Luxeon Star colour mixing

[Adrian]

So I fire-up a red, a green and a royal blue at .35A each and get a real heavy magenta/pink colour, no where near white:



(foget the spraycan lid that's mixing the light from the stars underneath - it's way over-exposed. Look at the reflection off the left side of the other cap in the photo. The right hand side of that one is lit by a 50W Halogen desk lamp for comparison)

So keeping green at .35A I back off red and blue until I get a reasonable white (compared against tungsten):



The shock is that now I've only got 0.1A in the red and 0.16A in the blue! Hardly worth going to the expense of the Stars for. A couple of superflux's would do.

You could say it's the green that's really lame compared to red and blue. Anyone at all unsuprised by this?

[greg_in_canada]

The formula for luminance in NTSC televsion is
Y Luminance = (Red x 0.30) + (Green x 0.59) + (Blue x 0.11)
(from this web page). That is based on the phosphors used
in TVs but your LEDs seem to be similar.

I guess it also relates to the relative efficiency of
the LEDs versus the TV phosphors.

Greg

[jtr1962]

Typical lumens for green and red are 30 and 44, respectively (in both cases I'm assuming high dome emitters). Lumens aren't given for the royal blue but let's say about 8. It's patently obvious that driving everything at the same current will result in a light that's far too red (i.e. magenta/pink). Backing off the red should have been enough to produce a reasonable white. You only needed to back off the royal blue because you were using a halogen lamp as your white point. What you actually are producing now is a yellowish-white light. Try increasing the royal blue back to 0.35 amp and see if what you get is closer to a regular white luxeon. I imagine it will be. In fact, you can probably vary color temperature across the board merely by playing around with the red/blue ratio.

BTW, I was able to produce "white" by combining ocean blue (~485 nm) and amber LEDs. What resulted appeared white to the eye, but of course had very poor color rendering. You get much better color rendering by using three colors. In fact, your color rendering is probably superior to a regular white luxeon.

I tend to think that once the kinks are worked out we'll be combining colors to make white for general lighting rather than using the other approaches (blue + YAG phosphor or UV + RGB phosphor). Provided you have a sensor to compensate for the variances in aging of each color, the color mixing approach is far superior because you can create any color temperature as well as any almost any color using one lamp. No need to stock multiple color temperatures as is done now with CFLs.

[Adrian]

Well to start with, what made me expect the intensities to be balanced was the Luxeon DCC backlight:



When I first saw that photo I incorrectly counted 18 devices and assumed 6xRGB (there are only 17!) After I posted this I did some more investigating and found the DCC data sheet. In it, it shows the weightings used in the different lengths available:



So that answers it... never assume [img]/ubbthreads/images/graemlins/grin.gif[/img]

[HarryN]

Hi Adrian - The newer greens are starting to come out in higher output to deal with your exact observation.

I am guessing that you are already up on the CIE curve, and I have seen some CIE curve calculators that let you plug in the wavelenght / lumen output ratio and they will chart where you end up.

An interesting one I use for some of my lights is a combo of red / orange (nominal 617nm) mixed with a blue / cyan (490nm) Done careully, it passes through white with only 2 colors. CRI is kind of low though.

[Adrian]

Thanks Harry, I'll keep an eye out for the higher o/p greens. At the moment I'm using a 3 Watt green suitably throttled to balance with the 1W red.