upc universitat politècnica de catalunya - emc design in industrial systems · 2020. 4. 27. ·...

18/11/2010

Course EMC Master DEE 1

EMC DESIGN IN EMC DESIGN IN INDUSTRIAL SYSTEMSINDUSTRIAL SYSTEMS

Dr. J. Balcells, Dr.D. Dr. J. Balcells, Dr.D. GonzGonzáálezlez, Dr. J. Gago, Dr. J. GagoDept. Dept. EnginyeriaEnginyeria ElectrònicaElectrònica UPCUPC

1.1. INTRODUCTIONINTRODUCTION

ELECTROMAGNETIC DISTURBANCES :Non desired changes in electric or magnetic characteristics of devices or systems, causing changes in their behaviour

ELECTROMAGNETIC INTERFERENCES, EMI:Electric or magnetic disturbances superimposed to a signal, causing malfunctioning of electrical and electronic devices or systems

ELECTROMAGNETIC COMPATIBILITY, EMC :Capability of a device, apparatus or system to properly work in a certain electromagnetic environment without being disturbed and without producing disturbances unacceptable for other systems in such environment

DEFFINITIONSDEFFINITIONS

18/11/2010


EMI EnergySources

PROPAGATION & VICTIMCanal de

acoplamiento

ASPECTS RELATED TO EMCASPECTS RELATED TO EMC

EMC DEPENDS ON 3 PARTS: EMI Source, Propagation path and Victim susceptibility

EMI SOURCES : – Natural– Man made– Non ideal behaviour of components

PROPAGATION:– Conduction (Common impedance sharing)– Close field: (Electric or capacitive coupling , Magnetic or inductive

coupling)– Far field: Electromagnetic field with E,H orthogonal

VICTIM– Active devices, both analog and digital

SUMMARY OF GENERAL CONCEPTSSUMMARY OF GENERAL CONCEPTS

18/11/2010


IMMUNITY / SUSCEPTIBILITY :Immunity is the property of a device or system to withstand a certain level of EMI in the environment without being disturbed.Susceptibility is the opposite property to immunity.

SUSCEPTIBILITY INDEX :Susceptibility is usually measured in terms of bandwidth in dBHzversus disturbance in dBmW or dBmV, at the threshold of circuit failure. It’s given by the susceptibility index SI

)dBm(D)dBHz(B

)level(dBmW eDisturbancdBHz)Bandwidth(SIW ==

)VdB(D)dBHz(B

V)level(dB eDisturbancdBHz)Bandwidth(SIV μμ

==


COMPATIBILITY MARGIN :The compatibility margin is the difference between the emission limit and the susceptibility limit (IEC-61000-1-1)

frequency

Disturbance level dBm or dBuV

CompatibilityLevel

SusceptibilityLimit

EmissionLimit

Immunity Margin

Emission Margin

Compatibility Margin


18/11/2010


ZERO?, GROUND? EARTH?. A real confusion !!These three terms are often used improperlyZERO : The reference point for all the potentials in a circuit. Usually this point is at the power supply. The preferred symbols are:

GROUND: Conductive parts surrounding an electric or electronic circuit. Usually the zero is connected to GND to avoid parasitic couplings. The preferred symbols are

EARTH: Conductive parts of buildings electrically linked to the soil earth. The preferred symbols are

REMINDER OF BASIC CONCEPTSREMINDER OF BASIC CONCEPTS

EMI PROPAGATION MECHANISMSEMI PROPAGATION MECHANISMS

Conduction or common impedance coupling– Share of supply cables or PCB paths– Share of signal paths

Close field coupling– Electric or capacitive coupling– Magnetic or inductive coupling

Far field coupling– Electromagnetic field with E,

18/11/2010


A NEW UNIT FOR LENGTHA NEW UNIT FOR LENGTH

From EMC point of view, the physical dimensions of a circuit (inmeters, cm, feet or inches) are not relevant by itself.Actually, the distance unit in EMC is the wave longitude λ

– If Dλ /2π → Waves Theory and transmission lines apply

Tfv 1λλ ==

v= propagation speed (v =c in vacuum or air)

l = wave longitude

f = frequency

Propagation speed only depends on the propagation media

EMI PROPAGATION MECHANISMSEMI PROPAGATION MECHANISMS

18/11/2010


CLOSE FIELD AND FAR FIELDCLOSE FIELD AND FAR FIELD

RADIATION (FAR FIELD)RADIATION (FAR FIELD)

Ωπεμ

377.120HEZ

0

oo ====

18/11/2010


CONDUCTED COUPLINGCONDUCTED COUPLING

It occurs when two circuits shares some impedanceNormally the impedance that is shared is the return path (ground)

ADVICES AND TIPS– Avoid long distance connections– Use a “dedicate” conductor to supply separately each circuit or system– Different grounds and safety earth connection must be done in only one

pointKeep the supply impedance as low as possible

Powersupply

Powersupply

LONG CABLES BEHAVIOUR:TRANSMISION LINESLONG CABLES BEHAVIOUR:TRANSMISION LINES

L L L L L

E C C C C Zf

+

-

L is the inductance per unit length H/m ; C is the capacitance per unit length F/m A piece of cable dx will have inductance L.dx and capacitance C.dx When the circuit is excited Cs are charged The charge will propagate a dx VCdxdQ = (1) Current will be VCvdt/VCdxdt/dQI === (2) The stablished flux in L will be LdxId =φ (3)

Substituting (2) in (3) LdxVCvd =φ (4) Electromotive force will be

2LCVvdt/LCVvdxVdt/de ====− φ (5)

From (5) we get the propagation speed LC/1v = (6)

And from (2) and (6) we get the characteristic impedance C/LI/VZ0 ==

18/11/2010


LONG CABLES LOSSY TRANSMISION LINELONG CABLES LOSSY TRANSMISION LINE

CAPACITIVE COUPLINGCAPACITIVE COUPLING

– It is due to stray capacitors between live parts of source and victim circuits– Stray capacitors are geometric characteristics– Actually, these capacitors are distributed but they are usually represented

with lumped parameters– This coupling is ruled by:

– Two kind of actions can be taken:• Reduce the voltage slope in the source (not always possible)• Change the geometry in order to reduce the capacitive coupling

dtdvCi sourcevictim =

18/11/2010




CONSEQUENCES

From capacitive coupling point of view, devices with low inputimpedance are better.

18/11/2010


INDUCTIVE COUPLINGINDUCTIVE COUPLING

– It is due to mutual inductance between two circuits– It is proportional to the surface of the source circuit– It is proportional to the surface of victim circuit– It induces a voltage in the victim circuit and the source is the

current derivative– The mutual induction coefficient is a geometric characteristic

dtdIM

dtdV 1121212 ==

φ

– Two kind of actions can be taken:• Reduce the voltage slope in the source (not always possible)• Change the geometry in order to reduce the capacitive coupling


EQUIVALENT CIRCUIT OF INDUCTIVE COUPLING

1121

12L VMjR

MjVω

ω+

=

Voltage induced in thevictim does not dependon its impedance

18/11/2010



CONSEQUENCES: How to connect the shield of a wire

Live wire

plus shield

RIGHT!

Shield NOT connected!

I leakage=0I leakage≠0

Shield grounded in one point only!


18/11/2010


SUMMARY OF CAP. AND IND. COUPLING SUMMARY OF CAP. AND IND. COUPLING

Capacitive coupling is modelled as a current source

Inductive coupling is modelled as a voltage source

EQUIVALENT CIRCUIT

EQUIVALENT CIRCUIT

RADIATION (FAR FIELD)RADIATION (FAR FIELD)

Ωπεμ

377.120HEZ

0

oo ====

18/11/2010


EMI PROPAGATION PATHS:CM, DMEMI PROPAGATION PATHS:CM, DM

CO

MM

ON

MO

DE

CO

MM

ON

MO

DE

DIF

FER

ENTI

AL

MO

DE

DIF

FER

ENTI

AL

MO

DE

CM

DM

DM EFFECTS DUE TO CM SOURCESDM EFFECTS DUE TO CM SOURCES

CM sources cause

DM effects due to

differences in cable

impedances

ZC1 ≠ZC2

and in stray

capacitances to GND

Cstr1 ≠Cstr2

CM EMI SOURCE

18/11/2010


EMC TESTSEMC TESTS

EMC tests EMC tests require professional require professional instrumentsinstruments•• Spectrum Analyzer with EMISpectrum Analyzer with EMIprepre--filter for BW limitationfilter for BW limitation

•• Calibrated EMI ReceiversCalibrated EMI Receivers•• Calibrated Probes and AntennasCalibrated Probes and Antennas•• Screened RoomsScreened Rooms•• Standard Impedance NetworksStandard Impedance Networks•• Mains analyzers Mains analyzers (harmonics, flicker)(harmonics, flicker)

•• Pulse generators + AmplifiersPulse generators + Amplifiers

DO NOT USE•Oscilloscope + probes•Oscilloscope using add/invert funct.•ADC’s for systems having bad

signal/noise performances•Badly screened cables•Randomly placed cores and filters

GENERAL CONCEPTS ON MEASUREMENTGENERAL CONCEPTS ON MEASUREMENT

TIM

E D

OM

AIN

TIM

E D

OM

AIN

FREQ

UEN

CY

DO

MA

INFR

EQU

ENC

Y D

OM

AIN

18/11/2010


GENERAL CONCEPTS ON MEASUREMENTGENERAL CONCEPTS ON MEASUREMENT

NOTICE! Sharp pulses have higher amplitudes at high frequencies, whileround pulses have lower amplitudes at high frequencies, f=1/T

SPECTRUM ANALYZERSPECTRUM ANALYZER

Input LP Filter: Low pass filter. Limits input signal BW to avoid aliasingLO : Local Oscillator. Is a VCOVCO: Voltage controlled oscillatorMixer: Consists of a signal multiplier (Modulator)IF BP Filter: Determines de RBW (Resolution Bandwidth)

Input Signal

Horizontal Sweep

Vertical Sweep

CRT

18/11/2010


SPECTRUM ANALYZER: SPECTRUM ANALYZER: MixerMixer

fsig : Input signal frequency ; fLO: Frequency of local oscillator

Output contents: fsig , fLO and harmonics hsig⋅fsig , hLO⋅fLOfLO ± fsig and harmonics fLO ± hsig⋅fsig , hLO⋅fLO ± fsig)

IF (Intermediate Filter) selects one of the sidebands , i.e. the signal fLO+hsig.fsig

SPECTRUM ANALYZER: IF SPECTRUM ANALYZER: IF FilterFilter

fLO must be above fRange+RBW

fLO must cover, at least the range

fLO+ fsig to fLO+hsig.fsig

Input Signal IF Filter Bandwidth

18/11/2010


EMI MEASUREMENTEMI MEASUREMENT

Broad band measurement:Δfmin < RBWWhere, RBW= Resolution Bandwidth

Narrow band measurement:Δfmin > RBWWhere, RBW= Resolution Bandwidth

frequency

AmplitudeResolution Bandwidth , RBW

Δffrequency

AmplitudeResolution Bandwidth , RBW

Δf

EMI MEASUREMENT: UNITSEMI MEASUREMENT: UNITS

dBT/kHz=20 log(B/1T)T=Tesla=1Weber/m2

Φ/SdBT=20 log(B/1T)T=Tesla=1Weber/m2

Φ/S

Power density dBm/m2/kHz=10 log(P/1mW/m2)

P/SPower density dBm/m2=10 log(P/1mW/m2)

P/S

Power, dBm/kHz=dBmW/kHzdBm/kHz=10 log(P/1mW)

PPower, dBm=dBmWdBm=10 log(P/1mW)

PdBA/m/kHz=20 log(V/1A/m)HdBA/m=20 log(H/1A/m)HdBV/m/kHz=20 log(V/1V/m)EdBV/m=20 log(E/1V/m)EdBμA/kHz=20 log(V/1μA)IdBμA=20 log(I/1μA)IdBμV/kHz=20 log(V/1μV)VdBμV=20 log(V/1μV)V

Units for RBW=1kHzMagAmplitude unitsMagBroad BandNarrow Band

18/11/2010


EMI MEASUREMENT: EMI MEASUREMENT: dBmdBm toto dBdBµµVV

dBm= dBµV-107

)(R10).V(V

)(R10).V(V.10)mW(W

921223

Ωμ

Ωμ −−

==

))(Rlog(10)10log(10))V(Vlog(20))mW(W.(log10 9 Ωμ −+= −

R of measuring instruments is usually 50Ω, then

1798,16)50log(10))(Rlog(10 ; 90)10log(10VdB))V(Vlog(20 ; dBm))mW(W.(log10

9 ≈==−=

==− Ω

μμ

EMI SOURCES AND THEIR FREQUENCYEMI SOURCES AND THEIR FREQUENCY

18/11/2010


EXAMPLE: WOLKSWAGEN FACILITIESEXAMPLE: WOLKSWAGEN FACILITIESFOR EMI TESTSFOR EMI TESTS

transmitting/receivingantennae

RF absorbingtiles for

cancellationof reflections

car under test

EMI TESTSEMI TESTS

EN-61000-6-1 Residential Immunity before EN-50082-1EN-61000-6-2 Industrial Immunity before EN-50082-2

EN-61000-6-3 Residencial Emission before EN-50081-1EN-61000-6-4 Industrial Emission before EN-50081-2

18/11/2010


EMI TESTSEMI TESTS

European Union DirectivesDirective 73/23/CEE (Low Voltage), modified 93/68/CEE (compliance mandatory for “CE” marking) Important when safety ground comes in conflict with EMC issues

Directive 87/404/CEE (Pressure vessels), modified 90/488/CEE and 93/68/CEE (mentioned for completeness)

Directive 2004/108, EMC. (compliance mandatory for “CE” marking) Supersedes : 89/336/CEE modified 92/31/CEE and 93/68/CEE CEE

Directive 89/392/CEE (Machinery), modified 91/368/CEE, modified 93/44/CEE and 93/68/CEE (compliance mandatory for electrical compressors, ventilation, electrical machine tools etc.)

Directive 89/686/CEE (Equipment for individual protection), modified 93/68/CEE and 93/95/CEE (mentioned for completeness)

Corresponding standardsEN 50065-1 Signal transmission on LV linesEN 61000-6-1&2 Generic EMC (emission)EN 61000-6-3&4 Generic EMC (immunity)EN 50091 Uninterruptable power suppliesEN 55011 ISM limits and measurements methods

EN 55013 broadcast receivers noiseEN 55014 Noise by analogue domestic apparatusEN 55015 Noise by Lighting ApparatusEN 55020 broadcast receiver immunityEN 55022 Noise: IT equipment (computers)EN 55104 Immunity of analogue dom. apparatusEN 60521 Energy countersEN 60555 harmonics & flicker induced by dom. e.

EN 60601 Medical apparatus general & EMCEN 61000 EMC standard collection covering all effects 11 sub-standards for immunity aloneLast but not least: Military Standards


Dr. Dr. J. Balcells, J. Balcells, Dr. Dr. D. D. GonzGonzáálezlez, , Dr. Dr. J. GagoJ. GagoDept. Dept. EnginyeriaEnginyeria ElectrònicaElectrònica UPCUPC

33. SOURCES AND VICTIMS OF EMI. SOURCES AND VICTIMS OF EMI

18/11/2010


PASSIVE COMPONENTS: RESISTORSPASSIVE COMPONENTS: RESISTORS

Resistors have a series parasiticinductance and a parallel parasiticcapacitance.Equivalent circuit depends ontechnology

Metal/Carbide film and SMD

Wound technology

PASSIVE COMPONENTS: CAPACITORSPASSIVE COMPONENTS: CAPACITORS

Capacitors have a series parasitic R called ESR (Equivalent series resistance) a parallel discharging R and L in series due to terminal leads

Typical equivalent circuit is as shown below.

18/11/2010


PASSIVE COMPONENTS: INDUCTORSPASSIVE COMPONENTS: INDUCTORS

Inductors have a series parasitic R and a parallelparasitic C

Typical equivalent circuit isas shown below.

ANALOG ANALOG ICsICs SUSCEPTIBILITYSUSCEPTIBILITY

VBEP

IBP

IBPO2IBPO1

t

EMI

ViP ViN

RCP RCN

+VCC

-VCC

IE

VO1ICP ICN

t

TP TN

Audio-rectification in diferential amplifier

Offset output caused by VBE anddifferences between transistors

18/11/2010



ViP ViN

RCP RCN

+VCC

-VCC

IE

VO1ICP ICN

t

TP TNViP ViN

RCP RCN

+VCC

-VCC

IE

VO1

IDP IDN

TP TNViP ViN

RCP RCN

+VCC

-VCC

IE

VO1

IDP IDN

TP TN

Bipolar FET

Input stage of OPAMP

MOS

Input stage of bipolar, FET and MOS behave similar


R1

-Vcc

+Vcc

R2

VO

Vi

GND

Entradasdel A.O.

A.O. configurado comoamplificador inversor

t

EMI

VO

t

Envolvente Función de laEnvolvente

AM demodulation effect in OPAMP

OPAMP output is disturbed by an EMI signal caused by AM demodulation of EMI coupled at the signal and power inputs

Envelope

OPAMP inputs

Inverter amplifierbased on O.A.

Function of envelope

18/11/2010



t

EMI

t

VOVoffVOM

frecf p f p + fmf p - f m

V s (f)

frec

Vo(f)

fm

VODC

VOM

fp

VOP

2fp

Rejected band

VIP

mVIP mVIP

EMI are usually RF signals modulated by random noise in AM

Output of AO contains the low frequency envelope and a continuous voltage

The RF components are negligible

Rejected band

EMISSION OF DIGITAL CIRCUITSEMISSION OF DIGITAL CIRCUITS

Ground bounce

dtdILV GNDGNDGND ⋅=

0 V VGND

IGND

LGND

ERROR

VOVO

The ground bounce is produced when a ground line is used by various digital IC

18/11/2010



Power Supply Switching Noise

dtdILV CCCCCC ⋅=

LCC

ICC

VCC

VO5 V

VO

ERROR

The supply switching noise is produced when a power line is used by various digital IC


Noise to power supply

TransformerL

F

E

5V

0V

EMI produced in digital ICs can be coupled to the power supply lines

18/11/2010



Reflections in data signal lines

ERROR

VOVO

ZO: Trace characteristic impedanceZout: IC output impedanceZin: IC input impedance

OUTOINO ZZorZZ ≠≠ Add termination R

Oin

OinZZZ-Z

+=ρ

Outo

Oout

ZZZ-Z

+=ρ

0=ρ

There are reflections in long and non-terminated lines

d > λ/2π


Crosstalk

V1

RS1

C1RL1

L1

+

V2

RS2

C2RL2

L2

+

C12M

Vout

Vin

Inductive and capacitive coupling together

+

+

Vout

VinRS1

RS2

V1

V2

RL1

RL2There is a crosstalk between small lines separated by short distances

18/11/2010



Differential mode radiation

S: Area ( cm2)

f in MHz IMD Current amplitude in mA

+

+

IDMS

E

Current loops IDM generate differential mode radiation

(V/m) I.S.f.10.263E DM212−=


Common mode radiation

L: cable length in mf in MHz ICM Current amplitude in mA

+

+

VN

E

Cable = Monopole

0V

ICMICM

Common mode current ICM trough cables connected to the PCB generates common mode radiation

Common mode current ICM is created by the ground noise in the GND plane

(V/m) I.L.f.10.26,1E CM4−=

18/11/2010


Factors that influence digital ICs emission

Logic Family– Digital signal spectrum

• Logic levels• Commutation times• Commutation frequency (less importance)

– IC housing and pinout

PCB layout of the digital circuit– Current loop areas– Trace inductance and parasitic capacitances– Separation distance between traces

Filters and decoupling capacitors


CHAPTER 4:PCB


Dr. J. BalcellsDr. J. BalcellsDept. Dept. EnginyeriaEnginyeria ElectrònicaElectrònica UPCUPC

8.8. EMI EMI ModelingModeling

18/11/2010


State of the Art in EMC ModelingState of the Art in EMC Modeling

There is NO ONE modeling technique that will do ‘everything’Problem must be carefully examined to choose a simulator

For Real-World applications EMC engineers need– Tools at various modeling levels– A variety of modeling techniques

Range of Modeling Levels─ Design Rules. Automated design rule checker─ Quasi-Static Tools─ Transmission lines or PEEC: L,C,R Extraction─ Voltage and current couplings due to near field─ Full Wave techniques

Suitable for Power Suitable for Power Electronics ModelingElectronics Modeling

Modeling Software PurposesModeling Software Purposes

18/11/2010


AtAt thethe beginningbeginning GOD GOD saidsaid

MAXWELL equations fully describe electromagnetic fields, but its solution in a real situation is very complex → NEED SIMPLIFICATION

ερ

=⋅∇ Er

0=⋅∇ Br

Ht

HE Mr

rrr

σμ −∂

∂−=×∇

E.tEH

rr

rrσε +

∂∂

=×∇

∫∫∫∫∫ =VS

dv.sdE ρε1rr

∫∫ =S

sdB 0rr

∫∫∫ ∂∂

−=SC

sdBt

ldE rrrr

∫∫∫ ∂∂

=SC

sdDt

ldH rrrr

Gauss’s Law

Faraday’s Law

Ampère’s Law

Differential form Integral form

Magnetic PolesLaw

Analytical Modeling SoftwareSolves specific problems with pre-defined geometries using closed form

equations.Provides fast solutions for a limited type of problems. The user must be able

to relate the geometry of the problem to a geometry that the software is capable to solve.

Models adapted to problem complexity. Suitable for Power Electronics Modeling

Numerical Modeling SoftwareSolves Maxwell's equations subject to appropriate boundary conditions. Provides very accurate solutions to very well-defined problems.Requires the user to be very familiar with the software, the limitations of the

technique, and the problem being analyzed.Require a very precise geometrical description of space. Difficult to

implement in Power Electronics

More Precise EMC More Precise EMC ModelingModeling MethodsMethods

18/11/2010


Choosing the Right Numerical Modeling SoftwareChoosing the Right Numerical Modeling Software

FDTD and FEM PreFDTD and FEM Pre--ProcessingProcessing

• The space needs to be meshed and problem boundaries must be stablished• μ, ε, σ must be defined for each small piece of space• Differential equations are substituted by finite difference equations• FDTD uses cubic mesh (volume based technique) and computes E in the edges and H in the center of each face. Temporal method (all frequencies in one single simulation). Brute Force Method → High computational demand• FEM uses average values in the volume of a mesh cell and usually employs frequency resolution methods (simulates a single frequency at a time) • Time step must be small enough so that fields do NOT propagate faster than the speed of light.

18/11/2010


FDTD (FDTD (Finite Differences Time DomainFinite Differences Time Domain) 3D) 3D

The FDTD method is a numerical technique based on the finite difference conceptthat is employed to solve Maxwell’s equations for the electric and magnetic fielddistributions in both the time and spatial domains.

Ht

HE Mr

rrr

σμ −∂

∂−=×∇

Faraday’s Law

E.tEH

rr

rrσε +

∂∂

=×∇

Ampère’s Law

Some examples of numerical solversSome examples of numerical solvers

• Static Field solvers

Fasthenry, Fastcap, Fastlap, Flux2D, Flux3D

• 2D Solvers

SUPERFISH, Quickfield

• Transmission Line Solvers

Microwave Explorer, EM

• 3D Full-Wave Solvers

NEC, XFDTD, EMA3D, Maxwell 3D, EMAP, EMIT, IE3D, HFSS, MiniNEC, MaxSIM-F, MSC EMAS, MagNet

18/11/2010



Dr. J. BalcellsDr. J. BalcellsDept. Enginyeria Electrònica UPCDept. Enginyeria Electrònica UPC

9. 9. EMC and Power ConvertersEMC and Power Converters

Power Converters are one of the main responsible for certaindisturbances produced on the supply network.

The most relevant problems are related with harmonics of mainsfrequency. EMC in this field requires some filters to avoid seriousproblems

Other problems related with HF switching frequency will be dealtlater

EMI CAUSED BY CONVERTERS ON THE MAINSEMI CAUSED BY CONVERTERS ON THE MAINS

18/11/2010


LOW FREQUENCY DISTURBANCESLOW FREQUENCY DISTURBANCES

• Most Power Converters are supplied by single phase or three phase rectifiers directly connected to mains → Causes EMI

Typical harmonics of a 6 pulse rectifiern=6k±1

LOW FREQUENCY DISTURBANCES : HARMONICSLOW FREQUENCY DISTURBANCES : HARMONICS

18/11/2010


Typical harmonics of a single phase rectifier

LOW FREQUENCY DISTURBANCES : HARMONICSLOW FREQUENCY DISTURBANCES : HARMONICS

HARMONICS IN THE POWER LINESHARMONICS IN THE POWER LINES

18/11/2010


FILTERS FOR CONVERTERSFILTERS FOR CONVERTERS

Static converters generate different types of disturbances in the net side as well as in the load side.Need of harmonic filters to solve such problems to fulfill with the standards EN-60000-4-3, IEEE-519 and EMC directive

LR FILTERS: REACTORSLR FILTERS: REACTORS

LR filtering reactors permit the reduction of harmonic currentgenerated by a converter from 20% up to 50%.

18/11/2010


SINGLE PHASE LR REACTORSSINGLE PHASE LR REACTORS

LR filtering reactors DO NOT WORK PROPERLY in single phase rectifiers

It is installed individually, upstream from the converter¿¿How How toto connectconnect??

M3

REACTANCIADE LA RED

CONVERTIDOR

L1 L2

L3

C

LCL Filter: 6 pulse convertersLCL Filter: 6 pulse converters

18/11/2010


SINUS FILTERSSINUS FILTERS

“Sinus” filters and du/dt filters are used between the converter and motor in inverters with PWM output to improve the waveform and to avoid overvoltages.

HF EMI CAUSED BY POWER CONVERTERS:HF EMI CAUSED BY POWER CONVERTERS:EQUIVALENT CIRCUITS FOR EVALUATIONEQUIVALENT CIRCUITS FOR EVALUATION

Dr. J. BalcellsDr. J. BalcellsDept. Dept. EnginyeriaEnginyeria ElectrònicaElectrònica UPCUPC

18/11/2010


EMI in Switched Mode Power Converters (SMPC) is characterized by:Different sizes, from low power high switching frequency (mobile telecom)

to high power (drives and energy conversion systems)Low power have definite lay-out (PCB). High power have a complex lay-

out and cablingMix of high power signals and weak signals in a very close spaceWide range of time constants: • Mains supply harmonics• Switching frequency harmonics• Rise and fall transients

Very strict design requirements:Low volume, low weight, low costConverters are more and more in domestic environment (exigence of

lower emission)Converters are also in industrial environment (exigence of higher

immunity)Design to avoid low frequency pollution in the mains (unit PF and low

harmonics consumption)

EMI IN POWER CONVERTERS: EMI IN POWER CONVERTERS: General General ConsiderationsConsiderations

Conducted EMI

EMI IN POWER CONVERTERS: EMI IN POWER CONVERTERS: General General ConsiderationsConsiderations

18/11/2010


Classical circuits theory is not enough to explain HF EMI phenomenaSimulation becomes difficult because of:

Geometrical complexity and many different boundariesDifficult to translate to simple mathematical modelsMany (known or unknown) variables involvedNon linear behaviourImpossible to apply FDTD, FEM, MoM methods to the whole system

Nevertheless those are applicable to obtain models of partsWide range of time constants (range 1:1000):

• Resolution in time domain requires very small time steps andcauses convergence problems• Large dimensions and high frequency: Involves conducted, near field and far field phenomena

Solution involves the combination of many different methods: High scaleequivalent circuits, Small scale equivalent circuits PEEC, Transmissionlines, etc. and combination of time domain and frequency domainprocedures

HOW TO DEAL WITH THE EMC PROBLEM IN HOW TO DEAL WITH THE EMC PROBLEM IN POWER SYSTEMSPOWER SYSTEMS

METHOD BASED ON: SOURCE METHOD BASED ON: SOURCE →→ PROPAGATION PATH PROPAGATION PATH →→ EMIEMI

Propagationpaths

(Z(jω))

t

v,i

Disturbances

(currents & voltages)

Time domain 1 CYCLE SIM

Frequency domain

Sources ofdisturbance

PasiveComponents

Controlparameters

τ 2

τ 1

d1t

d2t

I

Î 1

Î 2

Time domain PERIODIC SIM

Lay-outchanges

Active Components

f

⏐A⏐

18/11/2010


EQUIVALENT CIRCUIT MODELING METHODEQUIVALENT CIRCUIT MODELING METHODMatrixMatrix ConverterConverter ExampleExample

SOURCE MODEL: SOURCE MODEL: Long Long termterm simulationsimulation fromfrom MATLABMATLAB--SIMULINKSIMULINK

Example: Matrix Converter: Long term , low resolution simulation30Hz Output, 2kHz switching frequency

18/11/2010


SOURCE MODEL: SOURCE MODEL: SwitchingSwitching detailsdetails

Example: Matrix Converter: Short time , high resolution simulation with PSpice. Allows HF model refinement

EQUIVALENT CIRCUIT MODELING METHODEQUIVALENT CIRCUIT MODELING METHODMatrixMatrix ConverterConverter ExampleExample

Combine long term simulation and transient.Translate into frequency domain by means of FFT

18/11/2010


SOURCE MODEL: SOURCE MODEL: FirstFirst approachapproach

Power converters voltagewaves can be simulated in Matlab- Simulink

Due to rise and fall times, a first wave approximation istrapezoidal

Transients due to powersemiconductors delayscombined with snubbers mustbe introduced separately

Transients can be simulatedin Pspice or a more suitableprogram having semiconductor models

Results of both in time domain must be combined

Obtain sources model fromFFT of resulting wave

86

Square wave plus damped oscillation

)2

()2

()()()()( TtgTtutgtutfth −−−+=

)1)(()()( 20000

TjnejnGjnFjnH

ωωωω −+=

⎥⎦

⎤⎢⎣

⎡−

−+++−= −−− )1(

)()21(

22)( 2

02

20

πππ

αωωω

πω jn

r

rjnjn ejnAee

jnET

TjnH

)()( tsinAetg rt ωα−=

SOURCE MODEL: Analytic ModelSOURCE MODEL: Analytic Model

18/11/2010


EQUIVALENT CIRCUIT :PROPAGATION PATHSEQUIVALENT CIRCUIT :PROPAGATION PATHSMatrixMatrix ConverterConverter : : PartsParts ModelsModels

EQUIVALENT CIRCUIT :PROPAGATION PATHSEQUIVALENT CIRCUIT :PROPAGATION PATHSMatrixMatrix ConverterConverter : : PartsParts ModelsModels

Significative impedances of the system must be identified andmeasured (Real components and stray Ls and Cs)DRAWBACK: Parts can be measured, but for lay-out modeling a prototype must be built, because EMI propagation depends onthe geometry.STRENGTHS: Modeling helps improving the prototype, since the propagationpaths are identified.CM and DM effects can be seen separately, which helps in EMI filter design

18/11/2010


( ) ( ) ( )( )( )

( )( )

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⎣

⎡

−−−−−−−−−−−

++++−−−++++−−−++++++

−−

=

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⎣

⎡

+

+

0000

0000

0011000011

000000

111

111

11111

1111111

1111111

11111111111

motCmotBmotA

CBCCACMC

MEBMEAOCOBOA

CABCBCMEBMBLISNBMLBCABMB

CABMAMEACABCACLISNAMLAMA

MBMAOCmotCMCOBmotBMBmotAOAMA

ZZZ

ZZZ

ZZZZZ

ZZZZZZZZ

ZZZZZZZZ

ZZZZZZZZZZZ

bc

ab

vv

( ) ( )

( )( )( )

( ) ⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⎣

⎡

•

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

⎦

⎤

+++++++−

−++++++−−−−

−−−−−−−

+

8

7

6

5

4

3

2

1

1111

111111

11111111

11

11

1111111

2

1

000000

000000

vvvvvvvv

motCmotBmotAG

MCMECCBCCACLISNCMLCMEC

MECGMEBMEAOAOBMECOC

CBCMEB

CACMEA

motAmotBmotCMCOCOAOB

ZZZZ

ZZZZZZZ

ZZZZZZZZ

ZZ

ZZ

ZZZZZZZ

EQUIVALENT CIRCUIT MODELING METHODEQUIVALENT CIRCUIT MODELING METHODMatrixMatrix ConverterConverter HF HF ImpedanceImpedance MatrixMatrix

COMPARISON OF CM CURRENTSIMULATION AND MEASURED

Good agreement in the range 20kHz to 1MHz

Model must be refined for higher frequencies.

Peak values have errors less that 6dB. Considered a good simulation because of uncertainty

Intermediate frequencies depend on RBW of measuring instrument

The method allows to establish comparatives when lay-out or design changes are introduced.

The model allows prediction at a second attempt (after first prototype has been built)

EQUIVALENT CIRCUIT MODELING METHODEQUIVALENT CIRCUIT MODELING METHODMatrixMatrix ConverterConverter : CM : CM currentcurrent evaluationevaluation

18/11/2010


x

=

IL

FMOD

ID

-1OFFONONOFF2

+1ONOFFOFFON1

FMODQ4Q3Q2Q1State

Differential Mode EMI source

Cel

ec.

ID

Im3

2

ID

ZL

32

32

Im

Cde

cpIL

32

Q3Q1

Q2

LI

SN

Q4

ID reverses its sense when inverter diodesconduct, causing EMI due to L in DC bus

( )( )( )

o oL o

L o

V j nI j nZ j n

ωωω

⋅⋅ =

⋅

( ) ( )O o m oV j n f j n Eω ω⋅ = ⋅ ⋅ [ ] [ ][ ][ ]

( ) ( ) ( ) ( ) ( )

( ) ( )

( ) ( )Circular convolution

D L m L m

L L

m m

DFT I k DFT i k f k I n F n

I n DFT i k

F n DFT f k

= ⋅ = ©

=

=

©=

1

0

( ) ( ) ( ) ( ) ( )N

D L m L mm

I n I n F n I m F n m−

=

= © = ⋅ −∑

Load current calculation

mf

LI DI

DC current spectrum

Modelling ID from V0 , ZL

18/11/2010


IMD(jω)

iMD(jω)

ZL(jω) ZC(jω)

)j(Z)j(Z)j(Z)j(I)j(i

LC

CMDMD ωω

ωωω+

=

SINGLE PHASE INVERTER: DM MODELSINGLE PHASE INVERTER: DM MODEL

Important issues in DM:Pieces of the circuit where high di/dt exist must be identified. A typical circuit where this occurs is the DC bus of DC/DC and

inverters

400 kHz

Experimental ResultsExperimental Results

A band measurements10 – 150 kHz

B band measurements150 – 1MHz

Model & Real measurements comparison for D=0,5

18/11/2010


400 kHz

Experimental ResultsExperimental Results

A band measurements10 – 150 kHz

B band measurements150 – 1MHz

Model & Real measurements comparison for D=0,2

USE OF FIRST APPROACH MODELS USE OF FIRST APPROACH MODELS

Absolute values of conducted EMI obtained with the firstapproach models are quite good in band A (10kHz – 150kHz)

Absolute values of of conducted EMI obtained with the firstapproach models present important deviations in band B (150kHz to 30MHz)

Even being imprecise in absolute terms, the model results may be still useful to determine the improvements obtained by certainlay-out or control changes. Following we present an example

Precise models in band A and B require a modeling techniquewhere a first prototype is built and sources and propagationpaths can be measured, not simulated.

The first approach method is useful to improve the design frompoint of view of EMI emission, allowing the discrimination of CM and DM and discrimination of parts causing the problem

18/11/2010


EMC IN POWER CONVERTERS:EMC IN POWER CONVERTERS:MODELING PARTS FOR HF MODELING PARTS FOR HF RADIATION PREDICTIONRADIATION PREDICTION

Dr. J. BalcellsDr. J. BalcellsDept. Dept. EnginyeriaEnginyeria ElectrònicaElectrònica UPCUPCBased on the thesis of: Based on the thesis of: JJèèremieremie AimAiméé, , supervised by Dr. James Roudetsupervised by Dr. James RoudetLaboratoire de GLaboratoire de Géénie nie ElectriqueElectrique de Grenoblede Grenoble

MODELING HF IN POWER CONVERTERSMODELING HF IN POWER CONVERTERS

We shall present models based on approximate equivalent sources and equivalent circuits

We pretend the obtention of models valid in the whole range of conducted EMI (up to 30MHz). They will be based on:

Real measured waveforms for sources, including commutation details.

Real measured impedances of certain parts and stray propagation paths. If this is not available it can be sustituted by a refined simulation based on MoM or PEEC.

Detailed models based on PEEC or MoM

Precise models can be used to make radiated field predictions.

Usually the predictions of radiated EMI are based on the method of equivalent monopoles and dipoles

18/11/2010


RADIATED FIELD IN SMPCRADIATED FIELD IN SMPC

Far field radiation of SMPC is mainly due to CM currents in DC bus and cabling (more than DM)

CORRELATION OF FAR FIELD AND CM CURRENTCORRELATION OF FAR FIELD AND CM CURRENT

Far field and CM current correlation in a DC/DC converter

Far field radiation related with CM currents in DC bus and cabling

18/11/2010


EMC IN POWER CONVERTERSEMC IN POWER CONVERTERS

CM and DM current needed to obtain a field of 30dBμV/m (Limit of IEC-61800-3, CISPR 11 , C1 category) (Very low in CM, higher in DM)

L=10cm

d=1cm

DC/DC/DCDC CONVERTER:CONVERTER:SupplySupply cable cable lengthlength influenceinfluence onon radiatedradiated farfar fieldfield

Vertical Polarisation Horizontal Polarisation CM

Cables supplying the DC/DC converter have very low influence

18/11/2010


DC/DC/DCDC CONVERTER:CONVERTER:Load cable Load cable lengthlength influenceinfluence onon radiatedradiated farfar fieldfield

Vertical Polarisation Horizontal Polarisation CM

Cables supplying the load of DC/DC converter have high influence

CM LEAKAGE WITH AND WITHOUT SCREEN PLANECM LEAKAGE WITH AND WITHOUT SCREEN PLANE

Capacitance between signal track and GND decreases from 198pF to 18pF, thus reducing CM current

18/11/2010


Reduction of ICM due to screen plane. Notice the importance of a precise measure of Lcable to detect the resonance

CM LEAKAGE WITH AND WITHOUT SCREEN PLANECM LEAKAGE WITH AND WITHOUT SCREEN PLANE

CM LEAKAGE IN AN INVERTER LEG:CM LEAKAGE IN AN INVERTER LEG:InfluenceInfluence onon thethe farfar fieldfield radiationradiation

Model of an inverter leg with screen plane.

Current distribution in the different paths.

Less surface → Less far field radiation H,E

18/11/2010


Reduction of CM current in the LISN+ due to screen plane.

Conducted EMI reduction

CM current in the LISN -- due to screen plane.

Part of iCM goes through cable screen

CM LEAKAGE IN AN INVERTER LEG:CM LEAKAGE IN AN INVERTER LEG:InfluenceInfluence onon thethe farfar fieldfield radiationradiation

LISN + with screen planeLISN + without screen plane

LISN + with screen planeLISN + without screen plane

MODELING RADIATION OF A DC/MODELING RADIATION OF A DC/DCDC CONVERTERCONVERTER1st 1st TopologyTopology

First Topology Lay-out: 1 layer. GND plane very far from active tracks

Converter Schematic

18/11/2010


2nd Topology Lay-out: 2 Layers. Close Screen Plane not connected to GND

MODELING RADIATION OF A DC/MODELING RADIATION OF A DC/DCDC CONVERTERCONVERTER2nd 2nd TopologyTopology

Converter Schematic

MODELING RADIATION OF A DC/MODELING RADIATION OF A DC/DCDC CONVERTERCONVERTER3rd 3rd TopologyTopology

3rd Topology Lay-out: 2 Layers.

Converter Schematic

18/11/2010


MODELING RADIATION OF A DC/MODELING RADIATION OF A DC/DCDC CONVERTERCONVERTER4th 4th TopologyTopology

4th Topology Lay-out: 3 Layers

Converter Schematic

MODELING RADIATION OF A DC/MODELING RADIATION OF A DC/DCDC CONVERTERCONVERTER

LISN values

18/11/2010


CM current measurement in the different CM current measurement in the different topologiestopologies

Test configuration

Radiated field measurement in the different Radiated field measurement in the different topologiestopologies

18/11/2010


OTHER OTHER EXAMPLES:MotorEXAMPLES:Motor DriveDrive

OTHER EXAMPLES: Motor driveOTHER EXAMPLES: Motor drive

18/11/2010


OTHER OTHER EXAMPLES:MotorEXAMPLES:Motor DriveDrive

OTHER EXAMPLES: Motor DriveOTHER EXAMPLES: Motor Drive

18/11/2010



Influence of Cy_PA, between screen plane and DC_Bus+


18/11/2010




18/11/2010


CONCLUSIONSCONCLUSIONS

Models based on approximate equivalent sources and equivalent circuits, work well up to 3MHz

Models valid in the whole range of conducted EMI (up to 30MHz) are based on:

Real measured waveforms for sources, including commutation details.

Real measured impedances of certain parts and stray propagation paths. If this is not available it can be substituted by a refined simulation based on MoM or PEEC.

Detailed models based on PEEC or MoM can be used to make even radiated field predictions.

Usually the predictions of radiated EMI are based on the method of equivalent monopoles and dipoles

ReferencesReferences

[1] V. Jithesh and D. C. Pande, "A review on computational EMI modelling techniques," in Electromagnetic Interference and Compatibility,2003. INCEMIC 2003. 8th International Conference on, 2003, pp. 159-166.

[2] R. Scheich and J. Roudet, "EMI conducted emission in the differential mode emanating from an SCR: phenomena and noise level prediction," Power Electronics, IEEE Transactions on, vol. 10, pp. 105-110, 1995.

[3] D. Gonzalez, Balcells, et alt, "New simplified method for the simulation of conducted EMI generated by switched power converters," Industrial Electronics, IEEE Transactions on, vol. 50, pp. 1078-1084, 2003.

[4] J. C. Crebier, et al., "A new method for EMI study in boost derived PFC rectifiers," in Power Electronics Specialists Conference, 1999. PESC 99. 30th Annual IEEE, 1999, pp. 855-860 vol.2.

[5] F. Costa, et al., "Modeling of conducted common mode perturbations in variable-speed drive systems," Electromagnetic Compatibility, IEEE Transactions on, vol. 47, pp. 1012-1021, 2005.

[6] J. Balcells, Lamich, M. Bedford, D., "EMI Generation Models forSwitched Mode Power Supplies," The Smithsonian/NASA Astrophysics Data System ADS, vol. 416, pp. 421-426, 1998.

18/11/2010


ReferencesReferences

[7] L. Qian, et al., "Modular-Terminal-Behavioral (MTB) Model for Characterizing Switching Module Conducted EMI Generation in Converter Systems," Power Electronics, IEEE Transactions on, vol. 21, pp. 1804-1814, 2006.

[8] J. F. Kolar, T. Krismer, F. Round, S, "The essence of three-phase AC/AC converter systems," Przeglad Elektrotechniczny, pp. 14-29, July 2008

[9] H. Akagi and T. Shimizu, "Attenuation of Conducted EMI Emissions From an Inverter-Driven Motor," Power Electronics, IEEE Transactions on, vol. 23, pp. 282-290, 2008.

[10] U. T. Shami and H. Akagi, "Experimental Discussions on a Shaft End-to-End Voltage Appearing in an Inverter-Driven Motor," Power Electronics, IEEE Transactions on, vol. 24, pp. 1532-1540, 2009.

[11] Joao Pedro, A. Bastos, Nelson Sadowski, Electromagnetic Modeling by Finite Element Methods, (book) ISBN-0-8247-4269-9, Marcel Decker (2003)

[12] Josep Balcells, Rational Use of Electrical Energy, book, ISBN-84-699-2666-7 , Circutor SA (2001)

[13] Josep Balcells, Francesc Daura, Rafael Esparza , Ramon Pallas, Interferencias Electromagnéticas en Sistemas Electrónicos, book, ISBN-84-267-0841-2, Marcombo 1991

upc universitat politècnica de catalunya - emc design in industrial systems · 2020. 4. 27. ·...

Documents