of the 2014 International Symposium on Electromagnetic Compatibility (EMC Europe 2014), Gothenburg, Sweden, September 1-4, 2014
I.
II.
INTRODUCTION
Fig. 1. Radiating circuit and the victim grounded wire system understudy.
108
Proc. of the 2014 International Symposium on Electromagnetic Compatibility (EMC Europe 2014), Gothenburg, Sweden, September 1-4, 2014
Z(y=d)
u0(y=0)
u0(y=d)
Ludy u (y)
0
Fig. 3. Victim wire equivalent circuit under EM aggression based on the TL
LC-model.
k ( f ) = k x ( f )u x + k y ( f )u y + k z ( f )u z .
Z(y=0)
y=d
u(y=d)
A. NF/NF Transform
This EM computational transform is established from the
plane wave spectrum (PWS) theory applied to the planar NF
scanning [14-15]. For the analytical exemplification, we
y+dy
y
Cudy
u(y=0)
y=0
di ( y )
+ j Cu u0 ( y ) = 0
dy
(1)
Z ( y) Zc
,
(5)
Z ( y) + Zc
the reflection coefficients at the extremities of the wire, we
obtain the coupling voltages are generated from the BLT
equation [10]:
translation
( X ( z ) = f ( X ( z0 )) )
and
orthogonality
(Xz=f(Xx,Xy)), respectively, to NF scanned data by taking into
account the evanescent waves both in the frequency- and timedomain. Under the circumstance as stressed in Fig. 1, as the
measurement of EM wave in the surface area delimited by the
victim line and the GND plane is quite complicated, this
operation can serve us for extracting E-field components
necessary for the Agrawal model application [8] in order to
compute u(y=0) and u(y=d). We intentionally choose this
model thanks to its simplicity for extracting the induced
voltage from only the E-NF.
0 1 e d u1 / 2
u( y = 0) 1 + 1
, (6)
=
u( y = d ) 0
1 + 2 e d 2 u2 / 2
where is the TL propagation constant and :
z1
d
0
z2
,(7)
z1
d
(d y )
d
E y ( y , z1 )dy + ( E z (d , z ) e E z (0, z ))dz
u2 = e
0
z2
( y ) y = 0, d =
U ( j , y = 0) = Tv ( j , y = 0) FFT (ue (t ))
.
U ( j , y = d ) = Tv ( j , y = 0) FFT (ue (t ))
109
(8)
Proc. of the 2014 International Symposium on Electromagnetic Compatibility (EMC Europe 2014), Gothenburg, Sweden, September 1-4, 2014
|Ey(z0=3mm)|
60
y (mm)
60
20
20
x (mm)
40
4
2
0
20
40
x (mm)
[Ey(z0=3mm)]
60
40
20
-5
-10
0
20
x (mm)
40
|Ez(z0=3mm)|
y (mm)
60
60
15
40
10
20
40
20
-5
20
x (mm)
40
110
20
x (mm)
40
(rad)
(V/m)
y (mm)
III.
40
8
6
4
2
0
-2
-4
20
20
40
x (mm)
[Ex(z0=3mm)]
(V/m)
20
10
8
40
(rad)
40
(V/m)
y (mm)
10
8
6
4
2
(rad)
y (mm)
y (mm)
60
Proc. of the 2014 International Symposium on Electromagnetic Compatibility (EMC Europe 2014), Gothenburg, Sweden, September 1-4, 2014
|Ey (z=6mm)|
60
60
40
20
0
40
20
0
20
40
x (mm)
8
6
4
2
0
-2
0
40
20
(V/m)
y (mm)
y (mm)
1.2
1
0.8
0.6
0.4
0.2
40
20
-5
2.5
20
0
20
x (mm)
40
1.2
1.3
(b)
10 Tv(z=3mm)
(rad)
|Tv(y=0)|
10 Tv(z=6mm)
5
10 Tv(z=11mm)
0.8
0.9
1
1.1
Frequency (GHz)
1.2
1.3
Fig. 10. Coupling voltage transfer function magnitudes shifts for z={3mm,
6mm, 11mm}.
5
40
0
20
0
20
x (mm)
(a)
2
1
Tv(z=3mm)
Tv(z=6mm)
-1
-2
0.7
-5
0.5
0
0.9
1
1.1
Frequency (GHz)
0.5
60
(V/m)
y (mm)
y (mm)
1.5
0.8
20
40
x (mm)
[Ez (z=11mm)]
2
40
10 Tv(z=11mm)
3
0
20
40
x (mm)
|Ez (z=11mm)|
60
0
0.7
5
10 Tv(z=6mm)
1.5
Ex,y @
60
(a)
[Ey (z=11mm)]
|Ey (z=11mm)|
60
from
0
0.7
20
40
x (mm)
-10
20
40
x (mm)
[Ez(z=6mm)]
60
5
4
3
2
1
(V/m)
y (mm)
y (mm)
|Tv(y=d|)
20
40
x (mm)
|Ez (z=6mm)|
60
-5
[Tv(y=0)] (rad)
20
10 Tv(z=3mm)
(rad)
(rad)
20
40
(rad)
(V/m)
y (mm)
y (mm)
3
40
Tv(z=11mm)
0.8
40
0.9
1
1.1
Frequency (GHz)
1.2
1.3
[Tv(y=d)] (rad)
(b)
0
-5
Tv(z=3mm)
-10
Tv(z=6mm)
Tv(z=11mm)
-15
0.7
0.8
0.9
1
1.1
Frequency (GHz)
1.2
1.3
Fig. 11. Coupling voltage transfer function phase shifts for z={3mm, 6mm,
11mm}.
111
Proc. of the 2014 International Symposium on Electromagnetic Compatibility (EMC Europe 2014), Gothenburg, Sweden, September 1-4, 2014
ue'(t) (V)
-10
0
10
Time (ns)
(b)
0.8
1
1.2
Frequency (GHz)
u(y=0)
u(y=d)
u(y=0,z=6mm) (mV)
-2
(a)
0
10
15
Time (ns)
20
25
0.5
u(y=0)
u(y=d)
10
15
Time (ns)
20
25
50
u(y=0)
u(y=d)
0
(c)
-50
10
15
Time (ns)
20
-0.5
-1
(b)
0
10
15
Time (ns)
20
25
u'(y=0)
u'(y=d)
0.1
0.05
0
-0.05
(c)
-0.1
5
10
15
Time (ns)
20
25
u'(y=0)
u'(y=d)
10
0
-10
(d)
0
25
0
(b)
u'(y=0)
u'(y=d)
0.5
-0.5
4
Time (ns)
u(y=0,z=11mm) (V)
u(y=0,z=6mm) (mV)
u(y=0,z=3mm) (mV)
Fig. 12. Time-domain (a) and frequency spectrum (b) plots of the considered
excitation signal.
u(y=0,z=11mm) (V)
0.1
20
(a)
2
-2
u(y=0,z=3mm) (mV)
(a)
0
fft[ue(t)] (V)
ue(t) (V)
0.2
10
10
15
Time (ns)
20
25
Fig. 14. Transient coupling voltages from structure under 2ns excitation.
112
Proc. of the 2014 International Symposium on Electromagnetic Compatibility (EMC Europe 2014), Gothenburg, Sweden, September 1-4, 2014
D. Discussion
It is interesting to point out that the extraction of these
perturbation couplings can be improved for any excitation
signals by proceeding with extrapolation or resampling of the
frequency transfer function. One emphasizes that the CPU
time of the whole method implemented into Matlab run with
PC equipped by windows 7 having Intel Core i5-2467M
CPU @1.6GHz 4Go RAM was of about five seconds.
Compared to the most of existing EM computational
methods, the investigated modelling method is advantageous
in term of:
Flexibility to the EMC analysis for various families of
RF, digital and mixed PCBs placed in typically harsh
EM environment.
Possibility to integrate the EM aggression from nonuniform NF including the evanescent waves.
Possibility to be implemented both in frequency and
time domains. It makes the method promising for the
EMC susceptibility analysis when considering short
duration excitation as fast electrical transient
disturbances.
Computation speed very less compared to the 3D EM
commercial tools as HFSS from ANSYSTM, CST
MWSTM, EMCosTM
However, the proposed method presents drawbacks in
particular on the following points:
Dependence on several computational methods which
are susceptible to generate significant inaccuracies
according to the multiple physical parameters of
numerical resolutions x, y, z, f and t.
Limited to the EM quasi-static applications related to
the use of the quasi-TEM and TL theory.
IV.
[2]
[3]
[4]
[5]
[6]
[7]
[8]
CONCLUSION
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
113