Color changing, and even fixed color RGB sources, are very understandable in what they do and why they work well. If you can generate a 'native' light color, instead of filtering a wider spectrum source, you'd expect to get a lot more efficient solution. Present a source with a full or nearly full spectrum (think halogen or metal-halide) and then throw a blue colored lens in front of it, and I've been told that you're effectively disposing of 90% of the lumens that you started with. Feed a current to a blue LED, and you get 100% blue photons from start to finish, and at efficiencies that are as good or better than the full spectrum source was for generating all its light. It's a no-brainer to understand why Hollywood, theater and entertainment lighting leaped onto LED-based sources from early on. We all know red-green-blue generates what our eyes perceive as "white" since true white is simply a real healthy mix of the full spectrum. The funny thing is that an RGB solution is still a little "peaky", and while our eyes appreciate the rich color it returns, the instruments do not. That shows up as a penalty when you compare RGB-generated white-lumens to incandescent, or in the case of fluorescents, phosphor-generated lumens. This isn't about the "CRI" thing (an important ongoing discussion on its own), but about the "white energy" in my layman's terms.

A good example of this comes to mind in the projector applications. About a year and half ago, we reported on our subjective experience with a Samsung pocket-projector powered by one of Luminus Devices PhlatLight LEDs. In case you're not familiar with it, the Luminus chips are around half a business card sized monsters that contain big red, green and blue LED die that are about a quarter of an inch square. They scale up and down from there, and also have phosphor converted white solutions as well, but the RGB family made it's mark in the DLP television wave as the source that provided much richer colors, and a 50,000 to 100,000 hour life, instead of the optimistic 5000 that your standard metal-halide through a color wheel solution did. What my eyes saw from this 200-ish lumen RGB LED projector was overall brightness perception that came a good way towards matching our 2000 lumen conference projector. When it came to color quality, there was no contest at all. The movie on the LED projector didn't have that washed out color look, and it just 'felt' better. Mark McClear of Cree, in a talk at last summer's DOE meet in Chicago, posed the question, "Why can't the standards acknowledge what we see with our eyes?" Namely that LED light can provide a higher quality that currently isn't reflected in the numbers.

There are other interesting "tweens" that we're having to come to grips with now. Most recently, we've seen several new Edison-based A-lamp designs hit the market. When we think A-lamp, we picture our very familiar 60, 75 or 100w incandescents, with the visible addition of the heat sink there between the base and 'globe' in the designs of most LED challengers. With the virtual completion of the DOE-generated "Integral LED Lamp" Energy Star specification, there is finally a reference point on what a "replacement" for a number of standard incandescent Edison-based bulbs should do. The spec is pretty comprehensive, and places the emphasis in the right places. For PAR/R replacements, generally recognized as the easiest 'replacement bulb' challenge for LEDs to tackle, the standards are about smoothness in the distribution, the width of the beam angle, and the brightness (center beam candle power) on the target. The MR specs follow the same approach, and since you can measure those characteristics for the 'average' incandescent solution, the bar was set to meet the distribution and output, and do it at X number of lumens per watt or better. PAR/R/MR lamps need to beat 40-45 lm/watt (the lower number for the smaller lamps), decorative/candelabra base need to beat 40 lm/watt, and A-lamps need to beat 50 lm/watt for less 10w of LED power, or 55 lm/watt for those greater than 10w. (You can see PDF slide copies of the presentation that Marc Ledbetter of the DOE's Pacific Northwest National Laboratory gave at the January 2010 LA SSL Summit here).

Then there's the tweens. They have the same efficacy requirements as the A-lamps, but drop the requirement for a particular distribution which allows things like A-lamp shaped directional lamps, as a very specific 'for instance'. Why would you want one of those? For one big reason, to replace standard incandescent A-lamps, and CFLs in the zillions of pendants and cans that they have found themselves in. Would R's work as well? Seems like they would, but for whatever reason, whether for looks or cost, or because the fixture had an attractive way to leak some of the light out in other directions, omnidirectional lamps are in there, and mostly being asked to send light in one direction. What an ideal fit for LEDs, since they really do like to send light out directionally, and those sockets are being served by 10-40lm/watt omnidirectional solutions right now (I'm guessing the light loss is likely on the order of 25-50%, so consider the range to be 5-30lm/watt out of the fixture). Here's the part that's not easy when being tween... describing it.