It has long been common knowledge that no artificial light source can be properly assessed for its color performance simply by its color temperature alone. Depending on the spectral reflectance distribution (SRD) of the object or surface being illuminated, two lamps of the same color temperature may have vastly different effects on that object or surface. The challenge then is to develop a way of quantifying a lamp's ability to reproduce faithfully the colors inherent in any object.
As far back as 1931, the Commission Internationale de l'Eclairage (CIE)—an international authority on light, illumination, color, and color spaces—tried to strictly define color in scientific terms. It developed the chromaticity diagram, which became known as the universal coordinate system (UCS)—still in use today. The UCS is a means of plotting any pigment color by using the chromaticity diagram's X/Y coordinates as a function of that color's wavelength signature, something that has come to be known as spectral power distribution (SPD).
The CIE did not, however, solve the problem of color rendering. At the time, the issue was a small one. Remember, this was 1931; the lighting industry was dominated by the incandescent lamp, and color rendering had not yet been identified as an issue. The first lamp source other than incandescent, the mercury vapor lamp, was not available commercially until 1933. Fluorescent and sodium discharge lamps wouldn't follow until later that decade, and metal halide wouldn't come along until the late 1950s.
From the 1930s to the early 1960s, lamp quality improved significantly, especially in products that featured fluorescents, which, because of their lumen efficiency and reduced energy needs, became the lamp of choice for most commercial interior applications. The retail market was particularly keen to use the most energy-efficient means to light its products in an attractive manner.
Because of these new competing technologies, it became apparent to lighting designers that they needed a way to correctly quantify lamps in terms of color performance. Could every lamp be measured in a way that would be useful to designers of lighting systems? To answer that question, the color rendering index (CRI) was developed in 1964. The CIE came up with the idea of comparing each lamp to either an ideal or a natural source, depending on the color temperature of the lamp. When averaged, the shifts in the plotted points on the UCS would determine the color rendering number for that lamp.
To do this, the CIE chose 14 representative pigment color samples. The first set of eight are pastels, relatively low in chromatic saturation, evenly distributed across the range of hues, and used to calculate the general CRI number. The other six color samples are used to provide supplementary information about the color rendering properties of the light source being tested. Of those six, four are highly saturated solids, and the other two are meant to represent a generalized Caucasian skin tone and the color of typical plant foliage.
Each lamp source being tested is matched to the output of a black body radiator (the standardized reference light source that can serve as a basis for lamp rating) of the same correlated color temperature (CCT), as long as that CCT is 5000K or lower. Above 5000K, the reference source is one in a series of daylight spectral energy distributions. Once plotted, the lamp is assigned a number from one to 100 on the CRI. Halogen incandescent is the only lamp source to achieve a rating of 100 since it is essentially a black body radiator. A rating of 80 or above is generally acceptable for commercial interior applications.
But it is now being argued, particularly by the LED community, that a high CRI does not necessarily imply good rendering. This is because the lamp being tested may have an imbalanced SPD, such as fluorescents and “white” LEDs, or it could have an extreme CCT (below 1360K or above 5000K). LED manufacturers also feel that for some lamps the eight basic color samples used to determine a general CRI number are not enough to truly represent the optimal emission spectra. Some lamp sources are capable of accurately rendering the eight colors of low saturation, yet perform poorly against the four highly saturated colors. LEDs are being touted as an example of how unfair the current system can be since they tend to perform much better when the four saturated colors are included in the test.
For that reason, manufacturers of LEDs and luminaires incorporating LEDs have been calling for a way to refine how color quality is defined. Since 2006, the National Institute of Standards and Technology (NIST) has been developing a new metric, the color quality scale (CQS), that determines color performance using a method different from the CRI. When completed, the NIST will propose it as the new international standard. A different color space is used, and a new set of 15 reflective color samples, highly saturated and taken from the Munsell color system, replaces the 14 CRI samples and defines the difference between the test lamp and its reference. The NIST claims this should overcome hue and saturation shifts left out of the CRI calculation. This penalizes lamps of extreme CCTs, which frequently exhibit poor color quality. In the end, what will be familiar to users of the CRI is the CQS span, which will range from zero to 100.
Clearly, much study, discussion, and testing will need to be conducted before this new methodology is embraced by the lighting community. The criticisms leveled at the CRI system certainly are valid ones: it's old and was developed before many of the more sophisticated lamp sources came on the market. But for many in the lighting community, the CRI is still a useful tool, easy to understand and utilize. So will the CQS go too far and become a system that favors itself in its attempt to level the playing field for LED lamps? Interesting months lie ahead while we grapple with this issue. Let the debates begin!
Jeff Robbins is a commercial lighting specialist at the Lighting Design Lab in Seattle. He is lighting certified (LC), and serves as co-chairman of the NCQLP testing subcommittee. He also is the education chairman for the western district of the IES.
GLOSSARY Spectral power distribution: The output of a light source, characterized by its relative strength at each wavelength.
Spectral reflectance distribution: The response of an object to light, characterized by the wavelengths that it primarily reflects.
CIE chromaticity diagram: The visual spectrum bent to form a more or less triangular plane on which any color may be assigned an X/Y value. It provides a means by which color differences between sources or objects can be qualified, or by which color shifts of a single object when illuminated by different light sources can be qualified.
Correlated color temperature: The absolute temperature of a black body whose chromaticity most nearly resembles that of the light source being measured.
Black body radiator: A temperature radiator of uniform temperature whose radiant exitance (total light leaving an area) in all parts of the spectrum is the maximum obtainable from another temperature radiator at the same temperature.
Munsell color system: A system of surface-color specification based on perceptually uniform color scales of the three color variables: hue, value, and chroma.
Color rendering index: A measure of the degree of color shift that objects undergo when illuminated by the light source as compared with those same objects when [ illuminated by a reference source of comparable color temperature.