This article was published in the July/August 2011 issue of LEDs Magazine.
View the Table of Contents and download the PDF file of the complete July/August 2011 issue.
In early 2008, the American National Standards Institute (ANSI) published its Solid State Lighting (SSL) color standard, ANSI C78.377-2008, entitled “Specifications for the Chromaticity of Solid State Lighting Products.” For manufacturers, this standard defines how to communicate the chromaticity of white-light SSL products to end users.
After two years of practice using this LED standard, the industry recognized the need for improvements in its accuracy and the need to make the standard more user-friendly. In fall 2010, the ANSI Technical Committee TC78, Working Group of SSL Light Source WG09, formed an ad hoc task force to define the appropriate revisions to the document, focusing primarily on making improvements without requiring major thematic changes.
Similar to other lamp-color standards, this specification will provide recommendations on the white-color-variation ranges when SSL products are used for indoor lighting applications. The white-light chromaticity specified in the standard may deviate from “perceived” white, but are generally acceptable to most users.
Variations of white
The variations of white are primarily described in four so-called directions. These are yellowish (or warm) white; blueish (or cool) white; greenish white, and pinkish white. The variation from yellowish to blueish white occurs in the direction of the correlated color temperature (CCT), where the CCT ranges from lower temperatures (warm white) to higher temperatures (cool white), and the change is measured in delta T. Meanwhile, the variation from greenish to pinkish occurs as a deviation from the Plankian locus or blackbody curve, and is measured in delta uv (Duv). The Duv value on the curve equals zero, while the direction towards greenish color has a positive Duv, and conversely the direction toward pinkish color has a negative Duv value.
The CCT range and Duv range together form white-color boundaries that are detailed in the ANSI SSL chromaticity standard. Using this specification, a manufacturer of SSL products and the users of these products can achieve a common understanding of how any given white color will appear on the chromaticity diagram where the color point of the product fits within (or outside of) these color boundaries.
The ANSI standard does not imply that the whites that fit within the color boundaries are the “good” whites, or that those outside of the color boundaries are inferior. This is because the perceived whiteness of a light source is not directly related to its color rendering capabilities, or to personal preferences. Interestingly, if products’ color variations or tolerance fall within these ranges, they will not necessarily be consistent for observers. In other words, the specified color tolerance in the ANSI standard is large enough that color inconsistency from the same CCT products can be detected.
Chromaticity and product yield
During the development of the ANSI SSL color standard, two basic aspects were considered and the results are reflected in the current version of the document. The first aspect was the chromaticity specifications for existing technologies i.e. linear fluorescent and compact fluorescent lamps. In the SSL color standard, the centers of the defined color range, or the nominal CCTs, as well as the color variations or tolerances, are very similar to those of fluorescent lamps.
The other consideration in the standard was LED product yields. Instead of using ellipses, as is the case for the fluorescent lamp standards, the SSL color standard uses quadrangles so the LEDs in the gaps between ellipses can be used. The outlines of these quadrangles are defined as the overall color range or color boundaries for the SSL products. The ANSI standard recommends eight selected nominal CCTs and the flexible nominal CCTs from 2700K to 6500K with 100K for each increment. In other words, if a manufacturer or user chooses an SSL product claimed at 4200K, a close match to natural moon light, this product will meet the ANSI specification. With these nominal CCTs, the ANSI standard specifies the tolerances in CCT direction, or delta T, and in the perpendicular-to-CCT direction, or Duv. When the current version of the document was developed, the center points of the eight selected nominal CCTs were similar to the fluorescent lamp standards. However, five of the eight center points of the selected nominal CCTs are not on the blackbody curve.
For a given nominal CCT, the standard lists its center-point location with a chromaticity value (xy), as well as the tolerances for delta T and Duv. In the current version, the center points of the CCTs were approximated up to the third decimal of the Duv value. Because of the approximation used for the center-point values of Duv, the upper and lower color boundaries (positive and negative Duv) for the eight quadrangles are not continuous, rather they appear to follow a zigzag shape.
Issues and revisions
This may create two possible issues or inconveniences. One is the correlation between selected nominal CCTs and flexible nominal CCTs. Because both are nominal CCTs in principal, the tolerances for one or the other should be inclusive without conflict. For example, if a nominal CCT of 3800K is chosen, the upper or lower boundaries of the quadrangle should overlap some portion of the quadrangles for 3500K as well as for 4000K as both are selected nominal CCTs. However, in the current version of the standard, that is not the case.
Another inconvenience arises when SSL product manufacturers (LED package manufacturers in particular) divide the quadrangles into finer bins. The LED bins located near the zigzag points can lead to uncertainty or a lack of clarity for the users.
The revisions proposed by the ANSI ad hoc group are primarily focused on fixing these two issues in the current version. The revised technical matter is less complicated but the results can be more accurate. Instead of using three decimals for Duv values for the center points of the nominal CCTs, a fourth decimal is introduced. In addition, a new and more accurate mathematical expression for the Duv tolerance calculation is now presented. With this new expression, the Duv tolerance (+/- 0.006) is unchanged, but the derived upper and lower color boundaries for all the nominal CCTs now become continuous without zigzags. This change will make the tolerances of selected and flexible nominal CCTs more consistent with the hope of making the corresponding LED package binning simpler.
Other proposed revisions in the new version include adding more updated CIE chromaticity, or u'v' coordinates, into the specification. This is part of the continuous effort within ANSI to seek harmonization of international standards. Additionally, users have previously raised the question of how to convert given CCT and Duv values to chromaticity values xy or u'v'. The proposed revision will provide detailed explanations in one of the annexes for calculating these conversions. Technically, for the corner values of each defined quadrangle in xy or u'v', the connecting lines between the two upper corners and the two lower corners are not straight, but have some curvature similar to the blackbody curve. The new calculation method provided in the revision will explain how to derive any point on these curves. This will also serve to give a more accurate description when SSL products are to be certified for specific CCT values.
Two larger questions with no clear answers remain that are not addressed in this revision. The first is: what is "perceived" white, or where should the center points or curve be located for a white light source? Another question is: what will be the acceptable deviations from the center curve of the perceived white for observers? Ideally the color range or color boundaries specified in the ANSI standards should be based on the answers to these two questions.
When the ANSI SSL color standard was initially developed, some information was unavailable and certain assumptions were made, which is not unusual in standards development. Currently two studies are being conducted by research organizations in the US to better understand the relationship between these two questions. Like any other research into human factors, these studies may need to involve reasonably-adequate sample sizes and may not be completely conclusive. Human observers are often subjective, particularly when dealing with color-related questions. White light as well as its associated acceptable deviation to one group of observers may not be the same to another group of observers.
To add more complexity, convoluted answers can arise from observing the whiteness of a light source and then relating that whiteness to the color rendering of illuminated objects with a variety of colors. Hopefully the results from these studies may provide references or rationale to the standards community. Without convincing evidence or arguments, it is difficult to make further changes to any existing standards.