Journal of Transportation Technologies, 2012, 2, 78-83
http://dx.doi.org/10.4236/jtts.2012.21010 Published Online January 2012 (http://www.SciRP.org/journal/jtts)
LED Applications in Railway Signals: Wavelength and
Intensity vs Temperatur e Variation
Giuseppe Schirripa Spagnolo, Donato Papalillo, Andrea Martocchia
Department of Electronic Engineering, Università Degli Studi “Roma Tre”, Roma, Italy
Received October 12, 2011; revised November 17, 2011; accepted November 27, 2011
In recent years, light emitting diodes (LEDs) have entered the lighting market, offering consumers performance and
features exceeding tho se of traditional lighting technologies. LEDs (light-emitt ing diodes) are becoming more common
in safety signals for railroad, highway, automotive, and many other applications. In addition to having a longer life and
greater durability than incandescent bulbs, LEDs are much more energy efficient than their incandescent counterparts.
Since the heat from the junction must be dissipated into the ambient somehow, changing the ambient temperature af-
fects the junction temperature and hence the emitted ligh t. When the LEDs are used in the railway or traffic signals, the
optical proprieties of these have to maintain more rigorous specifications. Therefore the development of signals using
LED as light source, able to respect intensity specifications, is not simple. In this paper, we describe problems of the
temperature dependent changes of LED intensity and color shift. Besides we will introduce an innovative technique,
that we have developed, to allow the use of the LEDs in applications with rigorous specifications .
Keywords: LED Temperature-Depende ncy; LED Intensity Shift; Railway Signal
LEDs have been around since the 1960s, but they reach
significant levels of luminous output only recently. Power,
energy efficiency, maintenance savings, impact resistance
durability make LEDs suitable for safety signals for rail-
way, road traffic, and other applications in stationary
transportation [1,2 ].
The new systems of signaling go to replace the conven-
tional systems because they improve characteristics as
the maintenance and the time of failure.
Conventional signals produce their lit aspects using
either of following processes:
tungsten lamps which emit nominally yellowish-white
light and different combinations of colored and/or clear
focusing lenses, which collect some of the light from
the lamp, filter it into the appropriate color and focus
it into the requ ired angle.
tungsten halogen lamps which emit nominally white
light and with colored filters which develop the re-
The major drawbacks of this technology are:
limited duration of the light sources (about six mon-
low efficiency of conversion by electric input power
to optical power,
high risk of failure of signal i ng.
In particular, traditional light source fails catastrophi-
cally. Therefore, to guarantee the suitable safety stan-
dards, frequent interventions of maintenance and control
are necessary. The regular, and frequent, replacement of
filament lamps is labor intensive which results in high
maintenance costs. This can be particularly high for a
railway signal which requires a minimum of two staff
(for safety) and which may be located at some distance
from an accessible point along the railway. The reasons
behind the replacement of filament lamps on the railway
are thus reduction in maintenance costs. The replacement
of filament lamp road traffic signals also provides reduc-
tion in maintenance and improved readability, but the
main reason for utilizin g LEDs is the reduction in power
consumption and the consequent saving of energy [3,4].
In any case the use of traditional lamps has the advan-
tage that the intensity and chromaticity of emitted radia-
tion appear to be stable and not influenced by changes in
temperature. In contrast, the LED’s technology available
today has the defect that the intensity and dominant
wavelength are significantly temperature dependent.
Thermal management is one of the major issues to be
improved for implementing LEDs into lighting fixtures
because heat affects the performance and reliability of
those fixtures and LEDs. The optical proprieties of LEDs
depended on junction temperature [5,6]. So that the new
systems of signaling, using LEDs as light source, present
opyright © 2012 SciRes. JTTs
G. S. SPAGNOLO ET AL. 79
new critical problems in order to respect the limits im-
posed by the norms. For signals used in stationary trans-
portation, it is necessary to respect the prescriptions of
project on ample range of temperature (typically 248 -
348 K) [7,8]. The former consequence of temperature
variation affects the output spectrum of the light, latter
consequence is related to the intensity of radiation which
decrease as the temperature increase.
The main topic of this research project is to determine
a way to use the commercial available LEDs as light
source in railway signals (as example we have consid-
ered the Italian railway signals), so that they respect the
norms in each ambient temperature. To obtaining this
goal, we study the thermal behavior of the LEDs used in
those signals, like intensity and chromaticity. Then we
study how to drive th e chromaticity of th e signal through
intensity, in order to comply with the norms. At the end
we propose the circuit solution able to drive the LED’s
intensity in the right way.
2. Signal Characterization
The railway signal under exam (as shown in Figure 1) is
realized with some clusters of High-Power-LEDs of dif-
ferent type and color.
The clusters of LEDs are polarized with a constant
current that it is invariant to changes of temperature. In
order to verify the compliance of the norm at changing of
ambient temperature of the railway signal provided, we
implemented this test in a thermal chamber (see Figure
2). We change the ambient temperature while measuring
the optical proprieties of the signal, fo r the three d ifferent
Figure 1. Typical Italian railway signal and example of LED
matrix for different color aspects.
For what concerning the chromatic properties, we have
used spectroradiometric measurement techniques through
which we found the chromaticity coordinates of signals,
according to the International Commission on Illumina-
tion (CIE). Our experience, about the three color aspects,
pointed out that the yellow one has the most critical be-
havior. Using yellow LEDs realized in AlInGaP tech-
nology, it is difficult to achieve at extreme ambient tem-
perature because the chromatic shift is acceptable only in
a relative smaller range (263 - 323 K) instead of the one
specified by the norm (a typical example is shows in
From the previous spectroradimetric measurement, it
was also possible to extract intensity performance of the
three color aspects, which are represented in Figure 4. In
the picture, the inten sity curves are shown in straight line,
while norm limits are in dash line. It is easy to point out,
how the yellow aspect is the only one getting out of the
norm for both low and high temperature limits.
These experiences prove that the main problem in-
volving this railway signal is related to th e compliance of
the yellow aspect to Italian norm, which requires a strict
chromatic condition, nearly monochromatic.
Figure 2. Thermal chamber built to test railway signals.
Figure 3. Colour stability problems associated with yellow
Copyright © 2012 SciRes. JTTs
G. S. SPAGNOLO ET AL.
Figure 4. Intensity diagram for the yellow aspect as a func-
tion of ambient temperature.
Having a look to nowadays high-power LED’s tech-
nology, it’s simple to find out that GaN LEDs are prin-
cipally used as small wavelength sources (UV, violet,
blue, green), offering an inherent higher efficiency than
the AlInGaP technology, which is still impli ed to provide
wavelengths in the opposite side of the spectrum (see
Anyway, since GaN technology seems the flawless
solution regarding the intensity performance of the LED
source, this is not true anymore if compared to the emitting
spectrum for the yellow aspect (see Figure 6) [9-11].
AlInGaP provides a narrower spectrum able to accom-
plish with the Italian norm, and it still remains the main
technology used by the railway signals manufacturers for
the yellow aspect.
For this reason, all our efforts to have an acceptable
quality of the signal’s illumination, are applied to stabilize
the intensity of the AlInGaP-LED, as well as the thermal-
drift of the chromaticity outside the norm’s range.
2.1. Chromatic Compensation
For the chromatic aspect, our solution rely on a couple of
LEDs similar to the ones used by the railway signals,
which have different dominant wavelenghts:
Osram LED LY E6SF—594 nm (yellow);
Osram LED LO E6SF—605 nm (orange);
The purpose is to merge the two spectrum in order to
obtain the suitab le chromatic-b ehavior as the temperature
Then, we have built measurement setup similar to the
one proposed by A. Groh, S. Brückner and T. Khanh ,
which allow us to set the thermal control over the junc-
tion of the LED. At the beginning of the measurement
th e LEDs packages are so ldered onto the metal core board.
Figure 5. Output power of blue/green/amber InGaN and
amber/red AlInGaP LEDs as a function of the ambient tem-
perature from 233 K (–30˚C to 353 (+80˚C). The output
power of each LED was normalized to 1.0 at 25˚C .
Figure 6. Emission spectra of amber InGaN, amber AlIn-
GaP at a forward current of 20 mA .
It is important that the solder pad of the LEDs packages
exactly fit the solder pad of the metal co re board in order
to reach the lowest thermal resistance between LEDs and
board. In this way, the junctio n temperature is nearly the
same as the surface temperature of the die, being the
pn-junction temperature the cause of shift of dominant
wavelength. Afterwards the combination of the LEDs
and the board is mounted on the temperature-controlled
appliance. An outline of the whole structure is shown in
Copyright © 2012 SciRes. JTTs
G. S. SPAGNOLO ET AL. 81
Figure 7. Outline of the measurement system.
The next step is to apply the needed current to the
LEDs and wait until the structure is thermally stabilized.
Since we have registered a higher junction temperature
during the thermal characterization of the signal, we
measured the spectrums of the LEDs, in a different tem-
perature range from the one defined by the norm. The
temperature of the substrate is controlled by a PID-con-
troller combined with the Peltier element. The spectral
power distribution is measured from 380 nm up to 780
nm with the spectrometer. All photometrical and colori-
metrical parameters are calculated by software.
With this setup we have thermally characterized the
spectrum of the LEDs in the same configuration of po-
larization, used by the railway signal. The measurement
allows us to extract the chromatic coordinates of both
LEDs in some significant temperature.
Now, we consider a new railway signal made up by a
matrix of these two types of LED, so that its chromatic
behavior is represented by the combination of each cou-
ple of chromatic coordinates (yellow and orange). Then,
we use a software to manipulate the intensity ratio be-
tween the two types of LED, in order to obtain a result-
ing spectrum inside the norm limits, for every tempera-
ture conditions. So, if it is possible to define two electric
powers for each type of LEDs, we can obtain a suitable
optical characteristic for a signal, by mixing the chro-
matic coordinates weighted by the intensity of each LEDs.
Also, if these electric powers piloted through measuring
of temperature we could obtain a right behavior for all
range of temperature, defined by the specifications (see
2.2. Intensity Compensation
Generally, the datasheets supply information around the
electric/thermal behavior of the LED, and its optoelec-
tronic properties, su ch as spectrum and intensity stability
through thermal variations (see Figure 11) . It is pos-
sible to see the dependence of forward current vs forward
voltage at constant ambient temperature and forward volt-
age vs ambient temperature at constant current.
As we said before, an increasing temperature defines a
drop in intensity p erformance, as well as a diminishing in
the forward voltage. By measuring the forward voltage
Figure 8. It shows that using the amber LED with intensity
0.3 times the intensity of the yellow LED, working at tem-
perature of 298 K, the resultant color falls at the center of
Figure 9. Working temperature 246 K; amber LED with
intensity equal the intensity of the yellow LED, the resultant
color falls inside the color coordinates.
Copyright © 2012 SciRes. JTTs
G. S. SPAGNOLO ET AL.
Figure 10. Working temperature 346 K; amber LED is off,
with this temperature yellow led have color inside the color
Figure 11. Characteristics of Osram LEDs; (a) Forward
current vs Forward voltage; (b) Relative luminous intensity
vs Forward current; (c) Relative forward voltage vs tem-
perature; (d) Relative luminous intensity vs Temperature
drop, it is possible to have a te mperature reference of the
LED cluster. This reference allows us to estimate the
lowering in intensity, according to intensity-temperature
relation described in the datasheet. It will be possible to
compensate the intensity of radiation by increasing the for-
ward current, since those parameters are related propor-
tionally. However, if we increase the forw ard cu rrent, w e
lead the forward voltage to pursue this increment, so the
compensation system is not trivial. Our solution provides
a circuital feedback shown in Figure 12.
The core of the circuit is the red-marked op-amp, which
works to maintain the equality of its inputs defining the
polarization of the nMOS and so the current flowing in
The circuit is designed to provide the following equa-
tion between the parameters of the LED-cluster:
The first instrumentation amplifier measures the for-
ward voltage drop of the LED-cluster, due to thermal-
effect, while the second amplifier takes the RSIFW. The
red op-amp constrains the LED-cluster to work according
to the relation (1), by increasing/decreasing the current
only when the temperature raise/drop. The parameters of
the circuit RS and VREF define the work curve of the clus-
ter (see Figure 13), and must tune the electrical behavior
(voltage and current) to guarantee the thermal steadiness
of the intensity. Looking at the curve, it is possible to
understand the behavior of the circuit: as the temperature
rises, the voltage drops, and the operating poin t (Q) drifts
toward higher current, balancing the intensity loss.
Tuning these parameters lead us to merely compensate
the thermal intensity drop, or to obtain a new intensity
behavior for each temperature.
The ultimate goal of this study aims to design a new
railway signaling system that combines the two proposed
solutions for color stability and intensity. We have seen
Figure 12. Circuital Setup for the intensity stability.
Copyright © 2012 SciRes. JTTs
G. S. SPAGNOLO ET AL.
Copyright © 2012 SciRes. JTTs
 E. Hong and N. Narendran, “A Method for Projecting
Useful Life of LED Lighting Systems,” 3rd International
Conference on Solid State Lighting, Vol. 5187, 2004, pp.
 N. Zheludev, “The Life and Times of the LED—A 100-
Year History,” Nature Photonics, Vol. 1, No. 4, 2007, pp.
 GmbH, “LED Metrology Handbook of LED Metrology,”
Instrument Systems GmbH, München, 2000, pp. 1-40.
 D. G. Todorov and L. G. Kapisazov, “LED Thermal Man-
agement,” ELECTRONICS’ 2008, Sozopol, 24-26 Sep-
tember 2008, pp. 139-144.
Figure 13. Curve of the Operating Point (Q) defined by circuit.
 ITE, “Vehicle Traffic Control Signal Heads (VTCSH)”
and “VTCSH Part 2: LED Vehicle Signal Modules,” In-
stitute of Transportation Engineers, Washington DC.
that using a pair of yellow and orange LEDs, it is possi-
ble to adjust the chromaticity in the temperature range
indicated by the standard, only acting on their intensity
ratio. For what concern the stability of in tensity, we have
seen a circuital setup able to counteract the drop in lu-
minous efficiency with increasing temperature. The in-
teresting aspect of the proposed circuit is related to the
calibration of its parameters (RS and VREF), which are
able to provide any trend of intensity-temperature for a
cluster of LE D.
 CIE, “A Guide for the Design of Road Traffic Lights,”
Commission International De L’Éclairage, Vienna, 1988.
 T. Mukai, M. Yamada and S. Nakamura, “Characteristics
of InGaN-Based UV/Blue/Green/Amber/Red Light-Emit-
ting Diodes,” Japanese Journal of Applied Physics, Vol.
38, No. 7A, 1999, pp. 3976-3981.
 F. A. Kish and R. M. Fletcher, “AlInGaP Light-Emitting
Diodes,” In: G. B. Stringfellow and M. G. Crawford, Eds.,
Semiconductors and Semimetals, Academic Press, San
Diego, 1997, p. 149.
Assuming a signal consistin g of a matrix of yellow and
orange LED types, it will be possib le to use and calibrate
a couple of circuits for these two clusters of colors. As
we said, this configuration allows to secure the thermal-
stability of intensity, bu t also the inten sity trend to fo llow
a predetermined temperature pattern. In this way it is
possible to feed the clusters independently at any tem-
perature, according to our design, and to obtain the opti-
mal intensity ratio for the compliance of chromaticity
over the whole range of temperatures.
 T. Suski, P. Perlin and B. Monemar, “Gallium Nitride
(GaN) I,” In: J. I. Pankove and T. D. Moustakas, Eds.,
Semicondutors a nd Semi met al s, Academic Press, San Diego,
1998, p. 279.
 A. Groh, S. Brückner and T. Q. Khanh, “The Tempera-
ture-Dependent Changes of the Photometrical and Col-
orimetrical Parameters of Today High Power LEDs,” Pro-
ceedings of the 7th International Symposium on Auto-
motive Lighting, Munchen, 25-26 September 2007, pp.
REFERENCES  OptoSemicondutors OSRAM, LS E6SF, LR E6SF, LA
E6SF, LO E6SF, LY E6SF, datasheet.
 W. Nobuo, “Recent LED Display Devices,” Japanese
Railway Engineering, Vol. 119, 1992, pp. 6-8.
 E. F. Schubert, “Light-Emitting Diodes,” Cambridge Uni-
versity Press, Cambridge, 2003.