Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
I,
Sofiane Chabane
Abstract
Wiring still supports most of electric power fluxes as well as data transmission in various infrastructures and transport systems.
Hopefully, transmission line theory is a very helpful approximation to estimate interference propagation among cable harnesses. This
is definitely a useful tool for engineers, since it enables anticipating the probability of failure of equipment they are connected to.
Everything has been written from the initial roots of transmission line derivation, dated from the ancient telegraphist's equations
established by O. Heaviside back to... 1880! until the last developments of sophisticated and off-the-shelf transmission line solvers for
complex arrangements of cable networks. The fact is, that these tools are so familiar to many EMC engineers that it might come with
some misunderstanding of some results. In this paper, we take a look at the root assumptions of this approximate theory and examine
some of its potential weaknesses, through simple examples. Then, we investigate the question of the possible improvement of the
classical transmission line theory (when and if required). In principle, this would require the derivation of some kind of generalized
transmission line theory or even the examination of the super theory of transmission lines. We rather show that a much more simple
modification of transmission line equations is possible for a better approximation of the experimental observations.
66
Signal
Battery
also be found, but in very few numbers, mainly to connect antennas for RF applications or to shield power links of electric power
trains. Shielded harnesses, even partially, are practically never chosen because of their specific connector requirements, their over
all weight and cost.
From a wiring point of view, the electrical system of a vehicle often uses the metallic car body as part of the return currents' path
(Fig. 2) but the equipment is generally not directly connected to the car body, leading to portions of harnesses where the return
currents from the equipment are carried by wires in the harness (zero-volt conductors). To avoid increasing considerably the num
ber of wires connected to the vehicle body, several equipotential conductors in a same harness are very often connected by splic
es at a given position within the harness. Signals and power are therefore transmitted and received by the equipment either
between two wires of a same harness, between a wire and the car body, or both alternatively. Finally, parts of the harness do not
run close to the body of the vehicle (e. g. behind the dashboard), and can pass over slots and gaps (e. g. between the car body and
the engine). Therefore, the return currents are forced to deviate from beneath the harness. In such a real world context, one can
easily recognize that:
The loads (equipment) are not connected to ground at the end of the harness and the vertical wires at the end of the harness
(at the equipment location) do not exist
When connected to the reference ground plane (car body), the transition of the contact wires are progressive and do not look like
vertical wires
The harnesses have interleaved wires, that lead to non-uniform cross-sections that must be modeled by segmentation
The distances between the harnesses and the reference ground plane (car body) can be important compared to the wavelength
and therefore breaches the conditions of applicability of the multi-conductor transmission line theory.
Nevertheless, even with all these imperfections, multi-conductor transmission line theory remains extremely useful to model such com
plex harnesses in real electrical systems because it avoids having to represent the harness as a geometrical structure in the same way
as the car body (electromagnetic model), introducing multi-scale, model size and computation time issues. However, in order to produce
simulation results closer to those that can be obtained by an electromagnetic model, multi-conductor transmission line models have
been improved in the past decade to overcome most of their drawbacks, in particular the loss of energy by radiation which will be the
main focus of this article.
2014 IEEE Electromagnetic Compatibility Magazine - Volume 3 - Quarter 2
67
1-4.
Complete details of the following derivation may be found in 5 Maxwell's equations in an infinite domain surrounding the sources may be
sorted out in terms of the electric field vector (E) or the magnetic field vector (H) and equivalently in terms of scalar potential (cp) and
vector potential (A). Given the sources in terms of current (J) and charge densities (p) one obtain:
(1)
(2)
Note that these relations are derived after having adopted an appropriate gauge relation between the scalar and vector potential (here,
the L orenz gauge 5). Therefore, the following important relationships define cp and A:
H=-\7xA
(3)
fLo
E = -jwA
\7cp
(4)
The derivation of TLT equations from Maxwell's equations follows the procedure below. The excitation field gives rise to the scattered
field (E(s) which is the exact opposite to the applied field (E(e) at the wire's surface:
( 5)
Since the scattered field obeys Maxwell's equations, it satisfies (4) as well. Moreover, A and cp are governed by (2) and (1) respectively.
Given the general solution of Eq. 1 and of Eq. 2 and introducing the continuity equation that relates p and J Uwp+V.J=Ol, one can derive
the following coupled equations:
\7cp
Lu
f
I z')go(z, z')dz'
jWfLo
.fo (
L
1 1 9 dI z')
.
-]WEo
= E(h, 0, z)
(
dz' + cp = 0
dz' go(z, z')
--
(6)
(7)
go(z,z') is the free-space Green's function that governs the radiation at the observation point z given the source point z'. Equations
(6) and (7) already look like the Telegraphist's equations (see Eq. 10 and Eq. 11 in next SUb-section). The scalar potential will be directly
where
identified with the scattered voltage difference Vis) taken as the line integral of the electric field along the x-axis. To fully recover the TLT,
one has to determine the terms in the integral of Eq. 6 and Eq. 7. The Green's function for the case of a wire of radius a at a height hover
(8)
68
1(z)
11k
' 2
'z + h
z l-2h
)../27r ) we find the following approximate result for the integral of Eq.
90(Z, z')dz'
1 I 2h
)1(z)
n(
27T
a
-
6:
(9)
A similar result is provided for the integral in Eq. 7. This approximation of the integral of the Green's function is the basis of the classical
TLT. Note that the integral of the Green's function is therefore approximated as a pure real number. This is equivalent to say that TLT
does not consider radiation (antenna mode). The scattered field is confined and entirely guided along the wire axis: the transverse elec
tromagnetic (TEM) mode of propagation. Another consequence is that the intrinsic or characteristic impedance of the line is also a pure
real number (Zc=
Lo
1"0
27r
( 2h
n a)
Then, come the Telegraphist's equations for continuous waves with angular frequency CD:
dVCs)(z)
dz
d1(z)
jwLo1(z)
E; (h, 0, z)
jwCoV(z)
(10)
(11)
Approximation of Eq. 9 is in fact accepted if conditions 2 hf... and 3 Lgh apply. This point will be specifically addressed in section 3.
is applied at the extremity z=O of a single transmission line of diameter 2a at a height h over a PEG infinite
ground plane. The internal impedance of the source is labeled Zo whereas the other extremity of the line is loaded by the impedance Zv
69
20
Figure 3. A cylindrical wire of diameter 2a and length L and excited by a localized generator e over a PEC infinite ground plane
L et's have a look at two load configurations. L oad impedances are pure resistances given by ZO=ZL=50 Q in the first configuration and
Zo=O Q and ZL=l Q in the second configuration. In both cases, we extract the current crossing ZLthrough the classical TLT and com
pare it with a full-wave solution provided by a method of moments (MoM) software. Throughout this paper, the contribution of Joule's
losses of the copper wires has not been taken into account neither in the modified TLT simulations nor in the MoM simulations.The
length of the wire is L=lm and its height over the ground plane is h=10cm. However, the calculation by the MoM requires that a closed
circuit exists between the end of the horizontal wire and the ground plane. For this purpose, two vertical wires each of 10cm long, are
added using the same cylindrical wire of diameter 2a=1.5mm. In turn, these vertical wires are simulated in TLT by extension of the wire
length such as Lg=L+2h(1. 2 m) with identical characteristic impedance 6.
-44
--
- Method of Moments
-46
-48
:3
=
-50
P'l
"
U
-52
-54
/1
/1
/
/
/ I
/ I
/ I
/ I
/ I
/ I
I
/
/
/
I
/
I
/
I
/
I
/
I
/
I
/
I
/
I
/
I
I
\ /
J
_56 L-------L-L-L--L----L--L-
le+07
le+08
Frequency (Hz)
Figure 4: Current in ZL obtained through classical TLT and MoM for configuration 1: ZO=ZL=50 Q (e=0.632 V)
For both configurations (see Fig. 4 and Fig.5), we first note that there is a good agreement between MoM and TLT with a noticeable
exception around resonances. Focusing now on these resonance effects, it also appears that the higher the frequency, the worse is the
comparison between the maximum magnitudes predicted by TLT (theoretically constant) and the maximum level predicted by the MoM
which becomes lower and lower. Around resonances, the wire stores energy which is entirely lost in the load impedances by the
approximation of the classical TL l However, a part of this energy is radiated, a phenomenon which is taken into account in the MoM. In
configuration 1, since the load impedances are 50Q, the part of energy they absorb is the most significant and differences between max
imum at resonances between curves of Fig. 4 reach 5 dB at 500 MHz. In configuration 2, the load impedance in only lQ. Under these cir
cumstances, the electric losses are much smaller and the radiation losses are predominant. Resonances are much sharper, but the
Q-factor of resonances obtained from the TLT is much higher than the ones obtained from the MoM calculations. This time, radiation
losses provide a noticeable difference. At 500 MHz the predicted current magnitude given by the MoM is about 40 dB less than the one
provided by TL l
A widely accepted rule of thumb states that TLT is applicable if the condition hs A/l0 is fulfilled. This would correspond to an upper fre
quency of 300 MHz in this example. The above configurations show that the applicability of TLT, in its classical form may be questionable
and would probably require another condition at least around resonances. This condition concerns the relationship between the radia
tion resistance and the terminal loads. Fig. 6 illustrates this point showing the near-field mapping all around the transmission line for both
load configurations. At a frequency of 130 MHz, which corresponds to the first resonance of the line, the contribution of the radiated
70
electric field from the ends of the wire is clearly much more significant in the configuration ZO=O Q and ZL=l Q (right hand side). On the
contrary, at a frequency of 200 MHz, far away from a resonance, the two near field distributions are similar in both cases.
0 ,------,---,--,-,-,-,,-,---,-,
Transmission Line Theory
Method of Moments
-10
-20
le+08
Frequency (Hz)
Figure 5: Current in ZL obtained through classical transmission line theory and method of moments for configuration 1: ZO=OQ and ZL=lQ (e=l V)
X EFI.ld MJinilud* IclBV,m)
210
lfI.O
)'.0
12.0
'.0
"
"
00
"
"
"
.
.
.
( ) Frequency
a
130 MHz
.-J'
x
"
"
"
0.0
-1.0
.
.
(b) Frequency
200 MHz
"
"
( ) Frequency
a
130 MHz:
z
.-J
X
(b) Frequency
200 MHz
Figure 6: Near field mapping of the x-component (horizontal) electric field around the transmission line at 130 MHz and 200 MHz for two different
load configurations ZO=ZL=50 Q (e=0.632 V, left-hand side), ZO=O Q and ZL=l Q (e=l V, right-hand side)
references and others, a key element is to obtain a rigorous solution of Eq. 9 rather than its approximate solution. As already mentioned,
this approximation resulted to the conclusion that the integral of the Green's function (Eq. 8) was the product of a pure real scalar and of
the current distribution along the line. A rigorous calculation of this integral leads to a complex scalar. Its real part is a frequency depen
dent number accounting for the evolution of the Land C p.u.l. parameters of the line. Its imaginary part is related to the radiation proper
ties and more specifically to the Rand G p.u.l. parameters. Therefore it results in a new definition of the p.u.l. parameters.
A complete mathematical analysis 8 of the resulting p.u.l. parameters, so called RHF,LHF,CHF,GHF, was developed. These parameters
may be put into the following form :
(12)
(13)
(14)
(15)
2014 IEEE Electromagnetic Compatibility Magazine - Volume 3 - Quarter 2
71
(16)
In these expressions, CF(oo) accounts for an adding correction factor of the rigorous solution of the integral of the Green's function (Eq.
9). Its expression is given in 8. It is highly dependent on the angular frequency. CF(oo) tends to a null value when 00---'>0 and therefore the
p.u.l. parameters converge to classical ones : LHF(O)=Lo, RHF(O)=O, CHF(O)=CO' CHF(O)=O.
The existence of a radiation mode is equivalent to have RHF and CHF non-null. They have been established from a reference work 9.
Note that RHF is different from the definition of the radiation resistance Rrad that accounts for current distribution along the line where
as RHF is independent from the current distribution and is a p.u.l. parameter in Q/m. The most important consequences are the resulting
properties of the propagation constant and of the characteristic impedance of the transmission line. First of all, the propagation constant
is not affected by this change in the p.u.l. parameters and remains a pure imaginary number ( k
jwVJZj
jWy'foJLo). This is
explained by the balanced relationship RHFCHF+CHFLHF=O. However the characteristic impedance becomes
complex and is given by:
(17)
Therefore, the characteristic impedance of the transmission line is split into its classical TEM mode and a radiation mode that corre
sponds to its imaginary part which appears as the ratio of the p.u.l. resistance to the wavenumber.
We could therefore solve the transmission line equations under this enhanced form:
dV(z)
(RHF + jwLHF)J(z)
E(h, 0, z)
(18)
(19)
However, solving this new pair of equations leads to the same solution as the classical solution for the current and voltage induced on
the line. It is due to the fact that the TEM mode and the radiation mode both propagate without losses and no modification of the bound
ary conditions are introduced. To account for radiation losses at the termination of the line, a possible solution is to add the imaginary
part of the characteristic impedance as loss resistances at ends of the transmission line.
Another solution consists in inserting an additional p.u.l. resistance in series with RHF. This additional p.u.l. resistance, labeled R+ is
adjusted to account for the radiation losses. Therefore, the equivalent circuit for a short section of a transmission line, with respect to
the wavelength, becomes that of Fig. 7.
I
I
GHFdzl
I
I
IE
dz
Figure 7: Equivalent circuit of a short section of a lossless transmission line of length dz'A with the new definition of p.u.l. parameters and an addi
tional p.u.l. resistance (R+)
For small values of R+, the propagation constant may be written as:
72
(20)
Adjusting the value of R+ consists in equating the power losses along the length of line L g (i.e. 1-exp(2aLg)) and the inverse of the
Q-factor of the transmission line given by the ratio of the imaginary part of the characteristic impedance of the transmission line to its
real part:
(21)
We thus extract R+ as:
(22)
Therefore, a possible solution consists in using the enhanced version of Telegraphist's equations (i.e. Eq. 18 and Eq. 19) with the addition
of a p.u.l. resistance R + to account for radiation losses. The next section is dedicated to the application of this modified version of the
coupled equations of TLT to the previous examples.
--
-44
-46
-48
P'l
:3
"
g
"
-50
U
-52
"
"
"
Method of Moments
II
I
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
'I
, ,
, I
, ,
,
,
"
"
, ,
, I
, ,
, I
, ,
, ,
"
I
,
I
I
"
"
',
' ,
' ,
' I
' ,
I
I
I
I
' ,
I
I
I
, ,
I
I ,
I,
II
_ 56 ------L-L--L---L--
le+08
le+07
Frequency (Hz)
-54
""
"
\1
Figure 8: Current in ZL obtained through modified TLTand MoM for configuration 1: ZO=ZL=50 Q (e=O.632 V)
Fig.8 shows that the maximum current associated with successive resonance frequencies tends to decrease following the trend of the
MoM simulation. The modified TLT involves differences of only one or two dB: for this load configuration (Zo=ZL=50 Q). this modified ver
sion of TLT may be considered as unnecessary, in this frequency range. An opposite conclusion may be drawn as regards to the result of
Fig.9. The magnitude of current at resonances has been deeply reduced and is now much closer to the magnitude of the resonances
obtained by the MoM.
In both cases, two additional remarks may be provided. i) In the low frequency region, the response of the modified transmission line for
malism is not different from the classical one since, in that case the modified p.u.l. are close to classical ones: RHF", 0 , GHF", 0, LHF '"
LO, CHF", CO. ii) The results from the modified transmission line theory do not exactly match those of the MoM. The main explanation
for this is the role played by the vertical wires considered only as additional sections of the transmission line in TLT. Their own radiation
properties are fully taken into account in the MoM calculation only.
73
O ,------,---,--,-,-,-,,-,---,-
Modified Transmission Line Theory
-10
Method of Moments
_ 60 L------------L---L--L--------L------
IN08
le+07
Frequency (Hz)
Figure 9: Current in ZL obtained through classical TLTand MoM for configuration 1: ZO=O Q and ZL=1 Q (e=1 V)
An experimental validation is provided in the configuration Zo=ZL=50Q. For this purpose, the uncoated wire is installed over a ground
plane and ended into a connector which is fitted in the ground plane itself (see Fig. lO), thus being as close as possible to the configuration which
was simulated with the MoM method. The measurement was performed using a sine wave generator with a constant amplitude, tuned from
10 MHz to 500 MHz and a spectrum analyzer. This experiment confirms the general trend about the slight reduction of the maximum current of the
successive resonances on the line as observed with the MoM and as reproduced by the modified TLT formalism described in this paper.
I
\
'.
,
,
,
,
,
-48
is -50
"
"
I'
,
,
,
U -52
-54
- 56
le+07
le+08
Frequency (Hz)
Figure 11: Current in ZL obtained from experiment and comparison with modified TLTand MoM results for configuration 1: ZO=ZL=50 Q (e=0.632 V)
74
6. Conclusion
TLT is commonly used to determine currents and voltages induced within cable networks and at devices' inputs. Resonances are important phenomena
at the origin of radiated emission or radiated immunity issues. However, although TLT provides a rather excellent way for evaluating field to cable inter
ferences, it has some weaknesses. One of them is the accuracy of the estimation of currents and voltages specifically at resonances, depending on
transmission line boundary conditions. One may assume this estimation is likely to be incorrect at resonances specifically for low or high values of load
impedances (i.e. magnitude of reflexion coefficients close to 1). This is related to non negligible values of the radiation resistance that accounts for
antenna mode. This parameter tends to increase with the ratio of the height of the transmission line to the wavelength. A modified version of TLT is pre
sented in this paper, based on the calculation of the integral of the Green's function. This leads to enhanced p.u.l. parameters including per-unit-Iength
resistance and conductance associated with the radiation properties of the wire. Nevertheless, this leads only to the decomposition of energy in the
wire into a transmission line mode and an antenna mode (imaginary part of the characteristic impedance). Therefore, this one is still not radiated. Con
sequently, we introduce an additional loss resistance that dissipates the corresponding radiated energy. The modified equations of TLT obtained keep
the simplicity of the classical ones and are compatible with standard solvers, since they only involve a modification of per-unit-Iength parameters.
Results are shown to be much closer to the MoM calculation than the classical formalism at resonances.
7. References
1.
C Paul, "Frequency response of multiconductor transmission lines illuminated by an electromagnetic field," IEEE Transactions on Electromagnetic Compatibility, vol.
2.
A. Agrawal, H. Price, and S. Gurbaxani, "Transient response of multiconductor transmission lines excited by a nonuniform electromagnetic field," IEEE Trans. Elec-
3.
4.
F. Rachidi and S. Tkachenko, Electromagnetic field interaction with transmission lines. WIT Press, 2008.
6.
P. Oegauque and A. Zeddam, "Remarks on the transmission line approach to determining the current induced above-ground cables," IEEE Transactions Electromag
7.
Y. Bayram and J. Volakis, "A generalized MoM spice iterative technique for field coupling to long terminated lines," Electronic Letters, vol. 47, no. 3, pp. 222-223, February 2011.
5. J. Nitsch, F. Gronwald, and G. Wollenberg, Radiating non-uniform transmission line systems and the partial element equivalent circuit method. Wiley, 2009.
netic Compatability, vol. 30, no. 1, pp. 7 7-80, February 1988.
8.
9.
S. Chabane, P. Besnier, and M. Klingler, "Extension of the transmission line theory application with modified enhanced per-unit-Iength parameters," Progress in
8. Biographies
Philippe Besnier (M'04, SM'70) received the diplome d'ingenieur degree from Ecole Universitaire d'/ngenieurs de Lille (EUDIL),
Lille, France, in 1990 and the Ph.D. degree in electronics from the university of Lille in 1993. Following a one year period at ONERA,
Meudon as an assistant scientist in the EMC division, he was with the Laboratory of Radio Propagation and Electronics, University
of Lille, as a researcher at the Centre National de la Recherche Scientifique (CNRS) from 1994 to 1997. From 1997 to 2002, Philippe
Besnier was the Director of Centre d'Etudes et de Recherches en Protection Electromagnetique (CERPEM): a non-profit organiza
tion for research, expertise and training in EMC, and related activities, based in Laval, France. He co-founded TEKCEM in 1998, a
private company specialized in tum-key systems for EMC measurements. Back to CNRS in 2002, he has been since then with the
Institute of Electronics and Telecommunications of Rennes (IETR), Rennes, France. Philippe Besnier was appointed as senior
researcher at CNRS in 20 13 and has been co-head of the Antennas and Microwave devices department of IETR since 2012. His research activities are
mainly dedicated to interference analysis on cable harnesses (including electromagnetic topology), reverberation chambers, near-field probing and
recently to the analysis of uncertainty propagation in EMC modeling.
Marco Klingler was born in Zurich, Switzerland, in 1963. He received his Engineer's degree in computer science from HEI, Lille
(France) in 1989, his DEA (M.S.) degree in automatics/ robotics and his Ph.D. in electronics in 1989 and 1992 respectively, both
from the University of Lille. He then joined the French National Institute for Transport and Safety Research (INRETS) in Villeneuve
d'Ascq (France) as a researcher where he was in charge of the R&D activities in EMC of ground transportation systems. His main
_
interests were electromagnetic interferences on PCBs, behavior of electronic components in electromagnetic environments,
coupling to wire structures, test methods, and test facilities. In 2002, he joined PSA Peugeot Citroen in Vetizy-Villacoublay (France)
in the Development Division where he was successively in charge of the EMC design activities, the EMC/ antenna simulation
activities, and finally the EMC full vehicle validation activities. In 20 1 1, he moved to the Research Division where he is currently an
EMC Expert and responsible of the EMC/ antenna research activities. His main interests include EMC modeling and simulation of automotive electric
powertrains, multi-conductor transmission lines, new materials, and specific vehicle antennas.
Sofiane Chabane was born in Azazga, Algeria, in 1983. He received the Diplome D'etudes Universitaires Appliquees (B.S.) degree
in 2005, and the Engineer's degree in 2008 both in aeronautics and from Institut d'Aeronautique de BMa (Saad Dahleb University),
BMa (Algeria). In 2070, he received the master degree (M.S.) in physics of complex natural and industrial systems (SCNt), from
University of Rennes 1, Rennes (France). He is currently working toward the Ph. D. degree in electronics and telecommunications
at the Institute of Electronics and Telecommunications of Rennes (IETR), INSA of Rennes, France. His current main research inter
ests include electromagnetic compatibility and interference, electromagnetic field interaction with transmission lines, electromag
netic modeling and simulation.
2014 IEEE Electromagnetic Compatibility Magazine - Volume 3 - Quarter 2
EMC
75