Adv. Radio Sci., 9, 317–321, 2011
www.adv-radio-sci.net/9/317/2011/
doi:10.5194/ars-9-317-2011
© Author(s) 2011. CC Attribution 3.0 License.
Advances in
Radio Science
Reduction of heat sink common-mode currents in switching mode
power supply circuits
J. Kulanayagam, J. H. Hagmann, K. F. Hoffmann, and S. Dickmann
Helmut-Schmidt-University/University of the Federal Armed Forces Hamburg, Germany
Abstract. In this paper, a new filter design for a heat sink
is presented. The parasitic couplings between electric power
devices and the heat sink are responsible for common-mode
currents. The main focus is on the reduction of these currents
to reduce the heat sink radiation. For this purpose a new
filter design is proposed. In addition, experimental results
are shown to validate the proposed filter.
1
at different frequencies depending on the heat sink geometry
(Brench, 1994), (Das and Roy, 1998).
In particular, the heat sink can become an efficient antenna
when resonant phenomena occur (Damiano et al., 2004).
This paper investigates the origin of common-mode currents generated by SMPS and shows a filter design to reduce
common-mode currents of ungrounded heat sinks and an experimental setup for validating the input impedance and the
heat sink radiation.
Introduction
In general, power-switching semiconductors (MOSFET’s/IGBT’s) are mounted on heat sinks to keep the
power semiconductors within a given temperature range. A
parasitic capacitance can be formed between the case of the
semiconductors and the heat sink. The insulating thermal
compound (Damiano et al., 2004) is the dielectric of this
capacitance. It is shown as isolator 1 in Fig. 1.
Common-mode EMI is usually the result of parasitic effects (Tihanyi, 1995). Due to these capacitances commonmode currents are generated in Switching Mode Power Supply (SMPS) circuits. In consequence, these currents could
violate the EMC standards by their radiation. Therefore, the
heat sink should be well designed.
The drain-source voltage variation increases by decreasing
the commutation time of the transistor and by increasing the
insulation voltage level. Consequently, a high number of harmonic components are generated which makes the heat sink
an EMI generator (Sinclair et al., 1993).
The shape and the dimension of the heat sinks play a major
role in the power electronic system design. In general, as the
size of the heat sink increases, the radiation efficiency also
increases. However, the maximum radiation will take place
Correspondence to: J. Kulanayagam
(kulanayagam@hsu-hamburg.de)
2
EMI generator and heat sink radiation
A parasitic capacitance Cdh (see Fig. 2) exists between the
metal case of the semiconductors and the heat sink. The insulating thermal compound acts as a dielectric for this capacitance. Due to this capacitance and the time varying drainsource voltage a noise current inoise can be calculated according to the following equation, as a first approach for low frequencies and Cdh Chp Chg
inoise (t) = Cdh ·
duds
,
dt
(1)
where duds /dt is the time derivative of the drain-source voltage. Cdh can reach approximately 100 pF (Tihanyi, 1995).
This noise current inoise is split into the conducted EMI current ic and the radiated EMI current ih
inoise (t) = ih (t) + ic (t)·
(2)
The noise current paths through the heat sink in AC/DC converter are shown in Fig. 2 as an example.
In order to reduce the radiated emissions of the heat sink,
it may be grounded. Consequently, the conducted EMI current ic raises in the power supply (Tihanyi, 1995). The results
of the previous works (Li et al., 1993; Radhakrishnan et al.,
2000; Archambeault et al., 2001), have proved that grounding a heat sink can reduce radiated emissions.
Published by Copernicus Publications on behalf of the URSI Landesausschuss in der Bundesrepublik Deutschland e.V.
heat sink current. If the heat sink is not grounded, the conducted EMI current ic can be reduced but the radiated EMI
J. Kulanayagam
et al.:
Reduction
of heat
sink common-mode
current
ih to the
SMPS
raises
(Felic
and Evans,currents
2001).
nd S. Dickmann
Armed Forces
318Hamburg
sink
ower
mode
rents
new
esults
Heat sink
2
Cdh
Heat sink
J.Kulanayagam et al.: Reduction of Heat Sink Common-Mode Currents in Switching Mode Power Supply Circuits
Isolator 1
udsthe capacitance can be minimized. This screen is apdevice
cording to the following equation, as a first approach for low
plied as aCurrents
shield forinthe
heat sink.
Furthermore,
the sides
and Cdh >>etCal.:
Chg
hp >>
2 frequencies
J.Kulanayagam
Reduction
of Heat Sink Common-Mode
Switching
Mode
Power
C
hp Supply Circuits
of the screen are insulated with a well thermally conductdu
MOSFET/IGBT inoise (t) = Cdh · ds ·
ing dielectric material. This screening design is illustrated in
(1)
dt approach for low
device
capacitance
can be minimized.
cording to the following equation, as a first
Fig.
4. the
In this
work the radiated
EMI currentThis
loop screen
(see theis applied as a shield for the heat sink. Furthermore, the sides
frequencies
and
Cdh >>ofCthe
Chg
hp >>
The time
derivative
drain-source
voltage is duds /dt.
PCB
of the screen
are insulated with a well thermally conductCdh can reach approximately 100 pF (Tihanyi, 1995). This
MOSIsolator 2
duds
sink This screening design is illustrated in
noise currentPCB
inoise
is(t)
split
EMI current(1)
ic
ing dielectric Heat
material.
inoise
=into
Cdhthe
· conducted
·
p the
dt
Fig.
3.
Radiated
EMI
current
via the heat sink.
and the radiated EMI current ih
Fig. 4. In EMI
this work
the radiated
ange.
Fig. 3. Radiated
current
via the EMI
heat current
sink loop (see the
case
ating
ectric
g. 1.
ffects
mode
upply
could
e, the
The time derivative of the drain-source voltage is duds /dt.
inoise (t) = ih (t) + ic (t)·
(2)
Isolator 2
Cdh can reach approximately
100 pF (Tihanyi, 1995). This
Heat sink
The
current
paths
through
heat sink
in AC/DC
conefficient method
Fig.
1.noise
of
parasitic
capacitance
noise
current
inoise
isrepresentaion
split into
the
conducted
EMI
current
iAn
c
Screen to reduce radiated emissions of a heat
Fig.
1.Schematic
Schematic
representaion
of the
parasitic
capacitance.
verter
are shown
Fig. 2 as
and the
radiated
EMIincurrent
ih an example. In order to reduce
sink was introduced
Isolatorin
1 a previous study (Nagel, 1999). By in-
Heat sink serting
a copper screen
the heat sink and the
inoise
(t) = ih (t) + icon
(t)·the
(2)
Cdhheat sink geometry
at different frequencies
Isolator between
2
Rectifier depending
L
inoise
1994),
(Dasthrough
and Roy,
The(Brench,
noise current
paths
the1998).
heat sink
in AC/DC conMOSFET/IGBTScreen
C
C
x
In
particular,
the
heat
sink
can
become
efficient
antenna
y
s
verteruare
shown in Fig. 2 as an example.udsInan
order
to reduce
Isolator 1
when resonant phenomena occur (Damiano
etChpal., 2004).
Isolator 3
ih
Heat
sink
This paper investigates the origin of
common-mode
curPCB
Cdh
Chg
L shows a filter
Cy Rectifier
PCB design to reduce
rents generated
by SMPS and
i
asing
noise
ic
MOSFET/IGBT
common-mode currents of ungrounded
heat sinks and an exg the us
Cx
Cy
uinput
ds
perimental setup for validating
the
impedance
and
the
Ground
Fig. 4. Reduction of the common-mode current by screening
Chp
f harIsolator 3
heat sink radiation.
ih
sink
PCB
major
as the
also
place
Fig.
Fig. 2.
2. Noise
Noise current
currentpaths
pathsthrough
throughthe
theheat
heatsink
sinkininananAC/DC
AC/DCconconChg
Cy
PCB
verter
verter.
2 EMI generator and heatic sink radiation
the radiated emissions of the heat sink, it may be grounded.
the conducted
EMI
current
ic 2.
raises
in
thethe
Ground
The noise
current
loops
shown
in2)Fig.
The
current
AConsequently,
parasitic
capacitance
Cdhare(see
Fig.
exists
between
power
supply
(Tihanyi,
1995).
The
results
of
the
previous
path
i
via
C
and
ground
represents
the
common-mode
curc
hg
metal case of the semiconductors and the heat sink. The
inFig. 2.works
Noise
current
paths
through
the C
heat
sink et
in al.,
an
AC/DC
con(Li
et
al.,
1993),
(Radhakrishnan
2000),
(Arrent
and
the
current
path
i
via
and
the
PCB
is
the
heat
h
hp
sulating thermal compound acts as a dielectric for this capacverterchambeault
et al.,
haveisproved
that grounding
a heat
sink current.
the2001),
heat sink
not
conducted
itance.
Due toIfthis
capacitance
andgrounded,
the time the
varying
drainsink
can
reduce
radiated
emissions.
EMI
current
i
can
be
reduced
but
the
radiated
EMI
current
c
source
voltage
a
noise
current
i
can
be
calculated
acnoise
the radiated
emissions
ofloops
the heat
sink,
itagain
may
beFig.
grounded.
current
are
shown
in
3. The
to thenoise
SMPS
raises
(Felic
ih The
and
Evans,
2001).
Consequently,
conducted
EMI
current
ic raises
in athe
current
paththe
ic method
via
Chg to
and
ground
represents
the commonAn efficient
reduce
radiated
emissions
of
heat
mode
current
and
the
current
path
i
via
C
and
the
power
supply
(Tihanyi,
1995).
The
results
of
the
previous
h
hp 1999). PCB
sink was introduced in a previous study
(Nagel,
By inthe heat
sink
current.
The common-mode
ic (Arproworksis
(Li
eta al.,
1993),
(Radhakrishnan
et sink
al.,current
2000),
serting
copper
screen
between
the heat
and the
power
duces
conducted
EMI,
while
i
generates
radiated
EMI
h
chambeault
et al.,
2001), have
proved
that grounding
a heat
device the
capacitance
can be
minimized.
This screen
isvia
apthe
heat
sink
as
an
antenna.
If
the
heat
sink
is
not
grounded,
sink can
radiated
pliedreduce
as a shield
for emissions.
the heat sink. Furthermore, the sides
the
conducted
EMI
current
can bea reduced
the 3.
radiated
The
current
areicshown
againthermally
inbutFig.
The
of noise
the
screen
areloops
insulated
with
well
conductEMI
current
i
to
the
SMPS
raises
(Felic
and
Evans,
2001). in
h
ingpath
dielectric
screening
designthe
is illustrated
current
ic viamaterial.
Chg andThis
ground
represents
common4. and theHeat
modeFig.
current
current
path ih via Chp and Heat
the sink
PCB
sink
Cdh
Cdh
In this
work
the radiated
EMI current via current
the
heat isink
(see
is the heat
sink
current.
The common-mode
c pro3) is considered.
ducesFig.
conducted
EMI, while ih generates radiated EMI via
the heatudssink as an antenna. If the heatudssink is not grounded,
Chg
Chp
the conducted EMI
ih current ic can be reduced buticthe radiated
EMI current ih to the SMPS raises (Felic and Evans, 2001).
PCB
Cdh
Ground
Heat sink
Cdh
Heat sink
Fig. 3. Radiated EMI current (left) and conducted EMI current
(right)
loop Sci., 9, 317–321, 2011
Adv. Radio
uds
uds
An efficient method
a hgheat
Chp to reduce radiated emissions of C
ic 1999). By insink was iintroduced
in a previous study (Nagel,
h
left hand side of Fig. 3) between the heat sink and PCB is
considered.
Fig.4.4.Reduction
Reductionof
ofthe
the common-mode
common-mode current
Fig.
currentby
byscreening.
screening
3 Filter design
left hand
of Fig.
the heat sink
and PCB is
Due
to the side
extensive
use3)
of between
digital information
and commu3 Filter design
considered.
nication
equipment, EMC problems have increased. To mitigate
this
ferriteuse
cores
usually
attached and
on the
caDue toproblem,
the extensive
of are
digital
information
commuble/
wire
(see
Fig.
5)
to
suppress
electromagnetic
noise
emisnication equipment, EMC problems have increased. To mitisions
from
digital information
andonFuji3 gate
Filter
design
this problem,
ferrite coresequipment
are usually(Samir
attached
the cawara, 1999). This idea was integrated in the heat sink filter
ble/ wire (see Fig. 5) to suppress electromagnetic noise emisdesign
attenuate
the heat
radiated
emissions. and
The heat
Due
totofrom
the
extensive
use sink
of digital
information
commuFujisions
digital information
equipment
(Samir and
nication
equipment,
EMC
problems
have
increased.
To mitiwara, 1999). This idea was
Filter integrated in the heat sink filter
gate
thisto
problem,
cores
usuallyemissions.
attached on the cadesign
attenuateferrite
the heat
sinkare
radiated
ble/ The
wireheat
(seesink
Fig.is5)extended
to suppress
electromagnetic
emiswith a metallic ring noise
bar which
EMI Source
EMI Receiver
sions
from
digital
information
equipment
(Samir
and
Fujihas the same material characteristics like the heat sink. Furwara,
1999).
idea waspart
integrated
in sink
the heat
sinkcore
filter
thermore,
on This
the extended
of the heat
a ferrite
is attached.
The ferrite
core
is radiated
fixed between
the heat
design
to attenuate
the heat
sink
emissions.
Thesink
heat
Fig.
5. the
Block
diagram
of theThis
filterisdesign
and
power
device.
illustrated in Fig. 6.
Filteris attenuated due to the ferrite
The high frequency current
sink
is
extended
with
a
metallic
ring barby
which
has the same
core. This current can be calculated
the following
equamaterial
characteristics
like
the
heat
sink.
Furthermore,
on
tion
: Source
EMI
EMI Receiver
the extended part of the heat sink a ferrite core is attached.
The ferrite core is fixed between the heat sink and the power
device. This is illustrated in Fig. 6. The high frequency current
due to
ferrite
core. This current can be
Fig. is
5. attenuated
Block diagram
of the
thewww.adv-radio-sci.net/9/317/2011/
filter design
calculated by the following equation :
sink is extended with a metallic
U ring bar which has the same
I =
ds
·
(3)
power
. The
mmone PCB
c proMI via
unded,
diated
001).
ble/ wire (see Fig. 5) to suppress electromagnetic noise emissions from digital information equipment (Samir and Fujiwara, 1999). This idea was integrated in the heat sink filter
J. Kulanayagam et al.: Reduction of heat sink common-mode currents
design
to attenuateetthe
sink radiated
The heat Currents in Switching Mode Power Supply Circuits
J.Kulanayagam
al.:heat
Reduction
of Heatemissions.
Sink Common-Mode
J.Kulanayagam et al.: Reduction of Heat Sink Common-Mode Currents in Switching Mode Power Supply Circuits
Heat sink
Heat sink
319
3
3
Filter
Heat sink
Heat sink
EMI Source
EMI Receiver
sink
Ferrite core
Ferrite core
Fig. 5. Block diagram of the filter design.
Fig. 5. Block
diagram
Isolator
1 of the filter design
J.Kulanayagam
Power
Supply Circuits
Isolator
1
Isolator et
1 al.: Reduction of Heat Sink Common-Mode Currents in Switching Mode
Chg
current
a heat
By inpower
Isolator 1
sink is extended with a metallic ring bar which has the same
MOSFET
Heat sink
MOSFET
material
characteristics like the heat sink. Furthermore, on
the extended part of the heat sink a ferrite core is attached.
The ferrite core is fixed between the heat sink and the power
Isolator 2
device. This
is illustrated
in Fig. 6. The high frequency
Isolator cur2
Ferrite
core
PCB
PCB due to the ferrite core. This current can be
rent is attenuated
calculated
by the
Isolator
1 following equation :
MOSFET
Cdh
Cdh
Uds
IhZ=h
·
Zh Z
Heat sink
MOSFET
MOSFET
S11
S11
Network Analyzer
Network Analyzer
Isolator 2
Isolator 2
Isolator 1 PCB
PCB
(3)
Zf
Zf
Ih
Ih
Fig. 7.
7. Measurement
Measurement setup
setup for
for the
the input
input impedance
Fig.
impedance of
of the
the reference
reference
Fig. 7. Measurement setup for the input impedance of the reference
heat sink
sink design.
design
heat
MOSFET
heat sink design
Chp
ChpIsolator 2
Uds
Uds
PCB
Heat sink
Heat sink
S11
Network Analyzer
Ferrite core
Ferrite core
Fig. 6. Schematic representation of filter design and equivalent cirFig. 6. Schematic representation of filter design and equivalent circuit
cuit
Cdh
Zh
Zf
Ih
I is the heat sink current and Uds is the drain-source
voltage
Ihh is the heat sink current and Uds
is the drain-source voltage
of the transistor and Z is the impedance of the network.
Chp
Udstransistor and Z is the impedance of the network.
of the
The initial heat sink setup of Fig. 1 changes to the filter
The initial heat sink setup of Fig. 1 changes to the filter
design of Fig. 6. An additional impedance Z has to be added
design of Fig. 6. An additional impedance Zff has to be added
in series to the resonant circuit. Z is the impedance of the
in series to the resonant circuit. Zff is the impedance of the
ferrite core.
ferrite
core. representation
Fig.
6.Schematic
Schematic
representation
of
and
equivalent
cirFig.
6.This
of filter
filter
design
and
equivalent
cirnew design
has a higher
heatdesign
transfer
resistance
than
This
new design has a higher heat transfer resistance than
cuit.
cuit
the original setup, but it has a lesser heat transfer resistance
the original setup, but it has a lesser heat transfer resistance
than the screening design.
than the screening design.
Ih is the heat sink current and Uds is the drain-source voltage
U
ds
ofI 4the
transistor
and setup
Z is the impedance of the network. (3)
·
h = Experimental
4
Experimental
setup
Z
The initial heat sink setup of Fig. 1 changes to the filter
design
of
Fig. sink
6. An
additional
Zf has to be
added
Input
impedance
of
the
heat
sink
Ih4.1
is the
heat
current
and
Uimpedance
is the
drain-source
voltage
Input
impedance
of the ds
heat
sink
inof4.1
series
to
the
resonant
circuit.
Z
is
the
impedance
f
the transistor and Z is the impedance
of the network. of the
ferrite
core.
The initial
heat
sink setup ofwere
Fig.necessary
1 changestotoanalyze
the filter
Three
different
configurations
the
Threeof
different
configurations
were necessary
toto
analyze
the
design
Fig.
6. An
additional
impedance
Zf has
be added
input new
impedance
ofhas
theaheat
sink
designs.
This
design
higher
heat
transfer
resistance
than
input impedance of the heat sink designs.
in original
series
to setup,
the
resonant
circuit.
Zf issink
the design
impedance
of the
the
but
has
a lesser
heat
transfer
The first
setup
is anit original
heat
forresistance
measurThecore.
first setup is an original heat sink design for measurferrite
ingthe
thescreening
input impedance
than
design. as a reference. It was placed to the
ing
thenew
input
impedance
as a reference.
It was
placed tothan
the
This
design
has a to
higher
heatthe
transfer
resistance
vector
network
analyzer
measure
scattering
parameters
vector
network
analyzer
to
measure
the
scattering
parameters
the
it hasand
a lesser
heat transfer setup
resistance
S11original
(see Fig.setup,
7). Abut
second
third measurement
were
S11
(seescreening
Fig. 7). Adesign.
second and third measurement setup were
than
the
made
for the extended
heat sink design without ferrite core
for the extended
heat sink design without ferrite core
4 made
Experimental
setup
and
with ferrite core,
respectively. The scattering parameters
and with ferrite core, respectively. The scattering parameters
S11 of these setups were measured by a vector network anof these
setups were
by a vector network an11Input
4.1S
of measured
the
heat
sink
alyzer
(seeimpedance
Fig. 8). From
these
scattering
parameters of all
alyzer (see Fig. 8). From these scattering parameters of all
three configurations, the input impedances were calculated
three configurations, the input impedances were calculated
Three different configurations were necessary to analyze the
input
impedance of the heat sink designs.
www.adv-radio-sci.net/9/317/2011/
The first setup is an original heat sink design for measuring the input impedance as a reference. It was placed to the
vector network analyzer to measure the scattering parameters
3
PCB
Isolator 1
Isolator 1
Isolator 2
Fig. 7. Measurement setup
for the input impedance of the reference
MOSFET
MOSFET
heat sink design
Network Analyzer
Network Analyzer
Heat sink
S11
S11
PCB core
Ferrite
PCB
Isolator 2
Isolator 2
Fig. 8. Measurement setup for the
input impedance of the ferrite
Isolator
Fig. 8.
8. Measurement setup
setup for
for the
the
input1 impedance
Fig.
input
impedance of
of the
the ferrite
ferrite
heat sinkMeasurement
design
heat sink
sink design.
design
heat
MOSFET
by using (Diepenbrock et al., 2001)
4by using
Input(Diepenbrock
impedance ofetthe
al.,heat
2001)sink
1 + S11
S110 · 1 + S11
Zsetup
,
(4)
input = Z
4.1 Experimental
Network Analyzer
Zinput = Z0 · 1 − S11 ,
(4)
1 − S11
Three
were necessary
to analyze
the
wheredifferent
Zinput isconfigurations
the input impedance
to the network,
the used
Isolator
2
whereimpedance
Zinput
is the
input
impedance
toPCB
theThe
network,
the used
input
of
the
heat
sink
designs.
first
setup
is
an
normalization impedance Z0 is 50 Ohms, and the measured
normalization
impedance
Z
is
50
Ohms,
and
the
measured
0
original
heat
sink design
the input impedance
scattering
parameter
datafor
is Smeasuring
11 . In this work, the frequency
Fig.
8. Measurement
setup
the
input
impedance
of the ferrite
scattering
parameter
data
is Sfor
In this
work,
the frequency
11 .the
as
a reference.
wasup
placed
to
vector
network
analyzer
range
from 300ItkHz
to 1 GHz
was
considered.
heat from
sink design
range
300
kHz
up
to
1
GHz
was
considered.
to measure the scattering parameters S11 (see Fig. 7). A sec4.2 and
Radiation
of the heatsetup
sinkwere made for the extended
ond
third measurement
4.2 Radiation of the heat sink
heat sink design without ferrite core and with ferrite core,
using
et parameters
al., 2001)
Inby
order
to (Diepenbrock
validate
the attenuation
of the heat
sink
respectively.
The scattering
ofsink
theseradiated
setups
In order to validate
the attenuation
of theS11
heat
radiated
emissions
of
the
proposed
heat
sink
design
an
experimental
8).
were
measured
by
a
vector
network
analyzer
(see
Fig.
emissions of the proposed heat sink 1design
an experimental
+ S11
test
was done
inside
anZinput
anechoic
chamber
to ,avoid
the
pres=
Z
·
From
these
scattering
parameters
of
all
three
configu0
test was done inside an anechoic chamber
to avoid the pres- (4)
1EMI
− S11
ence
of
environmental
EMI
noises
in
measurements.
A
rations,
the input impedances
calculated
by using
ence of environmental
EMI noiseswere
in EMI
measurements.
A
simplified
schematic
of
the
experimental
setup
is
shown
in
(Diepenbrock
et is
al.,the
2001)
where Zinput
impedance tosetup
the network,
simplified
schematic
ofinput
the experimental
is shownthe
in used
Fig. 9. The RF signal generator was connected via a coaxFig.
9. The RF signal
generator
was
viathe
a coaxnormalization
impedance
Z0 is
50connected
Ohms, and
measured
scattering parameter data is S11 . In this work, the frequency
Sci., considered.
9, 317–321, 2011
range from 300 kHz upAdv.
to 1Radio
GHz was
4.2 Radiation of the heat sink
4
J.Kulanayagam et al.: Reduction of Heat Sink Common-Mode Currents in Switching Mode Power Supply Circuits
100
320
J. Kulanayagam
et al.: Reduction
ofthird
heat configuration
sink common-mode
currents
The
capacitances
the
heat Currents
J.Kulanayagam
et of
al.:in
Reduction
ofMode
Heat Power
Sink (extended
Common-Mode
J.Kulanayagam et al.: Reduction of Heat Sink Common-Mode
Currents
Switching
Supply Circuits
reference
J.Kulanayagam et al.: Reduction of Heat Sink Common-Mode
in Switching
ModeasPower
Supply
Circuits
sink withCurrents
ferrite core)
are the same
first and
second
configferrite core1
80 100
uration. But the inductances of
the heatChamber
sink are increased by
Anechoic
ferrite core2
The capacitances of the third
configuration (extended heat Das, S
100
the
ferrite
cores.
This
is
because
all
two
ferrite
core heat
consist
Thesink
capacitances
the third
configuration
(extended
without
ferrite core
reference
Antenna
hea
with ferriteofcore)
are the
same as first
and second
configreference
60 80
ferrite core1
sink
with ferrite
core) are
thetherefore
same
as first
and second
configof
different
materials
and
different
magnetic
charneti
Heat
sink
uration. But the inductances of the heat sink are increased by
ferrite
core1
ferrite
core2
80
uration. ButThe
the inductances
of of
thethe
heatsystem
sink are
increased
by the Diepen
acteristics.
quality
factor
is
reduced
by
ferrite
core2
the ferrite cores. This is because all two ferrite core consist
without ferrite core
the Signal
ferrite
This is because
all two ferrite
core consist
without ferrite core
gro
ferrite
core.cores.
Furthermore,
it depends
the ferrite
materials
40 60
of
different
materials and
thereforeon
different
magnetic
charGenerator
60
of different
materials and therefore different magnetic charEle
which
is
used.
acteristics.
The
quality factor
of the system
is reduced
by the
996
acteristics.
The
quality factor
of heat
the system
isincreased
reduced by
the
The
input
impedance
of
the
sink
is
because
20 40
ferrite
core.
Furthermore,
it
depends
on
the
ferrite
materials
Felic,
Ground
Plane
core.
Furthermore,
it depends
on the
ferrite
40
Spectrum
aferrite
ferrite
core
hasVa very large
impedance
due
to itsmaterials
permeabilwhich
is used.
circ
Analyzer
which
is used.
ity.
This
means
that
the
heat
sink
currents
will
be
minimized
The
input
impedance
of
the
heat
sink
is
increased
because
0 20
Com
The input impedance of the heat sink is increased because
20
by aadding
the
ferrite
core large
between
the heat
sink
and
power Li, K.
ferrite
core
has
a
very
impedance
due
to
its
permeabilaFig.
ferrite
core
has a very
large
due to its permeabilFig.
11.Experimental
Experimental
setup
emission
12.
setup
forimpedance
emission measurements.
measurements
device
and
the
heat
sink
radiation
is currents
attenuated
by
filter.
of F
This
means
the
heat
will
bethis
minimized
-20 5 00
ity.ity.
This
means
thatthat
the heat
sinksink
currents
will be
minimized
6
7
8
9
VL
10
10
10
10
10
by
adding
the
ferrite
core
between
the
heat
sink
and
power
by adding
the ferrite core between the heat sink and power
30
Frequency/ Hz
mag
Antenna vertical
and
Heat Sink by
frontal
device
and
heat
attenuated
this
filter.
without
ferrite
core
andsink
Vw/radiation
isisthe
received
RFfilter.
signal
device
and
the the
heat
sink
radiation
is attenuated
by this
Ferrite
-20 5
-20
Antenna
vertical
and
Heat
Sink
sideways
Nagel,
6
7
8
9
6
7
8
9
105
10
of the
sink with ferrite core. The measurement setups
10
10
1010
1010
10 10
25 heat
tron
30
Frequency/
Hz enhanced heat sink
Fig. 9. Input impedance of the Frequency/
reference,
Hzthe
andSink
Heat
frontal
were30configured for frontal Antenna
andAntenna
lateral
radiation.
InSink
this
work,
verticalvertical
and Heat
frontal
Wis
with and without ferrite core
andSink
Heat
Sink sideways
Antenna
and Heat
sideways
the frequency
range from 20
upAntenna
tovertical
300vertical
MHz
was
considered.
20
Radha
Input
Inputimpedance/
impedance/dB
dB
Input impedance/ dB
4
4
RRh h
LhLh
Uds
U
Uds
ds
Fig. 10. Equivalent circuit for original heat sink design
5.2 15Results and discussions
20 20
The attenuation of the heat sink radiation is calculated for
15
10
frontal 15
and lateral measurement of the heat sink is illustrated
in Fig.
11. The maximum attenuation value and the average
10
5 10
C
Ihhp
Ih attenuation
value of the frontal measurement of the heat sink
are
25
dB
and
15 dB, respectively. Furthermore, the maxiChp
5
Chp mum05attenuation
value and the average attenuation value of
0
the lateral
measurement
of the heat sink are 10 dB and 5 dB,
-5 0
0
0.5
1
1.5
2
2.5
3
respectively.
8
Frequency/ Hz
Ih
Fig.
Fig.10.
10.Equivalent
Equivalentcircuit
circuitfor
fororiginal
originalheat
heatsink
sinkdesign
design.
Fig. 10. Equivalent circuit for original heat sink design
impedance of the reference heat sink design can be expressed
as: impedance of the reference heat sink design can be expressed
1
1
impedance of the reference
heat sink design
can be expressed
as:
Zinput
·
(5)
1 +=
S11jωCdh1 + Zh + jωChp
1
as:
Zinput = Z0 · Zinput =
,
(4)
+
Z
+
·
(5)
h
1
1 −input
S11 =jωC1dh + Z
jωChp ·
(5)
h+
The impedance of Z
the heat
sink
is
given
by
jωCdh
jωChp
The
impedance
heatimpedance
sink is given
by network, the used
where
Zinput isof
thethe
input
to the
The
impedance
of
the
heat
sink
is
given
Z
=
R
+
jωL
·
(6)
h
h
h
normalization impedance Z0 is 50 Ohms,byand the measured
Zh = Rh + jωLh ·
(6)
scattering parameter data is S11 . In this work, the frequency
Zthe
Rh + jωLheat
h = original
h · sink design can (6)
The resonant
frequency
of
range
from 300
kHz up to 1 GHz was considered.
The
resonant
be expressed
as: frequency of the original heat sink design can
The
resonantas:
frequency of the original heat sink design can
be
expressed
4.2 Results and discussions
be expressed as:
1
1
fres1 = p
·
(7)
ploop · Cloop
=configurations
·
(7)9.
2π
L
The results of allfres1
three
are
illustrated
in
Fig.
1
2π p
Lloop · Cloop
f
=
·
(7)
res1
The curves of the first (reference heat sink design) and sec2π
L, loop
· Cloopimpedance beUp toUp
the
resonant
frequency
f
the
input
res1
ond
heat
sink without
core)
configuration
to (extended
the resonant
frequency
fres1 ,ferrite
the input
impedance
behaveshaves
like
athat
capacitance
and
above
the
resonant
frequency
show,
heat sink
structures
behave
like series
resonant
athe
capacitance
and
above
resonant
frequency
Up
to like
the
resonant
frequency
fres1
,the
the
input
impedance
befres1fhaves
the
impedance
behaves
likelike
anan
inductance.
The
resocircuits.
the
impedance
behaves
inductance.
The
resores1
like
a capacitance
and
above
the resonant
frequency
nant nant
fres2
of
second
configuration
The
result
of
thethe
configuration
can bewas
explained
by
frequency
fres2
offirst
the
second
configuration
wasshifted
shifted
ffrequency
behaves
like
an inductance.
The
resores1 the impedance
into into
the
lower
MHz
frequency
range
because
the
extended
the
following
equivalent
circuit
(see
Fig.
10.)
lower MHz
rangeconfiguration
because the was
extended
nantthe
frequency
fres2frequency
of the second
shifted
The
input
impedance
of
the
reference
heat
sink
design
can
heat heat
sink
design
has
a
larger
current
loop
than
the
reference
sink
design
has
a
larger
current
loop
than
the
reference
into the lower MHz frequency range because the extended
be
expressed
as:
design.
In design
other
words,
second
configuration
has
larger
design.
other
words,
second
configuration
has
heatInsink
hasthe
a the
larger
current
loop than
thea alarger
reference
inductance
than
the
original
heat
sink
setup.
In
consequence,
inductance
than
the
original
heat
sink
setup.
In
consequence,
design. In other
1 words, the1 second configuration has a larger
the
resonant
frequency
is ·smaller
than
resonantfrefre+
Zfhoriginal
+fres2
(5)
Zinput
=frequency
the resonant
is smaller
than
thetheresonant
res2
inductance
than
the
heat sink
setup.
In consequence,
jωC
jωCfrequencies
dh
hp
quency
f
.
The
resonant
of
the
first
and
secres1
quency
f
.
The
resonant
frequencies
of
the
first
and
secres1
the resonant
frequency fres2 is smaller than the resonant freond configuration
are 300
MHz
180
MHz,respectively.
respectively.
ond configuration
MHz
andand
180
MHz,
quency fres1 . are
The300
resonant
frequencies
of the first and secAdv. Radio Sci., 9, 317–321, 2011
ond configuration are 300 MHz and 180 MHz, respectively.
Attenuation/ dB
CCdhdh
25 25
Attenuation/
dB
Attenuation/
dB
Fig.9.
Impedance
of the of
reference,
the enhanced
heat sink
with
and
Fig.
thethe
enhanced
heat
sink
Fig.
9.9. Input
Inputimpedance
impedance
ofthe
thereference,
reference,
enhanced
heat
sink
without
ferrite
core.
Cdh ferrite
Lh
with
with and
andwithout
without
ferritecore
coreRh
-5
0 -5
0
x 10
3
2.5 8
3
x 10
8
12.Attenuation
Attenuation of
of the
the heat
frequency
x 10
6Fig.11.
Conclusions
Fig.
heat sink
sinkferrite
ferritedesign
designininthethe
frequency
0.5
0.5
1
1.5
2
1
1.5
Frequency/
Hz
Frequency/ Hz
2
2.5
rangefor
forfrontal
frontaland
and lateral
lateral measurements.
range
measurements
Fig. 11.
Attenuation
of the heat
sink ferrite design in the frequency
In Fig.
this 11.
paper
a
new
filter
design
introduced.
can
Attenuation
of
the
heat
sinkwas
ferrite
design in theItfrequency
range for frontal and lateral measurements
be range
used for
to frontal
reduceand
thelateral
heat measurements
sink common-mode currents in
The impedance
the heat
sink iscircuits.
given by The validation of
Switching
ModeofPower
Supply
5the Radiation
of the heat
input impedances
of thesink
heat sinks have verified that the
ZhRadiation
= Rh + jωLofh ·the heat sink
(6)
5input
impedances can be increased at some frequencies up to
5
Radiation
of
the
heat
sink
5.1
setup
45
dBExperimental
due to the
ferrite
core between the heat sink and the
The
resonant
frequency
5.1
Experimental
setupof the original heat sink design can
power
device.
Therefore,
the heat sink current can be minibe5.1
expressed
as:
Experimental
setup
In
order
toaddition,
validate
theradiated
attenuation
of thefrom
heat the
sink
radiated
mized.
In
the
emissions
heat
sink
In order toof
validate
the attenuation
of the
heatan
sink
radiated
emissions
the
proposed
heat
sink
design
experimental
1
are
attenuated
by
the
filter.
Again,
our
proposed
filter
design
In order
tothe
validate
the
attenuation
of the
heat sink radiated
emissions
ofp
proposed
sinkchamber
design
an
· heat
(7)
fres1
= done
test
was
inside
an
anechoic
to experimental
avoid the
the heat
presreduces
the heat
sink
noise
currents
anddesign
attenuates
2π
L
· Cproposed
loop
loop
emissions
of
the
heat
sink
an the
experimental
test
was done
inside
an
anechoic
chamber
to
avoid
presence
of
environmental
EMI
noises
in
EMI
measurements.
A
sink
radiated
emissions.
ence
ofwas
environmental
noises in EMI
measurements.
test
done
insideEMI
an anechoic
chamber
to avoid theA presUp
to
the
resonant
frequency
f
,
the
input
impedance
besimplified schematic of the experimental
setup is shown in
res1
simplified
of theEMI
experimental
setupmeasurements.
is shown in A
ence of schematic
environmental
noises
in EMI
haves
like
a RF
capacitance
and above
theconnected
resonant frequency
Fig.
12.
The
signal
generator
was
a coaxFig.simplified
12. The RF
signal generator
was connected
viavia
aiscoaxschematic
of
the
experimental
setup
shown
References
fres1
thetoimpedance
behaves
likedesign.
an inductance.
The
reso- ain
ial
cable
the
heat
sink
ferrite
On
the
other
ial Fig.
cable12.
to the
heat
sink ferrite
design.
Onconnected
the other via
sideside
The
RF
signal
generator
was
aacoaxthe second
configuration
was
shifted
nant frequency
fres2 ofperiodic
biconical/
logarithmic
antenna
receives
the
emitted
biconical/
logarithmic
periodic
antenna
receives
the
emitted
ial
cable
to
the
heat
sink
ferrite
design.
On
the
other
side a
Archambeault,
Pratapneni,
S.,range
Zhang,
L., andtheWittwer,
D.:
into the lowerB.,
MHz
frequency
because
extended
electromagnetic
fields
from
the
heat
sink.
This
received
RF
electromagnetic
fields
from
the
heat
sink.
This
received
RF
Comparison
of various
numerical
modeling
tools
against
biconical/
logarithmic
periodic
antenna
receives
thea stanemitted
heat
sink
design
has
a
larger
current
loop
than
the
reference
signal
Vproblem
detectedfields
byaaspectrum
spectrum
analyzer.
signal
is detected
by
dard
sink
emissions,
IEEE
electromagnetic
from
theanalyzer.
heat
sink.in:This
design.
In otherconcerning
words, theheat
second
configuration
has received
a Internalarger RF
From
these
RF
signals
of
with
and
without
ferrite
core
From
these
RF
signals
of
with
and
without
ferrite
core
tional
Symposium
on
Electromagnetic
Compatibility,
vol.
2,onpp.on
signal V than
is detected
by a heat
spectrum
analyzer.
inductance
the original
sink setup.
In consequence,
the1341–1346,
ungrounded
heat
sink,
the
attenuation
of
the
heat
sink
the
ungrounded
heat
sink,
the
attenuation
of
the
heat
sinkon
2001.
From these
RF signals
ofsmaller
with and
without
ferrite frecore
the resonant
frequency
fres2 is
than
the resonant
Brench, C.:
Heatsinkisis
radiation
as ausing
function
of geometry, in: IEEE
radiated
emissions
calculated
radiated
emissions
calculated
using
the ungrounded
heat sink,
the attenuation
thesecond
heat sink
quency
fres1 . The resonant
frequencies
of the firstofand
International
Symposium on Electromagnetic Compatibility, pp.
radiated
emissions
is
calculated
using
A
(dB)
=
V
(dBµV
)
−
V
(dBµV
),
(8)
configuration
are
300
MHz
and
180
MHz,
respectively.
The
hs(dB) =
Ferrite
A105–109,
Vw/o
Vw/
hs
w/oFerrite
Ferrite (dBµV ) −w/
Ferrite (dBµV ), (8)
1994.
capacitances
of
the
third
configuration
(extended
heat
sink
Damiano,
A.,
G., Ferrite
Marongiu,
I.,
A.:
A heat
sink
where
A(dB)
is Gatto,
the
ofofthe
heat
radiated
emisAhs
= Vattenuation
(dBµV
) heat
−Piroddi,
Vsink
(dBµV
), (8)
hs
w/
Ferrite
where
A
attenuation
theand
sink
radiated
emishs is the w/o
model
for
EMI
resonance
frequency
determination,
in:
Power
sions,
V
is
the
received
RF
signal
of
the
heat
sink
w/o
Ferrite
sions,
Vw/o
theConference,
receivedofRF
of the
heat
sink
Ferrite
where
Ahs
is theis attenuation
thesignal
radiated
Electronics
Specialists
vol.
1,heat
pp. sink
273–277,
2004.emiswww.adv-radio-sci.net/9/317/2011/
sions, Vw/o Ferrite is the received RF signal of the heat sink
gro
fere
189
Samir,
by
Pac
Sincla
of E
circ
Con
Tihany
K. E
J. Kulanayagam et al.: Reduction of heat sink common-mode currents
with ferrite core) are the same as first and second configuration. But the inductances of the heat sink are increased by
the ferrite cores. This is because all two ferrite core consist
of different materials and therefore different magnetic characteristics. The quality factor of the system is reduced by the
ferrite core. Furthermore, it depends on the ferrite materials
which is used.
The input impedance of the heat sink is increased because
a ferrite core has a very large impedance due to its permeability. This means that the heat sink currents will be minimized
by adding the ferrite core between the heat sink and power
device and the heat sink radiation is attenuated by this filter.
5
6
321
Conclusions
In this paper a new filter design was introduced. It can
be used to reduce the heat sink common-mode currents in
Switching Mode Power Supply circuits. The validation of
the input impedances of the heat sinks have verified that the
input impedances can be increased at some frequencies up to
45 dB due to the ferrite core between the heat sink and the
power device. Therefore, the heat sink current can be minimized. In addition, the radiated emissions from the heat sink
are attenuated by the filter. Again, our proposed filter design
reduces the heat sink noise currents and attenuates the heat
sink radiated emissions.
Radiation of the heat sink
References
5.1
Experimental setup
In order to validate the attenuation of the heat sink radiated
emissions of the proposed heat sink design an experimental
test was done inside an anechoic chamber to avoid the presence of environmental EMI noises in EMI measurements. A
simplified schematic of the experimental setup is shown in
Fig. 11.
The RF signal generator was connected via a coaxial cable
to the heat sink ferrite design. On the other side a biconical/ logarithmic periodic antenna receives the emitted electromagnetic fields from the heat sink. This received RF signal V is detected by a spectrum analyzer.
From these RF signals of with and without ferrite core on
the ungrounded heat sink, the attenuation of the heat sink
radiated emissions is calculated using
Ahs (dB) = Vw/o Ferrite (dBµV ) − Vw/ Ferrite (dBµV ),
(8)
where Ahs is the attenuation of the heat sink radiated emissions, Vw/o Ferrite is the received RF signal of the heat sink
without ferrite core and Vw/ Ferrite is the received RF signal
of the heat sink with ferrite core. The measurement setups
were configured forfrontal and lateral radiation. In this work,
the frequency range from 20 up to 300 MHz was considered.
5.2
Results and discussions
The attenuation of the heat sink radiation is calculated for
frontal and lateral measurement of the heat sink is illustrated
in Fig. 12. The maximum attenuation value and the average
attenuation value of the frontal measurement of the heat sink
are 25 dB and 15 dB, respectively. Furthermore, the maximum attenuation value and the average attenuation value of
the lateral measurement of the heat sink are 10 dB and 5 dB,
respectively.
www.adv-radio-sci.net/9/317/2011/
Archambeault, B., Pratapneni, S., Zhang, L., and Wittwer, D.:
Comparison of various numerical modeling tools against a standard problem concerning heat sink emissions, IEEE International
Symposium on Electromagnetic Compatibility, 2, 1341–1346,
2001.
Brench, C.: Heatsink radiation as a function of geometry, IEEE International Symposium on Electromagnetic Compatibility, 105–
109, 1994.
Damiano, A., Gatto, G., Marongiu, I., and Piroddi, A.: A heat sink
model for EMI resonance frequency determination, Power Electronics Specialists Conference, 1, 273–277, 2004.
Das, S. and Roy, T.: An investigation on radiated emissions from
heatsinks, IEEE International Symposium on Electromagnetic
Compatibility, 2, 784–789, 1998.
Diepenbrock, J., Archambeault, B., and Hobgood, L.: Improved
grounding method for heat sinks of high speed processors, Electronic Components and Technology Conference, 993–996, 2001.
Felic, G. and Evans, R.: Study of heat sink EMI effects in SMPS circuits, IEEE International Symposium on Electromagnetic Compatibility, 1, 254–259, 2001.
Li, K., Lee, C., Poh, S., Shin, R., and Kong, J.: Application
of FDTD method to analysis of electromagnetic radiation from
VLSI heatsink configurations, IEEE Transactions on Electromagnetic Compatibility, 35, 204–214, 1993.
Nagel, A.: Leitungsgebundene Störungen in der Leistungselektronik: Entstehung, Ausbreitung und Filterung [Dissertation],
Wissenschaftsverlag, Mainz, 1999.
Radhakrishnan, K., Wittwer, D., and Li, Y.-L.: Study of heatsink
grounding schemes for GHz microprocessors, IEEE Conference
on Electrical Performance of Electronic Packaging, 189–192,
2000.
Samir, A. and Fujiwara, O.: Calculation of load effect produced by
ferrite core attached to wire above a ground plane, Asia Pacific
Microwave Conference, 1, 1999.
Sinclair, A., Ferreira, J., and van Wyk, J.: A systematic study of
EMI reduction by physical converter layout and suppressive circuits, International Conference on Industrial Electronics, Control, and Instrumentation, 2, 1059–1064, 1993.
Tihanyi, L.: Electromagnetic Compatibility in Power Electronics, J.
K. Eckert & Company, Inc., Sarasota, Florida, 1995.
Adv. Radio Sci., 9, 317–321, 2011