CRI and efficacy of RGB arrays versus white LEDs?

[kaon]

Something I've been wondering:

In principle, for the purpose of making large quantities of white light, would RGB (or RGBY) arrays be cheaper and more efficient than "white" LEDs?

RGB arrays are usually wired so that each color group can be driven individually, since they are designed for variable colour.
If the goal were to make lots of white light, then addressability is not required.

Since "white" LEDs work depend on fluorescence, which has inherent inefficiencies, I am wondering if color mixing RGBY might produce white more efficiently.
In real life, "white" LEDs in production seem to have same or better efficacy than coloured ones, (of similar output / module), is this due to whites being in highest demand?

And what about CRI?

Of course, there's the problem of colored shadows, I expect them to be most pronounced in small arrays, but large ones with hundreds of interspersed RGB leds would not make the colored shadows.

 

[snarfer]

This is an interesting question. Something I've wondered about too. It certainly does appear that high CRI white LEDs provide more lumens per dollar and lumens per watt than RGB or RGBY arrays.

RGB arrays are usually wired so that each color group can be driven individually, since they are designed for variable colour.
If the goal were to make lots of white light, then addressability is not required.

I don't think this is true. White LEDs don't shift in color that much as you dim them, or as they change temperature. This is not true of colored LEDs. In order to have consistently high CRI white light with colored LEDs you would need to be able to vary the intensity of the different colors within the array.

Also some of the lights that use colored LEDs for high CRI applications actually have 6 or more different colors within the array. So the control circuitry is far more complex.

Of course, there's the problem of colored shadows, I expect them to be most pronounced in small arrays, but large ones with hundreds of interspersed RGB leds would not make the colored shadows.

I think the issue is not just colored shadows, which can be somewhat addressed with diffusion, but also colored reflections, which is a much harder problem. Any time you have a shiny surface you will see the reflection of those colored LEDs.

Since "white" LEDs work depend on fluorescence, which has inherent inefficiencies, I am wondering if color mixing RGBY might produce white more efficiently.
In real life, "white" LEDs in production seem to have same or better efficacy than coloured ones, (of similar output / module), is this due to whites being in highest demand?

This is just speculation on my part, but maybe it is due to higher efficiency of blue LEDs versus red LEDs. By that I mean that blue LEDs might tend to have higher emitted power power per input watt, so when you convert that energy to white light via a fluorescent material, even though you lose a little bit in the process, you end up with more white light per input power than if you had started out with a combination of different color LEDs.

 

[mahoney]

CRI of an RGB array would be poor, you only have 3 wavelengths to work with. The higher CRI white LEDs are using phosphor blends that emit over a broad range of wavelengths.

If colored light is desired, then RGB/RGBA arrays would be superior and more efficient than white LEDs and a filter, but a good quality white LED will be superior if white light is desired.

 

[2xTrinity]

Something I've been wondering:

In principle, for the purpose of making large quantities of white light, would RGB (or RGBY) arrays be cheaper and more efficient than "white" LEDs?

RGB arrays are usually wired so that each color group can be driven individually, since they are designed for variable colour.
If the goal were to make lots of white light, then addressability is not required.

Not true. Mixing red green and blue LEDs and getting an output that looks "white" is actually exceptionally difficult to pull off. Especially for an array bright enough to be used as an actual light source (not just using RGB LEDs as "pixels" on say a jumbotron screen). The different colors all have different responses to temperature. That is, if you were to just supply some set amount of current to red green and blue LEDs in an array, as the array were to heat up, the color would shift away from white and appear strongly tinted in a very short period of time, without some sort of temperature feedback that compensated by continuously adjusted the current to all the LEDs to maintain the balance.

In a blue + phosphor system, if the LED get less efficient due to heating, you just get dimming but no change in color, like you'd get if your different color LEDs respond to heat differently.

Since "white" LEDs work depend on fluorescence, which has inherent inefficiencies, I am wondering if color mixing RGBY might produce white more efficiently.
In real life, "white" LEDs in production seem to have same or better efficacy than coloured ones, (of similar output / module), is this due to whites being in highest demand?

Blue LEDs are currently the most efficient by far. The weakest link for RGB is the fact that green LEDs are not very efficient at all. The best are around 5-10% ( power out/ power in). The best blue are over 50% efficient. So generating green or yellow-green light using a blue LED with a phosphor coating is actually a more efficient way to do it.

Much of th reason why blue is so efficient is because there has been a lot of research money invested in blue LEDs, as they are the main component used to make white LEDs. However, the poor efficiency of green LEDs (The weak link in making an efficient RGB array) is a physics problem.

Both green LEDs and blue LEDs are made using InGaN semiconductor material. The color of an LED is determined by the bandgap energy, which is determined by the ratio of the dopant elements. Most semiconductor will have an optimal concentration of dopant -- too little and there will not be enough free electron-hole pairs for the device to work. Too much and your crystal lattice will have a lot more defects, and you will get higher resistance from the high concentrated impurities creating a "traffic jam" of sorts in your junction.

For the InGaN material, blue corresponds to a moderate (optimal) doping concentration. Both UV (shorter wavelength) and green (longer) LEDs are possible, but both are substantially less efficient.

Similarly, Red LEDs (based on the AlInGaP material) are also quite efficient, but yellow and yellow-green LEDs based on similar material are considerably less efficient (note again I'm still talking about power out/power in -- NOT lumens/watt)

Without a highly efficient green LED, RGB LEDs won't be able to even match, let alone exceed, the efficiency of a good phosphor white LED.

 

[kaon]

Very good explanations.
Thanks!

Quick OT question,
What (if anything) would best replace an MR16 50W IRC halogen? (retail lighting)

Last I looked, LED cannot match in terms of lumens per $, or lumens per MR16, and the lumens/W is about the same as CFL.
Any new developments in white LED for general lighting I should know about?

Do ceramic metal halide come in MR16 form factor yet?

 

[brickbat]

...Do ceramic metal halide come in MR16 form factor yet?

Yes.

http://www.gelighting.com/na/busine...ll_sheets/downloads/hid/cmhmr16_datasheet.pdf

Osram/Sylvania has a similar offering.

 

[blasterman]

Mixing red green and blue LEDs and getting an output that looks "white" is actually exceptionally difficult to pull off.

I have some fairly high end RGB LED PARSs, and it's impossible to dial a neutral white manually.

Note that cree *does* use two different color LEDs in their higher end ceiling fixtures to get the correct color balance. I believe the balance between the different LEDs is adjusted dynamically/electronically.

 

[cooyard]

What i know and done before, white LED using phosphor is cheaper than RGB(or RGGB). I used blue and green from CREE and red from Taiwan maker(forgot name). Need 1 blue, 2 green and 1 red die to get WHITE.
CRI better but lumen low.

 

[JoakimFlorence]

In principle, for the purpose of making large quantities of white light, would RGB (or RGBY) arrays be cheaper and more efficient than "white" LEDs?

The answer is no. The problem is that green LEDs are simply not that efficient, especially once you get into the yellow-green wavelength territory. And if only using three discrete wavelengths, then it is very important that the green be as close to 542nm (yellowish-green) as possible for maximum color rendering.

However, greenish-white LEDs that combine with a red LED to contribute to the spectrum are theoretically more efficient.
(There are a few complications caused by this design, which is why it is not in wider use)

see thread:
Theoretical limit for the efficacy of red, green, blue and white LEDs

 

[JustAnOldFashionedLEDGuy]

13 Year old thread. Are you trying to show off?