Comparing Metal Halide, Fluorescent, and Solid State Technologies Jim Dilbeck, May, 2012 Scope This study compares the color rendition characteristics of the three most common commercial lighting technologies; HID (high pressure sodium and metal halide), fluorescent, and solid state. This document addresses the color rendition of the white light used in commercial ambient lighting and the perception of color relative to the industry s only color measurement scale. This document is a cursory overview. The subject of color rendition is addressed conceptually here as it relates to the present measurement techniques and human eye perception. The science is extensive and addressed here in fundamentals only; much of this subject is beyond the scope of this document. Objective Providing a comparison between measured and perceived color, the consumer of commercial lighting will better understand how their spaces will appear, specifically how colors will look under each of the three commercial technologies considered herein. We intend to show that significantly too much emphasis is put on the Color Rendering Index (CRI) score. Few understand how the CRI works or how a light source with a CRI of 25 can be perceived as having better color rendition than a CRI of 80. This fact is the primary objective of this paper. Providing an understanding of how CRI is derived should simultaneously provide an understanding of where the scale applies, where it is misleading and, more specifically, why it should not be used to compare lighting sources across technologies. Overview Every human eye has a varying degree of color sensitivity; some of us see color extremely well and others not so. As we age, our ability to discern color decreases. As there is no standard human eye, the truth is that color rendition is subjective. The following details how color is rendered by the output of a light source being reflected off of an object to the eye. Supporting this is unbiased data comparing the color emissions of the three most common commercial ambient light sources; metal halide, fluorescent, and solid state. There is a lot of controversy regarding how color should be measured and how the results should be portrayed. This is because the present method, the Color Rendering Index, though suitable for the comparison of HID and fluorescent, is inadequate in the measurement of solid state lighting. We ll also touch on the pending change to the more recently developed Color Quality Scale (CQS). May, 2012 Page 1
Finally, we ll evaluate the three common commercial lighting technologies. In this, we ll go beyond the scales to provide an understanding of how the color rendering of each is not just measured but perceived, and how the color rendition of each technology changes over time. The Human Eye, The Ultimate Measuring Device The human eye consists of rods and cones. The rods provide our scotopic vision, which is primarily our night vision, useful in low light conditions but virtually non-existent in our perception of color. Conversely, the cones provide our photopic vision, which is useful in brighter light and provides the vast majority of our color perception. It is estimated that our photopic vision can differentiate about ten million colors (1). Color and White Light One unit of light measurement is wavelength. The visible spectrum of light is in the wavelengths roughly between 400 and 700 nanometers (nm). The shorter, higher frequency wavelengths are at the ultraviolet (UV) side of the visible spectrum while the longer, lower frequency wavelengths lean toward the infrared side. Color Violet Blue Green Yellow Orange Red Wavelength(nm) 400-450 450-490 490-560 560-590 590-635 635-700 Figure 1. Visible Spectrum and Wavelengths Pure white light is our basis for perfect color rendition. It is the presence, in equal intensity, of all wavelengths of visible light. Color Perception The core purpose of this document is to convey that the human eye can only perceive the specific color of an object when the wavelength of that color is emitted by the light source. When a wavelength is not emitted, the related color is not seen. Example: A ball that is red and green sits in the late morning sun. Both the red and the green are vibrant because the sun is delivering light across both the red and green wavelengths. Now, we take the ball indoors, into a light source that is weak in the spectral band between 520 and 540nm (see Figure 1). Since the green wavelength is missing from the light source, the green colors of the ball are now missing as well; the ball now appears to be red and something more brown or grey. Color is also subject to the intensity of a wavelength relative to the intensity of the other wavelengths present. This is called context. In context, a green will be perceived as being more green when it s adjacent red as opposed to a blue. It is only possible to adequately perceive a spectral band in context. May, 2012 Page 2
Spectral Distribution Spectral distribution charts the intensity of each wavelength relative to the highest intensity wavelength of the given light source. Some white lights have more red, others may emit more blue. Some have wide gaps in their distribution while others cover the visible wavelengths more consistently. METAL HALIDE FLUORESCENT SOLID STATE Figure 2. IES 2012 Design Guide for Choosing Light Sources for General Lighting (2) Figure 2 is an excerpt published by the Illuminating Engineering Society (IES), the unbiased governing body for standards in the lighting industry. This shows the spectral distributions of metal halide, fluorescent, and solid state, respectively. In reading spectral distributions, the intensity in each wavelength is relative to the maximum intensity of the highest single wavelength. A perfect spectral distribution would be flat, balancing the intensities across all wavelengths. As you can see, metal halide and fluorescent have a very spiked distribution; they are distributed in very narrow spectral bands and very weak in between. Conversely, the nature of the SSL delivers a more broad spectral distribution. Figure 3 shows an overlay of the IES spectral distribution charts from above, comparing fluorescent and SSL. The red illustrates the additional spectral bands provided by the SSL while the blue shows the additional provided by the fluorescent technology. Because the wavelengths are more evenly distributed, the SSL has significantly better color rendition. Figure 3. Spectral Band Comparison; Fluorescent to SSL If this is the case, how is it that these two sources can exhibit similar CRIs? May, 2012 Page 3
Color Measurement The two scales reviewed here are the Color Rendering Index (CRI) and the Color Quality Scale (CQS). The CRI was created by the International Commission on Illumination (CIE) in the 1960 s, prior to the advent of solid state, primarily as a device to compare HID and fluorescent lighting. The CRI value is a result of comparing the color rendering of a test source illuminating these colors to that of a perfect source at the same lighting level. It utilizes eight low saturation color samples each representing a spectral band of visible light and distributed over a range of hues. Figure 4. Colors of the Color Rendering Index As a side note, CRI is also subject to correlated color temperature (CCT). For warmer color temperatures (less than 5000K), the perfect source is a black body radiator (similar to an incandescent bulb). For the cooler temperatures above 5000K, the perfect source is a specifically defined spectrum of daylight. There are numerous drawbacks to the CRI, all of which make this scale extremely misleading. First, the CRI was created in the day of HID and fluorescent. In the forty years since the inception of the CRI, the competitive nature of the lighting industry has driven endless tuning of the phosphors used in these light sources for the sole purpose of scoring well on the CRI. You can see this in action in the charts below, where the spikes align with the specific CRI spectral bands. As evidenced by the spectral distributions of these sources, the CRI is a significant misrepresentation of color rendition. Figure 5. Laying CRI Over Fluorescent and SSL May, 2012 Page 4
Second, numerous marketing strategies exist to exploit this scale. Example: a 4100K CCT light source is measured against a 2700K perfect source (acceptable under CRI measurement criteria). The warmer 2700K perfect source, even though it has a CRI of 100, is weaker in intensity on the blue end of the visible spectrum while the 4100K test source is emitting more intensity there. This renders a higher CRI than if it were compared to the 5000K daylight spectrum despite the fact it is the closer perfect source. Color Quality Scale (CQS) The CQS is a more modern approach to color measurement. The National Institute of Standards and Technology (NIST), charged with the development of the CQS, states that the colors referenced by the CRI do not adequately span the range of normal object colors (4). Figure 6. Colors of the Color Quality Scale The samples of the CQS span the entire visible spectrum in evenly spaced wavelengths and provide for color saturation not addressed by the CRI. While the CIE has tested and concluded that the CQS is a more effective scale for the measurement of color rendition, it has yet to be implemented. Whether this is due to more development being required or the influence of incumbent lighting manufacturers protecting decades of work in tuning their phosphors is a subject beyond the scope of this document. Regardless, while the CQS continues in the basic vein of the CRI, it still does not encompass the roughly ten million colors the human eye discerns. Figure 5. Laying CQS Over Fluorescent and SSL Presently, the CRI is the only reported standard. While the two light sources above are similar in CRI, it s easy to see how this index misrepresents the truth in color rendition. When crossing technologies, the solution to the inadequacy of the CRI is personal interpretation of the spectral distributions; this becomes easy when comparing the obvious differences between fluorescent and SSL sources. May, 2012 Page 5
Color Shift The CRI and CQS are both initial measurements of color rendition. Color shift is a term that describes the change in the color rendition of a light source over time. This element is sorely missing from the color requirements and it is left to the consumer to bear in mind. The HID technology is long-known for its notorious shift in color and very few consumers remain unaware of this. (See Figure 6). For decades, the fluorescent technology has been superior to HID in color shift. Yet, while pink and green hues manifest as the fluorescent bulb matures, data pertaining to how much color shift is exhibited is up to the manufacturer to provide. As it is not in their best interest, the data is virtually non-existent. Figure 6. Example of Color Shift. In the solid state technology, color shift is measured as a part of IES LM-80 testing, performed by all of the reputable LED manufacturers (Figure 7). Figure 7. Color Shift Data Required in IES LM-80 May, 2012 Page 6
Summary The broadband phosphors used in the SSL emit much of their energy in the wavelengths that do not necessarily score well on the CRI while clearly outperforming in perceived color rendering. As it s necessary for a light source to emit a given wavelength for us to see that color (on an illuminated object), the outdated CRI scale inherently misrepresents true color rendition. It disfavors the SSL by not considering the multitude of wavelengths that fill the gaps (ref. figure 3) and bring us more of the ten million colors we can truly discern. For now, there is no scale for color rendition comparison across lighting technologies. Understanding this makes it clear that the optimum means of evaluating color rendition across technologies is in comparing the spectral distributions or, better yet, in side-by-side comparison of the technologies. May, 2012 Page 7
Addendum; US DOE Report, Color Rendering Index and LEDs May, 2012 Page 8
May, 2012 Page 9
(1) Judd, Deane B.; Wyszecki, Günter (1975). Color in Business, Science and Industry. Wiley Series in Pure and Applied Optics (3rd ed.). New York: Wiley-Interscience. p. 388. ISBN 0-471-45212-2. (2) IES DG-10-12, publication of the Illuminating Engineering Society, Copyright 2012, ISBN #978-0-87995-254-9 (3) U.S. Department of Energy, Energy Efficiency and Renewable Energy, http://cool.conservation-us.org/byorg/usdoe/color_rendering_index.pdf (4). Development of a Color Quality Scale, National Institute of Standards and Technology http://accreditedgemologists.org/lightingtaskforce/nistdevelopmentofcolorqualityscale.pdf May, 2012 Page 10