The Spectrally Tunable LED light Source – Design ... - PMOD/WRC

The Spectrally Tunable LED light Source – Design and Development Issue
I. Fryc 1 , S. W. Brown and Y. Ohno
National Institute of Standards and Technology, Gaithersburg, MD 20899, USA
1 Guest Scientist from the Bialystok University of Technology, Bialystok, Poland
Abstract. A spectrally tunable light source using a large
number of LEDs and an integrating sphere has been
developed at the National Institute of Standards and
Technology (NIST). The source is designed to have a
capability of producing any spectral distribution,
mimicking various light sources in the visible region by
feedback control of the radiant power emitted by
individual LEDs. The spectral irradiance or radiance of the
source is measured by a standard reference instrument; the
source will be used as a transfer standard for colorimetric,
photometric and radiometric applications. Source
distributions have been realized for a number of target
distributions.
Introduction
Photometric and colorimetric quantities of a light
source can be measured with a spectroradiometer
(calculated from its spectral power distribution) or with a
photometer or a colorimeter that have a relative spectral
responsivity approximated to the spectral luminous
efficiency function, V ( λ)
, or the color matching functions
defined by the CIE 1, 2 . The spectral correction filter is an
essential component of a photometer or colorimeter.
Tristimulus colorimeters typically use three or four
channels of detectors with filters to mimic the CIE color
matching functions. No photometer or colorimeter exactly
matches their spectral responsivities to V ( λ)
or the color
matching functions. Due to this imperfect matching of the
spectral responsivities, measurement errors are inevitable.
These measurement errors can increase dramatically when
the relative spectral power distribution (SPD) of a test
source is dissimilar from that of the calibration source.
Such spectral mismatch errors will be minimized by
calibrating the photometer or colorimeter with standard
sources having similar spectra as the source to be
measured. Spectroradiometers also have various sources of
error including stray light, and these errors can be
minimized by calibrating the instrument with a standard
source having similar spectra as the source being
measured.
A spectrally tunable source, capable of matching the
Spectral Power Distribution (SPD) of a variety of light
sources, would enable accurate and rapid calibration of
colorimeters or photometers used to measure various types
of sources. Such a spectrally tunable source using multiple
LEDs has been designed and constructed at the National
Institute of Standards and Technology (NIST). Utilizing
LEDs with different peak wavelengths and spectral power
distributions, it is possible to make a spectrally tunable
light source 3 (STS) capable of providing a high fidelity
spectral match to given "target" SPDs 4 . The STS is able to
approximate the distributions of a wide variety of light
sources, including daylight illuminants, with high spectral
fidelity. In this paper, we present the construction and
performance of the LED-based STS that has been
developed.
1. Construction of the STS
The STS (Fig. 1) is an integrating sphere source
illuminated by a large number of LEDs having different
spectral peaks and distributions. The sphere (30 cm
diameter) has an opening of 10 cm maximum diameter.
The LEDs can be driven and controlled individually with a
256 channel, computer-controlled power supply. The large
number of spectral channels helps facilitate accurate
matching of fine structure in target source SPDs.
KOMPUTER
COMPUTER
PC
PC
PO WER SUPPLY
LED
HEAD
INT EGRATING
SPHERE
OUTPUT
PORT
Fig. 1. Configuration of the NIST STS.
FIBER FIBRE OPTICS OPTICS
SPECTRORADIOMETER
A diode array spectroradiometer and a photometer are used
as monitor devices to determine the real-time radiometric
and photometric output of the STS, respectively. A
computer logs the output of the spectroradiometer and
individually controls the drive currents of the LEDs to
obtain the required target source distribution. The LEDs
are mounted in 8 to 10 heads equipped with a
temperature-controlled mounting plate to cool the LEDs
and stabilize their temperature.
2. Spectral matching algorithm
An algorithm was developed to achieve the best
matching of the output spectrum of the source to any input
target spectrum. In the optimization, the values of power
coefficients k i , i = 1...
n for the n different LEDs are
determined. The power coefficient k is strongly related to
the current applied to, or radiant flux from, the LED
belonging to that variable. The resulting spectrum is the
sum of all LED SPDs, weighted by the k ’s calculated
according to the optimization algorithm.
The optimization consists of the following steps:

1) Select the initial values of power coefficients
( 0
k
)
i , i = 1...
n of i-th LED; they can be any positive
number.
2) Calculate the j -th value of k i from the (j-1)-th
value, based on the partial derivative with respect to k i
of the square of the difference between the target
spectrum and the STS spectrum built up from the LED
spectra weighted by the variables k:
( )
( j−1
j
)
k = k − a
i
i
⎛
⎜
⎝
780 n
ki
380 i=
1
∂ ∑ ∑
( j −1) S ( λ) − S ( λ)
LED
i
( j −1
∂k
)
i
TARGET
where a should be between 0 and 1. To
achieve an optimal calculation time while still
ensuring convergence of the solution, a
parameter was set to 0.001.
3) Continue the iterations until the solution
converges, i.e. when
780 n
i LEDi
380 i=
1
780 n
= i LEDi
380 i=
1
( j) ∑∑k
S ( λ) − S ( λ)
TARGET
( j−1) ∑∑k
S ( λ) − S ( λ)
TARGET
The equality is obtained from a large number of iterations
where the individual LED currents (power coefficients k)
are increased or decreased in small increments. It can be
interpreted that the algorithm stops when the values k i
reach a steady-state solution. The individual LED currents
are obtained from the optimization. The integral-sum of
the differences at each wavelength is an indication of the
quality of the realized spectral match.
3. Realized STS spectral distributions
Figs. 2 to 4 shows examples of the output spectra of the
STS source matched to the given spectra. The deviations
from the given spectra are mainly caused by limitations in
the availability of LEDs with the appropriate peak
wavelengths, their finite spectral widths (SPD curves), as
well as drifts due to temperature dependence, instability,
and aging of the LEDs. The matching will be improved by
changing the selection of LEDs used. The average
luminance of the STS, which depends on the specific
spectral distribution desired, is between several hundred
-2
and one thousand [ cd ⋅ m ]. Further results of
characterizaion of the STS source developed will be
presented in the full paper.
4. Conclusions
A spectrally tunable LED source (STS) has been
developed. The source can approximate various CIE
illuminants (D65, etc.) as well as common source spectral
distributions for other photometric and colorimetric
applications, and may be useful for visual experiments on
colorimetry as well.
⎞
⎟
⎠
2
,
Relative distribution
1.2
1.0
0.8
0.6
0.4
0.2
0.0
350 450 550 650 750
Wavelength (nm)
Fig. 2. SPD of CIE illuminant D65 (dashed line) and the STS
output measured by spectroradiometer (solid line).
Relative distribution
1.2
1.0
0.8
0.6
0.4
0.2
0.0
350 450 550 650 750
Wavelength (nm)
Fig. 3. SPD of a CRT green (dashed line) and the STS output
measured by spectroradiometer (solid line).
Relative distribution
1.2
1.0
0.8
0.6
0.4
0.2
0.0
350 450 550 650 750
Wavelength (nm)
Fig. 4. SPD of a Cool White FL (dashed line) and the STS output
measured by spectroradiometer (solid line).
References
[1] ISO/CIE 10527-1991, CIE standard colorimetric observers,
1991.
[2] ISO/CIE 10526-1991, CIE standard colorimetric illuminants,
1991.
[3] S. W. Brown, C. Santana, and G. P. Eppeldauer, Development
of a tunable LED-based colorimetric source, J. Res. Nat’l. Inst.
Stands. Technol. 107, 363-371, 2002.
[4] I. Fryc, S. W. Brown, G. P. Eppeldauer, Y. Ohno, A spectrally
tunable solid-state source for radiometric, photometric, and
colorimetric applications, Proceedings of SPIE, 5530, 150 –
159, 2004.