Recent Patents: Lumped Color

This patent describes a simple way to achieve more control over color temperature and CRI without the expense of managing all constituent colors individually.

"Describing Two LED Colors as a Single, Lumped LED Color" seems like it ought to be the description of a research paper -- maybe even a high school paper, since one can find a great deal of information on the web on the subject of describing colors. But in reality, it is the title of a short patent (only eight pages long) recently issued to Peter Deurenberg, Johannes Ansems, and Christoph Hoelen, and assigned to Philips, US #8410723. We will look at: A) What is the problem the patent is addressing, B) How the invention works, and C) Where we might expect to see this patent employed in LED lighting.

The problem

Advanced lighting is usually done with RGB LEDs (red, green, and blue). But to control the color output accurately takes some doing. The patent explains previous art: "...light output and the color of the LEDs are controlled by measuring color coordinates for each LED... for different temperatures," storing all this information in a processor, "...calculating the color coordinates and lumen output fractions on-line," and then sending the right amount of current to each LED. And the problem with this is that "the demand on calculation power will increase cost," and adding more colors to improve the CRI makes the calculation burden even heavier.

The method

The inventors want to drive LEDs of different colors with the same current, so they can be in series, but still get the right color and light output, with decent CRI. How can that be done? Their plan is to switch from three colors (R, G, and B) to four, such as R, G, B, and amber. But there are still only three strings: One string is green, one string is blue, and the third string is red and amber LEDs together. The red and amber LEDs are driven with the same current.

And how does it work? "Earlier tests... showed that the differences in temperature behavior are too significant to just lump the red and amber LEDs in a single degree of freedom...." But if you combine the characteristics of the red and amber LEDs into a single, new "virtual" LED, then you suddenly get "very good [CRI] and easy driving."

As an example, they show two red LEDs (wavelength = 617nm) and six amber LEDs (wavelength = 593nm), all driven at 350mA; if they now call this a single virtual LED, it has light output, color, and temperature dependence intermediate between the original component LEDs, and so can be controlled as if it was a single device with wavelength = 598nm. But you actually have a wider volume of parameter space to range over, because the LEDs together have a broader range of wavelengths than the individual LEDs do alone. At first glance, you think "oh, that's obvious, how can that be patented?" But when you think about it a little more, it becomes less obvious. And so this is actually a pretty interesting little patent.

The likely application

Turning to the actual claims, there are only eight, with two independent. Both independent claims cite four colors of LEDs, with two of the colors being lumped together. They both also require a driving circuit for the LEDs, with inputs to determine what color is wanted, what the temperatures of the LEDs are, and a model to determine how to drive the LEDs such that they achieve the desired color. The second independent claim differs from the first only in specifying the wavelength range of the two LEDs being driven together, indicating that it's basically red and amber that are being combined in actual practice.

I observe that the claims specifically call out "producing white light." Maybe this system doesn't work so well for other colors? In any case, I conclude that Philips intends to use this patent (if at all) in things such as down lights, where they want to get a desired CCT and CRI with good accuracy, improved by using four colors instead of three, but without the cost of controlling all four individually. It won't be for situations requiring the best possible color, such as theatre -- for that, you still want individually to control all four colors, or even more.

As a final note, I observe that in some ways this patent is rather limited in scope. This suggests that, as often, there might be ways to design around it -- or to get an additional patent. If there's any interest, I'd like to hear from you.