Tomasz ÅšWI
TCZAK1, Bogusław WI
CEK1, Robert OLBRYCHT1, Jakub APOLINARZAK1, Gilbert De MEY2
Technical University of Lodz, Institute of Electronics, Lodz, Poland (1)
Department of Electronics and Information Systems, University of Ghent, Ghent, Belgium (2)
Pulse thermography measurements of heat dissipation from
high power LED
Abstract. Authors focused on the research of heat dissipation problem in electronic devices. For this purpose, a series of thermovision
measurements were carried out, and the object of the research was a high-power LED attached to a radiator. The thermal response in frequency
domain was calculated. Gained data consisting of amplitude and phase values of the thermal response of the diode-radiator set-up during work
allowed authors to estimate its cooling conditions, with particular interest set to the quality of thermal connection between the diode and radiator.
Streszczenie. Autorzy podjęli problem zbadania zjawiska rozpraszania ciepła w urządzeniach elektronicznych. Jako obiekt badań wybrano wysokiej
mocy diodę LED przymocowaną do radiatora i przeprowadzono szereg badań termowizyjnych. Wyznaczono odpowiedz
termicznÄ… w dziedzinie
częstotliwości. Otrzymane amplituda i faza odpowiedzi termicznej pozwoliły na wyznaczenie warunków chłodzenia badanego układu. (Badania
skuteczności chłodzenia diody LED wysokiej mocy metodą termografii impulsowej)
Keywords: thermal resistance, convection cooling, thermal response, thermovision measurements.
Słowa kluczowe: rezystancja termiczna, chłodzenie konwekcyjne, odpowiedz
termiczna, badania termowizyjne.
Introduction the allowed temperatures in the range between -30ºC and
This work is related to thermal issues in electronic +110ºC.
devices, in particular thermal processes associated to
connection between devices and radiators are taken into
consideration. This topic is of great importance, because
the power density increases significantly in newer, mostly
microelectronic devices when new technologies are put to
production. Ubiquitous miniaturization demands from
producers to use more and more efficient heat dissipation
systems from those devices. What is more, increasing
complexity of electronic devices and the amount of data
processed by it results in higher power consumption and
heat emission. Hence it is necessary to provide efficient
Fig.1. The subject of research - high power LED EDSW-3LA6
ways of cooling electronic devices. Problems related to
EDIXEON 3W
excessive heat dissipation in electronics are one of the
major reasons hampering the development of new
technologies for production of smaller generations of
electronic devices, such as processors. Besides of high
power densities in devices, equally important problem is the
way of its dissipation to ambience, often from the case
shared with other devices. Despite of searching alternative
methods of cooling the devices, still the most commonly
Fig.2. The subject of research - radiator A4291
used approach is based on attachment of the radiator and
its convective cooling, often with forced air circulation. In
such a case the device is mounted directly to a radiator.
The factors responsible for convective method are the
shape of the radiator, its active area, material, and very
important thermal contact between the device and radiator
(the way of attaching them). In case of the forced airflow,
additionally the fan efficiency and airflow direction are
important factors. During the works devoted to this article
authors focused mainly on the thermal response of
investigated set-up as a function of the quality of diode-
radiator connection (the method of diode to radiator
mounting), and additionally on the influence of airflow speed
Fig.3. LED diode 3LA6 attached to A4291 radiator
around the whole set-up. It was investigated if knowing the
thermal response phase value is sufficient to estimate the
There were three connection methods applied to mount
quality of connection between the high power LED and the
the diode to the radiator: direct contact, with
radiator, and in addition the influence of convective cooling
thermoconducting silicon paste H by AG Chemia and with
conditions of the radiator was determined. [1..3]
use of Arctic Silver 5 thermoconducting paste containing
silver particles.
Object of measurements
For research, authors have chosen white LED EDSW-
Measurements and calculations
3LA6 EDIXEON 3W by EDISON-Opto attached to the
The set-up of the radiator and diode, which was the
aluminum radiator A4291. This diode emits light similar to
subject of research, was placed in a wind tunnel that
the natural one, has 3W power and operation voltage from
enabled the user to change the cooling conditions thanks to
3.1V up to 4.3V. Maximum constant load is 700mA under
the possibility of adjusting the airflow speed. Series of pulse
138 PRZEGL
D ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 86 NR 11a/2010
thermography measurements were carried out during the cooling quality, which reflects the quality of diode to radiator
normal operation of this electronic device. Authors connection. [4..7, 10]
researched the thermal response of the set-up for three
cases of diode to radiator connection: direct contact, with
thermoconducting silicon paste H by AG Chemia and with
Arctic Silver 5 thermoconducting paste containing silver
particles. Each connection method provides different heat
transfer coefficient, hence the heat generated by the diode
is to be taken by the radiator differently. Authors collected
sequences of thermograms showing the thermal response
of the measured diode-radiator set-up in time, for all three
connection ways. For each method of connection, authors
recorded the sequence of thermograms for natural
convection and forced one, with airflow speeds: 0,5mps,
1mps and 1,5mps. Fig.6. The thermal response in time for forced convection 1.5mps
Figure 4 presents the comparison of the thermal
response in time domain for direct connection between the Using Matlab environment, FFT transform was
diode and the radiator for different cooling conditions. Blue calculated for each single pixel along all thermograms from
curve represents natural convection, other curves represent recorded sequences. The result was the frequency
forced convection (green curve is for 0.5 mps, red one is for response of the set-up for thermal excitation. The calculated
1 mps and black one is for 1.5 mps). phase and amplitude of the thermal response of the diode-
The thermal response in time domain for natural radiator set-up was subjected to the detailed analysis with
convection for different connection types between the diode particular interest set to phase diagrams.
and the radiator is shown in the Figure 5. Figure 6
represents the same comparison, but for forced convection
with 1.5 mps airflow speed.
In all cases the investigated circuit was connected to the
same radiator and remained in stationary environment
conditions. The differences in LED diode cooling quality
appeared solely as a result of the difference in diode to
radiator contact, whose properties were the subject of
change. In particular, authors applied three contact options:
direct, and with two different thermal pastes.
a) b)
Fig.7. Exemplary graphical representation of the amplitude (a) and
the phase (b) of the thermal response after FFT calculations
As it can be noted from Figure 8, the thermal response
phase of the investigated optoelectronic device is highly
dependent on its cooling conditions, that is diode to radiator
connection in this particular case. The better is the cooling,
the lower absolute values of thermal response phase may
be observed. What is more, the better the cooling
conditions are, the stronger this dependence is. For 1.5
mps airflow the difference of thermal phase response
between connections is more visible.
Fig.4. The thermal response in time for connection without
thermoconductive paste
Fig.8. Dependence of thermal response phase on the quality of
diode to radiator connection for natural convection (green plot) and
for forced convection: 0.5mps (black plot), 1mps (blue plot), 1.5mps
Fig.5. The thermal response in time for natural convection
(red plot). Point 1  connection without paste, point 2  connection
with silicon paste, point 3  connection with silicon paste containing
FFT allows to convert signal from time to frequency
silver
domain and calculate phase response for desired
frequency. The recorded sequence of one pulse lasted
For better comparison, in the table below one may find
1000 seconds, therefore authors have chosen the
thermal response phase values for natural convection as
frequency of the first harmonic (1 mHz), and for this
well as for forced airflow.
frequency plotted the thermal response phase versus the
PRZEGL
D ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 86 NR 11a/2010 139
Table 1. Comparison of thermal response phase values for natural Krajowej nt. "Termografia i termometria w podczerwieni".
and forced convection with 0.5mps, 1mps and1.5 mps airflow, Ustroń, 14 16 listopada 2002
calculated for 1 mHz (first harmonic) [3] G. De Mey, J. Pilarski, M. Wójcik, M. Lasota, J. Banaszczyk,
B. Vermeersch and A. Napieralski,  Influence of interface
Natural convection Forced convection 0.5mps
materials on the thermal impedance of electronic packages ,
Type of Phase value Type of Phase value
International Communications in Heat and Mass Transfer,
contact [deg] contact [deg]
Volume 36, Issue 3, March 2009, Pages 210-212.
Without paste -54 Without paste -44
[4] G.De Mey, B.Vermeersch, J.Banaszczyk, T.ÅšwiÄ…tczak,
B.Więcek, M.Janicki, A.Napieralski  Thermal impedances of
Silicon paste -53 Silicon paste -42
thin plates , International Journal of Heat & Mass Transfer,
Silicon paste Silicon paste ScienceDirect, Elsevier; Vol. 50 Issues 21-22 October 2007,
-49 -37
with silver with silver 2007, str. 4457-4460
[5] T.Świątczak, B.Więcek, K.Tomalczyk "Convective cooling
Forced convection 1mps Forced convection 1.5mps
evaluation of electronic devices using lock-in thermography",
Type of Phase value Type of Phase value
14 th International Conference, MIXDES 2007, Ciechocinek,
contact [deg] contact [deg]
Andrzej Napieralski, ISBN 83-922632-4-3 21-23 czerwca
2007, 2007, str. 363-368
Without paste -38 Without paste -28
[6] T.Świątczak, B.Więcek, K.Tomalczyk "Characterisation of
convection cooling for electronic circuits by frequency
Silicon paste -30 Silicon paste -22
analysis", ICSES 2006 International Conference on Signals
Silicon paste Silicon paste
-32 -19 and Electronic Systems, A
ódz
, A
ódz
, Materiały konferencyjne
with silver with silver
17-20 września 2006, 2007, str. 519-522 (nie uwzględniono w
2006)
Conclusion
[7] C. Ibarra-Castanedo, S. Sfarra, D. Ambrosini, D. Paoletti, A.
As it can be noted from above shown phase diagrams of
Bendada, X. Maldague Subsurface defect characterization in
the thermal response of the researched set-up, the better artworks by quantitative PPT, 9th Quantitative InfraRed
Thermography conference, QIRT2008, July 2-5, 2008,
was the diode to radiator connection (with or without
Krakow, Poland
thermoconductive paste), the higher was the thermal
[8] M. Klein, A. Bendada, M. Pilla, C. Ibarra-Castanedo, X.
response phase for particular frequency. In other words,
Maldague  Enhancing Infrared Images Contrast for Pulsed
when the amount of dissipated heat grows (for particular
Thermography , 9th Quantitative InfraRed Thermography
frequency), then phase of the thermal response also grows.
conference, QIRT2008, July 2-5, 2008, Krakow, Poland
What is more, the phase of this response (for particular
[9] M. Susa, X. Maldague, S. Svaic, I. Boras  The influence of
frequency) grows with the increase of airflow speed around
surface coatings on the differences between numerical and
the measured set-up. Hence the better the convective experimental results for samples subjected to pulse
thermography examination , 9th Quantitative InfraRed
cooling conditions are, the higher is the phase value of the
Thermography conference, QIRT2008, July 2-5, 2008,
set-up thermal response. It is worth noting that the most
Krakow, Poland
meaningful and precise information is corresponding to
[10] A. Gleiter, C. Spießberger, G. Busse  Phase Angle
lower frequencies of the spectrum, in the range of single or
Thermography for depth resolved defect characterization , 9th
several mHz. This is because of the pulse thermography
Quantitative InfraRed Thermography conference, QIRT2008,
method that was applied for the research, as the object was
July 2-5, 2008, Krakow, Poland
subjected to excitation in the form of rather long thermal
pulse with duration of 1000 seconds.
Authors:
To draw a final conclusion, one may state that using mgr inż. Tomasz Świątczak, Technical University of Lodz, Institute
applied pulse thermography technique, which is one of non- of Electronics, Electronics Circuits and Thermography Division,
Wólczańska 211/215 st., 90-924 A
ódz
.
destructive examination methods, it is possible to estimate
E-mail: [e-mail].
the quality of thermal connection of measured set-up.
dr hab. inż. Bogusław Więcek, Technical University of Lodz,
Recorded sequences of thermograms are to be transformed
Institute of Electronics, Electronics Circuits and Thermography
to the frequency domain using FFT in order to be used as a
Division, Wólczańska 211/215 st., 90-924 A
ódz
.
base to conclude about the conditions of cooling of
E-mail: [e-mail].
particular electronic device. In case of this article, using this
mgr inż. Robert Olbrycht, Technical University of Lodz, Institute of
method, it was possible to estimate the quality of thermal Electronics, Electronics Circuits and Thermography Division,
Wólczańska 211/215 st., 90-924 A
ódz
,
connection between the diode and radiator.
E-mail: [e-mail].
Positive results, that were obtained, allowed authors to
Gilbert De Mey, Department of Electronics and Information
conclude that proposed method may be widely applied in
Systems University of Ghent, Ghent, Belgium
optoelectronics and electronics to estimate the convective
E-mail: [e-mail].
cooling conditions and the quality of thermal connections.
Tomasz ÅšwiÄ…tczak thank to Ministry of Science and Higher
Education of Poland for the financial support (ref. no.
3601/B/T02/2009/36)
Tomasz ÅšwiÄ…tczak and Robert Olbrycht are scholarship
holders of the project  Innovative didactics without
limitations  integrated development of The Technical
University of Lodz  management of the university, modern
educational offer and increasing potential to employ also
people with disabilities co-financed by the European Union
within the European Social Fund.
REFERENCES
[ 1 ] Stefan Wiśniewski  Wymiana ciepła , WNT, 2000J.
[2] Nowakowski A.  Termografia i termografia dynamiczna w
zastosowaniach medycznych . Materiały IV Konferencji
140 PRZEGL
D ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 86 NR 11a/2010