Electromagnetic and Circuit Electromagnetic and Circuit Simulation of Injection Probes Simulation of Injection Probes for Bulk Current Injection for Bulk Current Injection 2009 CST European User Group Meeting March 16-18, 2009, Darmstadtium Congress Centre, Darmstadt, Germany POLITECNICO DI MILANO Dept. of Electrical Eng. EMC Group @ POLIMI Milan, Italy F. Grassi, L. Di Rienzo, F. Grassi, L. Di Rienzo, and S. A. Pignari and S. A. Pignari
21
Embed
Electromagnetic and Circuit Simulation of Injection · PDF fileElectromagnetic and Circuit Simulation of Injection Probes for Bulk Current Injection 2009 CST European User Group Meeting
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Electromagnetic and Circuit Electromagnetic and Circuit Simulation of Injection Probes Simulation of Injection Probes
for Bulk Current Injectionfor Bulk Current Injection
2009 CST European User Group MeetingMarch 16-18, 2009, Darmstadtium Congress Centre,
Darmstadt, Germany
POLITECNICO DI MILANO Dept. of Electrical Eng.EMC Group @ POLIMIMilan, Italy
F. Grassi, L. Di Rienzo,F. Grassi, L. Di Rienzo,and S. A. Pignariand S. A. Pignari
2 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
Research motivationResearch motivation
In real test setreal test set--upsups, the correlationcorrelation between noise levelsnoise levelsexpectedexpected and effectively injectedeffectively injected in the EUT may be seriously jeopardized by
electrically-long wire harness
mismatching at terminations
probe loading effects
multi-wire bundles
Need for unambigous correlationunambigous correlation
circuit/numerical simulation models
real/virtual calibration structures
possible extension to multi-wire cables
3 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
LumpedLumped--Pi circuit modelPi circuit model
Accurate lumpedlumped--PiPi circuit model
recently proposed
F. Grassi, F. Marliani, S. A. Pignari, F. Grassi, F. Marliani, S. A. Pignari, IEEE Trans. EMC,IEEE Trans. EMC, vol. 49, Aug. 2007.vol. 49, Aug. 2007.
represents the probe clamped on the
cable under test
stems from measurement datameasurement dataad hoc procedure of de-embedding
accounts for frequencyfrequency--dependent effectsdependent effects
ferrite complex permeabilty
parasitics, setup-related effects, etc.
C2 C2
L1w0
LC
L2(ω ) M(ω )
L1(ω )
VRF RS
C1CC+
4 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
Estimation of the inductive couplinginductive coupling requires the
knowledge of the complex permeability spectra of the corecomplex permeability spectra of the core
These spectra are retrieved via an experimental procedureexperimental procedureVNA measurementsVNA measurements of the probe input impedanceinput impedance in the absence of secondary circuit
dede--embeddingembedding of the effects due to the primary winding and the input connector/adapter
L1w0LN
L1(ω )C1+ CN
PROBE INPUTPROBE INPUTIMPEDANCEIMPEDANCE
MEASUREMENTMEASUREMENT
5 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
The The ““CoreCore”” problemproblem
)('')(')(ˆ ωµωµωµ rrr j−=
Complex spectraspectra of effectiveeffectivepermeability
inherent responseinherent response of a ferrite specimen
LL00 coreless self-inductance of the primary winding
Complex and frequencyfrequency--dependent inductancesdependent inductances
10 /)(ˆ)(ˆ NLM r ωµω = dr LNLL 22102 /)(ˆ)(ˆ += ωµω
106-100
0
100
200
300
400
500
107 108Frequency, [Hz]
||µµrr(f)|(f)|µµrr’’(f)(f)µµrr’’’’(f)(f)
)(ˆ)(ˆ01 ωµω rLL =
6 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
Rationale for CST MWS modelingRationale for CST MWS modeling
Why?Why?1.1. validationvalidation and extensionextension of the probe circuit model
2. simulation of set-ups involving multimulti--wire bundleswire bundlesCM vs. DM injection test procedures
3.3. virtual testingvirtual testing for EMC assessment, according to directive 2004/108/CE
4.4. optimizationoptimization of BCI probes designgeometrical dimensionsmaterial properties
How? How? A priori knowledge of :
1. geometrical dimensions of the probe (both outside and insideinside)
2. frequency spectra of intrinsic permeabilityintrinsic permeability of the ferrite core
7 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
CST MWS probe modelCST MWS probe model
Includes Includes probe metallic frame PECprimary winding (wound around the magn. core) PECinput connector/adapter pair PEC & dielectricstoroidal ferrite core freq. dependent material
L. Di Rienzo, F. Grassi, S. A. Pignari, L. Di Rienzo, F. Grassi, S. A. Pignari, ““FIT modeling of injection probes FIT modeling of injection probes for bulk current injection,for bulk current injection,”” in in Proc. ACES 2007Proc. ACES 2007, Verona, Italy , Verona, Italy ..
8 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
EM model of the input connectorEM model of the input connector
The adapter/connector pair is modeled as the chain connection of 3 coaxial lines with Z3 coaxial lines with ZCC = 50 = 50 ΩΩ
Model validationModel validation is obtained via preliminary simulation and VNA measurement of the connector/adapter series, in the absence of the probe ( ( SS1111 ---- 300 kHz 300 kHz –– 600 MHz600 MHz)
f [Hz]106 107 108
-60
-40
-20
0
|S11
| [d
B]
measurementprediction
|S11|
Imag(S11)Real(S11)
9 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
Modeling the ferrite coreModeling the ferrite core
InherentInherent response of the ferrite (i.e., of a ferrite specimenferrite specimen)
usually not available to endnot available to end--usersusersnot retrievable from meas. datanot retrievable from meas. data, due to superposition and interactionwith dimensional phenomena (effective permeability effective permeability spectra)
Iterative procedureIterative procedure for core characterizationguess a frequency modelfrequency modelfit model parametersfit model parameters by comparison vs. Zin meas. until optimal fitting
11stst order DEBYE MODELorder DEBYE MODELDISPERSION PHENOMENADISPERSION PHENOMENA
( )ωτµµµωµ
js
r +−
+= ∞∞ 1
ˆ105 106 107 108 109 10100
100
200
300
400
frequency [Hz]µ r
= µ r' -
j ωµ r''
|µr|
µr'
µr''
10 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
Modeling the ferrite core Modeling the ferrite core contcont’’dd
Preliminary validation of the model
reflection testreflection test, SSinin, at the input port of the probewaveguide port at the adapter input52080 hexahedral cells mesh
in
inin S
SRZ−+
=11
0
Sin
106 107 108-80
-60
-40
-20
0
f [Hz]
S 11[d
B]
Real(S11)
Imag(S11)
|S11|
measurementprediction
11 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
Different modeling strategiesDifferent modeling strategies
11stst order DEBYE MODELorder DEBYE MODEL(dispersion phenomena)(dispersion phenomena)
( )ωτµµµωµ
js
r +−
+≅ ∞∞ 1
ˆ
105 106 107 108 109 10100
100
200
300
400
frequency [Hz]
µ r = µ
r' - j
ω µ
r''
|µr|
µr'
µr''
CST MWS EM modelCST MWS EM modelinherent response of the materialinherent response of the material
0)(ˆ)(ˆ Λ=Λ ωµω rLL11((ωω), L), L22((ωω), and ), and M(M(ωω) ) proportional toOldOld model
ff--dependent behaviordependent behavior of the core: one parameter onlyone parameter only, i.e., L1(ω)coupling between winding: ideal transformerideal transformer (N1:1)leakageleakage inductanceinductance (L2d): analytically estimatedanalytically estimatedmore suited for SPICE implementation SPICE implementation via ABM modules
NewNew model
13 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
Implementation in SPICE Implementation in SPICE contcont’’dd
V2
I2
V1
I1
L1(ω )
N1:1 L2d
V2'
I1' I1
''V2''
gmV1V1
I2I1=0A VCCSVCCS with transfer function
is used to model the shunted branch, i.e., phenomena due to the core frequency behavior core frequency behavior
)(ˆ11
1
ωωLj
VIgm −=′
=
-++
-
E1
GAIN = 1
V(%IN+, %IN-)
GFREQ/GLAPLACE MODULE
OUT+OUT-
IN+IN-
F1
GAIN = -1
L2d
55 nH
0
2 possible2 possibleABMABM
1. GFREQ module
2. GLAPLACE module
14 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
)(ˆ11
1
ωωLj
VIgm −=′
=
ABMABM VCCSVCCS with gain assigned by a Laplace transform function
GLAPLACE:GLAPLACE: Main feature
• retrieved from measurement dataretrieved from measurement data (evaluation of the effective permeability spectra of the ferrite core)
• approximated by means of a proper transfer functiontransfer function with parameters obtained from measurement dataobtained from measurement data
Implementation in SPICE Implementation in SPICE contcont’’dd
106
107
108-100
0
100
200
300
400
500
f, [Hz]
| µr (f)|
µr '(f)
µr ''(f)
FROM MEASUR
LORENTZ
GLAPLACE:GLAPLACE: Application
VCCS gain
F. Grassi, F. F. Grassi, F. MarlianiMarliani, S. A. Pignari , S. A. Pignari ““SPICE modeling of BCI probes accounting for the SPICE modeling of BCI probes accounting for the frequencyfrequency--dependent behavior of the ferrite core,dependent behavior of the ferrite core,”” in in Proc.Proc. XIXth URSI GAXIXth URSI GA, 2008., 2008.
15 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
Model Validation up to 400 MHzModel Validation up to 400 MHz
SS1313= = --SS23 23 [dB][dB] SS33 33 [dB][dB]
107 108-80
-60
-40
-20
0
frequency [Hz]106
measurementprediction
Real(S13)
Imag(S13)
|S13|
106 107 108-80
-60
-40
-20
0
frequency [Hz]
measurementprediction
Real(S33)
Imag(S33)
|S33|
21 of 21
EMC Group @ POLIMI
2009 CST European User Group MeetingMar. 16-18, 2009, Darmstadt, Germany
ConclusionConclusion
Main stepsMain steps
1.1. EM EM model of themodel of the BCI probeBCI probe
CAD modeling of the probe & input connector/adapterCAD modeling of the probe & input connector/adapterCharacterization of the Characterization of the ff--response of the ferriteresponse of the ferrite corecore
first order first order DebyeDebye modelmodel for representing intrinsic properties
circuit modeling via SPICE for explaining the differences between differences between EM and circuital modelingEM and circuital modeling
2.2. EM EM model of themodel of the validation fixturevalidation fixture
Validation up to Validation up to 400 MHz400 MHz
Possible applicationsPossible applicationsVirtual test setVirtual test set--up up for EMC assessment EMC assessment & circuit modelcircuit model extension extension to the case of multimulti--wire bundleswire bundles
Design optimization Design optimization of injection devices