Ryzbor wrote:

4000K+4500K as a neutral tinted (?) 4250K

Not much reason to mix such close color temperatures. The difference between 4000K and 4000+4500K will be practically indistinguishable.

Tested the difference. CIEDE2000 color difference is 1.44 which means there’s no practical difference. DeltaE >3 starts to be obvious.

Aren’t they just glorious (sorry, chart doesn’t reach warm enough to show 2000K, 2200K, 2000+3000K).

CRI data for all the emitters and some tint mixes. All except one mistake at 700mA per emitter.

2000K
2200K
2700K
3000K
3500K
4000K
4500K
5000K
5700K
6500K

2000+3000K
3000+4000K
3000+4500K
3500+4500K
4000+4500K (350mA per emitter, oops)
4000+5000K
4000+5700K
2000+6500K

Thanks a lot maukka and clemence.

This is really cool, and I should definitely try some E21A emitters.

Any idea how a 2700K+6500K or 3000K+6500K combo would look? Ideally I’d hope to hit near 4200K with duv -0.008 or so.

Edit: On second thought, perhaps I should just go for the 4500K emitters by themselves. They seem just about right, aside from being a bit less rosy than desired. With the flat surface, it’d probably have a fantastic beam.

Here’s data in a big table:
https://docs.google.com/spreadsheets/d/1CERSuF2JvYtjsoyPh-wEZ0dG6fSI-FDC...

I haven’t measured lumens yet, so the Y column is just relative brightness.

ToyKeeper wrote:

Any idea how a 2700K+6500K or 3000K+6500K combo would look? Ideally I’d hope to hit near 4200K with duv -0.008 or so.

Open the chart in a photo editing program and draw a line between the emitters you’d like to mix and eyeball the xy coordinates at the middle point. Enter them to a calculator such as this one: https://www.waveformlighting.com/tech/calculate-duv-from-cie-1931-xy-coo...

edit: as it happens, in my image the duv (with a bit of a multiplier) is pretty much the 1:1 pixel distance from the BBL, so that helps too

maukka, could you hold total current fixed but change current of components that mixing?
It is interesting would it be trajectory of such kind or not.

Here’s the spectra and CIE plot for the E17A color LEDs

SRGB color space for reference

Only the CRI for the amber LED is of interest

Thanks Maukka for providing those extremely useful test data for us. I have a few questions.

1) Why does tint mixing drop R9 so much on the 2000k+ 6500k

2) Could the low R values for 2200k be a fluke?

3) If I remember correctly the batch of E21A you tested last year was much rosier. Is that correct? This batch seems to hug the BBL similar to the Optisolis.

Also a request if it doesnt require too much effort is if you can test 2000k+2700k, 2000k+4000k, and 2000k+4500k. Of it require much effort to setup the test then nevermind.

Thanks again for your hard work and contribution.

These two scenes are amazingly similar!

What’s that game called, Clemence?

clemence wrote:

AFAIK it starts with something “…find hidden objects…”

It seems your image is a screenshot from a game called ‘Criminal Case.’

clemence wrote:

Amber looks like a low CRI extremely warm LED. Mixing red and amber can produce something interesting…I request a test for this Maukka.

Amber+Red

RGBA. Bad Rf, huge Rg. Very magenta.

maukka wrote:

RGBA. Bad Rf, huge Rg. Very magenta.

Thanks for making that measurement. I suspected it would end up being unbalanced. Trying to build anything around this mix would result in a freakshow light if current to each channel can’t be managed separately. One the other hand, being able to adjust each channel individually would be what would make an RGBA light fun.

It seems like blue would need to be dialed back quite a bit, and perhaps the amber increased a little in order to get a more natural result.

Sure, by individually addressing the emitters, you can tune it to anywhere within the color space quadrangle they cover.

It’s trippy.

Added lumens and efficacy numbers to the Google Sheet.

Individual Current-Flux-Voltage curves

2000K
2200K
2700K
3000K
3500K
4000K
4500K
5000K
5700K
6500K

Tint in an integrating sphere without any optics, i.e. how these measure diffused and how the binning is done (much closer to advertised CCT values)

Individual CRI data

2000K
2200K
2700K
3000K
3500K
4000K
4500K
5000K
5700K
6500K
2000K+6500K

Great job as usual Maukka, thank’s for measuring the requested mixes

maukka wrote:

Ryzbor wrote:

4000K+4500K as a neutral tinted (?) 4250K

Not much reason to mix such close color temperatures. The difference between 4000K and 4000+4500K will be practically indistinguishable.

The thought behind this is to get the tint of a d240 4000K Wizard Pro below the bbl on the medium modes since the current setup is above it even on M3 ~250lm as bob_mcbob measured. And I see 4000K+4500K is slightly below the BBL at around 450lm.

Speaking of, Maukka could you add info about which flux binning you used in cases more than one is available? Eg. your previous the 4000K you measured abut a year or more ago had the d220 bin but now you’ve measured the d240 bin.

I find the results a bit optimistic besides the 5000K (which bin?). At 700mA/led, thus around 990lm it’s pretty obvious the tint is going to be below BBL and the E21A is well known for that. Even the XHP50 WW in the Wizard is below BBL on turbo. 990lm is more a turbo mode than a regular used and sustainable mode in a typical host. What really matters is how every setup performs a medium mode and a low mode which are going te be used by fat the most in real world usage. IMO it would be more interesting to know how they perform at lets say 200lm or even less. That’s where the “big table” helps.

clemence wrote:

Normally, the middle rank is the safest for good output at or below BBL tint. The higher the output the farther up from BBL the tint. [Clemence]

I couldn’t agree more…

But I’m actually open for a positive duv setup for special tasks. For walking in the woods I actually prefer positive duv above a pink tint, maybe because trees aren’t pink and leaves look less green under a pink tint

Or maybe a 4500K (rosy)+5000K(greenish) mix to create a true neutral 4750K…

Ryzbor wrote:

Speaking of, Maukka could you add info about which flux binning you used in cases more than one is available?

Clemence probably knows the flux bins. I just got the bare leds on mcpcbs.