1
Emi filter design based on the separated electromagnetic interference in switched
2
mode power supplies
3
Samet YALÇIN1,*, Şükrü ÖZEN2, Selçuk HELHEL2
4
1
Electrical and Electronics Engineering Department, Faculty of Technology, Suleyman
5
6
Demirel University, Isparta, Turkey,
2
Electrical and Electronics Engineering Department, Engineering Faculty, Akdeniz
7
University, Antalya, Turkey
8
*Correspondence: sametyalcin@sdu.edu.tr
9
Abstract: Usage of switch mode power supplies (SMPS) has been increased
10
tremendously in recent years since their advantages compared to conventional ones. In
11
spite of their advantages, SMPSs cause conducted and radiated emissions due to the fact
12
that they switch on and off at specific frequencies. Electromagnetic interference (EMI)
13
filters are mostly preferred equipment for the reduction of conducted emission for
14
coupled circuits.
15
Improved EMI filter in the literature to suppress both common and differential mode
16
noises sourced by ATX (a sample of SMPS) power supply has been proposed. A
17
proposed filter was designed by use of AWR Microwave Office and MATLAB. The
18
power supply was tested according to CISPR22 that 30dB worse noise level compared
19
to limit was observed. A new improvement has also been proposed in order to separate
20
resultant EMI into its common and differential modes for proper EMI filter design.
21
Designed filter by considering common and differential modes suppress those noises by
22
37dB.
23
Key words: Conducted Emission, Electromagnetic Compatibility, Electromagnetic
24
Interference (EMI) Filter, Noise Separator, Switch Mode Power Supply (SMPS)
1
1
1.
2
Switching mode power supplies (SMPS) have found wide range of application area,
3
since 1960. SMPSs have both low power dissipation and stepped voltage outputs that
4
they became basic part of electric and electronic devices [1].
5
In spite of advantages it is known that SMPSs cause conducted and radiated emission at
6
switching frequencies [2-7]. Also Shin [4] reported that conducted and radiated
7
emissions can result in system malfunctions and electromagnetic interference (EMI) at
8
switching operation. Radiated emissions usually appear above 30 MHz and reasons of
9
them are circuit structure and/or cables. Conducted emissions that are main noises
Introduction
10
appear between 150 kHz and 30 MHz [8].
11
Issues related to analysis and attenuation of electromagnetic interference (EMI) have
12
been addressed on lots of platforms [9-11]. Electromagnetic shielding (ES) is most
13
commonly used basic method to attenuate the radiated emission [12,13], but it is not
14
sufficient for attenuation of conducted emission. In order to suppress the conducted
15
emission, noise components need be separated first. Separation of components is
16
followed by grounding, shielding or filtering methods [14].
17
Measurements show that ATX SMPS units present on computers are emitting above the
18
limits of CISPR22 that this is the major part of computer related conduction emissions.
19
This study proposes a new filter to reduce SMPS based electromagnetic interference,
20
and it needs to be used between main and ATX SMPS units.
21
22
1.1.
23
Conducted EMI consists of two components called common mode (CM) noise and
24
differential mode (DM) noise according to current direction (Figure 1). Common mode
Components of conducted emission
2
1
noise is the sum of currents that flow from both phase and neutral to ground. The reason
2
of that noise is inductive impulses during switching on/off or that consist from filter
3
capacitors and mosfet’s heatsinks’s capacitive impacts. Differential mode noise is the
4
current that results from the difference between phase and neutral voltages. Origin of
5
differential mode is an interaction between circuit components [15-17].
6
Voltages (vcm) and currents (icm) are calculated for common mode noise as indicated in
7
Eq.1 and Eq.2.
vcm 
8
2
(1)
icm  i p  in
9
10
11
v p  vn
(2)
Voltages (vdm) and currents (idm) of differential mode noise described as indicated in
Eq.1 and Eq.2 and Figure 1.
12
vdm  v p  vn
13
idm 
i p  in
(3)
(4)
2
14
15
Figure 1.Conducted emission components
16
17
During separation (measurement) 50 Ω loads of line impedance stabilization network
18
(LISN) are connected as parallel in common mode and as serial in differential mode.
19
Thus Equation (1) and (3) produce Eq.5, Equation (2) and (4) produce Eq.6 [18,19].
20
21
vcm 
v p  vn
2

50  i p  50  in
2

50
 (i p  in )  25  icm
2
vdm  v p  vn  50   i p  50   in  2  50   (i p  in )  100   idm
(5)
(6)
22
vp is phase voltage, vn is neutral voltage, ip is phase current and in is neutral current. Eq.
23
5 and Eq. 6 show the equation between CM-DM noise voltages and currents. CM noise
3
1
currents flow in the same direction. Thus as shown in Eq. 5 LISN’s resistors are
2
considered as parallel. Conversely DM noise currents flow in the opposite direction. For
3
that reason LISN’s resistors are considered as serial in Eq.6. Conducted emission is
4
assayed by military or civil standards. If the noise exceeds the limits it will have to be
5
separated noise components by current probes or separation methods. The importance of
6
noise separation is underlined by literature to achieve an appropriate filter design [20-
7
23]. Obtained noise components allow us to determine EMI filter components both for
8
common and differential mode.
9
2.
Conducted emission measurements
10
2.1.
CISPR 22 measurement
11
CISPR 22 is adresed to determine conducted emission of device under test (DUT)
12
between 150 kHz and 30 MHz under certain conditions. This standard divides
13
equipment, devices and apparatus into two classes. Class A equipments are not intended
14
to be used in the domestic environment, and Class B equipment are intended to be used
15
in the domestic environment. In this paper, chosen SMPS under test was assumed as in
16
Class B [24].
17
Figure 2 is a screen shot of spectrum analyzer [25] that shows CISPR 22 limits as well
18
as both conducted emissions come from phase and neutral lines. Observed signals need
19
to be separated in order to suppress exceeded levels.
20
21
Figure 2: Conducted emission results of (a) phase and (b) neutral lines
22
23
2.2.
Conducted Noise Separation
4
1
There are few methods to determine both common and differential mode noise
2
components emitted by device under test (DUT) by measuring voltage [21, 26-28]. One
3
of these methods is the noise separator circuit that offered by Shou Wang in 2005 [26].
4
Resistance of 50Ω presents in the circuit given in Figure 3 has parallel connection for
5
common mode and serial connection for differential mode such that those connections
6
satisfy resistance response of HEDD for either common or differential modes. Binding
7
of T1 transformers’ windings allow us to block parasitic capacitances that may appear
8
between them. These aforementioned advantages are new capabilities according to
9
previous designs present in the literature. The results of offered design are here: -50 dB
10
for S11, -50 to -90 dB for S21, constant -6 dB for S31 and S41 (Figure 4).
11
12
Figure 3: Scheme of noise separator
13
Figure 4: Noise separator schematic results
14
15
In order to improve previously proposed circuit, larger ground and narrower gap
16
between the circuit components have been applied. Making circuit’s paths’ shape as
17
curvature instead of sharp is another solution that we also applied. S-parameters of
18
improved noise separator were measured by network analyzer [29].
19
Figure 5 indicates both Shou Wang’s results and proposed circuits’ response. Instead
20
emitted noise from DUT exceeded the limits between 150 kHz and 4 MHz bands,
21
proposed circuit has a capability to suppress them at the same band interval.
22
23
Figure 5: (a) Designed and (b) improved noise separator’s results
24
5
1
3.
2
Figure 6 shows both common and differential modes separated from emitted noise by
3
SMPS exceeded limits. The common mode noise points are 20.7dBµV, 27.55 dBµV
4
and 27.75 dBµV at 168 kHz, 235 kHz and 305 kHz. The differential mode noise points
5
are 22.26 dBµVand 30.55 dBµV at 202 kHz and 404 kHz.
EMI Filter Design
6
Figure 6.SMPS’s a) common mode and b) differential mode noises
7
8
9
3.1.
Determination of the cut-off frequencies
10
Selection of 2-level filter structure to have 40 dB/decade suppression ratios has
11
followed by determination of cutoff frequencies to suppress noises to aimed level.
12
Moreover noise attenuation must be 6 dB much more than exceeded noise given at
13
Eq.7.
(Vatt ) dB  (Vexc ) dB  6dB
14
(7)
15
Vatt is noise attenuation, Vaxc is exceeded noise and the 6 dB is correction factor [30].
16
The correction factor is added to exceeded noise in order to ensure that noise is fully
17
suppressed.
18
By using Equation (7) suppression values are 26.7 dBµV, 33.55 dBµV and 33.75 dBµV
19
at 168 kHz, 235 kHz and 305 kHz for common mode, and 28.26 dBµV and 36.55 dBµV
20
at 202 kHz and 404 kHz for differential mode. Cutoff frequencies that would allow us
21
to suppress noises were calculated by using MATLAB, and it is presented in Figure 7.
22
34 kHz for common mode and 39 kHz for differential modes are determined as cut-off
23
frequencies, respectively.
24
6
1
Figure 7. a) Common mode and b) differential mode cut-off frequencies
2
3
3.2.
Calculation of EMI filter components’ values
4
3.3.
It is well known in EMC point of view that a capacitor in any filter may behave
5
parasitic inductor and an inductor in any circuit may behave parasitic capacitor at
6
certain cases. For that reason, a low pass filter may behave as a band reject filter in
7
some cases. This behavior forces engineers to determine first cut-off frequency, middle
8
frequency and second cut-off frequency of that filter [31]. Second cut-off frequency is
9
usually about 20 MHz. The noise to be suppressed ends at 4 MHz in this paper. Thus
10
the middle frequency and the second cut-off frequency might be ignored for designed
11
EMI filter that we ignored.
12
The designed EMI filter’s differential mode equivalent circuit was shown in Figure 8.
13
The values of those circuit components are calculated using Equation (8). Values of C x
14
and Cy are selected 4,7nF to prevent leakage currents more efficiently [31].
15
16
Figure 8.Filter’s differential mode equivalent circuit
17
18
f c , DM 
[Hz]
1
2 2 LDM  (C X 
19
20
21
CY
)
2
fc,DM is differential mode cut-off frequency and calculated as 39 kHz, Cx and Cy are 4,7
nF. So if LDM is taken from the Eq.8;
LDM 
1
[H]
3
2  (2  39  10 )  4,7  10 
2
3 2
22
(8)
9
LDM (differential mode inductor) is 1200 µH.
7
1
The designed EMI filter’s common mode equivalent circuit is shown in Figure 9. LCM
2
value of the circuit is calculated using Equation (9).
3
4
Figure 9.Filter’s common mode equivalent circuit
5
6
1
f c ,CM 
2 ( LCM
[Hz]
(9)
L
 DM )  2CY
2
7
According to (9) equation fc,CM (common mode cut-off frequency) is calculated as 34
8
kHz, LDM is differential mode inductor (1200 µH) and LCM is common mode choke. So
9
if LDM is taken from the Eq.9;
10
11
LCM  (
1
1,2 10 3

)
3 2
9
2
(2  34 10 )  9,4 10
[H]
LCM is 1800 µH. So structure of the EMI filter is shown in Figure 10.
12
13
Figure 10.Structure of the EMI filter
14
15
Interaction between different circuit levels designed for different noise components has
16
been shown. It has to be noted that DUT connected to filter’s output has high
17
impedance and meanwhile LISN connected to filter has low impedance level. So when
18
the filter is designed, capacitors and then inductors are placed for impedance matching.
19
The CM and DM equivalent circuits of filter are shown in Figure 11 with determined
20
components simulated at Microwave Office.
21
22
Figure 11. (a) Differential mode and (b) common mode equivalent circuits simulation of
23
EMI filter
8
1
4.
2
A response of the filter designed to suppress CM and DM noises produced by ATX
3
power supply was performed and analyzed at Microwave Office platform. Besides the
4
filter’s common mode and differential mode simulation, s-parameter results are
5
indicated in Figure 12 and in Figure 13, respectively. According to the results, the EMI
6
filter is able to suppress common mode noises by 28.56, 32.54 and 35.46dBµV at 168,
7
235 and 305 kHz, and able to suppress differential mode noises by 30.46 and 38.25
8
dBµV at 202 and 404 kHz.
Experimental results
9
10
Figure 12. Common mode simulation frequency response of EMI filter
11
Figure 13. Differential mode simulation frequency response of EMI filter
12
13
5.
14
In this paper, newly designed EMI filter based on separated noise components for
15
common and differential mode interference of ATX power supply (a sample of SMPS)
16
has been proposed. For this purpose, noise separator circuit previously designed by
17
Shou Wang in 2005 has been improved. Both S11 and S21 for improved noise separator
18
vary between -46dB and -37dB at the bandwidth where the noise exceeds the limits.
19
The performance of the designed filter has been proved to be sufficient to reduce the
20
interference sourced by CM and DM, and equivalent circuit simulations were carried
21
out on Microwave Office and MATLAB. During the improvement of noise separator
22
design, the effect of both LISN and SMPSs on filter design has also been discussed. It
23
was shown that proposed improvement is good enough to suppress CM and DM noises
24
sourced by SMPSs type ATX power supplies. Comparison of active and passive EMI
Conclusion
9
1
filter performances on SMPSs type ATX power supply based interference suppression
2
is our future goal.
3
ATX SMPS units present on computers are emitting above the limits of CISPR22, and
4
this is the major part of computer emissions. We may certainly conclude that proposed
5
filters need to be used between main and computer SMPS units to get rid out of
6
electromagnetic interference.
7
8
Acknowledgement
9
This study was supported by Akdeniz University Scientific Research Supporting Unit
10
(BAP), and measurements were carried out at EMUMAM EMC Laboratories (Akdeniz
11
University) granted by State Planning Organization (2007K120530-DPT). We also
12
would like to thank to Asst. Prof. Evren EKMEKÇİ for his valuable contributions and
13
measurement support.
14
References
15
[1]
16
USA: Elsevier, 2012.
17
[2]
18
Bifurcations, chaos, control, and applications: Wiley-IEEE Press, 2001.
19
[3]
20
(SMPS). IEE conference publication; 1999; Institution of Electrical Engineers: 180-185.
21
[4]
22
Guide of Active EMI Filter in a Compact Package for Reduction of Common-Mode
23
Conducted Emissions. IEEE Transactions on Electromagnetic Compatibility 2015; 57:
24
660-671.
Brown MC. Practical switching power supply design. San Diago, California,
Banerjee S, Verghese GC. Nonlinear phenomena in power electronics:
Nagrial M, Hellany A. EMI/EMC issues in switch mode power supplies
Shin D, Kim S, Jeong G, Park J, Park J, Han KJ, Kim J. Analysis and Design
10
1
[5]
2
from near field measurements for EMI evaluation of switch mode power supplies
3
components by method of moments. 4th Power Electronics, Drive Systems and
4
Technologies Conference (PEDSTC), 2013: 421-425
5
[6]
6
reducing conducted emission in switched mode power supplies. IEEE Electromagnetic
7
Compatibility Magazine 2013; 2: 67-71.
8
[7]
9
DC-DC converter using chaos. 2011 International Conference on Emerging Trends in
Mirtalaei S. M. M, Sadeghi S. H. H, Moini R. Radiated emission determination
Premalatha L, Raghavendiran T. A. A simple approach of generating CPWM for
Premalatha L, Ravichandran C. Reduction of electromagnetic interference in
10
Electrical and Computer Technology; 2011: 236 - 239.
11
[8]
12
switched
13
Electromagnetic Compatibility; 2009: 241 - 244.
14
[9]
Ozen S, Arı N. Elektromanyetik Uyumluluk. Ankara: Palme Yayıncılık, 2008.
15
[10]
Çömlekçi S, Özen Ş, Köklükaya E. Optimization of resistive loading of
16
EMI/EMC near field probe. Robot Auton Syst. 2004;49:245-51.
17
[11]
18
Electromagnetic Interference Emitted from Compact Fluorescent Lamps According to
19
CISPR-15. The 36th Progress In Electromagnetics Research Symposium in Prague:
20
PIERS Proceedings; 2015: 1133 - 1136
21
[12]
22
Simulation Analysis of Metalic Enclosures with Apertures.
Chen H, Cheng L, Qiu X, Zhao Y. Conductive EMI noise measurement for
reluctance
drive.
20th
IEEE
International
Zurich
Symposium
on
Kocakusak A, Cakır M, Yalcın S, Ozen S, Helhel S. Evaluation of
Basyigit IB, Caglar MF, Helhel S. Magnetic Shielding Effectiveness and
9th International
11
1
Conference on Electrical and Electronics Engineering; Bursa, TURKEY: ELECO;
2
2015: 1-5.
3
[13]
4
Aperture Wıdth Ratio On Broadband Shielding Effectiveness.
5
URSI; 2011: 1-4.
6
[14]
7
challenges and perspectives. Turk J Elec Eng & Comp Sci 2009; 17: 273-278.
8
[15]
Ozenbaugh RL, Pullen TM. EMI filter design: CRC Press, 2011.
9
[16]
Tihanyi L. EMC in power electronics: Newnes, 1995.
10
[17]
Sánchez A. M, Pérez A, Regué JR, Ribó M, Cepeda RP, Pajares FJ. A Modal
11
Model of Common-Mode Chokes for Conducted Interference Prediction. IEEE
12
Transactions on Electromagnetic Compatibility 2010; 52: 749-752.
13
[18]
14
scattering parameters: Theory and simulation. IEEE Transactions on Microwave Theory
15
and Techniques 1995; 43:1530-9.
16
[19]
17
Impedance of Common-Mode and Differential-Mode Noise Separators. IEEE
18
Transactions on Industry Applications 2015; 51: 2352-2360.
19
[20]
20
differential mode noise: design and realization. Seventeenth Annual IEEE Applied
21
Power Electronics Conference and Exposition 2002; 1: 100-105.
22
[21]
23
mode-conducted EMI noise. IEEE Transactions on Power Electronics 1996; 11:480-
24
488.
Basyigit IB, Tosun PD, Özen S, Helhel S. An Affect Of The Aperture Length To
URSI GASS 2011:
Sevgi L. Electromagnetic compatibility engineering education: problems,
Bockelman DE, Eisenstadt WR. Combined differential and common-mode
Kostov K. S, Schroth S, Krismer F, Priecinsky M, Nee HP, Kolar JW. The Input
Caponet MC, Profumo F. Devices for the separation of the common and
Guo T, Chen DY, Lee FC. Separation of the common-mode-and differential-
12
1
[22]
2
separator with impedance matching approach. IEEE Applied Power Electronics
3
Conference and Exposition - APEC 2014. 2014: 3392-3396.
4
[23]
5
2006.
6
[24]
7
characteristics-Limits and methods of measurement. CISPR 22. 1997.
8
[25]
Techologies K. ESA-E Series Spectrum Analyzer. USA2014.
9
[26]
Wang S, Lee FC, Odendaal WG. Characterization, evaluation, and design of
Chang X, Chen W, Yang X. Performance improvement of CM/DM noise
Paul CR. Introduction to Electromagnetic Compatibility: John Wiley & Sons,
Commission
IE.
Information
technology
equipment-Radio
disturbance
10
noise separator for conducted EMI noise diagnosis. IEEE Transactions on Power
11
Electronics 2005; 20: 974-82.
12
[27]
13
differential mode noise separation: comparison of two different approaches. IEEE 32nd
14
Annual Power Electronics Specialists Conference 2001; 3: 1383-1388.
15
[28]
16
Industry Applications Society Annual Meeting (IAS), IEEE; 2006; 3: 1209-1214.
17
[29]
Technologies K. FieldFox Handheld Analyzers. USA2014.
18
[30]
Lu Y-Y, Chen Z-H, Huang H-C. The System for the Reduction of Conducted
19
Electromagnetic Interference Emission from Switching Power Supply. Third
20
International Conference on Intelligent Information Hiding and Multimedia Signal
21
Processing, IEEE; 2007: 2: 397-402.
22
[31]
23
the passive components. 15th IEEE Workshop on Signal Propagation on Interconnects.
24
IEEE; 2011: 143-146.
Caponet MC, Profumo F, Ferraris L, Bertoz A, Marzella D. Common and
Chen P-S, Lai Y-S. Software-based separation of conductive EMI signals., 41st
Kotny J, Duquesne T, Idir N. EMI Filter design using high frequency models of
13
1
Figures and Tables
2
DUT
LISN
ip
vdm
50 Ω
50 Ω
3
4
vcm
icm,p
idm
Zcm
in neutral
icm,n idm
icm
icm
DM Noise
Supply
CM Noise
Supply
phase
idm
Zdm
Vdm
Vcm
ground
Figure 1.Conducted emission components
5
6
(a)
7
8
(b)
9
Figure 2: Conducted emission results of (a) phase and (b) neutral lines
10
14
MUC2
ID=M1
L1=780 uH
R1=0.32 Ohm
L2=780 uH
R2=0.32 Ohm
K1_2=1
PORT
P=1
Z=50 Ohm
1
PORT
P=2
Z=50 Ohm
RES
ID=R1
R=50 Ohm
2
1
2
MUC2
ID=M2
L1=780 uH
R1=0.32 Ohm
L2=780 uH
R2=0.32 Ohm
K1_2=1
3
1
4
2
1
2
RES
ID=R2
R=50 Ohm
PORT
P=4
Z=50 Ohm
3
4
PORT
P=3
Z=50 Ohm
1
2
Figure 3: Scheme of noise separator
3
4
5
Figure 4: noise separator schematic results
6
15
1
2
(a)
0
-10
s11
s21
s31
s41
X: 9.596e+005
Y: -6.086
-20
dB
-30
-40
-50
X: 2.156e+006
Y: -43.94
X: 9.596e+005
Y: -46.13
-60
-70
-80
6
7
10
10
Frequency (Hz)
3
4
(b)
5
Figure 5: (a) Designed and (b) improved noise separator’s results
6
16
50
45
40
35
X: 2.35e+005
Y: 27.55
dBuV
30
X: 3.05e+005
Y: 27.75
25
20
X: 1.68e+005
Y: 20.7
15
10
5
0
5
10
1
2
6
10
Frequency (Hz)
(a)
50
45
40
35
X: 4.04e+005
Y: 30.55
dBuV
30
X: 2.02e+005
25 Y: 22.26
20
15
10
5
0
5
10
3
6
10
Frequency (Hz)
4
(b)
5
Figure 6.SMPS’s a) common mode and b) differential mode noises
6
17
50
45
40
35
dBuV
30
25
20
15
10
5
0
4
10
5
6
10
10
Frequency (Hz)
1
2
(a)
50
45
40
35
dBuV
30
25
20
15
10
5
0
4
10
5
6
10
10
Frequency (Hz)
3
4
(b)
5
Figure 7. a) Common mode and b) differential mode cut-off frequencies
6
phase
phase
2LDM
neutral
LLEAKAGE
CY/2 CX
neutral
7
8
Figure 8.Filter’s differential mode equivalent circuit
18
1
phase neutral
LCM
phase neutral
LDM/2
2CY
ground
ground
2
Figure 9.Filter’s common mode equivalent circuit
3
4
EMI Filter
phase
LISN
neutral
1200
µH
1800
µH
phase
4.7
nF
4.7
nF
1200
µH
TAC
neutral
4.7
nF
ground
ground
5
6
Figure 10.Structure of the EMI filter
7
8
9
(a)
19
1
2
(b)
3
Figure 11. (a) Differential mode and (b) common mode equivalent circuits simulation of
4
EMI filter
5
6
7
Figure 12. Common mode simulation frequency response of EMI filter
8
20
1
2
Figure 13. Differential mode simulation frequency response of EMI filter
3
21