Guest post by Jim Brodrick, reprinted with permission from Postings: from the desk of Jim Brodrick
If you’ve ever shopped for white paint, you know there’s not just one universal white, but hundreds of different shades. White paint can be bold and bright, have hints of yellow or blue, or have subtleties that are hard to describe. The same thing holds true for white light, and defining its different shades as well as its ability to accurately render the colors of illuminated objects — collectively referred to as its color quality — is an important aspect of general illumination and something the SSL industry is working hard to refine. Why? Because the measures we’ve been using to define the color of conventional lighting simply aren’t adequate for SSL.
The two most frequently used metrics to quantify color for lighting are Correlated Color Temperature (CCT) and Color Rendering Index (CRI). Both have been in use for years, and it’s important to understand the difference between them. I think of it as the difference between looking at the light and looking at the things that are lit.
CCT describes the hue of the “white” light itself, which might be perceived as bluish to yellowish to reddish, or “cold” to “warm.” It’s represented by degrees Kelvin, with warm yellowish light in the 2500-3500K range, and cold bluish light typically at 5000K or more. The mid-range is often referred to as “neutral” white. For reference, all light sources can be described using this numbering system. For example, a match flame is about 1700K, an incandescent “warm white” bulb is usually 2700-3300K, and daylight on an overcast day is about 6500K.
CRI, on the other hand, was created to describe how well a light source renders colors, on a scale of 0 to 100. For example, how does that red carpet look in candlelight, compared with how it looks under an incandescent bulb or in daylight? When you bring a paint chip sample or fabric swatch home to see how it will really look in that room, what color you actually perceive is much influenced by the CRI of your lighting.
While both CRI and CCT are useful descriptors, we’ve found that with SSL they’re not descriptive enough. At DOE’s SSL Market Introduction Workshop in Chicago this past June, Ron Steen of Xicato demonstrated one part of the problem by showing us three pairs of lights on a screen, all of which were labeled as 3000K CCT, which is in the warm white range. The first pair had CCTs of 3004K and 3025K, which might seem like they’re close enough to one another. But when viewed side-by-side, the colors could be seen to differ significantly. He then showed another pair of lights that were even closer to one another in actual CCT, but once again, if they were placed next to each other facing the same wall, I doubt a customer would be happy. Finally, Ron showed us a pair of lights with CCTs of 3013K and 3018K, which provided what most of us in the room considere
Ron’s point was this: What happens when a customer places a product with a CCT of 3005K +/- 5 beside a product with a CCT of 3020K +/- 5? Chances are they would look substantially different in color, despite both boxes saying “3000K.” I have to agree with Ron’s advice on how to avoid dissatisfied customers: Consider the color sensitivity of the application. But a smaller CCT range by itself doesn’t guarantee matched color rendering. You can have two light sources with identical CCTs that give very different color rendering of a white wall because they have different colors that together make up the white light.
CRI, which was developed by the International Commission on Illumination many years ago, is also useful, but it too lacks fine-tuning. CRI’s inherent problems describing solid-state light sources stem from the fact that it depends on comparison to a somewhat arbitrary and limited library of specific standard colors. That selection of colors does an adequate job with fluorescent lighting, but it misses the mark with LEDs because the LED spectrum is quite different from the fluorescent spectrum.
To address this issue and to meet the new needs for communicating color quality of lighting products, the National Institute of Standards and Technology (NIST) is developing a Color Quality Scale (CQS) that is calculated using a different method from CRI and when completed will be proposed as a new international standard.
Finally, there’s the important question of color shift, or what happens to a luminaire’s color over time. Because so many LED products — even good ones — have shown poor color stability over time, it’s important to take it into consideration and figure out how it might play into product failure. The CALiPER testing program is looking at trends in color shifts, and the issue is being discussed in the context of product lifetime considerations by a special working group on SSL reliability and lifetime that DOE has set up. Composed of a wide range of experts in reliability, lighting, and LED technology, the group is under the guidance of the Quality Advocates oversight committee — a joint body of DOE and the Next Generation Lighting Industry
Even though an SSL luminaire’s lifetime is often described only in terms of having adequate light output, it’s clear there are some applications where excessive color shift would make the luminaire unusable and thus render a shorter useful life. For that reason, the reliability and lifetime working group is taking a careful look at color shift and is including it in the new set of recommendations.
SSL is a more advanced technology that offers us a whole new world of lighting, but it’s still evolving. As that process continues at a rapid pace, technological improvements will make LED products increasingly better light sources. We’re always looking for ways to make SSL better and to use it smarter, and refining the way in which we define its color will help us do that.
