Panel Discussion. Alfred Borden Principal, The Lighting Practice; Naomi Miller Senior Lighting Engineer, Pacific Northwest National Laboratory; Willem Sillevis Smitt,- Xicato; Kevin Willmorth, Lumenique LLC

2. Credit(s) earned on completion of this course will be
reported to AIA CES for AIA members. Certificates of
Completion for both AIA members and non-AIA
members are available upon request.
This course is registered with AIA CES for
continuing professional education. As such, it does
not include content that may be deemed or
construed to be an approval or endorsement by the
AIA of any material of construction or any method or
manner of
handling, using, distributing, or dealing in any
material or product.
___________________________________________
Questions related to specific materials, methods, and services will
be addressed at the conclusion of this presentation.

3. The method for describing and quantifying the color of white lighting has always been insufficient.
Metrics such as color temperature and color rendition never quite satisfied designers or specifiers of
legacy sources using filaments, phosphors and gaseous excitation. The practical result is that it can
be nearly impossible to find different sources in a lighting design that truly match color. The myriad
of possibilities and pitfalls for white lighting offered by emerging solid-state technology demands a
new vocabulary of descriptive terms, redress of manufacturing protocols, creation of new test metrics
and rebuilding of specification standards. This panel will address these issues from the viewpoint of
product developers, marketers, lighting designers and visual researchers. The presentation will
review the roots and limitations of current methods, the opportunities for building new standards and
market pressures that impede their adoption. The panelists intend for this presentation to push the
industry into changing how it answers the basic question: What Color is White solid-state Lighting?

4. 1. Understand the deficiencies in current metrics as a means to describe, test and communicate the
color of white solid-state lighting, including Correlated Color Temperature, Color Rendering Index, LM-
79 and MacAdam Ellipses.
2. Understand the use of color metrics to quantify color similarities and differences, and how to specify
different lighting products in a single room and get them all to match.
3. Understand the importance of manufacturing quality and testing methods in the development of
products that can be reliably specified to produce a specific color performance
4. Develop an appreciation for the effect of lighting color on human visual performance and physical
health.
5. Understand the attributes of hue, value and saturation that must be considered in order to formulate
an effective color metric.

5. Presenters:
Alfred R. Borden, The Lighting Practice;
Naomi J. Miller, Pacific Northwest National Laboratory;
Willem Sillevis Smitt, Xicato;
Kevin Willmorth, Lumenique, LLC

6. Alfred R. Borden

7. Early problems with color shift and consistency

8. Problems with 3000 K MR-16 LED replacement lamps

9. Recent examples of the shift and consistency problems

10. Test of the proposed solution shows the same problem

11. Test of the proposed solution shows the same problem

12. 3000 K CCT, +90 CRI

13. Are the most accessible standards of any value?

14. Naomi J. Miller

15. COLOR…
• is one of the key attributes of lighting
quality
• is rooted in human perception
COLOR METRICS…
• allow for communication of color
attributes
• attempt to characterize human
perception, but aren’t always perfect
• have changed and improved over
time
• establish standards for specifying
products and holding manufacturers
accountable

16. Halogen
99 CRI , 2917 K, Duv 0.000
Compact Fluorescent
82 CRI, 2731 K, Duv 0.003
LED
84 CRI, 2881K , Duv 0.000
Metrics aren’t perfect!

17. Perceiving Color
Spectral Power Distribution (SPD)
Object Spectral Reflectance
Human Observer Response and Interpretation

18. Spectral Power Distributions

19. Spectral Power Distributions

20. Quantifying Color
The CIE System of Colorimetry
Chromaticity Diagrams & Chromaticity Coordinates
Color Mixing
Color Spaces

21. 470
475
480
485
490
495
500
505
510
515
520
530
540
550
560
570
580
590
600
610
620
700
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
y
x
• Used for color of light,
not objects
• Colors of the spectrum
appear around upper
edge (in nm)
• Bottom edge displays
non-spectral colors;
“purple line”
• Colored background is
theoretical only; cannot
be displayed accurately
• Equal energy point at
(0.33, 0.33)
CIE 1931 (x, y) Chromaticity
Diagram
[Adapted from NIST Spreadsheets]

22. 470
475
480
485
490
495
500
505
510
515
520
530
540
550
560
570
580
590
600
610
620
700
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
y
x
Spectrum Locus
Planckian locus
CIE 1931 (x, y) Chromaticity
Diagram • Black body locus
(also called
Planckian locus)
• Chromaticity is the x-
y point on the
diagram, but does
not specify a
spectrum

23. 470
475
480
485
490
495
500
505
510
515
520
530
540
550
560
570
580
590
600
610
620
700
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
y
x
Spectrum Locus
Planckian locus
Illuminant A
CIE 1931 (x, y) Chromaticity
Diagram • Use for plotting
chromaticity
• Use for light sources,
not for determining
absolute
appearance of
objects!

24. • MacAdam ellipses
plot the just-
noticeable-
differences (those
shown are 10 JNDs)
• X-y color space is not
perceptually
uniform!!!
CIE 1931 (x, y) Chromaticity
Diagram

25. 0.0
0.1
0.2
0.3
0.4
0.5
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
v
u
u = 4x / (-2x + 12y + 3)
v = 6y / (-2x + 12y + 3)
CIE 1960 (u, v) Chromaticity
Diagram
• Simply linear
transformation of CIE
1931 x-y
• u-v coordinates
• Intended to be more
uniform (although not
perfect)
• Used for calculating
CCT and Duv

26. u’ = 4x / (-2x + 12y + 3)
v’ = 9y / (-2x + 12y + 3)
CIE 1976 (u’, v’) Chromaticity
Diagram
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
v'
u'
• Further transformation
of CIE 1960 UCS
(multiply v by 1.5)
• u’-v’ coordinates
• Is the most uniform
available (still does not
apply to objects)
• Used for calculating
Δu’v’ for color shift

27. Color Appearance
Correlated Color Temperature
Duv
MacAdam Ellipses
Δu’v’

28. Can you picture (0.4369, 0.4041)?
Can you picture 3000 K?

29. 0.26
0.28
0.30
0.32
0.34
0.36
0.38
0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30 0.32
u
575 580
570
585
Correlated Color
Temperature
CIE 1960 (u, v) Chromaticity Diagram
• Iso-CCT lines are
perpendicular to
Planckian locus in CIE
1960 UCS
• CCT and chromaticity
are not the same
• Two sources that
appear very different
can have the same
CCT!

30. 0.26
0.28
0.30
0.32
0.34
0.36
0.38
0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30 0.32
u
575 580
570
585
+
Duv
- Duv
CCT + Duv
CIE 1960 (u, v) Chromaticity Diagram
• Duv adds a second
dimension to better
convey appearance
• Iso-CCT lines shown
are ± 0.02 Duv
• Typical limits for white
light are -0.006 to
0.006 (but depends on
CCT)

31. Flies in the Color
Ointment• Illuminance adaptation
(Hunt Effect)
• Chromatic adaptation
• And a million other
perceptual quandaries…

32. Consistent chromaticity
When does consistent chromaticity
between multiple light sources really
matter?
• When objects and surfaces are
white or light in color
• When they are viewed side-by-side
• When the scene is not complex or
the light source spills beyond the
frame
• When the application is color-
critical

33. Color Rendering
Color Rendering Index (CRI, Ra)
R9
IES TM-30-2015 Color Metrics

34. Color Rendering Index
(CRI) Ra• Intended to be a fidelity metric
• Reference is blackbody radiation (< 5000 K) or a
representation of daylight (> 5000 K) at same CCT
as test illuminant. “Reference” does not mean
“ideal” illuminant.
• Compares chromaticity of eight (pastel) test color
samples under test illuminant to reference
illuminant
• Averages (and scales) differences of each sample
to result in single number
• Maximum score of 100 if all samples match exactly
• CRI is part of a larger system that includes 14 (now
15) total samples
• Applicable to sources near blackbody locus

35. Because color space is skewed at red…
R9=0+ is Good; R9=50+ is Very Good; R9=75+ is Excellent
[Equivalent R9 CRI = 100 – (100-R9)/4 ]
Special Color Rendering Index R9
• Same calculation method as CRI (Ra)
• Saturated red
• Red is particularly important for human skin complexion
• Often considered a valuable supplement to CRI (Ra)
• Can be gamed by manufacturers to get higher scores
• Doesn’t communicate color saturation
• Does not work well for very discrete SPDs (i.e., RGB LED)
• Red colors seem to get short-changed
Limitations of CRI

36. IES TM-30-2015 Color
Rendering Metrics
• TM-30 two-metric system (fidelity [Rf] and
gamut [Rg])
• Rf quantifies average color rendition of 99
color evaluation samples (CES) selected to
represent real objects uniformly distributed
in color space, still related to the reference
source
• Rf ranges from 0 to 100
• Rg quantifies the average increase or
decrease of color saturation. 100 means
identical saturation to reference. Can range
above and below 100.

37. TM-30-2015 Color Rendering
Metrics• TM-30 two-metric system
(fidelity [Rf] and gamut [Rg]).
• Harder to game.
• Rf and Rg are still averages.
• Spreadsheet tool offers
information and graphics on
specific colors and hue bins.
0%
20%
40%
60%
80%
100%
380 430 480 530 580 630 680 730 780
RelativePower
Wavelength (nm)
Reference Source LED Hybrid Blue Pump (2)

38. Conclusions +
Notes It is important to understand the limitations/intended use of the various
color metrics.
 Learn to use standard color photometry and tools to calculate a large
range of metrics.
 Even if light sources match when new, they may shift apart over time.
 Metrics get you in the ballpark, but if you are a designer, you must
evaluate color with your own eyes.

39. Thanks!
with a special nod to Dr. Michael Royer
Pacific Northwest National Laboratories
Portland OR
Michael . Royer @ PNNL . gov

40. Willem Sillevis Smitt

41. 44
Do you know which colors these are?
(0.4599, 0.4106)
(0.4369, 0.4041)
(0.4053, 0.3907)
(0.3804, 0.3767)

42. 45
Do you know which color these are?
(0.4599, 0.4106)
(0.4369, 0.4041)
(0.4053, 0.3907)
(0.3804, 0.3767)
CIE 1931, “x,y” coordinates
Or these?
(0.2625, 0.5274)
(0.2505, 0.5214)
(0.2357, 0.5113)
(0.2251, 0.5015)

43. 46
Do you know which color these are?
(0.4599, 0.4106)
(0.4369, 0.4041)
(0.4053, 0.3907)
(0.3804, 0.3767)
CIE 1931, “x,y” coordinates
Or these?
(0.2625, 0.5274)
(0.2505, 0.5214)
(0.2357, 0.5113)
(0.2251, 0.5015)
CIE 1976, “u’, v’” coordinates

44. 47
Do you know which color these are?
(0.4599, 0.4106)
(0.4369, 0.4041)
(0.4053, 0.3907)
(0.3804, 0.3767)
CIE 1931, “x,y” coordinates
Or these?
(0.2625, 0.5274)
(0.2505, 0.5214)
(0.2357, 0.5113)
(0.2251, 0.5015)
CIE 1976, “u’, v’” coordinates
How about these?
2,700K Duv 0.000
3,000K Duv 0.000
3,500K Duv 0.000
4,000K Duv 0.000
CCT and CIE 1960 Duv
More intuitive metric for
white

45. 48
CIE 1931, “x,y” coordinates CIE 1976, “u’, v’” coordinates Simple metric specific for white

46. 49
Duv positive: Greenish / Yellowish
Duv negative: Pinkish

48. 1. Process Control and Product Design
2. Measurement Accuracy
3. Colorimetric Framework

49.  Accurately Matching Primaries
 Keeping them consistent

50.  Accurately Matching Primaries
 Keeping them consistent

51.  Accurately Matching Primaries
 Keeping them consistent

52.  Accurately Matching Primaries
 Keeping them consistent

53.  Accurately Matching Primaries
 Keeping them consistent

54.  Accurately Matching Primaries
 Keeping them consistent

55.  Results from external LM-80 testing on Xicato Modules
 At maximum rated current and temperature
 10,000 hours
 Plot shows individual parts at 0h and 10,000h
 (0 and 10,000h connected by a line)
 Worst case shift:
 CCT +37K
 Duv +0.0016
58

56.  Found on LED datasheet:
“[Manufacturer] maintains a
tolerance of ±0.007 on x and y
color coordinates in the CIE 1931
color space”

57. Proposed Tester Accuracy:
Duv +/- 0.0005 (0.5mDuv)
CCT +/- 20k
At application temperature

59.  Problem: Same Measurement Results (CCT, duv)  clearly different appearance
 Cause: Color matching functions used to quantify color from spectral data
[Csuti, Shanda, Harbers and Petluri, PLDC 2011, Getting Colour Right: Improved
Visual Matching with LED Light Sources ]

60.  Improved color matching
functions significantly
improve consistency
between measurements
and visual observations
[Csuti et al, PLDC 2011]

61.  Pick the
“Standard
Observer” ;-)
 Aging

62.  Initial Color Consistency (LM-79)
 ≤ 1x2 SDCM or
 mDuv ≤ +/- 1 and CCT variation ≤ +/- 50K
 Maintained Color Consistency –
 Ask for LM-80 data represented in CCT and Duv
 Ask for worst case conditions
 Ask for largest shifters (B0) – not averages!
 Light Source Manufacturer Measurement Consistency
 ≤ +/-20K (CCT) and ≤ +/-0.5 mduv
 Test 80 and 95CRI parts of same CCT from same manufacturer to
validate their testing capability
 on a white wall (color matching functions)

63. K. Willmorth

64. Stepping beyond single level classification

65. CCT is listed in
rounded values
(3000K, 4000K, etc..
Virtually no product
delivers exactly that
value

66.  A “3500K” product
delivering 3358K with a
98CRIe will not appear
the same as a “3500K”
product with a 98CRIe
delivering 3795K
 TM-30 does not
resolve this
Range of
“3500K”

67. 3K
LED
3K
LED
3K
LED
3K
LED
3K
LED
3K Output +/- 0K ~-100K
-50 ~ -
150K
-150 ~ -
300K
Clear TIR
Optic
Diffuse R +
Clear Lens
Specular R
+ Diffuser
Diffuse R +
Diffuser

68.  The steps are from a center point
2X Value
1X Value

69.  Without a universal center comparison point,
representations are irrelevant
Manufacturer A 2 Step
Manufacturer B 2 Step
Manufacturer C 2 Step

70.  Two sources with identical represented performance do
not appear the same
 Without any center anchor point for comparisons, there
is no way to determine if products from disparate
providers will appear the same
 Averaged color performance conceals color distortion
within spectral output

71.  We all know incandescent distorts color, yet accept it
as having a high color rendering value
 Daylight is assumed to be a singular color, and assign
it as perfect in color performance
What would happen if we used a central
neutral white (5K?) to compare
EVERYTHING to?

72. Incandescent CCT
Matching Basis

73. Incandescent to
Neutral White Model

74. 65K Ideal Daylight
CCT Match Basis

75. 65K Ideal Daylight to
Neutral White Model

76.  No one quality of light is universal to all needs
 No single metric value can describe white
 Over simplifying leads to confusion
 Reliance on averaged values describing one facet of
quality leads to errors between observation and metric
representation

77.  Uniformity
 CCT specific and range within production
 Duv
 McAdams steps from color point
 Quality
 Fidelity
 Gamma effect
 Lowest performance value
 Human Factors
 Flicker
 S/P ratio

78. A Concept for multi-dimensional description of white lighting
qualities

79. Dust
Water Impact
Q
• Size of particle
• Environmental
(movement)
• 1-6 classification (6
highest)
• Size of object
• Impact energy
• 1-9 classification (8
highest)
• Volume of water
• Pressure/direction/immersio
n
• 1-8 classification (8 highest)
A standard delivering
classification information with
some depth

80. Uniformit
y
Light
Quality
Human
Factors
LQ
C
• CCT Variation / COA
• Duv
• MacAdams Variable
• S/P Ratio
• Flicker
• Color Fidelity
• Saturation Effects
• Spectral Consistency
(lowest R value)
Aggregate multi-dimensional
quality classification
1-5 values – 5 highest

81.  Applied Metrics
 CCT Limits (variation from the stated value)
 Duv – Deviation above or below the Plankian locus
 MacAdam steps from center point of stated CCT value @
Plankian locus intersection
 Rating 1-5, based on combined results of values
 1 delivers the least uniform performance
 5 delivers optimal uniformity

82.  Applied Metrics
 Average CRIe or TM-30 Rf
 Lowest specific “R” value included in average
 TM-30 Rg value
 Rating 1-5, based on combined results of values
 1 delivers the lowest color quality
 5 delivers optimal color performance and rendering

83.  Applied Metrics
 S/P Ratio
 Flicker rating (Frequency, % and Index combined)
 Rating 1-5, based on combined results of values
 1 delivers the lowest visual performance
 5 delivers optimal visual performance

84.  LQC-235
 Factory space with medium color demand, high visual
performance demands
 Places emphasis on economy and visual performance
 LQC-33
 Budget specification with no human factors consideration
 LQC-55
 High color performance for retail/museum
 LQC-555
 Inspection task light for critical visual performance

86. This concludes The American Institute of Architects
Continuing Education Systems Course