Volume 103, Number 6, November–December 1998

Journal of Research of the National Institute of Standards and Technology

[J. Res. Natl. Inst. Stand. Technol. 103, 615 (1998)]

Spectral Response Based Calibration Method

of Tristimulus Colorimeters

Volume 103 Number 6 November–December 1998

George Eppeldauer A new method is described to calibrate power distributions can be measured with
tristimulus colorimeters for high accuracy an accuracy dominated by the sub tenths
National Institute of Standards and color measurements. Instead of traditional of a percent uncertainty of novel spectral
Technology, lamp standards, modern, high accuracy response determinations.
Gaithersburg, MD 20899-0001 detector standards are suggested for calibra-
tion. After high accuracy absolute spec- Key words: chromaticity coordinates;
tral response determination of the tristimu- color calibration; color temperature; col-
lus receivers, color (spectral) correction orimetry; detector standard; photometry;
and peak (amplitude) normalization can spectral response; tristimulus values.
minimize uncertainties caused by imper-
fect realizations of the Commission Interna-
tionale de l’Eclairage (CIE) color match- Accepted: August 19, 1998
ing functions. As a result of the corrections,
stable light sources of different spectral Available online: http://www.nist.gov/jres

1. Introduction
Tristimulus colorimetry is based on light measure- (k = 2)1 for the disseminated spectral irradiance stan-
ment using three or more receivers with spectral respon- dard lamps in the visible range and a relative 0.59 %
sivities matched to the Commission Internationale de long-term reproducibility [3]. Research is being con-
l’Eclairage (CIE) x̄(l ), ȳ(l ), and z̄(l ) color matching ducted at NIST to decrease the 0.67 % uncertainty by a
functions [1]. To achieve accurate measurements for a factor of three still using standard lamps [4]. The accu-
large variety of light sources, the spectral matches racy of the standard lamp influences the photometric
should be as close as possible to the color matching [receiver matched to ȳ(l )] accuracy of the tristimulus
functions. The receivers are usually realized with silicon colorimeter. The wavelength dependent (e.g., burning
photodiodes and attached filter packages [2]. Usually, time caused) changes of the standard lamp influence the
the spectral mismatch between the realized and the color colorimetric accuracy of the tristimulus meter.
matching functions give the dominant uncertainties in Several commercially available colorimeters were
tristimulus color measurements. compared for accuracy by measuring nine different
At present, tristimulus colorimeters are calibrated laser lines (saturated colors) by Berns et al. [6]. The
with standard lamps. The calibration of the most fre- differences between the theoretical and the measured
quently used color temperature standard lamps is chromaticity coordinates were reported. The lowest rms
derived from source-based spectral irradiance scales. 1
NIST reported a 0.67 % relative expanded uncertainty Throughout this paper uncertainties are given as relative expanded
uncertainties with a coverage factor k = 2 unless otherwise stated [5].

615
 Volume 103, Number 6, November–December 1998
Journal of Research of the National Institute of Standards and Technology

errors in x and y (0.0042 and 0.0051, respectively) were 2. Theoretical Basis

observed on a tristimulus colorimeter employing partial
(mosaic) filters. The errors are expected to decrease to In order to determine x , y chromaticity coordinates of
a large extent for white light sources, such as tungsten a light source, the CIE tristimulus values of the source
lamps. The reported errors were dominated by spectral are to be obtained by
response deviations (mismatch) of the receivers relative
to the CIE recommended color matching functions. The X = k el S (l ) x̄(l ) dl
uncertainties of the calibrating standard lamps also con-
tributed to the results. Y = k el S (l ) ȳ(l ) dl (1)
In contrast to source standards, primary standard de-
tectors (cryogenic radiometers) can measure optical (ra- Z = k el S (l ) z̄(l ) dl ,
diant) power with an uncertainty of 1024 [7, 8]. Certain
where S (l ) is the spectral power distribution of the
type silicon detectors, in a light-trap arrangement, can
source to be measured; x̄(l ), ȳ(l ), and z̄(l ) are the 1931
be calibrated against primary standard radiometers us-
CIE color-matching functions; and k is a normalization
ing several intensity stabilized lasers [9]. With interpo-
factor. In practice, x̄(l ) is realized by two receivers,
lation between the laser lines, a 0.06 % uncertainty in
x̄L(l ) and x̄S(l ):
spectral responsivity was reported on light-trap Si de-
tectors between 406 nm and 920 nm [10]. This uncer-
x̄S(l ) = 0 and x̄L(l ) = x̄(l ) if the wavelength is
tainty is about one fourth of the uncertainties reported
longer than 504 nm, and
for traditional monochromator based detector spectral
x̄L(l ) = 0 and x̄S(l ) = x̄(l ) if the wavelength is
response measurements [11].
shorter than 504 nm.
Achievable total uncertainty of spectral transmittance
measurements on high quality color filters is reported to (Note that the subscripts L and S mean long and short,
be 2 3 1024 [12]. respectively.)
A narrow-band filter-radiometer was calibrated for The CIE tristimulus value X can be written as:
spectral irradiance response in the visible region against
a trap detector with an uncertainty of 0.07 % [13] for X = X1 + X2, (2)
measured irradiance. In that report, the dominant uncer-
tainty components were the uncertainties of the trap where
detector response, 0.036 %, the area of the trap aper-
ture, 0.034 %, and the wavelength reproducibility, X1 = k el S (l ) x̄L(l ) dl , and
0.026 %.
If the uncertainties of the receiver response measure- X2 = k el S (l ) x̄S(l ) dl .
ments are very small, the spatial response non-uniformi-
ties of the filter-detector packages could limit the accu- Y in Eq. (1) will give an absolute photometric quantity
racy of the tristimulus measurements. This problem can (e.g., in lux) [15] if
be avoided if apertures are used in front of the filter-de-
tector packages and the apertures are overfilled with the k = Km = 683 lm/W. (3)
uniform field of the (point) source to be measured. The
area of the apertures can be measured with an uncer- The measured photodiode output currents of the four
tainty of 0.026 % [14]. separate receivers are
The motivation behind the development of a detector-
IX1 = el S (l ) sXL(l ) dl
based calibration method for tristimulus colorimeters
was to utilize the significantly lower uncertainty of new
IX2 = el S (l ) sXS(l ) dl
detector standards compared with traditional lamp stan-
dards. The goal of the method described in this paper is
IY = el S (l ) sY (l ) dl
to determine broad-band calibration factors for all re-
ceivers in a tristimulus colorimeter to minimize mea- IZ = el S (l ) sZ (l ) dl . (4)
surement uncertainty. If the spectral response determi-
nation of the receivers is accurate, application of the where sXL(l ), sXS(l ), sY (l ), and sZ (l ) are the absolute
calibration factors for different source distributions will spectral responsivities of the receivers.
result in high colorimetric accuracy. This multiple re- When measuring a light source of known spectral
ceiver method is an extension of the single receiver power distribution S (l ), the receiver calibration factors
spectral mismatch correction used in our detector-based can be determined from the ratio of Eq. (1) to Eq. (4):
illuminance scale realization [15, 16].
616
 Volume 103, Number 6, November–December 1998
Journal of Research of the National Institute of Standards and Technology

X1 Kmel S (l ) x̄L(l ) dl Once the tristimulus colorimeter is calibrated for kX1,

kX1 = =
IX1 el S (l ) sXL(l ) dl kX2, kY , and kZ , the tristimulus values of a test light
source can be measured as
X2 Kmel S (l ) x̄S(l ) dl
kX2 = =
IX2 el S (l ) sXS(l ) dl X' = X1' + X2' where X1' = kX1IX1' and X2' = kX2 IX2'
(5)
Y Kmel S (l ) ȳ(l ) dl Y' = kY IY' (8)
kY = =
IY el S (l ) sY (l ) dl
Z' = kZ IZ'
Z K e S (l ) z̄(l ) dl
kZ = = m l .
IZ el S (l ) sZ (l ) dl where IXl' , IX2' , IY' , and IZ' are the measured output
currents of the receivers.
By normalizing the color matching functions to their The calibration procedure can be applied to various
peak values, the receiver calibration factors can be writ- measurement geometries (e.g., illuminance, luminance,
ten as: luminous flux, or luminous intensity) depending on the
units in which sXL(599), sXS(442), sY (555), and sZ (446)
1.06291 Km FX1 are expressed.
kX1 =
sXL(599)

kX2 =
0.3501 Km FX2 3. Achievable Accuracy
sXS(442)
(6) In order to obtain the highest color measurement ac-
K F curacy, the receiver calibration factors are to be redeter-
kY = m Y
sY (555) mined for all S (l ) source distributions to be measured.
The spectral mismatch of the receivers, relative to the
1.78297 Km FZ CIE functions, should be small to allow for relatively
kZ =
sZ (446) large uncertainties when determining S (l ). It was
shown in an earlier work [16] that the change of FY with
by introducing the color correction factors : a high quality spectral match of f1' = 1.43 % [2], was
0.1 % for a color temperature change from 2600 K to
el S (l ) x̄Ln(l ) dl 3200 K of a Planckian radiator. With a lower quality
FX1 =
el S (l ) sXLn(l ) dl spectral match of f1' = 3.4 %, which is typical for the red
and blue receivers, the change in FY would be larger, still
el S (l ) x̄Sn(l ) dl allowing for a large enough uncertainty of S (l ). The
FX2 =
el S (l ) sXSn(l ) dl final S (l ) for tungsten lamps, which are more or less
(7) similar to Planckian radiators [17], can be obtained by
e S (l ) V (l ) dl iterating the Planckian function (at different tempera-
FY = l
el S (l ) sYn(l ) dl tures) and the tristimulus measurements, until the
highest color measurement accuracy is reached. For
el S (l ) z̄n(l ) dl other types of sources with smoothly varying spectral
FZ = ,
el S (l ) sZn(l ) dl power distribution (e.g., many kinds of paints, color
tiles, etc.), S (l ) can be measured with a low accuracy
where sXL(599), sXS(442), sY (555), and sZ (446) are the spectroradiometer, and the color measurement accuracy
absolute responses of the realized receivers at the peak still remains high.
wavelengths of the color matching functions; and According to the references in the Introduction, the
sXLn(l ), sXSn(l ), sYn(l ), and sZn(l ) are the relative re- presently achievable relative expanded uncertainty of
sponses of the realized receivers normalized also at the absolute spectral response determinations is on the or-
peak wavelengths of the color matching functions. The der of 0.1 %. The uncertainty of relative spectral re-
peak wavelengths of the realized receivers are not neces- sponse measurements can be lower because of the
sarily equal to the peak wavelengths of the color match- smaller number of uncertainty components. The uncer-
ing functions. tainties of the tristimulus values propagate to the uncer-
A color correction factor will be unity if the normal- tainties of the chromaticity coordinates. The chromatic-
ized channel response is equal to the normalized CIE ity coordinates can be calculated from the tristimulus
color matching function. values via

617
 Volume 103, Number 6, November–December 1998
Journal of Research of the National Institute of Standards and Technology

X Y ance trap detectors (for point sources) is about 0.03 %.

x= , y= . (9)
X+Y+Z X+Y+Z All of our existing illuminance-type meters can be cali-
brated against the irradiance trap detectors with substi-
Now, assume that the relative uncertainties of X , Y , and tution in the uniform field of the tunable monochromatic
Z are all 0.1 %. For a Standard Illuminant A, where point source. After spectral response calibrations, the
x = 0.4476 and y = 0.4074, the worst case scenario described correction method will be applied. A tristimu-
(DX = + 0.1 %, DY = 2 0.1 %, and DZ = 0 %) shows a lus colorimeter calibrated this way will have a chro-
chromaticity coordinate change of ≅0.0004 for both x maticity coordinate measurement uncertainty (ex-
and y , resulting in a ≅10 K change in the correlated panded but not relative) of about 0.0003. The accuracy
color temperature. These expanded uncertainties are of matrix corrected tristimulus colorimeters [22, 23, 24]
similar to those reported for the source-based NIST can also be improved if they are calibrated against the
color temperature scale [18]. With improvements of suggested response corrected reference tristimulus col-
spectral response determinations (as suggested below), orimeters. Further analysis is suggested to establish a
the accuracy of detector-based tristimulus color mea- relationship between the f1' of the receivers and the col-
surements can be further increased. orimetric accuracy when spectrally structured and
changing source spectral power distributions (e.g., color
TV monitors) are measured.
4. Suggested Realizations

The described detector-based calibration method can 5. Conclusion

be applied to the calibration of existing tristimulus col-
orimeters where the spectral response of the receivers The significant decrease of uncertainties in detector
can be measured. The achievable color measurement spectral response measurements in the past 5 years mo-
accuracy will depend on the uncertainty of the spectral tivated the development of a detector-based calibration
response measurements. method for tristimulus colorimeters.
Standard quality tristimulus colorimeters can be con- A method for color (spectral mismatch) correction
structed using high accuracy detector standards such as and peak (amplitude) normalization of color measuring
trap detectors. The advantage of using silicon trap de- tristimulus receivers has been developed. Broad-band
tector standards is that the spectral response of the trap calibration factors, based on spectral response measure-
detector can be determined directly against a primary ments of the receivers, can be determined to minimize
standard cryogenic radiometer [10]. Further advantages the color measurement uncertainties caused by the im-
of trap detectors are relative response non-uniformities perfect receiver response realizations. The method uti-
of less than 0.02 % [19] and very low reflectance in the lizes the lower uncertainty of new detector standards
visible wavelength range [9], especially for transmission relative to traditional lamp standards. The calibration
type versions [20, 21]. The filter packages can be mea- factors can be determined for sources of different spec-
sured separately if they are used with transmission-type tral power distributions.
trap detectors because of zero back reflection from Application of the described detector-based calibra-
these detectors. tion method will result in chromaticity coordinate ex-
We have started developing a new facility to calibrate panded uncertainties (not relative) of less than 0.001.
illuminance measuring photometers and tristimulus col- This corresponds to a color temperature measurement
orimeters against irradiance measuring trap detectors. accuracy equal to or better than that of presently used
These trap detectors will be calibrated against the cryo- primary lamp standards.
genic radiometer in the power (detector is underfilled by
the laser beam) measurement mode. The trap detectors Acknowledgments
are equipped with precision apertures and will measure
the well-collimated radiation (within the aperture) from The author thanks Y. Ohno and M. Racz for their
point sources to obtain the highest accuracy. The point helpful advice.
sources are being realized with small integrating
spheres illuminated by tunable lasers. The aperture ar-
eas are also measured with point sources. The accuracy 6. References
of our cryogenic radiometer is being improved to
achieve radiant power measurements with an uncer- [1] Publication CIE No 15.2 (1986). Central Bureau of the CIE,
tainty of between 0.01 % and 0.02 %. The expected A-1033 Vienna, P.O. Box 169, Austria.
[2] Publication CIE No 69 (1987). Central Bureau of the CIE, A-
spectral irradiance response uncertainty of the irradi-
1033 Vienna, P.O.Box 169, Austria, pp. 5, 9.

618
 Volume 103, Number 6, November–December 1998
Journal of Research of the National Institute of Standards and Technology

[3] J. H. Walker, R. D. Saunders, J. K. Jackson, and D. A. McSpar- [24] Y. Ohno and J. E. Hardis, Improved Matrix Method for Tristim-
ron, Spectral irradiance calibrations. NBS Special Publication ulus Colorimetry of Displays, AIC Color’97, Kyoto, May 26-30,
250-20 (1987). 1997.
[4] B. K. Tsai, Developments in the New Spectral Irradiance Scale
at the National Institute of Standards and Technology, NCSL About the author: George Eppeldauer is a Ph.D. Elec-
Workshop and Symposium, Session 3E (1996) pp. 325–341.
[5] B. N. Taylor and C. E. Kuyatt, Guidelines for Evaluating and
tronics Engineer in the Optical Technology Division of
Expressing the Uncertainty of NIST Measurement Results, the NIST Physics Laboratory. The National Institute of
NIST Technical Note 1297 (1994). Standards and Technology is an agency of the Technol-
[6] R. S. Berns, M. E. Gorzynski, and R. J. Motta, CRT Colorime- ogy Administration, U.S. Department of Commerce.
try, Part II: Metrology, Color Res. Appl. 18 (3), 315–325 (1993).
[7] J. E. Martin, N. P. Fox, and P. J. Key, Metrologia 21, 147–155
(1985).
[8] K. D. Stock and H. Hofer, PTB primary standard for optical
radiant power: transfer-optimized facility in the clean-room cen-
ter, Metrologia 32(6), 545–549 (1996).
[9] T. R. Gentile, J. M. Houston, J. E. Hardis, C. L. Cromer, and A.
C. Parr, National Institute of Standards and Technology high
accuracy cryogenic radiometer, Appl. Opt. 35, 1056–1068
(1996).
[10] T. R. Gentile, J. M. Houston, and C. L. Cromer, Realization of
a scale of absolute spectral response using the NIST high accu-
racy cryogenic radiometer, Appl. Opt. 35, 4392–4403 (1996).
[11] T. C. Larason, S. S. Bruce, and C. L. Cromer, The NIST High
Accuracy Scale for Absolute Spectral Response from 406 nm to
920 nm, J. Res. Natl. Inst. Stand. Technol. 101(2), 133–140
(1996).
[12] K. L. Eckerle, J. J. Hsia, K. D. Mielenz, and V. R. Weidner, NBS
Special Publication 250-6, Regular Spectral Transmittance
(1987) p. 34.
[13] R. Friedrich, J. Fischer, and M. Stock, Accurate calibration of
filter radiometers against a cryogenic radiometer using a trap
detector, Metrologia 32(6), 509–513 (1996).
[14] E. Ikonen and P. Toivanen, A new optical method for high-accu-
racy determination of aperture area (preprint).
[15] C. L. Cromer, G. Eppeldauer, J. E. Hardis, T. C. Larason, and A.
C. Parr, National Institute of Standards and Technology Detec-
tor-based Photometric Scale, Appl. Opt. 32(16), 2936–2948
(1993).
[16] C. L. Cromer, G. Eppeldauer, J. E. Hardis, T. C. Larason, Y.
Ohno, and A. C. Parr, The NIST Detector-Based Luminous
Intensity Scale, J. Res. Natl. Inst. Stand. Technol. 101, 109–132
(1996).
[17] Y. Ohno and M. Navarro, Improved Photometric and Colorimet-
ric Calibrations at NIST, Proc. of the NCSL Workshop and
Symposium, Session 1E (1997) pp. 99–110.
[18] Y. Ohno, Photometric Calibrations, NIST Special Publication
250-37 (1997).
[19] R. Kohler, R. Goebel, M. Stock, and R. Pello, An international
comparison of cryogenic radiometers, in Optical Radiation Mea-
surements III, J. M. Palmer, ed., Proc. SPIE2815 (1996) pp.
22–30.
[20] J. L. Gardner, Transmission trap detectors, Appl. Opt. 33(25),
5914–5918 (1994).
[21] T. Kubarsepp, P. Karha, and E. Ikonen, Characterization of a
polarization-independent transmission trap detector, Appl. Opt.
36(13), 2807–2812 (1997).
[22] G. Eppeldauer and J. Schanda, Colorimeter with matrix transfor-
mation, AIC Conference on Color Dynamics, Budapest, Hun-
gary, Proc. (1976) pp. 403–413.
[23] E 1455-92, Standard Practice for Obtaining Colorimetric Data
from a Visual Display Unit Using Tristimulus Colorimeters,
ASTM Standards on Color and Appearance Measurement, 4th
Ed. (Available from ASTM, 100 Barr Harbor Dr., Wet Con-
shohocken, PA 19428-2959.)

619