conducted noise prediction method for inverter & converter...

1 © 2016 ANSYS, Inc. November 20, 2017

Conducted noise prediction method for Inverter & Converter

ANSYS KOREASeungjoo Kim

[email protected]


CONTENTS

I. System simulation Introduction (CE)

II. Key point- Switching Device model

- PCB & Power Module model

- Passive component model

- Magnetic device model

III. Simulation Result

IV. Case Study

V. Conclusion


System simulation Introduction for EMI (CE)


Systems Context: Physics + Controls

Electronic Control

Operating Conditions

Safety Requirements

Operational Profiles

Embedded SoftwareSensors

Actuators

How does the system perform?


System Simulation for Power ElectronicsParasitic Effects in Power Electronic Systems

IGBT based inverter

000

0

0

PhaseBPhaseA

AA

A

A

A

Amp_4

+

V

+

V

+

V

+

V

+

V

+

V

A

+

V

+

V

+

V

+

V

+

V

+

V

+

V

+

V

+

+

Q3D ROM

System-Level Use Case• Validate compliance with EMC

requirements (conducted emissions)• Optimize the performance of power

electronic systems, including AC drive system, DC links, and IGBT modules

Keys to the System Model • Reduced-Order Models of package,

busbar, and cable parasitics from ANSYS Q3D and Maxwell


Motivation of Simulation


Three Factors – EM Noise Issue

1) Noise Source(EMI) – Radiated or Conducted Emission

2) Load (Immunity) – immunity response against EMI

3) Noise Path

EMISource

Immunity

ConductiveNoise

InductiveNoise

CapacitiveNoise

RadiationNoise

Conduction Path

Space Path

High Frequency(HFSS)

Low Frequency

Space

“EMC ANALYSIS METHODS AND COMPUTATIONAL MODELS”: Frederick M. Tesche, Michel V. Ianoz:John Wiley & Sons, Inc. 1997 p34-36

Non-Radiation


What is CM/DM noise?

DM noise cause Voltage&Currentdifference between both lines supplying electric power

Differential Mode

Common Mode

CM noise is occurred due to current flow to Ground and makes Voltage&Current difference between Ground and power paths


Equivalent circuit of LISN➢ LISN (Line Impedance Stabilization Network)


Key pointsfor Precise Simulation


Simulation Model

Tri-Phase 400Vrms Inverter

Input : DC source, Output : Resistance

Cy : Stray Cap of Solar panels


Measure Environment

SiC inverter

Load Resistor

(12.5ohm*3)

LISN

600Vdc

3phase Normal mode filter


System Overview➢ STEP1. Switching Device Model


Half-Bridge Power Module


Transfer Characteristic

Key point – Switching Device Model(Simplorer)

Output Characteristic

ANSYS Simplorer

Dynamic Characteristic


Key point – Switching Device Model(Simplorer)

Equivalent Circuit for SimulationExperimental setup

Turn off Turn on

100ns/di

v

100ns/di

v

---simulation---Experimental

---simulation---Experimental


System Overview➢ STEP2. Circuit board & Power module & Cable Model


Influence of Parastic LCR

rA

SA

SYTR

VL

dt

dILV

Surge Noise

VY: Surge Voltage(V)、Ls: Line Self inductance(nH)ΔV: Pulse Voltage(V)、Ra: Driver side impedance(ohm)Tr: Pulse Rise time(ns)

rA

Minduc

TR

LCrosstalk

r

MBcap

T

CRCrosstalk

High-Speed Digital Design”:Haward Johnson, Martin Graham:Prentice Hall PTR 1993 P25-P36

capinductotal CrosstalkCrosstalkCrosstalk

Crosstalk Noise

LM: Line Mutual Inductance , CM: Line CapacitanceRA: Driver side impedance(ohm)RB: Receiver side impedance(ohm)


Key point - Circuit Board & Power Module & Cable (Q3D)

Simplorer Schemetic

Self Inductance


Key point - Circuit Board & Power Module & Cable (Q3D)

The Value of Measurement and Simulation are almost identical


Importing Q3D model to Simplorer


System Overview➢ STEP3. Passive components & LC filter model


Key point – Magnetic device

B-H NL Char.- Flux density rely on the magnetic field- Permeability of core rely on the Freq

Inductance rely on Freq&Current

Skin/Proximity effect- Current distribution change rely on the Freq

Resistance & Inductance rely on the Freq

Coupling Maxwell and Simplorer

- Magneto static (saturation) Table model

- Eddy current (skin/proximity, Freq dep) SS model

-Transient (saturation, skin/proximity) Co-simulation

Coupling HFSS/Q3D and Simplorer

- State Space model ( S-parameter )


Key point – Magnetic device (Maxwell)

B-H Curve :Saturation characteristic of Core

Convert to Table model

1/14 Model


Key point – Magnetic device (HFSS/Q3D)

Simplorer’s System Circuit


Key point – Magnetic device (HFSS/Q3D)

Without Choke filter With Choke filter

115dBuV

118dBuV100dBuV

73dBuV

17dBuV

Vdm (dBuV)

Vcm (dBuV)

Vdm (dBuV)

Vcm (dBuV)


Key point – Passive Components (Simplorer)

1.00E-001 1.00E+000 1.00E+001 1.00E+002 1.00E+003 1.00E+004 1.00E+005F [kHz]

-90.00

0.00

90.00

Phase [deg]

-50.00

-25.00

0.00

25.00

Gain

[dB

]

00_nonideal_RImpedance

Curve Info

-E1.V/E1.IAC

R1C1

L1

n

p

+

V

VM1

A

AM1

p

n

E1

Resistor model

n

p

R1

C1

L1

+

V

VM1

A

AM1

Inductor model

1.00E-001 1.00E+000 1.00E+001 1.00E+002 1.00E+003 1.00E+004 1.00E+005F [kHz]

-90.00

0.00

90.00

Phase [deg]

-40.00

-15.00

10.00

25.00

Gain

[dB

]01_nonideal_LImpedance

Curve Info

-E1.V/E1.IAC

n

p

R1

C1

L1

+

V

VM1

A

AM1

1.00E-001 1.00E+000 1.00E+001 1.00E+002 1.00E+003 1.00E+004 1.00E+005F [kHz]

-90.00

0.00

90.00

Phase [deg]

-40.00

-15.00

10.00

35.00

60.00

Gain

[dB

]

02_nonideal_CImpedance

Curve Info

-E1.V/E1.IAC

Capacitor model

Passive Components with Frequency dep. S-parameter model or Equivalent circuit model Ignore Saturation characteristic


Result


Simulation and Experimental result

Similar trend Less than 15dB error


Case study


Case study – Flyback converter



Capacitance Matrix

(30 min) – 1 >10pF

– 11 5-10pF

– 24 1-5pF

–428 <1pF

•Inductance Matrix: R,L,M

(30 min) –63 Paths (23 Pri & 40 Sec)

–31 Paths (16 Pri & 25 Sec)



0.15 0.3 1 3 10 30-20

0

20

40

60

80

100

120Simulated -Black vs Measured -Red CM EMI Spectrum

CM

Nois

e (

dBV

)

0.15 0.3 1 3 10 30-20

0

20

40

60

80

100

120Simulated -Black vs Measured -Red DM EMI Spectrum

DM

Nois

e (

dBV

)

Frequency (MHz)

Simulated

Measured


n

p

R1

C1

L1

+

V

VM1

A

AM1



0.15 0.3 1 3 10 30-20

0

20

40

60

80

100

120Simulated Total EMI Spectrum

Noi

se (d

B V

)

Frequency (MHz)

Nominal EMI

Filtered EMI

Filtered/Shielded EMI

CISPRA

CISPRB



Conclusion

- Switching device, one of noise source

- Need to Detail Switching device model Ideal < SPICE < Dynamic

- influence to noise of Magnetic device

- Passive component with Frequency dependent impedance

- With accurate model, Possible EMI prediction.

- Estimate result after enhancement


Thank You

Discussion & Next Steps

conducted noise prediction method for inverter & converter...

Documents