EMC Filter Analysis

Dr David Pommerenke with the help of many members of the EMC laboratory

November 2018

1

2

EMC laboratory Missouri SampT

MMiissssoouurrii SSampampTT EEMMCC LLaabboorraattoorryy

St LouisKansas City

Memphis

Missouri SampT13

13

EMC Laboratory13

13

13

The group

6 Tenured faculty 8 Research faculty 4 Adjunct faculty 6 Visiting scholars Laboratory technician

gt40 studentsndash 11 PhDndash 31 MSndash 4 undergraduate

2 Administrative support

3

May 2016 EMC Center Meeting

Research partners from industry and government

Most projects are in close cooperation with industry

4

bull Intel Corporation (Total 19 9-PhD10-MS)ndash Oleg Kashurkin MSEE 2016ndash Ben Orr PhD 2016ndash Wei Qian MSEE 2016ndash Xinyao Guo MSEE 2016ndash Jayong Kool PhD 2008ndash Vijay Kasturi MSEE 2007ndash Juan Chen MSEE 1999 (Princ

Engin)bull Altera (Total 5 4-PhD1-MS)

ndash Xinyun Guo MSEE 2016ndash Jianmin Zhang PhD 2007ndash Shishuang Sun PhD 2006ndash Kundan Chand MSEE 2006

bull Apple Computer (Total 20 10-PhD10-MS)

ndash Abhishek Patnaik PdD 2018ndash Satyajeet Shinde PhD 2017ndash Kyoungchoul Koo PostDoc 2017ndash Guangyao Shen PhD 2016ndash Guanghua Li PhD 2015ndash Xu Gao PhD 2014ndash Tianqi Li PhD 2014ndash Andriy Radchenko PhD 2014ndash Kam Keong PhD 2012ndash Frederico Centola MSEE 2003

bull NVidiandash Aukit Bharagava MSEE 2010ndash Fan-Keung Chang MSEE 2006ndash Chen Wang PhD 2004

bull Cisco (Total 13 3-PhD10-MS)ndash Kartheek Nalla MSEE 2016ndash Chunchun Sui PhD 2015ndash Natalia Bondarenko PhD 2015ndash Xiangyang Jiao MSEE 2015ndash Soumya De PhD 2012ndash Amendra Koul MSEE 2010

bull Motorolandash Bivin Varghese MSEE 2003ndash Prem Ganeshan MSEE 1998

bull Panasonicndash Xiao Luo MSEE 1999ndash Janet Luo MSEE 1998

bull Dell Computerndash Arun Chada PhD 2014ndash Lauren Zhang MSEE 1999

bull Googlendash Suyu Yang MSEE 2016ndash Jingnan Pan PhD 2015ndash Ketan Shringarpure PhD 2014ndash Weifeng Pan PhD 2008ndash Sandeep Chandra MSEE 2007

bull Sandia National Laboratoryndash Ben Conley MSEE 2015ndash Matthew Halligan PhD 2014ndash Matt Schepers MSEE 2010ndash David Carter MSEE 2004

And other companieshttpwwwemclabmstedustudfacalumniht

ml

Where are Our Graduates 105 MS 53 PhD 12 Postdocs

bull Broadcom Corporationndash Hongyu Li PhD 2012ndash Liehui Ren PhD 2011

bull Agilent Technologies ndash Joe Bishop MS 2011ndash Kuifeng Hu PhD 2007

bull Juniper Networksndash Qian Liu MSEE 2016ndash Hui He MSEE 2015

bull Mentor Graphics Corpndash Aleksandr Gafarov MSEE

2012ndash Praveen Anmula MSEE

2009bull IBM Systems

ndash Matteo Cocchini MSEE 2008

ndash Michael Cracraft PhD 2007ndash Xiaoxiong Gu MSEE 2002

bull Research in Motionndash Gang Feng PhD 2010ndash Qing Cai MSEE 2007

bull Lockheed Martinndash Jim Feucht MSEE 2010

bull Amazonndash Liang Li MSEE 2016

bull Semtech Corpndash Li Zhen MSEE 2011

bull Texas Instrumentsndash Haixiao Weng PhD 2006

6

Introduction EMC Filters in Power Electronics

12V network

DCDC converter

ACDC charger

Charging station

E-motor PE module HV battery pack

Low- current HV components

LVDC

3-PHASEAC HVDC+

HVDC-

FILTERS

Inverter

EMC Laboratory MSampT

Structure of the talk

Different solution path Circuit Circuit with parasitics full wave Film Capacitor modeling SMT Capacitor modeling CMC modeling Material property extraction Examples of properties of magnetic materials used in EMC

7

One capacitor

8

log(f)

log(|Z|)

log(f)

log(|Z|)

1jwC

1jwC1jwL

ESR

With increased modeling requirement more and more EM effects need to be considered

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

2

EMC laboratory Missouri SampT

MMiissssoouurrii SSampampTT EEMMCC LLaabboorraattoorryy

St LouisKansas City

Memphis

Missouri SampT13

13

EMC Laboratory13

13

13

The group

6 Tenured faculty 8 Research faculty 4 Adjunct faculty 6 Visiting scholars Laboratory technician

gt40 studentsndash 11 PhDndash 31 MSndash 4 undergraduate

2 Administrative support

3

May 2016 EMC Center Meeting

Research partners from industry and government

Most projects are in close cooperation with industry

4

bull Intel Corporation (Total 19 9-PhD10-MS)ndash Oleg Kashurkin MSEE 2016ndash Ben Orr PhD 2016ndash Wei Qian MSEE 2016ndash Xinyao Guo MSEE 2016ndash Jayong Kool PhD 2008ndash Vijay Kasturi MSEE 2007ndash Juan Chen MSEE 1999 (Princ

Engin)bull Altera (Total 5 4-PhD1-MS)

ndash Xinyun Guo MSEE 2016ndash Jianmin Zhang PhD 2007ndash Shishuang Sun PhD 2006ndash Kundan Chand MSEE 2006

bull Apple Computer (Total 20 10-PhD10-MS)

ndash Abhishek Patnaik PdD 2018ndash Satyajeet Shinde PhD 2017ndash Kyoungchoul Koo PostDoc 2017ndash Guangyao Shen PhD 2016ndash Guanghua Li PhD 2015ndash Xu Gao PhD 2014ndash Tianqi Li PhD 2014ndash Andriy Radchenko PhD 2014ndash Kam Keong PhD 2012ndash Frederico Centola MSEE 2003

bull NVidiandash Aukit Bharagava MSEE 2010ndash Fan-Keung Chang MSEE 2006ndash Chen Wang PhD 2004

bull Cisco (Total 13 3-PhD10-MS)ndash Kartheek Nalla MSEE 2016ndash Chunchun Sui PhD 2015ndash Natalia Bondarenko PhD 2015ndash Xiangyang Jiao MSEE 2015ndash Soumya De PhD 2012ndash Amendra Koul MSEE 2010

bull Motorolandash Bivin Varghese MSEE 2003ndash Prem Ganeshan MSEE 1998

bull Panasonicndash Xiao Luo MSEE 1999ndash Janet Luo MSEE 1998

bull Dell Computerndash Arun Chada PhD 2014ndash Lauren Zhang MSEE 1999

bull Googlendash Suyu Yang MSEE 2016ndash Jingnan Pan PhD 2015ndash Ketan Shringarpure PhD 2014ndash Weifeng Pan PhD 2008ndash Sandeep Chandra MSEE 2007

bull Sandia National Laboratoryndash Ben Conley MSEE 2015ndash Matthew Halligan PhD 2014ndash Matt Schepers MSEE 2010ndash David Carter MSEE 2004

And other companieshttpwwwemclabmstedustudfacalumniht

ml

Where are Our Graduates 105 MS 53 PhD 12 Postdocs

bull Broadcom Corporationndash Hongyu Li PhD 2012ndash Liehui Ren PhD 2011

bull Agilent Technologies ndash Joe Bishop MS 2011ndash Kuifeng Hu PhD 2007

bull Juniper Networksndash Qian Liu MSEE 2016ndash Hui He MSEE 2015

bull Mentor Graphics Corpndash Aleksandr Gafarov MSEE

2012ndash Praveen Anmula MSEE

2009bull IBM Systems

ndash Matteo Cocchini MSEE 2008

ndash Michael Cracraft PhD 2007ndash Xiaoxiong Gu MSEE 2002

bull Research in Motionndash Gang Feng PhD 2010ndash Qing Cai MSEE 2007

bull Lockheed Martinndash Jim Feucht MSEE 2010

bull Amazonndash Liang Li MSEE 2016

bull Semtech Corpndash Li Zhen MSEE 2011

bull Texas Instrumentsndash Haixiao Weng PhD 2006

6

Introduction EMC Filters in Power Electronics

12V network

DCDC converter

ACDC charger

Charging station

E-motor PE module HV battery pack

Low- current HV components

LVDC

3-PHASEAC HVDC+

HVDC-

FILTERS

Inverter

EMC Laboratory MSampT

Structure of the talk

Different solution path Circuit Circuit with parasitics full wave Film Capacitor modeling SMT Capacitor modeling CMC modeling Material property extraction Examples of properties of magnetic materials used in EMC

7

One capacitor

8

log(f)

log(|Z|)

log(f)

log(|Z|)

1jwC

1jwC1jwL

ESR

With increased modeling requirement more and more EM effects need to be considered

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

The group

6 Tenured faculty 8 Research faculty 4 Adjunct faculty 6 Visiting scholars Laboratory technician

gt40 studentsndash 11 PhDndash 31 MSndash 4 undergraduate

2 Administrative support

3

May 2016 EMC Center Meeting

Research partners from industry and government

Most projects are in close cooperation with industry

4

bull Intel Corporation (Total 19 9-PhD10-MS)ndash Oleg Kashurkin MSEE 2016ndash Ben Orr PhD 2016ndash Wei Qian MSEE 2016ndash Xinyao Guo MSEE 2016ndash Jayong Kool PhD 2008ndash Vijay Kasturi MSEE 2007ndash Juan Chen MSEE 1999 (Princ

Engin)bull Altera (Total 5 4-PhD1-MS)

ndash Xinyun Guo MSEE 2016ndash Jianmin Zhang PhD 2007ndash Shishuang Sun PhD 2006ndash Kundan Chand MSEE 2006

bull Apple Computer (Total 20 10-PhD10-MS)

ndash Abhishek Patnaik PdD 2018ndash Satyajeet Shinde PhD 2017ndash Kyoungchoul Koo PostDoc 2017ndash Guangyao Shen PhD 2016ndash Guanghua Li PhD 2015ndash Xu Gao PhD 2014ndash Tianqi Li PhD 2014ndash Andriy Radchenko PhD 2014ndash Kam Keong PhD 2012ndash Frederico Centola MSEE 2003

bull NVidiandash Aukit Bharagava MSEE 2010ndash Fan-Keung Chang MSEE 2006ndash Chen Wang PhD 2004

bull Cisco (Total 13 3-PhD10-MS)ndash Kartheek Nalla MSEE 2016ndash Chunchun Sui PhD 2015ndash Natalia Bondarenko PhD 2015ndash Xiangyang Jiao MSEE 2015ndash Soumya De PhD 2012ndash Amendra Koul MSEE 2010

bull Motorolandash Bivin Varghese MSEE 2003ndash Prem Ganeshan MSEE 1998

bull Panasonicndash Xiao Luo MSEE 1999ndash Janet Luo MSEE 1998

bull Dell Computerndash Arun Chada PhD 2014ndash Lauren Zhang MSEE 1999

bull Googlendash Suyu Yang MSEE 2016ndash Jingnan Pan PhD 2015ndash Ketan Shringarpure PhD 2014ndash Weifeng Pan PhD 2008ndash Sandeep Chandra MSEE 2007

bull Sandia National Laboratoryndash Ben Conley MSEE 2015ndash Matthew Halligan PhD 2014ndash Matt Schepers MSEE 2010ndash David Carter MSEE 2004

And other companieshttpwwwemclabmstedustudfacalumniht

ml

Where are Our Graduates 105 MS 53 PhD 12 Postdocs

bull Broadcom Corporationndash Hongyu Li PhD 2012ndash Liehui Ren PhD 2011

bull Agilent Technologies ndash Joe Bishop MS 2011ndash Kuifeng Hu PhD 2007

bull Juniper Networksndash Qian Liu MSEE 2016ndash Hui He MSEE 2015

bull Mentor Graphics Corpndash Aleksandr Gafarov MSEE

2012ndash Praveen Anmula MSEE

2009bull IBM Systems

ndash Matteo Cocchini MSEE 2008

ndash Michael Cracraft PhD 2007ndash Xiaoxiong Gu MSEE 2002

bull Research in Motionndash Gang Feng PhD 2010ndash Qing Cai MSEE 2007

bull Lockheed Martinndash Jim Feucht MSEE 2010

bull Amazonndash Liang Li MSEE 2016

bull Semtech Corpndash Li Zhen MSEE 2011

bull Texas Instrumentsndash Haixiao Weng PhD 2006

6

Introduction EMC Filters in Power Electronics

12V network

DCDC converter

ACDC charger

Charging station

E-motor PE module HV battery pack

Low- current HV components

LVDC

3-PHASEAC HVDC+

HVDC-

FILTERS

Inverter

EMC Laboratory MSampT

Structure of the talk

Different solution path Circuit Circuit with parasitics full wave Film Capacitor modeling SMT Capacitor modeling CMC modeling Material property extraction Examples of properties of magnetic materials used in EMC

7

One capacitor

8

log(f)

log(|Z|)

log(f)

log(|Z|)

1jwC

1jwC1jwL

ESR

With increased modeling requirement more and more EM effects need to be considered

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Research partners from industry and government

Most projects are in close cooperation with industry

4

bull Intel Corporation (Total 19 9-PhD10-MS)ndash Oleg Kashurkin MSEE 2016ndash Ben Orr PhD 2016ndash Wei Qian MSEE 2016ndash Xinyao Guo MSEE 2016ndash Jayong Kool PhD 2008ndash Vijay Kasturi MSEE 2007ndash Juan Chen MSEE 1999 (Princ

Engin)bull Altera (Total 5 4-PhD1-MS)

ndash Xinyun Guo MSEE 2016ndash Jianmin Zhang PhD 2007ndash Shishuang Sun PhD 2006ndash Kundan Chand MSEE 2006

bull Apple Computer (Total 20 10-PhD10-MS)

ndash Abhishek Patnaik PdD 2018ndash Satyajeet Shinde PhD 2017ndash Kyoungchoul Koo PostDoc 2017ndash Guangyao Shen PhD 2016ndash Guanghua Li PhD 2015ndash Xu Gao PhD 2014ndash Tianqi Li PhD 2014ndash Andriy Radchenko PhD 2014ndash Kam Keong PhD 2012ndash Frederico Centola MSEE 2003

bull NVidiandash Aukit Bharagava MSEE 2010ndash Fan-Keung Chang MSEE 2006ndash Chen Wang PhD 2004

bull Cisco (Total 13 3-PhD10-MS)ndash Kartheek Nalla MSEE 2016ndash Chunchun Sui PhD 2015ndash Natalia Bondarenko PhD 2015ndash Xiangyang Jiao MSEE 2015ndash Soumya De PhD 2012ndash Amendra Koul MSEE 2010

bull Motorolandash Bivin Varghese MSEE 2003ndash Prem Ganeshan MSEE 1998

bull Panasonicndash Xiao Luo MSEE 1999ndash Janet Luo MSEE 1998

bull Dell Computerndash Arun Chada PhD 2014ndash Lauren Zhang MSEE 1999

bull Googlendash Suyu Yang MSEE 2016ndash Jingnan Pan PhD 2015ndash Ketan Shringarpure PhD 2014ndash Weifeng Pan PhD 2008ndash Sandeep Chandra MSEE 2007

bull Sandia National Laboratoryndash Ben Conley MSEE 2015ndash Matthew Halligan PhD 2014ndash Matt Schepers MSEE 2010ndash David Carter MSEE 2004

And other companieshttpwwwemclabmstedustudfacalumniht

ml

Where are Our Graduates 105 MS 53 PhD 12 Postdocs

bull Broadcom Corporationndash Hongyu Li PhD 2012ndash Liehui Ren PhD 2011

bull Agilent Technologies ndash Joe Bishop MS 2011ndash Kuifeng Hu PhD 2007

bull Juniper Networksndash Qian Liu MSEE 2016ndash Hui He MSEE 2015

bull Mentor Graphics Corpndash Aleksandr Gafarov MSEE

2012ndash Praveen Anmula MSEE

2009bull IBM Systems

ndash Matteo Cocchini MSEE 2008

ndash Michael Cracraft PhD 2007ndash Xiaoxiong Gu MSEE 2002

bull Research in Motionndash Gang Feng PhD 2010ndash Qing Cai MSEE 2007

bull Lockheed Martinndash Jim Feucht MSEE 2010

bull Amazonndash Liang Li MSEE 2016

bull Semtech Corpndash Li Zhen MSEE 2011

bull Texas Instrumentsndash Haixiao Weng PhD 2006

6

Introduction EMC Filters in Power Electronics

12V network

DCDC converter

ACDC charger

Charging station

E-motor PE module HV battery pack

Low- current HV components

LVDC

3-PHASEAC HVDC+

HVDC-

FILTERS

Inverter

EMC Laboratory MSampT

Structure of the talk

Different solution path Circuit Circuit with parasitics full wave Film Capacitor modeling SMT Capacitor modeling CMC modeling Material property extraction Examples of properties of magnetic materials used in EMC

7

One capacitor

8

log(f)

log(|Z|)

log(f)

log(|Z|)

1jwC

1jwC1jwL

ESR

With increased modeling requirement more and more EM effects need to be considered

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

bull Intel Corporation (Total 19 9-PhD10-MS)ndash Oleg Kashurkin MSEE 2016ndash Ben Orr PhD 2016ndash Wei Qian MSEE 2016ndash Xinyao Guo MSEE 2016ndash Jayong Kool PhD 2008ndash Vijay Kasturi MSEE 2007ndash Juan Chen MSEE 1999 (Princ

Engin)bull Altera (Total 5 4-PhD1-MS)

ndash Xinyun Guo MSEE 2016ndash Jianmin Zhang PhD 2007ndash Shishuang Sun PhD 2006ndash Kundan Chand MSEE 2006

bull Apple Computer (Total 20 10-PhD10-MS)

ndash Abhishek Patnaik PdD 2018ndash Satyajeet Shinde PhD 2017ndash Kyoungchoul Koo PostDoc 2017ndash Guangyao Shen PhD 2016ndash Guanghua Li PhD 2015ndash Xu Gao PhD 2014ndash Tianqi Li PhD 2014ndash Andriy Radchenko PhD 2014ndash Kam Keong PhD 2012ndash Frederico Centola MSEE 2003

bull NVidiandash Aukit Bharagava MSEE 2010ndash Fan-Keung Chang MSEE 2006ndash Chen Wang PhD 2004

bull Cisco (Total 13 3-PhD10-MS)ndash Kartheek Nalla MSEE 2016ndash Chunchun Sui PhD 2015ndash Natalia Bondarenko PhD 2015ndash Xiangyang Jiao MSEE 2015ndash Soumya De PhD 2012ndash Amendra Koul MSEE 2010

bull Motorolandash Bivin Varghese MSEE 2003ndash Prem Ganeshan MSEE 1998

bull Panasonicndash Xiao Luo MSEE 1999ndash Janet Luo MSEE 1998

bull Dell Computerndash Arun Chada PhD 2014ndash Lauren Zhang MSEE 1999

bull Googlendash Suyu Yang MSEE 2016ndash Jingnan Pan PhD 2015ndash Ketan Shringarpure PhD 2014ndash Weifeng Pan PhD 2008ndash Sandeep Chandra MSEE 2007

bull Sandia National Laboratoryndash Ben Conley MSEE 2015ndash Matthew Halligan PhD 2014ndash Matt Schepers MSEE 2010ndash David Carter MSEE 2004

And other companieshttpwwwemclabmstedustudfacalumniht

ml

Where are Our Graduates 105 MS 53 PhD 12 Postdocs

bull Broadcom Corporationndash Hongyu Li PhD 2012ndash Liehui Ren PhD 2011

bull Agilent Technologies ndash Joe Bishop MS 2011ndash Kuifeng Hu PhD 2007

bull Juniper Networksndash Qian Liu MSEE 2016ndash Hui He MSEE 2015

bull Mentor Graphics Corpndash Aleksandr Gafarov MSEE

2012ndash Praveen Anmula MSEE

2009bull IBM Systems

ndash Matteo Cocchini MSEE 2008

ndash Michael Cracraft PhD 2007ndash Xiaoxiong Gu MSEE 2002

bull Research in Motionndash Gang Feng PhD 2010ndash Qing Cai MSEE 2007

bull Lockheed Martinndash Jim Feucht MSEE 2010

bull Amazonndash Liang Li MSEE 2016

bull Semtech Corpndash Li Zhen MSEE 2011

bull Texas Instrumentsndash Haixiao Weng PhD 2006

6

Introduction EMC Filters in Power Electronics

12V network

DCDC converter

ACDC charger

Charging station

E-motor PE module HV battery pack

Low- current HV components

LVDC

3-PHASEAC HVDC+

HVDC-

FILTERS

Inverter

EMC Laboratory MSampT

Structure of the talk

Different solution path Circuit Circuit with parasitics full wave Film Capacitor modeling SMT Capacitor modeling CMC modeling Material property extraction Examples of properties of magnetic materials used in EMC

7

One capacitor

8

log(f)

log(|Z|)

log(f)

log(|Z|)

1jwC

1jwC1jwL

ESR

With increased modeling requirement more and more EM effects need to be considered

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

6

Introduction EMC Filters in Power Electronics

12V network

DCDC converter

ACDC charger

Charging station

E-motor PE module HV battery pack

Low- current HV components

LVDC

3-PHASEAC HVDC+

HVDC-

FILTERS

Inverter

EMC Laboratory MSampT

Structure of the talk

Different solution path Circuit Circuit with parasitics full wave Film Capacitor modeling SMT Capacitor modeling CMC modeling Material property extraction Examples of properties of magnetic materials used in EMC

7

One capacitor

8

log(f)

log(|Z|)

log(f)

log(|Z|)

1jwC

1jwC1jwL

ESR

With increased modeling requirement more and more EM effects need to be considered

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Structure of the talk

Different solution path Circuit Circuit with parasitics full wave Film Capacitor modeling SMT Capacitor modeling CMC modeling Material property extraction Examples of properties of magnetic materials used in EMC

7

One capacitor

8

log(f)

log(|Z|)

log(f)

log(|Z|)

1jwC

1jwC1jwL

ESR

With increased modeling requirement more and more EM effects need to be considered

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

One capacitor

8

log(f)

log(|Z|)

log(f)

log(|Z|)

1jwC

1jwC1jwL

ESR

With increased modeling requirement more and more EM effects need to be considered

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Two capacitors

9

Capacitive coupling exists between components

In reality there will be chassis traces and other components involved making the circuit model more complex

Building a full wave model helps with more accurate evaluation of the coupling and easier to implement if the modelling strategy is clarified

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

CMC characterization Sounds too simple

10

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Port4Port3

Port2Port1

M1 C6C5 C4C3

C2

R10

L6 R9

R8

L5R7

L4L3

R6 R5

R4R3

R2R1

L2L1

L7

C1R11

R12 C1

C7

C8

R13

C9 R14

L8

C10

C11

Circuit Model for Choke as Proposed

11

Characterization of EMI Chokes

Leakage Impedance Magnetizing

ImpedanceRC L1

R1

Ce Cc

C2rsquo

R2rsquo

Rc2

C2

L2

R2L2rsquo

Cp

Cs

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Using different connections open and shorts the value of individual elements or groups of elements can be determined

Alternative methods use S-parameters however if one element has only a small effect on the S-parameters the extraction is error prone

12

Characterization of EMI Chokes

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Configuration T2

Equivalent Circuit

13

Characterization of EMI Chokes

1198621198622

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Configuration T1

Equivalent Circuit

14

Characterization of EMI Chokes

1198621198621 41198771198771 41198711198711

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Obtain 119885119885119897119897119897119897119897119897119897119897 119862119862119887119887

15

This configuration only provides the leakage

Ω

The extracted parameters119877119877119897119897119897119897119897119897119897119897 = 367119896119896Ω119871119871119897119897119897119897119897119897119897119897 = 285120583120583120583120583119862119862119901119901 = 38119901119901119901119901

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Configuration T4

Equivalent Circuit

16

Characterization of EMI Chokes [1]

1198621198622prime

1198711198712prime

1198771198772

1198771198771198881198882

1198621198622 1198771198772prime

1198711198712

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Simplified Circuit

17

Characterization of EMI Chokes

Port1 Port2

Port3 Port4

C1

L1 L2R1 R2

R3 R4

R5R6

L3 L4

R7 L5

R8

R9L6

R10

C2

C3 C4C5 C6M1

L1

R1

RC

Ce Cc

Cp

L2R2

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

18

Extracted Circuit for LLF-16Magnetizing Impedance Leakage Impedance Parasitic Capacitances

L2= 045 mH L1= 047 uH Cc= 119 pF

Rrsquo2= 2453 kΩ R1= 9372 Ω Ce= 06 pF

C2= 0 pF Rc= 0013 Ω Cp= 045 pF

MeasurementsSimulation

MeasurementsSimulation

1198791198795119875119875119875119875119875119875119875119875119875 (1198791198795)

119875119875119875119875119875119875119875119875119875 (1198791198796)

1198791198796

100 kHz 1 GHz 100 kHz 1 GHz

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Real filter

19

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

20

1st filter EMI filter

EMI filterndash Usually consists of film capacitors ceramic capacitors and inductorsndash The response of the filter can be degraded by the mutual parasitic

parameters

One-stage EMI filter

[1] Shuo Wang W G Odendaal and F C Lee Extraction of Parasitic Parameters of EMI Filters Using Scattering Parameters Pg2672 IAS 2004

Circuit model including self parasitic Parasitic coupling in an EMI filter [inductive]

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Inductive Coupling

Accurate Analytical Prediction of EMI Filter Attenuation by Considering Intercomponent Coupling PhenomenaMarine Stojanovic Freacutedeacuteric Lafon Richard Perdriau Mohamed Ramdani 21

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

2nd filter EMI filter

Common Mode Choke

CX Capacitor

3

CX Capacitor

CY Capacitor

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Parasitic associated with PCB layout

The layout introduces both capacitive and inductive coupling to components The layout also introduces time delay relevant at higher frequencies and loss which helps to attenuate resonances

23

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

DM results of EMI Filter

24

MeasurementImproved measurementQ3D simulationIdeal components simulation

No method is easy

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

3-D modelling

25

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

3D modelling

Allows for dealing with more compact filter design

Allows for 3D placement of components

Take into account coupling connectors and wires

CMC Film Capacitor

Ceramic Capacitor Other metal

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

P1

CM_choke2

L11

13n

C4

47n

L13

13n

C3

47nR2

1MEG

L1

05n

P2

L10

13n

L16

20n

L15

13n

L2

05n

L12

05n

890324023025_2

Dv_2

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14

P_out

Net15

N_out

Net16GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0via

2_2

Net2

1

via3_

1

Net2

2via

3_2

Net2

3

27

Methodologies for the Modeling of EMC Filters

3Dsimulation

Stray RLC parameters of the PCB traces included to

the circuit calculation with s-parameters

Circuit calculation of the filterwith s-parameters of the components

CMC

R1

1MEG

7448

0407

07_1

L4

13n

L3

13n

Cy_1

47n

P_in

N_outN_in

L1

13n

8903

2402

3025

_1

P_outCx

Cy_2

47n

L2

13n

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

CEMC IAB MeetingMay 8-10 2012

UniversityCenter Confidential

Example Film Capacitor

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR high VAC ratings capacitance

stabilityndash SymmetricAsymmetric rolling

29

At least two methods to roll it

Asymmetric

Symmetric

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Film capacitor

Film capacitorndash Used in EMI filter for the benefit of low ESR

high VAC ratings capacitance stabilityndash SymmetricAsymmetric rolling

30

Other types of rolling can also be handled using similar strategy

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

31

Proposed modelling structure

The structure is made up ofbull Core Abull Core Bbull Dielectric filling the gapbull Outer shell dielectricbull Leadsbull [Lumped element RC series]

Cutting the capacitor to obtain dimensions

The roll can be considered as centered inside the shell

Materials Structure

∎Lossy dielectric Dielectric in the gap

∎PEC Lead of the capacitor

∎Lossy Metal Core Bodies

∎Lossless Dielectric Shell of the capacitor

Symmetric structureAsymmetric structure

Lumped RC series

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Approach for determining the inner structure

32

The coupling to other components depends on the symmetry of the construction of the capacitor- symmetric then mounting is not relevant- assymetric mounting direction changes coupling

flip flip

119909119909 = 0 119909119909 = 0

Potential along the axis Potential along the axis

119909119909 coordinate 119909119909 coordinate

119909119909 = 0 119909119909 = 0

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Move an active high impedance probe along the capacitor to obtain a measure of the surface potential

Determining the inner structure by measurement

HP 85024A

Probe

VNAX=0 X

Asymmetric

Capacitor

Flip the capacitor monitor the change in the trend

Symmetric

33

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

00 1 2 3 4 5 6 7 8 9 10

dB(S

21)

location x

Potential trend (dB(S21)

10nF Body left

10nF Body right

1nF Body Left

1nF curves are identical after flipping

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Three validation cases Validation S21 sim vs meas

capacitor to wire coupling

Copper Wire Diameter 07 mmWire length 46 mmWire height 83 mm (from GND)Distance to capacitor surface 12 mm

Measured S21

Parameter Different loading combinations are tested

34

The sheet has length of 46 mm height 12 mm

capacitor to sheet coupling capacitor to capacitor coupling

1 nF capacitors placed as close as possible

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Capacitor to wire coupling (various loading combinations)

Port1

Port2

Port1

Port2

Port1

Port2

Port1

Port2

In open condition element 12 are set as a small capacitor representing the via capacitance

In shorted loading condition the wirepin are extended to make contact with ground

Short-ShortSS

Open-OpenOO

Short-OpenSO

Open-ShortOS

Calibration plane

bull Element 12 are used to adjust the termination conditionbull FR4 is hidden

port extension to here

35

Element 2Element 1

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

1E6 1E7 1E81E5 1E9

-100

-80

-60

-40

-20

-120

0

freq Hz

Measured S21 [dB]Simulated S21 [dB]

Capacitor to wire coupling Numerical model is goodOS

OO SO

SS

Port1

Port2

Port1

Port2

Port1

Port2

3dB difference at 20MHz

Port1

Port2

Variation due to loading capacitance

36

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Capacitor to capacitor (symmetric) coupling

OO caseMeasured S21Simulated S21

SO caseMeasured S21Simulated S21

Port1 Port1

Port1 Port1

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

39

dB (S

21)

dB (S

21)

OO

SO

capacitive inductive

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

ESR of film capacitor

40

119916119916119916119916119916119916 = 119916119916119956119956 +119916119916119941119941

1 + tan2 120575120575119889119889

119877119877119889119889119894119894119894119894 =119877119877119889119889 lowast 119877119877119894119894119894119894119877119877119889119889 + 119877119877119894119894119894119894

asymp 119877119877119889119889 =tan 120575120575119889119889120596120596119862119862

prop1119891119891

httpswwwvishaycomdocs26033gentechinfofilmpdf

119916119916119916119916119912119912119912119912 =11990311990302

1205751205752119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus1+119897119897

minus1199031199030120575120575 119877119877119863119863119863119863 = 120588120588119897119897

12058712058711990311990302

A high frequency approximation can be made as

In a round wire having radius r0 and skin depth 120575120575

119877119877119878119878 = 119877119877119878119878 119877119877119878119878119863119863119863119863 1199031199030 120575120575

119845119845119845119845119845119845119955119955120782120782120633120633 rarrinfin

119916119916119916119916119912119912119912119912 =11990311990302120575120575 119877119877119878119878119863119863119863119863 prop 119891119891

Dielectric resistance119877119877119889119889

119877119877119889119889119863119863 =119916119916119941119941

1 + tan2 120575120575119889119889

1199031199030

Conductive resistance119877119877119894119894

10 0 10 5 10 10

freq Hz

10 -2

10 -1

10 0

10 1

Res

ista

nce

RA C

exact formula

10dBdec

Dielectric dissipation angle not skin depth

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Components of ESR

41

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-5

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E-6

2E2

freq Hz

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximatio119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 =119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is significantly higher than B then a lsquoflatrsquo region appears

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

Skin depth becomes important

Dissipation angle is constant

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

1E3 1E4 1E5 1E6 1E7 1E8 1E91E2 1E10

1E-4

1E-3

1E-2

1E-1

1

1E1

1E2

1E3

1E-5

2E3

freq Hz

Components of ESR

42

LC Z1P_Eqn

Z1P_Eqn

Term

L5C5 Z1P12

Z1P11

Term8

R=L=L nH

C=C Z[11]=12pifreqCtand

Z[11]=RDC(k2)2(k-1+e(-k))Z=50 OhmNum=8

119916119916119941119941119916119916119912119912119912119912 with high freq approximation119916119916119912119912119912119912 exact formula119916119916119916119916119916119916 = 119916119916119956119956 + 119916119916119941119941

1+tan2 120575120575119889119889

Point A Point BIf A is close to B (119877119877119863119863119863119863 is small) there wonrsquot be a lsquoflatrsquo region

119916119916119916119916119912119912119912119912 =119903119903021205751205752 119877119877119878119878119863119863119863119863

2 1199031199030120575120575 minus 1 + 119875119875minus

1199031199030120575120575

119877119877119889119889 =1

120596120596119862119862 tan 120575120575119889119889

Dielectric resistance

DC resistance is smallno flat region

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Measure ESR of the capacitor using Impedance Analyzer

HP4294A amp HP4291A are used at MSampT

By combining their results we can obtain the ESR of a film capacitor from 40Hz to 1GHz

101 102 103 104 105 106 107 108 10910-1

100

101

102

103

1041nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

101 102 103 104 105 106 107 108 10910-1

100

101

102

103 10nF

Frequency (Hz)

R (O

hm)

HP4294A dataHP4291A data

HP4294AHP4291A

HP4294AHP4291A

Calibration port extension and standard validation ensures good data

Calibration plane

Port extension

Measuring 01 Ohm SMD resistor for validation

ESR of 10nF and 1nF film capacitor

Flat region

No flat region

43

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

With lumped element

Lumped element 119877119877119863119863119863119863 Lossy Dielctric 119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

Without lumped element

Lossy Dielctric 119912119912119877119877119889119889 Lossy Metal 119877119877119860119860119863119863

44

ESR can be modelled in two means

Lossy metal

PECLossy dielectricAccounting for C119877119877119889119889

Lumped capacitor

Lossy metal

PEC

Lossy DielectricAccounting for 119877119877119889119889

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Model without lumped element

Model without lumped element makes the tuning procedure easier When the capacitance to is too large (gt1uF) solver inaccuracy problems may occur The time cost and mesh size is comparable with the model using lumped element The mutual coupling is not affected when switching from lumped element to without

lumped element

The gap is set as 1120583120583120583120583 to handle large capacitance case and avoid solver accuracy issues

The material inside the gap determines both the capacitance and low frequency ESR of the modelOnce the dimensions of the model are fixedbull The relative dielectric constant determines the capacitancebull The electrical loss tangent of the material tan 120575120575119889119889 determines the tan 120575120575 or 119864119864119864119864119877119877 at low frequency

119862119862 =1205761205760120576120576119903119903 119888119888119888119888 119888119888119875

119892119892119875119875119901119901

tan 120575120575119889119889 = tan 120575120575 = 119864119864119878119878119877119877120596120596119863119863

45

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Full wave simulation requirements

For full wave simulation the inner structure needs to be know It maybe not necessary to reproduce the inner structure Material properties need to be known The complexity needs to be acceptable from the memory demand and

simulation time

46

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Simulation choices

Due to the shape of the structure CST time domain solvers FIT or TLM are not suitable

HFSS CST frequency domain EMCOS MoM

47

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

SMT capacitor

48

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Transversal view Lateral view

height

widthlength

width

height

length

height

asymp 43120583120583120583120583

asymp 3120583120583120583120583

asymp 23120583120583120583120583

SMD capacitor

49

1nF capacitance

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

3D model for 1nF capacitor CST MW studio

Created by a discrete port between two soldering terminals adjusted 120576120576119903119903 to reach 1nF capacitance120634120634119955119955 = 120783120783120783120783120783120783120783120783

E-field distribution

Impedance of 1nF capacitor

50

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Coupling between two capacitors (HH VV)

HH VV HV cases are all modeled in CST MW studio Each terminal is assigned a port the loading conditions can thus be easily altered in post processing

For example vertically placed capacitor having loading combination short-short is denoted as VV SS

51

Horizontal-HorizontalHH

Vertical-VerticalVV

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

The HH case comparison

OOSOSS

OOSOSS

Simu Meas

The SS case has maximum 3dB difference between simulation and measurement in capacitive and inductive part of the couplingOther two cases are well matched

Flat20dBdec 40dBdec

Time for the simulation 8min 26s

52

HH

Only 1 nF capacitance Low freq no short

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

bull 3D model is made explicitly following the measured dimensions of the MLCC‒ In MW studio the tuning yields 120634120634119955119955 asymp 120783120783120788120788119916119916120788120788 to

reach a capacitance of 1nF

The simulated coupling between two capacitors in different orientation and loading condition combinations are compared with the measured result Works

Summary of capacitor modeling

53

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

54

CMC

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

CMC

Common mode chokendash Used in EMI filter to suppress common mode noisendash Core come in different types Mn-Zn Ni-Zn Nano

Crystalline etcndash The core and the wire are both coated with insulation layerndash Wires can overlap in actual windings

55

Coating layer is relevant

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

3D model example

17 turn CMCInner Diameter 10mmOuter Diameter 182mmWidth 107mm[with paint]

The covered angle of each winding is about 130deg

Inner Diameter 9mmOuter Diameter 20mmWidth 98mm[without paint]

The paint has thickness of around 04mm

A model is made using all measured dimensions

56

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Challenges during the modelling

1 WireThe wire has very thin insulation coatingThe wires are squeezed together and often overlap in reality

2 The Material properties120583120583lsquo 119891119891 120583120583primeprime 119891119891120576120576prime 119891119891 120576120576primeprime 119891119891120590120590

ndash Frequency dependence of the material parameters is not easy to measure

ndash Materials may not be homogenous

57

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Modelling of the wire (wire shape)

58

Round wireMesh Count 537kMesh time 27mSolver Time 6h15m total frequency range

Square wireMesh Count 59kMesh time 16sSolver Time 20m

Using square wires instead of round wires reduces mesh count in the frequency domain solver

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Coating of the wire

In actual windings the parasitic capacitance contributed by the adjacent wires can be large due to the small distance between them

In simulation many gaps of 120583120583120583120583 scale need be avoided A solution to this problem will be published

59

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Assigning material properties

60

Material Assigned parameter

Core 120583120583prime 119891119891 120583120583primeprime 119891119891 120576120576prime 119891119891 120576120576primeprime 119891119891

Insulation layer 120576120576119903119903

Painting of core 120576120576119903119903 = 1

Wire 119875119875119864119864119862119862

Seat 120576120576119903119903 = 2

bull The base is not modelled however the wire lengths need to be exact as they influence coupling and leakage inductance

bull PEC is used for the wires because the DC resistance can be easily assigned using lumped element or be placed elsewhere to save solver time besides the skin effect is negligible in the frequency below 100MHz

bull The painting is at first not modelled

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Extract 120583120583prime 120583120583primeprime one coil the other left open

61

10 2 10 3 10 4 10 5 10 6 10 7 10 8

freq

-2000

0

2000

4000

6000

8000

CST Debye real

meas real

CST Debye imag

meas imag

120583120583119903119903 = 120583120583infin +120583120583119894119894119904119904119897119897119904119904119894119894119888119888 minus 120583120583infin

1 + 119895119895120596120596120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903120583120583prime = 2120587120587120587120587 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

120583120583primeprime = 2120587120587119877119877 119891119891

12059612059611987311987321205831205830119867119867119864119864 ln119863119863119864119864119863119863119868119868

Extracted from measurement 1st order Debye model

120583120583119894119894119904119904119897119897119904119904119894119894119888119888 = 7000 120583120583infin = 100120591120591119903119903119897119897119897119897119897119897119903119903119897119897119904119904119894119894119903119903119903119903 = 35119875119875 minus 7119875119875

The permeability of some cores can be approximated using 1st

order Debye model

An approximated value

`

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

62

Core Magnetic Material Properties - measurement

Fig6 EMCoS test fixture for the measurement of core permeability

Oslash Test fixture is used to measure magneticparameters of toroidal cores

Oslash Measurement of input impedance of thefixture with and without toroidal core is done

Oslash Complex permeability is calculated from theformula (1)

(1)

Reference Agilent Solutions for Measuring Permittivityand Permeability with LCR Meters and ImpedanceAnalyzers

μ = 1 +2π(119885119885119888119888119885119885119905119905119875 119888119888119888119888119903119903119875119875 minus 119885119885119891119891119885119885119909119909119905119905119909119909119903119903119875119875)

119895119895ωμ0119875 ln119863119863119888119888119909119909119905119905119863119863119885119885119894119894

100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

1

10

100

1k

10k

100k

Perm

eabi

lity

Measured complex permeability vs frequency

Measurements of the complex permeability

Measurements of complex permeabilityare performed in EMCoS laboratoryusing impedance analyzer in thefrequency range from 40 Hz up to 110MHz

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

63

Coil to wire coupling

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Test case for mutual parasitic prediction [P6 open]

64

12

3

4

5 6

Core defined by 1st order Debye model120583120583119894119894119904119904119897119897 = 7000120583120583infin = 1 120591120591 = 35119875119875 minus 7119875119875

MeasurementSimulated

1E6 1E71E5 1E8

-80

-60

-40

-20

-100

0

freq Hz

dB(S

(25

24))

dB(S

(53

49))

S51 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(27

26))

dB(S

(53

50))

S52 6_Open

1E6 1E71E5 1E8

-100

-80

-60

-40

-20

-120

0

freq Hz

dB(S

(29

28))

dB(S

(53

51))

S53 6_Open

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(51) S(52)

S(53) S(54)

dB T

rans

Coe

ff

dB T

rans

Coe

ffdB

Tra

ns C

oeff

dB T

rans

Coe

ff

A resonance is missed it is related to the core permittivity

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Mismatch between simulation and measurement

Take the far-end coupling S54 as example

65

1E6 1E71E5 1E8

-120

-100

-80

-60

-40

-140

-20

freq Hz

dB(S

(31

30))

dB(S

(53

52))

S54 6_Open

S(54) port6 open

dB T

rans

Coe

ff

1E6 1E71E5 1E8

-70

-60

-50

-40

-30

-20

-80

-10

freq Hz

dB(S

(44

43))

db(S

(58

57))

S54 6_Short

S(54) port6 short

dB T

rans

Coe

ff

12

3

4

5 6

The major trend shows good match while the resonance doesnrsquot show up in the simulation result

Eps_r needs to be known

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Parametersrsquo Impact on Coupling [P6 open]

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq HzS(51)

EPC=1 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

EPC=36 pF

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

1E6 1E7 1E81E5 1E9

-80

-60

-40

-20

-100

0

freq Hz

C_couple=02 pF C_couple=1 pF

freq 50M rarr100M

67

12

3

4

5 6

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Parameters that affects the resonance

68

Parametersto tune Influence on resonance

Leakageinductance Increase the leakage inductance will lower the resonance frequency

EPC Increase the value of EPC will lower the resonance frequency

C_couple Increase C_couple will increase the level of curves

Important to determine εr (key parameter for EPC inter coil coupling capacitor)

Use 3D model to investigate the impact of εr

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Investigate the impact of 120576120576119903119903 (open-ended plate)

69

12

3

4

5 6opened port 6

S54

120634120634119955119955 lt 120783120783120783120783120782120782

120634120634119955119955 ge 120783120783120783120783120782120782

120576120576119903119903 = 100120576120576119903119903 = 120

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Summary

The CMC model can represent the actual CMC and the coupling between the CMC and metal plate up to 10MHz

From the circuit analysis the EPC and coupling capacitance affects the coupling between the CMC and the plate

By adjusting the 120576120576119903119903 and 120583120583infin itrsquos seen that the higher order resonance above 50MHz is contributed by both the inductance and the capacitance associated with the core material

70

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

bull Both inner and outer surface is completely covered by silver paint

bull DC resistance measured using mustimeter (point to point) is around 19Ω

bull Copper tape at tip of the probe touches the surfaces of the core

bull The contact is enhanced by inserting foam in the inner space and tape at the outer surface

Measure the core with silver paint on outer and inner surface

Extract the 120576120576 and 120583120583 of the corendash Implement in 3D simulation scenario firstndash Try extraction using measured data

Attempts on extracting 120576120576 and 120583120583infin

RDC = 19 Ohm

71

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Extracting the 120576120576

10 5 10 6 10 7 10 8 10 910 2

10 3

10 4

10 5

1st order Debye

1st order Debye

The 1st order Debye model is not enough to represent the rapid decrease of 120576120576primeThe 120576120576120576 at above 10MHz remains unknown But can be solved under another configuration

120576120576119864119864119905119905119875119875 = 120788120788120783120783120782120782120782120782120782120782 120576120576119868119868119894119894119891119891 = 100 120591120591119864119864 = 385119875119875 minus 7

RDC = 19 Ohm

The misalignment with the Debye model indicates the static conductivity of the core as well as the small DC resistance measured across the toroid

72

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Filter modeling at EMCOS

73

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

74

Modeling of the Filter Components EMCOS

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

10m

100m

1

10

100

1k

10k

Impe

danc

e [

]

WCAP-FTX2 890 324 023 025

10kHz 100kHz 1MHz 10MHz 100MHz

Frequency

-60

-40

-20

0

S pa

ram

eter

s [d

B]

WCAP-FTX2 890 324 023 025

S11 = S22

S12 = S21

Calculated impedance (Pin length = 3mm)

Calculated S parameters (Pin length = 3mm)

Electrical parameters

Nominal capacitance C 015 microF

Equivalent series resistance ESR 60 mΩ

Equivalent series inductance ESL 80 nH

Rated voltage V (AC) 275 V

WCAP-FTX2 MKP Film Capacitor 890324023025

Film capacitor F3D model of film capacitor

Geometrical parameters

Max dimensions (L times W times H) 13 times 7 times 12 mm

Distance between pins 10 mm

Pins diameter 06 mm

Electrical parameters of film capacitor

Geometrical parameters of film capacitor

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

75

EMCoS Studio Filters Components LibraryOslash Integrated library of pre-assembled

standard filter components of differenttypes provides possibility to easily add adesired component to 3D Circuit filterlayout

Oslash The library is based on OEM datasheetsand measurements of actual components

Oslash The components are grouped by typesand each component type has own librarycatalog

CoresRodsBarsBobbinsToroidsUampE CoresSuppressors

CapacitorsFilmFeedthroughSMD

InductorsDM ChokeCM Choke 1-phaseCM Choke 3-phase

General Structure of Library Catalogs

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

76

Simulation Models of EMC Filter in EMCoS Studio

Common Mode

Differential Mode

Circuit calculation of the filter with s-parameters of the components

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

N_out

P1

Cx

Cy 47n

R

1MEG

Cy 47n

CMC

8903

2402

3025

7448

0407

07

P_outP_in

P2

N_in

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-80

-60

-40

-20

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Speed-oriented EMCoS circuitsolver TSReady with the modifiednodal analysis (MNA) approach isused for circuit calculations

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

77

3D PEEC Solution of Stray RLC Parameters of the PCB

L1 L2

C1 C3C2

Pi-scheme

L1 L2

C1 C2 C3

Extracted equivalent

circuit

Equivalent Circuit Extraction

Evaluation board from Wuumlrth Elektronik design kit

3D PEEC model of the evaluation board

bull Model is divided into partial elementssurface geometry patch elements (for CGcalculation) and branches (for RLcalculation)

bull Pins are corresponding to partialcapacitance (C) and conductance (G)elements Each pin has coordinate andcorresponding surface geometry patch

bull Branches are corresponding to partialinductance (L) and resistance (R)elements Branches are based on pins

bull Automatic partitioning is supported

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

78

Simulation Models of EMC Filter in EMCoS Studio

Circuit Calculation of the Filter with S-Parameters of the Components and Stray RLC Parameters of the PCB Traces

P2

L8

05n

CM_choke1

L7

13n

L4

05n

L3

20n

R1

1MEG

P1L5

13n

L2

13n

C2

47n

L6

13n

L1

05n

890324023025_1 C1

47n

Dv_1

P_in

Net1

N_in

Net2

R_1

Net3

R_2

Net4

Cx_1

Net5

Cx_2

Net6

CMC_4

Net7

CMC_2

Net8

CMC_3

Net9

CMC_1

Net10

Cy2_2

Net11

Cy2_1

Net12

Cy1_1

Net13

Cy1_2

Net14P_out

Net15

N_out

Net16

GND

Net17

via_1

_1

Net1

8

via1_

2

Net1

9

via2_

1

Net2

0

via2_

2

Net2

1

via3_

1

Net2

2

via3_

2

Net2

3Net-list Device

Equivalent circuit of the evaluation board

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0Fi

lter D

M a

ttenu

atio

n [d

B]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

79

Simulation Models of EMC Filter in EMCoS Studio

3D simulation with full-wave MoM solver

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r DM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

10kHz 100kHz 1MHz 10MHZ

Frequency

-100

-50

0

Filte

r CM

atte

nuat

ion

[dB]

Measurement

Circuit calc(S-parameters of components)

Circuit calc(S-parameters of components and PEEC of PCB)

Full-wave 3D simulation

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

80

3D Current Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

Current distribution at 1 MHz

IN

OUT

IN

OUT

Current distribution at 9 MHz

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

81

Magnetic Near Field Distribution

3D simulation with full-wave MoM solver TriDCommon Mode Differential Mode

H-field distribution at 1 MHz

IN

OUT

IN

OUT

H-field distribution at 9 MHz

H-field distribution at 1 MHz

INOUT

INOUT

H-field distribution at 9 MHz

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

82

Conclusions ndash MoM modeling

Oslash 3D models of the components as well as their circuit elements based on theS-parameters can be used for filter simulations

Oslash Different simulation levels for the modeling and validation of EMC filtermodels are presented and show good matching with measurements

Oslash Comparison of the simulation results with measurements is performed inthe frequency range from 10 kHz up to 50 MHz and up

Oslash 3D PEEC solution helps to reproduce effects caused by interaction andconductive coupling between PCB traces FEM solutions are also suitable

Oslash 3D full-wave MoM solution takes into account conductive coupling as wellas EM field interaction between components

Oslash Demonstrated techniques can be used for the investigations andimprovement of the EMC filters performance

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Measurement of core parameters

7 mm air line Using turns 16454A magnetic material test fixture

83

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Solutions for measuring ε μ

84

III Permeability EvaluationInductance measurement methodRelative permeability of magnetic material

derived from the self- inductance of a cored inductor that has a closed loop (such as the toroidal core) is often called effective permeability The conventional method of measuring effective permeability is to wind some wire around the core and evaluate the inductance with respect to the ends of the wire Effective permeability is derived from the inductance measurement result using the following equations

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Solutions for measuring ε μ

85

III Permeability EvaluationMeasurement system using the 16454A magnetic material test fixture

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Material properties comparison

Mn-Zn ferrites Nanocrystalline Flexible Material

86

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Overview on 120583120583120576 vs frequency

87

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Overview on 120583120583120576120576 vs frequency

88

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Nanocrystalline ----- 120583120583prime 120583120583primeprime vs frequency

89

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Flexible Material ----- 120583120583prime 120583120583primeprime vs frequency

90

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

MnZn ----- 120583120583prime 120583120583primeprime vs frequency

91

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

NiZn ----- 120583120583prime 120583120583primeprimevs frequency

92

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93

Thank you

93

• EMC Filter Analysis
• Slide Number 2
• The group
• Research partners from industry and government
• Where are Our Graduates 105 MS 53 PhD 12 Postdocs
• Introduction EMC Filters in Power Electronics
• Structure of the talk
• One capacitor
• Two capacitors
• Slide Number 10
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Characterization of EMI Chokes
• Obtain 119885 119897119890119886119896 119862 119887
• Characterization of EMI Chokes [1]
• Characterization of EMI Chokes
• Extracted Circuit for LLF-16
• Slide Number 19
• 1st filter EMI filter
• Inductive Coupling
• 2nd filter EMI filter
• Parasitic associated with PCB layout
• DM results of EMI Filter
• Slide Number 25
• 3D modelling
• Methodologies for the Modeling of EMC Filters
• Example Film Capacitor
• Film capacitor
• Film capacitor
• Proposed modelling structure
• Approach for determining the inner structure
• Determining the inner structure by measurement
• Three validation cases Validation S21 sim vs meas
• Capacitor to wire coupling (various loading combinations)
• Capacitor to wire coupling Numerical model is good
• Capacitor to capacitor (symmetric) coupling
• ESR of film capacitor
• Components of ESR
• Components of ESR
• Measure ESR of the capacitor using Impedance Analyzer
• Slide Number 44
• Model without lumped element
• Full wave simulation requirements
• Simulation choices
• Slide Number 48
• SMD capacitor
• Slide Number 50
• Coupling between two capacitors (HH VV)
• The HH case comparison
• Summary of capacitor modeling
• Slide Number 54
• CMC
• 3D model example
• Challenges during the modelling
• Modelling of the wire (wire shape)
• Coating of the wire
• Assigning material properties
• Extract 120583 prime 120583 primeprime one coil the other left open
• Core Magnetic Material Properties - measurement
• Slide Number 63
• Test case for mutual parasitic prediction [P6 open]
• Mismatch between simulation and measurement
• Parametersrsquo Impact on Coupling [P6 open]
• Parameters that affects the resonance
• Investigate the impact of 120576 119903 (open-ended plate)
• Summary
• Attempts on extracting 120576 and 120583 infin
• Extracting the 120576
• Slide Number 73
• Modeling of the Filter Components EMCOS
• EMCoS Studio Filters Components Library
• Simulation Models of EMC Filter in EMCoS Studio
• 3D PEEC Solution of Stray RLC Parameters of the PCB
• Simulation Models of EMC Filter in EMCoS Studio
• Simulation Models of EMC Filter in EMCoS Studio
• 3D Current Distribution
• Magnetic Near Field Distribution
• Conclusions ndash MoM modeling
• Measurement of core parameters
• Solutions for measuring ε μ
• Solutions for measuring ε μ
• Material properties comparison
• Overview on 120583prime vs frequency
• Overview on 120583primeprime vs frequency
• Nanocrystalline ----- 120583 prime 120583 primeprime vs frequency
• Flexible Material ----- 120583 prime 120583 primeprime vs frequency
• MnZn ----- 120583 prime 120583 primeprime vs frequency
• NiZn ----- 120583 prime 120583 primeprime vs frequency
• Slide Number 93