100 CRI obtainable in a triple XML (3,700K - 5,000K - 8,300K)?

[SmurfTacular]

According to Cree XML's PDF

• Typical CRI for Cool White (5,000 K – 8,300 K CCT) is 65.
• Typical CRI for Neutral White (3,700 K – 5,000 K CCT) is 75.
• Typical CRI for Warm White (2,600 K – 3,700 K CCT) is 80

What If I made a 3D, Direct Drive, triple XML, with all three different color temperatures... Or what if I used a Cool White, Warm White, and an RGB LED? Would it be possible to obtain a CRI of 100? Wouldn't that be the more efficient/higher quality?

 

[DimeRazorback]

I would imagine the beam tint would look strange... so even if you did obtain the desired CRI the strange tint may be off putting to your senses.

 

[mvyrmnd]

I've wondered this myself before. I came to a realisation recently:

Low CRI emitters are usually that due to the lack of red/amber. You will probably not be increasing the CRI of the warm emitter by adding a cooler one. I suspect the final result will still be a CRI of not much more than 80.

If you were to repeat the experiment shown here, you could probably work out which parts of the spectrum were missing and find an emitter that would replace them accordingly.

People often get a CRI increase by adding Red (and sometimes cyan) emitters to the mix.

 

[SmurfTacular]

Wow, thats good to know. What I'm trying to make is a three LED Maglite that is capable of producing ~99 CRI Neutral/Cool white light. What LED would I have to use to balance the CRI of two Neutral/Cool White XML's?

EDIT:

(from Cree XML PDF)

It looks like if I used a ~480nm LED, it would fill in the gap in the spectrum...

 

[mvyrmnd]

480nm is still blue though... :thinking:

Have a look at the graph for the 80-CRI XM-L Easy White

And this very cool chart, from this very informative photography guide

What you're really trying to achieve is the dotted line. The 3000K emitter is clearly lacking enough blue, so I think it's time for me to retract my earlier statement. I'm not so sure including a 4000K emitter will help much, but a 6000K and 3000K emitter should produce a wider spectrum of light.

460-530nm is the crossover from blue to green. A cyan emitter would fill that hole nicely.

If you could get a High CRI 3000K emitter, a 6000K emitter and a cyan emitter together, and run them at the required power to have them produce the correct spectrum, and successfully blend them so there's no funky bits in the beam, you might just get your 95-100 CRI...

Also, have you read this thread?.

 

[Harold_B]

I would think that it is not likely with three LEDs and very unlikely with the standard XM-L to exceed 90 CRI. There are a few things working against you for the XM-L one of them being color separation over the viewing axis. When we tested our prerelease samples we found that they varied about 4000 K over a 45 degree sweep from the viewing axis. Sylvania uses a mix of cool white and red in their A19 replacement (12 OSLONs total) for a CRI in the 80's. Cree uses an active feedback in their easywhite down lights to maintain the color balance over time since the cool white and the red LEDs will degenerate at different rates. I was not aware until I saw an article linked by EVO that the Philips LPrize entry has red LEDs in it. That would make it a remote phosphor plus red LEDs to get the required 90+ CRI for LPrize.

I'm not trying to be overly negative about your project I'm just trying to add a different perspective so that you know how difficult your goal is.

 

[ma_sha1]

Not possible with triple XML, but might be possible with proper RGB mix?

 

[CKOD]

The problem with RGB is its 3 relatively narrow spectrums. While we can produce any apparent color to the eye with the RGB combination, it doesnt mean we can use it to illuminate everything perfectly. I.E. if a pigment in something has a narrow response to a cyan color, and you only have a blue and a green peek, the color of that pigment could be washed out, even though you could fool the eye into seeing cyan with green and blue.

The chart mvyrmnd posted makes it make sense to me why we see all these hi-CRI leds in disappointing 2700k and 3000k tints, instead of a nice white 4500-5500k sun-like tint. The warm white output spectrum with its relatively low output in the blue-green area matches up with 3000k incan spectrum fairly well. If you can find color LEDs with a decently wide spectrum, a neutral or warm white LED with green and blue (or just cyan) added as needed to smooth out the spectrum, and possibly a deep/far-red.

 

[CarpentryHero]

Have you thought about Sst-90's ? There a little more expensive but they've been around longer so more tints are available like red.

 

[SmurfTacular]

If you could get a High CRI 3000K emitter, a 6000K emitter and a cyan emitter together, and run them at the required power to have them produce the correct spectrum, and successfully blend them so there's no funky bits in the beam, you might just get your 95-100 CRI...

You think two 10 watt XML's (3K + 6K), plus a 3 Watt cyan would be the correct balance? The reflector assembly is one piece so all LED's will hit the same hotspot.

 

[mvyrmnd]

You think two 10 watt XML's (3K + 6K), plus a 3 Watt cyan would be the correct balance? The reflector assembly is one piece so all LED's will hit the same hotspot.

I wouldn't have a clue. Trial and error?

Personally I'd put it all behind a diffuser to make sure it's properly blended. You shouldn't have chunks in your cocktail.

 

[Th232]

Adding cyan would help, but getting 95-100 is still a pretty big ask IMO. Remember that as you get closer to 100, you're trying to fill increasingly small gaps and regulating the current to get the right proportions of each wavelength. I suspect that there's also going to be a certain point where you'll have to get a deep/far red LED. How are you planning to measure this anyway? Is something that "looks close enough to daylight" going to be sufficient, or are you going to put this through a spectrometer to see how far off you are?

Also, with only specifying XM-Ls and that you want it to be direct drive, without knowing the Vf or flux bins figuring out the proportions is going to be a tricky exercise. Especially as the batteries die down, the proportions will probably be thrown out.

 

[mvyrmnd]

I just had an experience that highlighted the lack of useful green light in the LED output. I was using my High CRI Triple to look for ripe lemons in the dark. I could easily see the yellow lemons thanks to the high amber output, but I couldn't differentiate the slightly unripe lemons from the fully ripe ones. After guessing and taking them inside I discovered at least half of the ones I picked weren't quite ripe.

The extra cyan would definitely help here.

I suspect that there's also going to be a certain point where you'll have to get a deep/far red LED.

Absolutely. I've been focusing so far on filling in the trough in the spectrum at the lower (higher?, shorter?) end. As you can see form the two graphs I posted, there's a massive falloff in red output above (below?, longer?) 650nm. The daylight graph still has over 80% relative output at 650-700nm

 

[foxtrot824]

I had a similar idea using XP-Gs in 3K 4K and 6.6K and didn't end up with an even color like I though would have balanced. See thread here
If you look at Cree's LR6 light fixture that scores a 93 or better on the CRI they you'll see they mix 5 cools with 3 red under a diffuser to create a 27-35K light. I'm not sure if a ratio of 3/5th would look as good when focused to a point but it might be a different approach.

 

[Al Combs]

I also think you'd get better results mixing in color LED's instead of different shades of white. Mac's Customs used to make a light called the Sceptre, a real beauty. In post #103 of that thread, he mentions future versions (over 3 years ago) would have a single red LED in the center of the other 6 for enhanced color rendition. I wouldn't know if it was the first of its kind. Just the first time I saw the idea.

I was surprised to find that Cree doesn't even make a cyan LED. I looked at the Rebel data sheet after looking at mvyrmnd's link to the high-CRI flood light thread. The cyan peak of 505 nm is much closer to the hole in the spectral power distribution graph you have in post #4. Even though the XR-E is available in colors beside shades of white, only the XR-C and XP-E are available in red. They both have max drive currents of 700ma, so you couldn't wire them in series with the Cool White XM-L's. You'd need a separate regulator for the Red.

Sorry, it's not very helpful to your project.