters are done by me, domes removed, and they are much more intense than a HI bin LED.

I can elaborate some more about what is going on with the XP-G3 beam/intensity. First of all, it is fully dedomed by me, there is no coating on it that I apply. What I see is the original phosphor layer applied by Cree after I dedome. OK, now…! It almost looks like the phosphor on a G3 is like a wet paint when they push and cure the dome onto the LED. I am able to see that the phosphor moved around under the dome, just like a liquid would. Maybe a high-viscosity liquid, etc. Anyways, on the XP-G3s I have dedomed, they all look the same. The “square” (die) is slightly raised up under the phosphor layer—you can clearly see its outline without the dome on. The interesting thing, is that the phosphor appears to be mashed outward from the die center, forming a concave surface. As would be expected, the high sides of the concave are around the edges of the raised die. However, the valley of the concave surface is at die center. Most of us know one thing about LEDs: phosphor creates the warmth of colors in the white mixture of light; colors that are beyond the blue pump wavelength (such as green/yellow/orange/red). Typically a thinner phosphor coating gets a lower number color bin (0D, 1A, 1D etc). Thicker phosphors, and phosphors with more pigments in them, render more total colors (usually higher CRI value, too) on the warmer end, but very much blue light is blocked from leaving—resulting in a lower lumen output bin. Hence, you won’t find 90CRI emitters with the same top bin numbers for lumens as 70CRI emitters in cooler tints that have less phosphor on top. Makes sense to those reading so far, right? Fairly common sense I think for many who know their tints and flux bins.

Great then, this should get even easier for me to explain if that all made sense.

I think, and this is just my opinion, but I think the XP-G3 S5 possibly “cheats” in some way to obtain its bin. Before you pounce all over me, let me try to explain the idea coming across my head. The very hotspot of the XP-G3 S5 in my photos shows high-intensity right near target center, not all across the hotspot like the XP-G2 does. For me to know why, that seems simple. The XP-G2 is one smooth square die face emitting light forward after dedoming it. On the other hand, the XP-G3 S5 is NOT smooth, remember that it has a concave dip on the die phosphor. Since the valley of the concave is almost smack dead center on the die square, more light of higher kelvin temp will escape the G3 from that small area at die center! Ah ha, now the hotspot is explained based upon physical characteristics which are real, not based on some unknown magical effect. :wink: Lumens are lumens of course, and if an emitter makes a certain number of lumens, it gets an intensity bin achieved accordingly. But, lumens operate on the principle of the CIE1939 color standard—a standard that takes into account very much the sensitivity of specific wavelengths that human eyes are most responsive to. Lux, just like lumens (real lux measured on a reference spectrometer), also uses the CIE1939 color coordinate system to gain potential light value when much green light can be found.

Now I’m starting to head away from my original subject and go on a tangent (yes, I’m aware I do this all of the time), but something interesting could further be concluded which I don’t often hear brought up. Since the CIE1939 standard places greens as the color to achieve lumens most easily, this topic circles and heads back to pose questions first about certain Cree emitters. There has been so much talk about XPG2-S4 2B emitters, yet many do not like them because, they have an ugly green tint! The highest bins that came from earlier G2s, and they were green. Why wouldn’t Cree have a S4-1A, or S4-1D on the previous emitter cycle with the G2 S4-2Bs? If you ask me, the answer almost seems clear. They were only able to obtain that bin as an S4, for the very reason that it contained so much green! Green content would produce optical testing results that showed higher lumens, and for you, if you used a wavelength band-gap spectrometer, you too would see lux values far above none-green emitters at alike power output levels.

Of course, this isn’t really cheating in any true way, because standards are standards, and the human eye is most responsive to greens. The question really is; would you prefer a tint that achieves a high bin because it mimics a single color the most like green, or would you be happier with a truer white, that may provide more total subject color information back to your eyes, though without getting the same lux/lumen response rate when tested?

Going all the way back to the XP-G3s now, here’s what I think. Cree could have done one smooth color, quite easily too. Just like a G2, they could have done it with a consistent phosphor coat. When someone with very precise means of component manufacturing does something a certain way, the only way I can think about it is, what’s the reason they did it that way? With the G3 having the concave phosphor, that leaks much blue light at the center of the die, what may Cree be achieving by producing such a vast mixture of tints from the XP-G3? It could be used to sway tint one direction or another, or, it could even be a way to extract a hotspot of extra lumens, even though they have a strange tint from that area. Even with all of the green color talk, just lowering the phosphor thickness in an area will still boost lumens, even if it doesn’t appear to be color biased. Whatever the reason, I suspect there very well is one.

Just sayin’! :smiley:

:beer:

P.S., Thanks to light modder extraordinaire Dale aka DB Custom, I was able to actually get the scraped and measured die size on the XP-G3 emitter earlier from him. I don’t know if this has ever been posted yet. I was using a picture attempting to measure in pixels, and when I came up with 1.36-1.38mm, Dale confirmed that the XP-G3 die is indeed ~1.4mm². It’s smaller than a standard XP-G2 by about 0.1mm²!

So…

XP-G2 = 1.5mm² = 2.25 square mm die area.

XP-G3 = 1.4mm² = 1.96 square mm die area.

Meaning, I must contest what is going on here. That’s 622 lumens per square mm on the XP-G2, based on 1400 lumen max. On the XP-G3, that’s 908 lumens per square mm, based on 1780 lumens. What the heck is going on? Almost 50% more lumens coming from the same area is simply more intense. I looked at it with dark weld shades; the phosphor glows at the die surface, the whole phosphor surface isn’t lighting up.

Time to get to the bottom of this, once and for all. :smiling_imp:

amazing work MEM! But how is it possible that all throw results from the xp-g3 are lower than those from the xp-g2?

If you check out what I just wrote above at the end, the mathematics at first do not make sense.

I suspect there is a third hidden element to the outcome. I believe that it very well could have to do with total, 3-dimensional surface area. Look at a map of the Rockies, the map is flat. Examine them though, and a great difference presents itself. A non-flat phosphor surface is certainly something the G3 has but the G2 does not.

I will get to the bottom of it, though. Might take me a week.

This is assuming the flux doesn’t decrease after dedoming. Also, one can’t assume that two different LEDs have the same behavior upon dedoming, as we have seen with the 219c, which is close to the XP-G2 in total flux before dedoming, but has a much lower luminance than the XP-G2 after dedoming.

Still looking forward to some lux measurements.

you mean the uneven surface of the xp-g3 has a much larger total area than the xp-g2, thus reducing its overall brightness ? or spatial orientation of the light emitted from the die to be the cause- since it comes from a curved die, vectors of photons vary more than usually, and are hard to focus in a traditional reflector???

Underlined: That was my assumption/guess as to why these exhibit the different k hotspot/corona, the different thickness phosphor around the edges.

As far as the XP-G2, has anyone seen one measured in a spectrometer? It would show the dominant and peak wavelengths and answer the “green” question, I’m not sure I have ever seen any posted.

Thanks for the informative post.

-Michael

The XP-G3 doesn’t have visible bond wires. Not with the dome on, not with the dome off, not with the phosphor removed. It’s made more similarly to the MT-G2 than to the other XP series emitters, with the exception of it’s stack structure, which is entirely new.

And once the phosphor is removed from the silicon chip, the chip itself will freely pop off the die substrate and underlying architecture! Never seen anything like it.

MEM

I mean wow! Very informative reply. Plenty of your replies are very informative some of them more like expert to expert orientated.

“The question really is; would you prefer a tint that achieves a high bin because it mimics a single color the most like green, or would you be happier with a truer white, that may provide more total subject color information back to your eyes, though without getting the same lux/lumen response rate when tested?”

MEM old xp-g2 s4 2b had neutral white color with slightly green on the edges:
I must show this picture(and this is taken under the angle by me so it is not top pic)

Left is new G2 s4 2b and right it is old G2 S4 2B (on this picture although new G2 loos smaller cause i took angle pic it has bigger dice projection than old G2 with significant lux drop and ugly green tint) They are both de domed.

So yes if they can produce something that would have nice neutral white color(after de doming of course or lets say in HI version) like old G2 S4 2B dd had… Then yes… I would really prefer it over lets say lower performing cool white color.

Cool white does not render as well as neutral white imho.

I don’t get all this “mystery” around XP-G3.
Its die size is certainly bigger than XP-G2,check pictures in this thread if you don’t believe:

And this is confirmed by djozz,Richard(MTN)-official seller and others.

How much xg-g3 die is bigger compared to old xp-g2? You don’t even have to measure it,it’s in datasheet numbers:
Current 2A vs. 1.5A,
Thermal resistance 3 vs. 4C/W,
Viewing angle 125 vs 115 deg
Forward voltage 2.89 vs 3.00V @1A

2A/1.5A = 1.33

(3C/W)/(4C/W)=0.75, or 1/0.75= 1.33 times better

Those two numbers are very tight correlated with only die size,especially thermal resistance.

So,that means XP-G3 die size is most probably ~1.33 times bigger than xp-g2 die size,but it could be slightly lower if we consider that cree decided to drive xp-g3 just a little bit harder,so my best bet is 1.25 times bigger,or ~2.5mm^2,a nice “round” number.

And it’s not hard to achieve better lumen per package performance with bigger die at same current,so it’s unfair to compare xp-g3 to xp-g2 at for example 1Amp,just like it would be unfair to compare XP-L to XP-G3 at 1Amp - bigger die wins.

If we talking about performance, xp-g3 top available bin is S5,while top bin for xp-g2 is S4. Difference is only one bin,or 7%,and if xp-g3 is in same performance class,difference should be at least 3 bins,not just one.

What it’s most weird is that Cree didn’t change “G” letter in XP-G3, since that letter shows die size,well at least until now, E was 1mm^2, G was 2mm^2, L was 4mm^2,my opinion is that is just commercial trick,illusion of improvement in lm/mm^2 at the same current,but in reality there isn’t any improvement in that area,higher lumens are mainly because of larger die/lower current density.

Also answer to question why there was xp-g2.5 version with old looking style,but with new bigger die can be found in xp-g3 press release:
http://www.cree.com/News-and-Events/Cree-News/Press-Releases/2016/April/Cree-Introduces-the-Next-Generation-of-XLamp-XPG-Platfor

“XP-G3 LED has 6,000 hours of LM-80 data immediately available”

That means they tested new die for almost one year before xp-g3 announcement, so most probably they made a test batch of xp-g2.5 for LM80 testing,and I doubt they were intended for commercial use.How and why they end up for sale,we should ask Chinese :money_mouth_face:

As for greenish tint in almost all Cree leds,they do this from day one,and you explained the reason: more green - more lumens !
I actually think name Cree originate from Green,anyone else see the coincidence? :confounded:

led4power,

Man you did mistake :slight_smile:

In my thread I am speaking about production differences of same emitters so it is XP-G2 new process vs XP-G2 old production process.

XP-G3 is completely different than them. So MEM is probably right. I like expert to expert confrontations we can learn a lot from you guys :+1:

You’ll have to edit your reply a bit :slight_smile:

I know, xp-g2.5 has the same die size as xp-g3, so everything should be correct.

Oh… I got you now my apologize :blush:

XP-G3 die size, y’all are funny.

Talk about it all you want, extrapolate from the numbers, do whatever you think ya gotta do, or simply take one apart and measure it with digital calipers…