I have completed the RGB Light experiment, and decided it is inferior to the white LEDs in terms of display quality. My digital camera cannot get good enough pictures of the display to illustrate what is going on, so a text description will have to do. I'll still send you some cool pictures.

Attached is a picture of my setup. Since Radio Shack only had 2 potentiometers, I ran the blue LED through a fixed resistor. I adjusted its brightness by adjusting the voltage, and then adjusted red and green LEDs with the two 10-turn 1K potentiometers shown. I was able to adjust the current in increments of less than a milliamp, which worked very well.

If you look closely in the picture, you can see that there are 4 wires going into the Pelican light instead of two. The 3 LEDs in the RGB LED share a common annode (+ side).

First, the good news. As I mentioned before, the light is very uniform in both brightness and color. Also, I could easily create any color of light I wanted, from warm-white to cool-white, to any color of the rainbow. Also, this LED is BRIGHT. Also, the GBA display has incredible contrast when illuminated with the RGB LED. Lastly, reds are much, much more saturated than with any other light source. They are truly a deep, deep red, instead of that orangish color.

Now, the bad news, and the reason I've concluded that the RGB LED is inferior to white LEDs for illuminating a GBA.

• Many colors are darker
• Yellow is less saturated
• While some colors are better with the RGB LED, overall, the colors do not look as good as with the white LEDs.

I have several possible theories as to *why* it doesn't look as good.

1. In the low light levels (compared to bright sunlight) where the Pelican light is used, human vision includes both rod vision and cone vision.The RGB LED does a good job of covering the cone's responses, but does not have spectral peaks that line up with the rods. The rods have a peak at a different wavelength than the human eye's red, green, and blue cones. There are recent experiments that imply that at these kinds of light levels, the rods also contribute to color vision, and I'm convinced that my experiment is right in this zone. The white LED has a continuous spectrum, rather than 3 peaks like the RGB LED. While the white LED's spectrum is not flat, it still emits significant energy throughout the visible spectrum. It's a very strange sensation looking at a piece of white paper illuminated by the RGB LED. I can adjust the red, green, and blue to produce white, but it still doesn't look right. It actually looks like colored white. You almost have to see it to see what I'm talking about. By colored, I don't mean that it seems to have any sort of hue, like blue, or cyan, or yellow. What I mean is that it looks like colored white. I did numerous experiments with pairs of colors, and also noticed an anomaly when mixing red and green light to get yellow. I couldn't get a yellow that really looked right. It's hard to describe what was wrong. I could continuously adjust the red and green to get any hue I wanted, but for some reason, it looked like red and green mixed together, not yellow! I took out a number of diffusers (some of them very thick milky plastic) to make sure there wasn't some variation in the light pattern, but the diffuser had no effect (other than to dim the light).

2. While the light from the RGB LED matches the red, green, and blue cones in the human eye, it may not be well matched to the dyes used in the GBA display. This may explain why some GBA colors get very dark and saturated. It's neat that they're saturated, but they become too dim to see. I played games for about 10 minutes, then switched lights, played 10 minutes, then switched back, to get used to each light, and get a good feel for it's strengths and weaknesses. When all was said and done, I clearly like the white LED better.

I'll continue playing with it for a few days just for fun before moving on to the next experiment.

(As always, feel free to post to stimulate discussion)

-Andy

---

Pictures and explanations
The Nichia RGB LED comes in a small, white rectangular package with two contacts on each side. The package is 3 mm on a side. When the LEDs are turned on, they appear to come from almost a single point, even though there are 3 tiny LEDs inside.

The original Pelican reflector was sanded down to remove the back side. Next, and exacto knife was used to scrape away the conductive reflective area around the part of the reflector that would contact the PC board and LED.

A PC Board was made to mount the LED for several reasons. First, it made it easier to attach the wires. Second, the board mounts flush with the bottom of the modified reflector, pointing the LED towards the GBA display. Third, I originally intended to mount resistors on this board to balance the 3 colors so that I would only have to run 2 wires from the light control unit, at the bottom of the lid. After several experiments with resistors, it became obvious that I needed to have a more robust test jig. I globbed solder on to the PC board to short out the places where the surface mount resistors used to go, and attached 4 wires - one for +, and one each for Red, Green, Blue.

To get wires that would be flexible and still fit in the Pelican Light Shield's wire channel in the lid, I took an old GBA 4-player game cable that I had already dissassembled, and carefully split it open and removed the wires. Four of these easily fit in the wire channel. To make assembly easier, I glued the 4 wires together at a few points with superglue. The wires don't naturally stay as tightly grouped as they are in this picture! :-)

-Andy