PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 84 NR 8/2008 129
Andrzej PAWLAK
1
, Krzysztof ZAREMBA
2
Central Institute for Labour Protection - National Research Institute (1), Bialystok Technical University (2)
Light quality improvement for white electroluminescent diodes
Abstract. The light emitted by the white electroluminescent diodes (LEDs) is typically characterized by relatively high color temperature (daylight)
and low general color rendering index Ra (approximately 70). The conducted research resulted in a mixture of ten white light LEDs, two red-orange
light LEDs and one orange light LED, characterized by very good color rendering index Ra = 90 and color temperature Tc = 2 912 (warm-white
color). The presented solution is further characterized by the luminous efficiency of approximately 30 lm/W.
Streszczenie.
Światło diod elektroluminescencyjnych barwy białej charakteryzuje się zazwyczaj wysoką temperaturą barwową (barwa dzienna)
i
niskim wskaźnikiem oddawania barw Ra (ok. 70). Przeprowadzone badania doprowadziły do uzyskania promieniowania mieszaniny dziesięciu
diod białych, dwóch czerwono-pomarańczowych i jednej pomarańczowej, charakteryzującej się bardzo dobrym ogólnym wskaźnikiem oddawania
barw Ra =
90 i temperaturą barwową Tc = 2 912 K (barwa ciepło-biała). Przedstawione rozwiązanie charakteryzuje się skutecznością świetlną
wynoszącą ok. 30 lm/W. (Poprawa jakości światła diod elektroluminescencyjnych).
Keywords: high power LED, light quality, correlated color temperature, color rendering index.
Słowa kluczowe: diody LED o dużej mocy, jakość światła, temperatura barwowa, wskaźnik oddawania barw.
Introduction
The standard PN-EN 12464-1:2004
– „Light and
lighting. Lighting of work places. Part 1. Indoor work places
”
requires in the majority of cases the utilization of light
sources with the general color rendering index Ra equal to
or superior to 80. There are however locations, where the
application of light sources with the color rendering index
Ra of at least 90 is deemed necessary. Such applications
include all work places where ideal color identification is
necessary e.g. in printing, paint industry etc. The currently
manufactured white light LEDs (Light Emitting Diodes) are
characterized by a low general color rendering index,
typically not exceeding 80 [1,2], which in practice prevents
their application for the general lighting purposes. In this
work, we attempt to find a solution allowing for improving
the color rendering index for white light LEDs by adding
color LEDs.
Light emitted by LXHL-BW03 LED
LXHL-BW03 LED is currently one of a few available
solutions characterized by high color rendering index. The
analysis started with the examination of the spectral
distribution (see Figure 1
– continuous line) and the
examination of the parameters of the LXHL-BW03 LED with
a correlated color temperature of 3300 K (distributed
between 2850 K and 3800 K) and the general color
rendering index not smaller than 90, as indicated by the
manufacturer (see www.lumileds.com). The spectro-
radiometric analysis of the LED in question further
confirmed its catalogue parameters. Based on the
measured spectral distribution, the general color rendering
index Ra was calculated to be equal to 91 and the color
temperature was evaluated to Tc = 3328 K. These
parameters are almost identical with the values provided by
the manufacturer.
Figure 1. Spectral distribution of the relative radiant power P
e
r
emitted by the LXHL-BW03 type LED with high color rendering
index (continuous line) and a typical white LED (dotted line).
The spectral distribution of the LXHL-BW03 LED may
be therefore treated as the prototype for comparisons of
any other LEDs in the attempt to achieve high value of the
color rendering index Ra. However, the spectral distribution
of a typical white LED (see Figure 1
– dotted line)
significantly differs from the said prototype and may not be
altered sufficiently by the addition of radiation emitted by
color
LEDs.
Moreover,
the
LXHL-BW03
LED
is
characterized by very low luminous efficiency and the
energy saving properties of the target solution are critical.
Light emitted by color LEDs
Spectral distributions of the available color LEDs are
narrow and do not cover the whole spectral range of
interest. The previous experience indicates that the
analysis should be based on the real spectral distributions
rather that the catalogue parameters (see Figure 2).
Therefore, prior to the following analysis, the measurement
of the spectral distributions for examined color LEDs was
carried out.
Figure 2. Spectral distributions of relative radiant power P
e
r
emitted by the color LEDs: catalogue distribution (continuous line)
and measured distribution (dotted line).
As anticipated, the measured spectral distributions (see
Figure 2
– dotted lines) are different from the catalogue
distributions (see Figure 2
– continuous lines). The
deviation in question differs between various LED groups
as well as within a single group. In the LED group
characterized by warm radiation: amber, red-orange and
red colors, the deviation is smaller and independent from
the LED color. In the LED group characterized by cold
radiation: blue, cyan and green colors, the deviations are
larger and depend on the LED color (royal-blue LED was
not measured). The spectral characteristics of the blue LED
is shifted towards longer wavelengths, while the cyan and
green LEDs have the spectral characteristics shifted
towards shorter wavelengths.
130
PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 84 NR 8/2008
Table 1. Selected measured and catalogue parameters of the color
LEDs.
LED TYPE
Luminous flux
[lm]
Dominant wavelength
[nm]
Measured Catalogue Measured Catalogue
BLUE
18,1
23
473,6
470
CYAN
45,3
64
495,7
505
GREEN
64,2
64
523,6
530
AMBER
39,8
110*
591,2
590
RED-ORANGE
63,4
190*
**
617
RED
57,9
140*
625,6
627
* - catalogue parameters measured for the current of 1400 mA
(measured values collected for the current of 700 mA)
** - value not calculated by the program
The comparison of the basic lighting parameters of the
color LEDs i.e. their luminous flux and the dominant
wavelength, is included in Table 1. The measured values of
the dominant wavelength deviate from the catalogue values
more in the cold light LEDs than in the warm light LEDs,
which originates from the aforementioned differences in the
spectral distributions. The values of the luminous flux may
be compared directly only in the group of cold light LEDs,
which have the catalogue values provided for the supply
current of 700 mA. The luminous flux of the green LED is
approximately equal to the catalogue value, while the
luminous flux of the blue and cyan LED is smaller by
approximately 25% when compared with the catalogue
value. The luminous flux for warm light LEDs may not be
directly compared with the catalogue values, since the
measurement was carried out for the supply current of 700
mA, while the catalogue values were collected for the
supply current of 1400 mA. Since the warm and cold light
LEDs are connected in parallel branches in the luminaire,
both types of LEDs had to be examined for the same
supply current. The luminous flux of the warm light LEDs
measured at the supply current of 700 mA ranges between
33.4% (red-orange LED) and 41.4% (red LED) of the
catalogue luminous flux provided for the two times larger
supply current of 1400 mA. The measured values of the
luminous flux of the color LEDs are exploited in the
remainder of this paper when calculating the spectral
distributions of the radiation mix of white and various color
LEDs.
Analysis of the radiation mix of a typical white LED
and color LEDs
In the following analysis, the spectral distributions of 5
white light LEDs (W5) and one red-orange light LED (O1)
will be used (see Figure 3
– dotted line), which - based on
the experimental data - is characterized by the general
color rendering index of 86 while maintaining high luminous
efficiency. The distribution in question was altered by
adding light emitted by other color LEDs. By comparing the
resulting spectral characteristics of the mix W5-O1 with the
measured spectral distributions of the color LEDs (see
Figure 2
– dotted lines), it was decided to start the
computer simulations by adding one blue (B1) or one cyan
(C1) light LED radiation to the input W5-O1 distribution
(see Figure 3
– continuous lines). The radiation emitted by
these LEDs fills the missing bands in the input distribution
W5-O1 for the wavelength range spanning from 460 to 510
nm, while a more balanced distribution is achieved for the
mix designated as W5-O1-B1. By adding the cold light
LEDs to the input mix of W5-O1 with the color temperature
of 3055 K, the resulting radiation has increased color
temperature, following the expectations. The W5-O1-B1
mix is characterized by the color temperature of 4180 K,
while the W5-O1-C1 mix has the color temperature of
4049 K.
Figure 3. Spectral distributions of the relative radiant power P
e
r
of
the radiation mix of 5 white light LEDs and one red-orange light
LED (black, dotted line) and one blue light LED (black, continuous
line) or one cyan light LED (grey, continuous line).
Successful reproduction of the cold-white light, used
most commonly in practical applications, could be
considered a success if it was accompanied by the
increase in the color rendering index Ra. It was not the case
since the W5-O1-B1 mix is characterized by the general
color rendering index of 77, while the W5-O1-C1 mix has
the color rendering index of 80. This indicates a significant
deterioration when compared with the input W5-O1 mix with
the color rendering index of Ra = 86.
Table 2. Individual color rendering indices Ri for the light mix
emitted by 5 white light LEDs and one red-orange light LED ((W5-
O1) in connection with one blue light LED (W5-O1-B1) or one cyan
light LED (W5-O1-C1).
Sample
number
Sample Description
Special color rendering
index (Ri) of light mix
emitted by the LEDs
W5-
-O1
W5-
-O1-
-B1
W5-
-O1-
-C1
1
Light greyish red
89
68
75
2
Dark greyish yellow
95
79
89
3
Strong yellow green
76
92
93
4
Moderate yellowish green
75
74
68
5
Light bluish green
89
70
74
6
Light blue
95
73
82
7
Light violet
84
94
89
8
Light reddish purple
82
65
71
9
Strong red
79
12
36
10
Strong yellow
85
68
76
11
Strong green
75
73
61
12
Strong blue
78
59
59
13
Light yellowish pink
91
69
79
14
Moderate olive green
83
91
94
Achieving optical radiation with high color rendering
index Ra by simple addition of the light emitted by a blue
LED (W5-O1-B1) or a cyan LED (W5-O1-C1) to the input
mix W5-O1 failed. Therefore, further studies were based on
the analysis of the individual special color rendering indices
(Ri) for individual mixes (see Table 2). The said analysis
accounted not only for samples number 1
–8, the average
value of which is equal to the general color rendering index
Ra, but also for samples number 9–14, which represent
basic colors with higher saturation as well as reflect the
colors of skin and leaves.
The input mix W5-O1, characterized by the overall color
rendering index Ra = 86, had the indices R3 = 76 (strong,
yellow-green sample) and R4 = 75 (medium, yellow-green
sample) smaller than 80 as well as two indices R7 = 84
(strong, yellow-purple sample) and R8 = 75 (light, red-
purple sample) smaller than the Ra. The increase in these
indices without simultaneous decrease in the remaining
values could potentially achieve the target characteristic i.e.
the general color rendering index of at least 90. By adding
PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 84 NR 8/2008 131
the radiation emitted by a blue LED to the input mix W5-O1,
the values of the R3 and R5 indices increased significantly
(see Table 2
– W5-O1-B1 – R3 = 92; R7 = 94) while the
remaining indices decreased. Similar situation was
observed when adding the radiation emitted by a cyan LED
(see Table 2
– W5-O1-C1 – R3 = 93; R7 = 89), while the
decrease in the remaining indices was notably smaller.
Further analysis focused on the verification of the
aforementioned radiation mixes with a smaller luminous flux
emitted by the blue or cyan LEDs. It was speculated that
there could exist such a level of the luminous flux emitted
by the additional color diodes which could potentially
assure increase in the R3 and R7 color rendering indices
while not deteriorating the remaining ones. This could result
in the increase in the general color rendering index Ra for
the input mix W5-O1. The obtained results (see Table 3)
were not satisfactory, since along with the decrease in the
luminous flux emitted by the additional color LEDs, the
color rendering index increased monotonically to Ra = 86,
characteristic for the input mix W5-O1. Under no conditions
was the obtained result better than Ra = 86.
Table 3. General color rendering indices Ra for the input mix W5-
O1 mixed with the radiation emitted by a blue or a cyan light LED
with limited luminous flux.
Fraction of the nominal
luminous flux of the LED
Color of the additional LED
blue
cyan
50%
83
83
20%
85
84
10%
86
85
Prior to commencing the search for solutions
comprising more than 3 types of LEDs, it was decided to
verify the parameters of the mix of W5-O1 and one green
(G1) light LED or one orange (A1) light LED (see Figure 4,
Table 4). Such mixtures exploit all potential combinations of
the white light LED and two color LEDs.
Figure 4. Spectral distributions of the relative radiant power P
e
r
of
the radiation mix of 5 white light LEDs and one red-orange light
LED (black, dotted line) and one green light LED (black,
continuous line) or one orange light LED (grey, continuous line).
The W5-O1-A1 (with the additional orange light LED)
mix featured the color rendering index increased to Ra = 89,
while the W5-O1-G1 (with the additional green light LED)
mix featured the color rendering index decreased to Ra =
81. The W5-O1-A1 mix features the smallest color
rendering index of 80 for sample number 8 (light reddish
purple). The following analysis featured changing the
quantity of the luminous flux emitted by the orange light
LED in the W5-O1-A1 mix. The increase in the quantity of
the luminous flux emitted by the orange light LED by 20%
(to 120%) caused decrease in the color rendering index to
Ra = 88. However, a decrease in the quantity of the
luminous flux emitted by the orange light LED to 80%
caused an increase in the color rendering index to Ra = 90,
thus achieving the minimum target of the elaborated
research. Further decrease in the quantity of the luminous
flux emitted by the orange light LED resulted in the color
rendering index of Ra = 90 for the luminous flux ranging
between 50% and 80% of the nominal value. Especially the
lower bound of this range is interesting, since its practical
implementation would feature one orange light LED diode
placed in the mixture of 10 white light LEDs and 2 red-
orange light LEDs.
Table 4. Individual color rendering indices Ri for the light mix
emitted by 5 white light LEDs and one red-orange light LED (W5-
O1) in connection with one green light LED (W5-O1-G1) or one
orange light LED (W5-O1-A1).
Sample
number
Sample Description
Special color rendering
index (Ri) of light mix
emitted by the LEDs
W5-
-O1
W5-
-O1-
-G1
W5-
-O1-
-A1
1
Light greyish red
89
86
96
2
Dark greyish yellow
95
93
94
3
Strong yellow green
76
73
88
4
Moderate yellowish green
75
66
84
5
Light bluish green
89
86
93
6
Light blue
95
95
94
7
Light violet
84
77
84
8
Light reddish purple
82
73
80
9
Strong red
79
63
61
10
Strong yellow
85
82
84
11
Strong green
75
62
86
12
Strong blue
78
72
78
13
Light yellowish pink
91
93
96
14
Moderate olive green
83
82
91
Conclusions
The conducted research resulted in a mixture of ten
white light LEDs, two red-orange light LEDs and one
orange light LED, characterized by very good general color
rendering index Ra = 90 and color temperature Tc = 2912
(warm-white color). Therefore, using only 3 types of LEDs,
it was possible to achieve radiation quality comparable with
the previously presented parameters of the LXHL-BW03
LED. The presented solution is however significantly better
in terms of luminous efficiency, since the LXHL-BW03 LED
is characterized by the luminous efficiency of at most 20
lm/W, while each LED used in the examined mix has the
luminous efficiency not worse than 30 lm/W.
REFERENCES
[1] B r o w n D . , N i c o l D . , F e r g u s o n I . : Investigation of the
spectral properties of LED-based MR16 bulbs for general
illumination, Opt. Eng., 44(11), 111310-1-4 (2005)
[2] O h n o Y . : Spectral design considerations for white LED color
rendering, Opt. Eng., 44(11), 111302-1-9, (2005)
Authors: Andrzej Pawlak, M.Sc. (E.Eng.), Central Institute for
Labour Protection - National Research Institute, ul. Czerniakowska
16, 00-701 Warszawa, Poland; phone 48 22 623-46-75, fax 48 22
623-3695, anpaw@ciop.pl;
Krzysztof Zaremba, Ph.D. (E.Eng), Bialystok Technical University,
Chair of Optical Radiation, ul. Wiejska 45D, 15-351 Bialystok,
Poland; phone 48 85 746 94 47, zaremba@pb.edu.pl;