testing two of KD's new Nichia 219b leds (4500K'92CRI' and 5500K)

I am still confused about which 219B is which, Illumination Supply had supposedly the best tint and bin of the new Nichia219B 4500K 92CRI, but Kaidomain also showed up with some new Nichia 219B's, with mistakes in the type nrs. IS was directly sold out, I ordered the KD one, and also a 5500K (non-highCRI) one. These are my tests. The subjects (the first two are the same led type):

In the description further in this post I just stick to the type nr. of Kaidomain, I will update if I know anything for sure.

This new 219B looks a bit different from the 219A I was used to, the most important difference (apart from performance of course) is that it has a more shallow dome (most of the Nichia 119 and 219 leds have this shallower dome also), actually it is more comparable to a XP-G2 now. You can see that reflected in the emitting angle, it went up from 110 degrees to 120degrees. (left old, right new):

As usual I test output and voltage at various currents with the leds on 16mm Sinkpads mounted on a aluminium block, with a reflector on the led, not extemely accurate, but good for a general idea .I think at least I can compare results well to my earlier results with other leds. I ran out of Sinkpads so these are second hand and quite dirty, I guess they still work alright:

These new Nichia's test very well, but, as I can not measure it, I can not guarantee that the colour temperature and CRI of the leds, as given by the Kaidomain site, are correct (as I understood, Kaidomain is notorious for wrong descriptions). However, as far as I can guess by eye (I have seen a lot of led tints), the high CRI one could really be high CRI (nice tint and colour reproduction), although the tint is a tiny bit cooler than the 'old' Nichia 219A 92CRI, when compared directly (if I mount it in a flashlight one of these days, I can see it better). For real answers on tint and CRI maybe we should ask texaspyro for some thorough colour testing of this led on his fancy equipment?

In the 'high CRI' graph I compare the 92CRI 219b to the old 92CRI 219A (measurements from some time ago) because that is the most interesting comparison. In the 'cool white' graph I compare the 5500K 219b to the XP-G2 R5 0D-tint (measurements from some time ago) because that would be its closest competitor. Here they are:

If this led is the real thing (4500K and 92CRI), it performs way better than the old 219 in every aspect: considerably higher output at any current (up to 3.4A not that far from a cool white XP-G2 actually), it reaches maximum at a higher current, and the voltage is way lower, lower even than the xp-g2.

The Nichia 219 5500K compared to Cree XP-G2:

Also this is impressive, up to 3.2A its output is a bit over the XP-G2 R5, but at a lower voltage :-) . Up to 3.4A the Nichia is more efficient than the Cree, and current regulation in a single Li-ion flashlight will be easier. The XP-G2 is still lord and master at the higher currents, so for the extreme flashlight builds it will be around for a while.

But does the Nichia dedome? I had a go some time ago on a 219A 92CRI led and that dedome worked ok, but a shiny thin layer of silicone was leftover on the die. Well, this time it went no different. I heated the 5500K 219b board up on my heatblock 'till the solder melted, took the board off, pushed it down with a piece of cardboard, cut the dome in from the side with a scalpel and popped the dome slowly off. Same stiff silicone as the other 219 I dedomed, much stiffer than Cree. Some silicone was left on the die, tried to remove some of it with a scalpel, but not everything came off. So this Nichia does not dedome as neat as the Cree dies (perhaps the gasoline methos works a bit better??), but the dedome was done and the led still works. First a comparison of the die size with a dedomed xp-g2 (the Nichia die is bigger, not promising for throw..), and then the die at 10mA, the tint has turned a bit greenish .


Next is build it in a C8 pill at 3A for a real life test. Hang on, that may take a week, I'm busy and tired...

And I guess after confirmation of the CRI (how? :-( ), it is also time to rebuild my 1x18650 Nichia 219 high CRI triple for some more than doubling the output....

Thanks again djozz. How did you like the tint?

Thanks again MRsDNF for the quick reply (to all my threads actually :-) )

I liked the tint of the '4500K 92CRI' one very much, very neutral white without the 'rosy' tint that the 219 is famous for. I have not payed much attention to the 5500K led tint yet. I dedomed one (so that one is destroyed for tint asessment), I will have a look at the other one soon.

So this were true ! Impressive ! Maybe one day Nichia will make a bigger LED to compete against XM-L ?

KD also sell triple Nichia! This is Cheap ! This should be the ultimate high quality flood light . :heart_eyes:

Interesting! Thanks once again for your curiosity djozz!
I’d like to see a mod with the triple nichia board, will old-lumens do one soon perhaps?

Thanks for the graphs!

Thanks for your wotk. Is there any chance for some beamshots? Quick mount with any free reflector, pointed at some colorful object so everyone could judge is it really high CRI.

WOW.. If the Nichia 219B really is 92 CRI then its a HUGE step forward.

Being able to easily get 500+ high cri NW lumens out of a single cell light... Me likey!

The 5500K Nichia looks really good on paper too!

This also open up new extreme possibilities with those 20mm triple TIR optics.. J)


Thanks djozz! Looking forward to hear more about your subjective thoughts when you get the time.. :)

Looks like I may be replacing a few old Nichia’s in some of my lights.
Thanks for sharing your work here Djozz, much appreciated. :wink:

Nicha chip is obviously bigger than 2mm^2.Could someone calculate exact die size from pic? It looks like 3mm^2 at least.

I took out my trusty ruler, the surface area of the Nichia219b die from the picture measures 1.3 times the XP-G2 die. so with the same output, throw should be 23% less.

Thanks Djozz for the tests!!! Highly appreciated!

Now I'm looking forward to hear more about the CRI.......

I devoted quite some time in finding out how to reproduce tints and CRI on the internet, and although it is possible to show differences between leds, it is just not what is happening in real life. I had a discussion with RaceR about the usefulness of an as good as possible representation of led tints on colour charts and photo's, and although despite the quite disappointing results I still thought it was worth the effort, RaceR thought not. But it gives an idea how hopeless it is, especially if you want to demonstrate CRI with pictures, high CRI leds look nearly as good as any neutral led in such a picture while in real life it is so much nicer...

Mind that the triple that KD is selling is the with old type 219A, great emitter up till now, but with this new option I would not want it anymore. Great price though :-)

lower forward voltage, and more lumen!


You must have misunderstood me. I personally found the charts very interesting. :) I was the last one posting in that thread too..

I just did not think that using the same white balance for emitters in the 7000K and 3000K gives the best comparison when comparing various and completely different emitters ability to render colors (CRI).

If the white balance is correct for two different tints (basically white should be proper white) it will be easier to compare their ability to render colors correctly. Especially if you have a benchmark (100 CRI picture) where the WB is correct too (again, proper white).

If you want to get some sort of visual point of view, then the approach with a fixed daylight setting is "ok". Im saying "ok" because there is no great way capture tints like the eyes perceives and adapts to various tints. For a CRI comparison, id say correct WB on all pictures is easily the best way. But there as still challenges......

At the end of the day, it would not be easy to distinguish a 92 CRI and a 80 CRI emitter with certainty anyway. So your subjective thoughts would be good enough for me. You have lots of experience with various emitters and tints. The best way is always to personally experience tints in real life and in various environments anyway. :)

Thanks for explaining RaceR86, in the end this all means: tint photography can be useful, but only bleakly reflects what we see in real life..

And even if you manage to get a picture that accurately reflects what you see in real life, it may look completely different on somebody else's monitor. There's likely a bigger difference between two different monitors than between two similar LEDs.

That is huge variable indeed, different panels (TN, various IPS), personal settings (brightness, Kevin of the monitor), browsers often times do not display colors well. Most monitors are bad anyway even if you can say "I see rich colors"

My pictures, with my camera and as displayed by my monitor, only look 'right' with the white balance on the fluorescent setting. They may look like harsh blue-purple utter crap on everybody else's monitors, who knows. Using the daylight setting on the camera gives something that looks like it's from the 1940s.