From my experience, the 3 step nichias are more likely to be more yellow/green (above BBL) and the 5-step ones are more neutral to rosy (below BBL). The only way to know for sure which side of the black body locus it’s going to be on is to get the ‘70e’ bins (219c sm3070e, sm4070e, sm5070e, etc.) or those 9080 219b emitters Clemence has.
good info, I agree the e bins or B nichias are more likely to be below the BBL, into the pink I like
otoh, every led varies, I have two sm305, one is slightly pink, the other slightly yellow, as are the sm303
a 1/8 minus green filter helps the non pink LEDs 
note about white balance
my photos use auto white balance, so the sw40 being coolest, was captured as white, thus the other led colors are only relative, not absolute
you can compare them to each other, but not to a photo taken with a cooler white balance. the color of the sw40 is different when white balance is set to 6000k than when it is set to a white balance of 4000k as in the two images above
Hi Jon, I adjusted the white balance to match my tintshot:
Hint:
What I’ve learnt so far….
Higher voltage bin = slightly brighter LED. For example D240 M1 can be actually D260ish, while D240 L1 placed at the bottom of the D240 bin rank.
This is easy to check by light them up under moonlight mode (very low current) in series config.
Higher voltage or any brighter LED within the same bins usually clumped at the upper zone of each Macadam’s island. For example most of the 219C sm403 M2 will be most likely to be brighter and greenish than 219C sm403 L1.
But logically the voltage increase over higher current should be less in warmer temp LEDs due to increased die temperature.
We have no control over voltage selection, only the flux bin and Macadam steps requests are available. Well, at least we know what to expect. Will confirm later if this finding can be used reliably as a guidance or not. I need to check and post the results.
- Clemence
That 1/8 minus green filter is nice…is the lower output noticeable?
ThankYou! Super helpful and appreciated.
Ive been trying, and failing, to figure out how to do that
well… its a drop of just under 15%, or in the case of a 200 lumen light, it would become a 170 lumen light…
if the question is, can you notice a drop of 30 lumens… not very easily
its just another way to “improve” tint that is too yellow
but there is no free lunch… nichia got more lumens by tolerating more yellow lumens… if they had stuck to red lumens, we would still have Nichia 219b, and they give up more brightness…
basically, putting a 15% lumen reducer on a yellow nichia, makes it act like a pink nichia, if that makes any sense
as to whether you would notice, that depends on how you feel about pink lumens vs yellow lumens
I happen to love pink and hate yellow, but Im neither you, nor normal… LOL
its cheap enough to try, and won’t harm anything… contact lee filters for a sample pack…
oh, and according to maukka, the filter does raise CRI, slightly… see this post that shows his measurements
I’m experimenting with a “corona killer”.
Basically a collar around the emitter that i painted flat white with ‘white out’ (a.k.a. ‘tipp ex’)
You can use a high LED gasket or a section of a TIR holder.
What it does is mix the more yellow light coming from the sides of the LED with the spill, because it no longer hits the reflector, but is scattered by the flat white collar.
I use ‘white out’ (‘tipp ex’) because it’s the whitest and best covering (least translucent) flat white paint that i know of.
It takes away some yellow from the beam and gets rid of most of the corona.
The spill gets a little warmer (more yellow).
Basically it reduces the tint shifts significantly.
My Cree XM-L2 4D in a C8 sized light (actually a Luma power MRV) with OP reflector and a mild diffuser (glued onto the glas lens with a very thin layer of ‘Kafuter’ UV curing glue) looks a lot better now, it looks pretty nice actually.
My UltraFire WF-502B (OP reflector) with Nichia 219C 3000K with basically the same adaptations improved similarly.
Both have slightly less throw now, and a bit more spill in stead, but not dramatically.
Can you check the output difference?
No i can’t…
I have no lux meter…
I had a good result with more of less same approach. Will do it again and post the result tonight.
Been having fun with these nichias Clemence! Built an M1 quad today using 2S2P emitters. All Nichias I got from you.
It has one of the prized 219B SW45K R9080 in series with a 219C SM305 R9050, and a 219B SW40 R9080 in series with a 219C SM4070e R9050.
The resulting beam is around 3800k-ish I would guess, with a lovely rosy tint and very little to no yellow. Probably the best looking tint I have ever seen, and my new favorite light. The beamshot really doesn’t do it justice at all.
I did a similar experiment mostly by accident. I 3D printed my own spacer/centering ring for an XHP70.2 and messed the dimensions up a bit. As a result the white plastic ring came up the sides of the emitter quite a bit. I noticed though that afterwards much of the tint shift expected with the new Cree emitters was gone, and decided that my mistake wasn’t a bug, it’s a feature.
Whoaa, Nichia cocktails! Interesting….
If you can select any of them individually it would be a nice moodlight!
Tint blending and mixing Vf has been a fun project for me. It started just to show that using boost drivers for triples we can mix whatever emitters we want regardless of Vf because they are in series. It ran a 219B SW40, 319A SM355, and XP-G3 30G 80+, and I was amazed by the beam.
After that first light I got hooked on the idea that by mixing different tints we can basically tailor the net light output to almost anything the user desires in multi-emitter setups.
I think I’ve proven the concept to myself at least, so it is likely there will be more lights like this from me in the future. I really don’t see any downsides. It certainly makes it much more interesting deciding what to build. I want to try one with tints further apart in the future, maybe a 6500k 319A with a 2700k 219C and a 4500k 219B in a triple.
Your driver will be a good candidate for the next E21A quad boards. It’s a true DIY quad and multi CCT board with much better beam control than currently popular quad optics. It’s very similar to XML colour but the user can freely decide which to put inside.
3V, 6V, 9V, and 12V configurable
Each LED are individually addressable for greater freedom.
- Clemence
Reading through this recently, my dream light would now have a 4 FET driver hooked up to 4 different tints from 2700K (or lower if possible) to 5000K or so for smooth ramping of both color temperature and intensity. I doubt I’ll ever have time to make it happen though 
I am impressed. Might have to pick up some E21As once the boards go on sale. Sort of like a DIY 144A. I do wish I could address the emitters individually with the driver I am using now, would be cool to have 4500k to 3000k on command.
I might be missing something, but outside of only populating 3 spots, how would your board be 9V? 3, 6, and 12 are what I would expect for 4 emitters.
Here's the redo test:
I roughen the inside of the vertical LED cavity only (left the top untouched) using cotton swab and 320grit silicon carbide mixed with soapy water. Then polished the entire internal cavity using clear plastic polish (car headlight polish) to get the smooth diffused surface. The result were excellent but unfortunately I dropped this pretty optics and it cracked internally.
The center collimator already implemented sandblasted finish by Gaggione.
For the brief test I used Sctch's magic tape (cleaned the adhesive first with IPA). FYI, this is the one of the most unforgiving 35mm TIR optics dedicated for maximum candela. Even the good old XPG cast die image with this one.
The magic tape turned the light yellowish. Even if it's applied in front of the optics/reflector. But this should give you the idea.
The best result still achieved with the super expensive Luminit's 5° LSD
- Clemence
9V for 2S2P config. This might not appeal most flashaholics. But in automotive world or wall washing lighting you can get almost perfect projected beam profile from an oval optics using asymmetrically lit LED array. Projected at an angle simple oval optics would make the beam profile overly bright for close range and insufficient lux at farther range. With complex asymmetric TIR/multi faceted reflector any kind of beam shapes are possible but at the expense of additional R&D cost. Simple off the shelf oval optics paired with asymmetric LED array produces nearly perfect beam pattern instantly without having to spend more R&D money.
- Clemence
I would like that, IF I could set Low to use only the 2700k, Medium to use only the 4500k, and high to use only the 6500k…
I suppose with the right UI you could add other options like pairing the 2700 and 4500… pairing the 4500 and 6500, and of course turbo would fire all 3 at once…
but to me, mixing leds that turn on simultaneously is just a compromise, that I won’t make. When I want 2700k light, I absolutely do not want 6500k light, and vice versa…
respect for your creativity, forgive me for failing to understand why it is desirable to mix color temperatures… Im open to learning
Brilliant!!!