IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 56, NO.

6, DECEMBER 2014 1295

Fast Prediction of Transmission Line Radiated

Emissions Using the Hertzian Dipole Method
and Line-End Discontinuity Models
Jin Meng, Yu Xian Teo, David W. P. Thomas, Senior Member, IEEE, and Christos Christopoulos, Fellow, IEEE

Abstract—High-frequency signals on interconnects can cause normal functioning of solid-state circuits, control, and massive
significant radiated electromagnetic emissions. An intermediate data commutation networks, the electric field from interesting
level modeling method aimed at providing a faster solution with sources in both the far-field and near-field should be accurately
less computing resources to allow designers to obtain rapid approx-
imations is desirable. This paper presents a modeling technique to calculated and controlled.
speed up the evaluation of radiated fields from interconnect cables. Basically, two techniques have been used to compute the
Based on the Hertzian dipole radiation theory and transmission- radiated electromagnetic field from a transmission line (TL):
line frequency-dependant solutions, the radiating source is mod- closed-form analytical solutions [5]–[7] and the Hertzian dipole
eled by the sum of a large number of short dipoles. This model approximation approach [8]–[11].
allows the contributions of line-end discontinuities to be included
through a vector network analyzer measurement together with The first approach resolves the travelling current signal into a
a monopole approximation. The proposed method is verified by series of transient functions. The resultant field is then a super-
open-line and RG 58 coaxial cable measurements. position over all these functions [5]. This task reduces minutes
Index Terms—Dipole radiation, electromagnetic compatibil- or hours of computer time to solve a full-wave model of the
ity, radiated electromagnetic emissions, transmission-line (TL) TL. Although this solution is very powerful and accessible, the
modeling. radiation due to the discontinuities introduced by the line ends
(e.g., vertical supports and connection transitions) was assumed
to be negligible and no direct comparison with measured radi-
I. INTRODUCTION ated fields was provided. A closed-form solution for the total
APID advances in science and technology have led to in- power radiated by a multiconductor line above a ground plane
R creased requirements for high-speed clock rates. At higher
frequencies, energy is more difficult to contain. Factors like en-
was proposed in [7]. Since the ohmic losses were neglected in
this modeling, several decibels discrepancy can be occurring
closure slots, length of individual component interconnections, between the experimental data and numerical ones. Meanwhile,
and radiation directly from terminated high-speed circuit boards the minimum frequency for the measured data was limited by the
all are potential causes for unwanted radiation. Electromagnetic chamber’s dimension when the reverberation chamber methods
field radiation from the electrical and electronic apparatus is an were used [5], [7].
important topic in electromagnetic compatibility (EMC). The The second approach is to segment a TL into many short
standard EMC method [1], [2] is to measure the electric field Hertzian dipoles. The radiated field by this line is then the sum-
radiated by the source over the frequency range from 30 to mation of all contributions from each dipole. In time-dependant
1000 MHz, performed typically in an anechoic chamber or an transient studies, the radiation current and the radiated fields
open-area test site. Recently, researchers have suggested the are both calculated and expressed in the time domain [8]–[11].
substitution of radiated power measurements for the present EMC designers usually solve problems using frequency-domain
electric field measurements [3]. If the total radiated power and observations as typically provided by a spectrum analyzer.
the maximum directivity of the source are known, the max- Therefore, a fast frequency-domain calculation model is more
imum electric field over a sphere of arbitrary radius can be suitable for radiated field verifications. Most of the aforemen-
easily determined [4]. This is essentially a far-field expression tioned methods focus on the differential-mode radiation from an
of the radiation by an antenna. On the other hand, to ensure the open-wire or single PCB trace was evaluated with a closed-form
solution elaborated by a TL model and a modified Green’s func-
tion [12]. However, when shielded cables are used, the common-
mode radiated fields differ from those from bare wires, and
Manuscript received August 8, 2013; revised November 28, 2013; accepted
February 14, 2014. Date of publication May 9, 2014; date of current version therefore, warrant investigation.
December 11, 2014. This work was supported by the National Natural Science In this paper, an intermediate level modeling method aimed
Foundation of China under Grant 51077131. at providing a faster solution with less computing resources, al-
The authors are with the George Green Institute for Electro-
magnetics Research, the University of Nottingham, Nottingham, NG7 lowing designers to obtain fast but accurate approximations,
2RD, U.K. (e-mail: [email protected]; [email protected]; is discussed. The starting point of the method presented is
[email protected]; [email protected]). to calculate the current distribution along the line using the
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. TL approximation. The line-end discontinuities are character-
Digital Object Identifier 10.1109/TEMC.2014.2318720 ized by fitting to the measured impedance curve from a cable

0018-9375 © 2014 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution
requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
1296 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 56, NO. 6, DECEMBER 2014

prototype. After that, to calculate the radiated field, the method

of small Hertzian dipoles is used, requiring numerical sum-
mation. A computation code in MATLAB has been used for
the implementation of the proposed numerical technique and
all of these calculations are very faster than the 3-D full-
wave simulation. The calculated results of the radiated emis-
sions from nonshielded transmission as well as braided shielded
TLs are compared with the measurement results using antenna
measurements.

II. HERTZIAN DIPOLE ANTENNA METHOD

The Hertzian dipole is well known and corresponds to a short-
length wire (of length l << λ, where λ is the wavelength of the
Fig. 1. Experimental setup for wire radiation measurements.
signal) driven by an ac current of frequency ω. The wire is taken
along the z-axis in a rectangular coordinate system. The current
flowing in the dipole is assumed at the start to be a time-varying
current given by I(t, r = 0) = I0 cos(ωt) at the dipole. The
magnetic- and electric-field components of the Hertzian dipole
are [13] as follows:
 
μ0 I0 l sin θ cos(ωt − βr) β sin(ωt − βr)
H= − nφ (1)
4π r2 r
 
2I0 l cos θ sin(ωt − βr) β cos(ωt − βr)
E= + nr Fig. 2. Equivalent TL circuit of Fig. 1 when a bare wire is used.
4πε0 ω r3 r2

I0 l sin θ sin(ωt − βr) β cos(ωt − βr)
+ + copper supports, whereas the line ends were connected to two
4πε0 ω r3 r2 N-type connectors. The first connector was used for loading the

β 2 sin(ωt − βr) line with N-type loads, and the other one was connected to a
− nθ (2) coaxial cable for the line excitation. An antenna was placed at
r
the height of Hant above the wire to detect the electric field
where ε0 and μ0 are the permittivity and permeability for free radiation.
space, respectively. In Fig. 1, if the metal plate was infinite, this line would be
In general field calculations, which may contain many radi- equivalent to a two-wire line, because of the image theorem.
ating structures, summation of the field components can best When the ground plane is finite, the equivalence is only approx-
be achieved using Cartesian coordinates. The electric fields for imate, but if its size is much larger than the distance h between
each dipole element in Cartesian coordinates are as follows: the wire and the plane, the errors are negligible [14].
 
lxzI0 ej ω (t−r /c) 3 3 jω
Ex (ω) = + + 2 (3) A. TL Model With Line-End Discontinuities
4πε0 r2 cr2 jωr3 c r
  When an open line is investigated as the radiating line, the
lyzI0 ej ω (t−r /c) 3 3 jω
Ey (ω) = + + 2 (4) transmission network circuit can be modeled as shown in Fig. 2,
4πε0 r2 cr2 jωr3 c r
where Ct1 and Ct2 are the equivalent parallel capacitances for
 
lz 2 I0 ej ω (t−r /c) 3 3 jω the feeding coaxial connector and the load coaxial connector,
Ez (ω) = + + 2 respectively. The per-unit-length parameters of an open line over
4πε0 r2 cr2 jωr3 c r
  ground are defined as Row , Low , Cow , and Gow . The line current
lI0 ej ω (t−r /c) 1 1 jω anywhere on the line in closed form in terms of the reflection
− + + 2 . (5)
4πε0 cr2 jωr3 c r coefficients at the source and the load [15] is as follows:
1 − ΓL e−γ o w l e e2γ o w z Vth
III. RADIATION FROM AN OPEN LINE OVER GROUND I(z) = e−γ o w z (6)
1 − ΓS ΓL e−γ o w l e Zth1 + Zow
A line model configuration like the one in Fig. 1 has been
considered. The measurements were carried out in an anechoic with
chamber to reduce reflections and improve the experimental
accuracy. A single round bare copper wire le in length, rw Zth1 − Zow Zth2 − Zow zG
ΓS = , ΓL = , Zth1 = ,
in radius, and h in height above a 2 m × 0.5 m aluminum Zth1 + Zow Zth2 + Zow 1 + jωCt1 ZG
sheet was constructed as the radiation model. The aluminum
sheet was placed on the floor of the chamber and served as ZL VG
the ground plane. The line was kept suspended by two vertical Zth2 = , Vth = .
1 + jωCt2 ZL 1 + jωCt1 ZG
MENG et al.: FAST PREDICTION OF TRANSMISSION LINE RADIATED EMISSIONS 1297

Fig. 3. Short line measurement for line-end discontinuities characterization.

Fig. 5. Measured and simulated OC input impedance profile for the 6-cm open
wire.

Fig. 4. Equivalent LC network for the 6-cm-long line. (a) Lumped circuit
model and (b) physical dimensions for the N-type connector.

The exact per-unit-length high-frequency resistance for a

round wire over ground is [15]
  Fig. 6. Measured and simulated SC input impedance profile for the 6-cm open
RS h/rw wire.
Row = (7)
2πrw (h/rw )2 − 1
where a1 = 3.05 mm is the diameter of the core conductor,
where RS = 1/(σδ) is the surface resistance and b1 = 8.08 mm is the diameter of the dielectric material, Poly-
δ = 1/(πf μ 0σ) is the skin depth.
In Fig. 2, the horizontal part of the line is regarded as a tetrafluoroethylene, and d1 = 24.59 mm is the length of the
uniform transmission line (TEM mode) with a characteristic connector. Using the actual dimensions, the values of Ct and Lt
impedance Zow and a complex propagation constant γow . The are given by
line-end discontinuities such as the connectors are approximated 2πεPTFE d1
by two 1-pF capacitors [19]. To accurately determine the current Ct = = 2.80 pF (9)
ln (b1 /a1 )
distribution by using (6), the line-end capacitances related to the
coaxial connectors should be taken into account. μ0 d1 b 
Lt = ln 1/a1 = 4.82 nH. (10)
The model for line-end discontinuities have been determined 2π
using a 6-cm long and 5-cm high line configuration shown in Alternatively, an experimental characterization, in terms of
Fig. 3. An equivalent model for this structure is given in Fig. 4(a), scattering parameters S11 , has been performed to derive the
showing two terminal capacitances Ct and two inductances Lt LC model of Fig. 4(a). It relies on the knowledge of the input
to add at the line ends. Fig. 4(a) also shows that the short line impedance Zin of the whole circuit under test, which is extracted
is represented by connecting two short sections in cascade. A from the S11 parameter of the circuit obtained from a vector
good approximation of the highest frequency range represented network analyzer (VNA).
by the two cascaded nominal circuits is given in [16]
1 + S11
Nυ 2 × 3 × 108 Zin = ZVNA (11)
fm ax = = ≈ 3.18 GHz. (8) 1 − S11
π π × 0.06
where ZVNA = 50 Ω is the nominal impedance of the VNA.
This means that the representation of this distributed effect The validation of the cable impedance with frequency of a
by the equivalent lumped model is a useful approximation up to 6-cm line length in open circuit (OC) and short circuit (SC) are
a frequency of 3.18 GHz. represented in Figs. 5 and 6, respectively. At low frequencies,
The values of Ct and Lt will depend on the physical dimen- the SC impedance profile is basically inductive, while the OC
sions of the N-type coaxial connectors, as shown in Fig. 4(b), impedance is capacitive.
1298 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 56, NO. 6, DECEMBER 2014

From the measurement data shown in Figs. 5 and 6, the

equivalent total capacitance and total inductance through the
impedance profile can be determined as
1
Ceq = 2 (Ct + C0 ) = ≈ 6.19 pF (12)
ω · |Zin−OC |
|Zin−SC |
Leq = 2 (Lt + L0 ) = ≈ 74.75 nH. (13)
ω
The total inductance and total capacitance have been assumed
to be constant over the range of interest at 74.75-nH and 6.19-pF,
respectively.
The capacitance and inductance parameters [15] for each
segment are given by
0.5 πε0
C0 = = 0.315 pF (14)
cosh−1 (h/rw )
0.5 μ0
L0 = cosh−1 (h/rw ) = 31.8 nH . (15)
π
Using (12)–(15), we can obtain
Ct = Ceq /2 − C0 = 2.78 pF (16)
Lt = Leq /2 − L0 = 5.57 nH . (17)
It is obvious that the value of Ct from (15) and (9) agrees
very well. It is interesting to note that the value of Lt from (17)
is larger than the evaluated result from (10). This is likely due
to the inductance of vertical support was neglected in the model
of Fig. 4.
The results obtained through this representation are shown Fig. 7. Validation of the line impedance in 50-Ω load configuration. (a) 30-cm
in Figs. 5 and 6. The empirical model was proposed in [17] length and (b) 1-m length.
by estimating about 1 pF for the line-end transition. In fact, as
shown in Fig. 6, the empirical model predicts the first resonance
frequency at about 570 MHz; whereas the actual resonance takes
place around 310 MHz. Also, as shown in both Figs. 5 and 6,
the proposed model better approximates measurement, which
demonstrates the models effectiveness.
As further validations, Fig. 7(a) and (b) plots the computed
results of two cases for wire over ground both in 50-Ω loaded
configuration compared to the experimental data, where kept the
test fixture same as in the 6-cm case of Fig. 3 but the length in
Fig. 7(a) is 30-cm, while (b) is for 1-m case. The results obtained
with this proposed model demonstrate good description of the
line-end discontinuities behavior.
Fig. 8. Current segmentation of short dipole assumption for field calculations.
B. Model for Radiated Field Calculation
To determine the radiated field generated by the TL system
shown in Fig. 1, one can regard the TL as being made up of a
load side vertical support current. By using Kirchhoff’s current
large number of short dipole radiators. As shown in Fig. 8, the
law, we can obtain
length of each dipole is chosen to be short enough to approxi-
mate a Hertzian dipole. Currents I1 − Iξ , which are in general
frequency dependant, are obtained from (6). Currents I1 − Iξ Ivsd1 (ω) = I1 (ω)ny and Ivs2 (ω) = −IN (ω)ny (19)
are the components of the mirror wire current.
Iξ (ω) = −Iξ (ω) = I(ξle /N , ω)nz , ξ = 1, 2, . . . N (18)
where ny is the unit vector in the y-direction.
where N is the segment number. Given the observation point coordinates P (x, y, z), according

Current Ivs1 = Ivs1 defines the real part and mirror part of the to the Hertzian dipole field theory addressed in Section II, we

source side vertical support current. Current Ivs2 = Ivs2 defines can compute the radiated fields from all the radiators in Fig. 8.
MENG et al.: FAST PREDICTION OF TRANSMISSION LINE RADIATED EMISSIONS 1299

Fig. 9. Photograph of the experimental validation.

Fig. 10. Current spectra for the test line with measurement position z = 50
cm. (a) Z L = infinite (open circuit) and (b) Z L = 50 Ω.
After that, the total radiated field from the entire TL system
can be calculated using the following expressions:

N 
N sured by a biconical antenna (Eaton Corp., model 94455-1, 20–
 total (ω) =
E  real (ξ, ω) −
E  im ag (ξ, ω)
E 200 MHz) connected to a spectrum analyzer. Above 200 MHz, a
ξ =1 ξ =1 log-periodical antenna (Electro-mechanics Corp., model 3146,
200–1000 MHz) was used.
 vs1 (ω)+ E 
 vs1  vs2 (ω)− E 
 vs2
+E (ω)− E (ω) (20) Alternatively, a 3-D time-domain TL modeling method was
utilized to confirm the analytical result of straight open wire.
where Ereal and Eim ag are the radiated electric fields from the
The 3-D full-field program called wireSolver was developed by
real current segment and mirror current segment, respectively,
John Paul [18]. A broadband model has been developed in wire-
and Evs1 and Evs2 are the radiated electric fields from the real
Solver to obtain the transient response of voltage and current
source side vertical support and the real load side vertical sup-
  distributed on the line. After running for sufficient time steps,
port, respectively, while Evs1 and Evs2 are the radiated electric
voltage and current responses approach steady state. Applying
field from the mirror source side vertical support and the mirror
a fast Fourier transform to the time domain current approach-
load side vertical support, respectively.
ing steady state, the frequency spectrum of current on the line
can be obtained. After that, the radiation spectrum can also be
C. Measurement Results for Open Wire determined referring the process of Section III-B.
In order to validate the field model derived previously, the Fig. 10(a) and (b) shows the results from the calculated current
radiated fields from a bare wire above ground plane (le = 1 m, spectrum using (6), 3-D time-domain simulation, and measured
rw = 0.5 mm, and h = 5 cm) were calculated and measured. data using the current probe. The computation by (6) was carried
An open wire under test in Fig. 9 was realized with the source out by dividing the line into 100 segments. Comparisons were
feeding cable connected to a wideband swept sinusoidal sig- made in accordance with the overall feature selective validation
nal excitation source whose amplitude is 0.224 V (107 dBμV) criteria defined in IEEE P1597.1 [19]. Both the OC load and
over the entire range 10 MHz–1 GHz frequency. The wire load the 50-Ω resistive load results give good agreement between
was alternately disconnected and connected to a 50-Ω resistance the model and the measurement. The small resonant frequency
to realize the open and resistive load. A current probe (Eaton shifts above 400 MHz can be observed. These errors are due to
Corp., model 94111-1, 1–1000 MHz) clipped on the wire at the artifacts introduced by the current probe, scattering from the
an arbitrary position was used to measure the RF current. At termination plates, laboratory objects, etc. These effects were
frequencies from 20 to 200 MHz, the radiated field was mea- not considered in the model.
1300 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 56, NO. 6, DECEMBER 2014

Fig. 12. Shielded coaxial cable model for the TL analysis.

Fig. 11. Validation of radiation model by comparing calculations and exper-

imental data with the antenna height H a nt = 1 m. (a) Z L = infinite (open
circuit) and (b) Z L = 50 Ω. Fig. 13. Equivalent P.U.L. TL circuit of Fig. 16.

Fig. 11(a) and (b) shows the comparison of the measured and
computed radiated electric fields at the observation point P (0, tional shielded cable analysis method, Fig. 12 gives the three-
1 m, 50 cm). The radiation field was studied in the frequency conductor system with reference directions and notation [20],
range 30 MHz—1 GHz. These results show that radiated fields [21]. The whole system is interpreted as two TL loops, which are
from the line-end vertical supports are important although usu- coupled via a mutual conductor, the shield. Adopting the current
ally they are neglected in calculations. In both the OC load notations shown in Fig. 12 for the inner and outer system cable
and 50-Ω resistive load cases, major discrepancies occur be- representation, the outer braid current flowing between the cable
tween 30 and 900 MHz due to the presence of the vertical sheath and the ground plane can be treated as the common-mode
supports. Indeed, the currents flowing through them contributed radiated field excitation current.
to a monopole-like field, which is not negligible. In this case, In Fig. 12, the inner current anywhere on the core conductor
the current distribution significantly affects radiation at the fre- is given by
quency below 100 MHz, approximately 300, 600, and 900 MHz.
1 − ΓL e−γ 1 l e e2γ 1 z Vth
The computed results that include the vertical support contribu- IS (z) = e−γ 1 z (21)
tions match the measurement results very well. The difference at 1 − ΓS ΓL e−γ 1 l e Zth1 + ZC1
most frequencies is less than 3 dB. The proposed computational
where γ1 and ZC 1 are the complex propagation constant and
time required for the emission calculation problem on an Intel
the complex characteristic impedance of the coax inner TL,
Core2Duo E8400 4.0G computer is only 6.75 s and modeling
respectively.
accuracy is indicated by the comparison results, which indicate
In this paper, the transfer mechanism between the internal
that the proposed model can predict the radiated emissions.
voltages and currents, and the external voltages and currents
is modeled by the transfer impedance only, i.e., the transfer
IV. RADIATION FROM A COAXIAL CABLE OVER GROUND admittance has not been included as for an optimized cable
A. TL Models of Shielded Cable shield it has a negligible effect [22], [23]. According to the
coupling model shown in Fig. 13, the braid shield leakage source
To investigate the radiated fields from shielded coaxial ca- VT can be expressed in terms of the transfer impedance and the
bles, the bare wire in Fig. 1 was replaced by a RG 58 coaxial inner excitation current
cable together with two SMA-type connectors (introducing two
2.6-pF equivalent parallel capacitances). Based on the conven- VT (z, ω) = ZT IS (z, ω) (22)
MENG et al.: FAST PREDICTION OF TRANSMISSION LINE RADIATED EMISSIONS 1301

Fig. 14. Comparison of the measured and simulated shield current for the test Fig. 15. Validation of radiation model by comparing calculations and exper-
coaxial cable with measurement position z = 60 cm. (a) Z L = infinite (open imental data with the antenna height H a nt = 1 m. (a) Z L = infinite (open-
circuit) and (b) Z L = 50 Ω. circuit) and (b) Z L = 50 Ω. Solid line represents the measurement, dotted line
is for the calculated result without vertical support contributions, and dashed
one is for the proposed model.

where the approximate formula for the transfer impedance ZT

of single braided shields developed by Kley is given in [24]. C. Measurement Results for the Coaxial Cable
The current flowing between the braid and the ground return
is obtained by [25], [26]: By using the same measurement configuration as shown in
Fig. 1, the radiated fields from a RG 58 coaxial cable above
 le ground plane (le = 1.2 m, h = 5 cm) were studied. The coaxial
0 cosh [γ2 (z − ξ)] VT (z, ω)dξ
ISH (z) = cosh(γ2 z) cable under test was fed by a source feeding cable connected
ZC 2 sinh(γ2 le )
to a wideband swept sinusoidal signal excitation source whose
z
1 amplitude is 0.224 V (107 dBμV) over the entire range 10 MHz–
− sinh [γ2 (z − ξ)] VT (z, ω)dξ (23)
0 ZC 2 1 GHz frequency.
Fig. 14(a) and (b) shows the comparison between the calcu-
where γ2 and ZC 2 are the complex propagation constant and lated shield current spectrum using (23) and measured results
the complex characteristic impedance of the coax outer TL, using the current probe. The coax internal and external TL equa-
respectively. tions were discretized by dividing the line into 120 segments.
The results from OC load and 50-Ω resistive load cases both give
a very good agreement between the model and the measurement.
B. Model for Coax Radiated Field Calculation The difference at most frequencies is less than 3 dB.
To determine the radiated field generated by the coaxial cable Fig. 15(a) and (b) shows the comparison of the measured and
system shown in Fig. 12, we regard the outer shield as being computed radiated electric fields at the observation point P(0,
made up of a large number of short dipole radiators. Like the 1 m, 60 cm). The radiation field was studied in the frequency
analysis in Section III-B, see Fig. 8, the shield current distri- range 30 MHz–1 GHz. The computational time for solving the
bution along the cable can be analytically calculated according emission calculation problem on an Intel Core2Duo E8400 4.0G
to (23). The total radiated field from the entire coaxial cable computer is only 26 s. The computed electric field also includes
system can then be calculated by combining contributions from the contributions from the cable-end vertical supports. Notice
the sheath current and the line-end vertical supports current, as that at the most frequencies, both the open load results and
shown in Fig. 8. 50-Ω resistive load results show good agreement between the
1302 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 56, NO. 6, DECEMBER 2014

measured and simulated results, with respect to both the reso- [13] C. A. Balanis, Antenna Theory: Analysis and Design. Hoboken, NJ,
nant frequencies and the peak values. This generally good match USA: Wiley, 2005, pp. 151–154.
[14] N. N. Rao, Elements of Engineering Electromagnetics, 3rd ed. Engle-
provides a further validation of the radiated emissions model wood Cliffs, NJ, USA: Prentice Hall, 1992.
in the case of coaxial cable applications. Major discrepancies [15] C. R. Paul, Analysis of Multiconductor Transmission Lines, 2nd ed.
occurring around 450 MHz and 900 MHz may be due to inac- Hoboken, NJ, USA: Wiley, 2007, pp. 240–281.
[16] Y. Tang, H. Chen, H. Wang, F. Dai, and S. Jiang, “Transmission line models
curacies in the modeling of the coaxial braid current. Another used in travelling wave studies,” in Proc. IEEE Transmiss. Distrib. Conf.,
contributing factor for the discrepancies observed may be due 1999, vol. 2, pp. 797–803.
to the uncertainty in the position of the antenna during measure- [17] A. Cozza, F. Canavero, and B. Demoulin, “High-frequency extension for
the transmission-line theory for an open line,” in Proc. IEEE Int. Symp.
ment procedures. Electromagn. Compat., Mar. 16–18, 2004, vol. 1, pp. 165–170.
[18] J. Paul, C. Christopoulos, D. W. P. Thomas, and X. Liu, “Time-domain
TLM simulation of electromagnetic wave interaction with thin-wires,”
V. CONCLUSION IEEE Trans. Electromagn. Compat., vol. 47, no. 3, pp. 447–455, Aug.
2005.
A modeling technique has been presented to speed up the [19] IEEE Standard for Validation of Computational Electromagnetics Com-
evaluation of radiated fields from interconnect cables. Based puter Modeling and Simulations, IEEE Standard P1597.1, 2008.
[20] S. Caniggia and F. Maradei, “SPICE-like models for the analysis of the
on the Hertzian dipole radiation theory and a TL frequency- conducted and radiated immunity of shielded cables,” IEEE Trans. Elec-
dependant current solution, the radiating cables are modeled by tromagn. Compat., vol. 46, no. 4, pp. 606–616, Nov. 2004.
a large number of short dipoles and the radiated field generated [21] R. Tiedemann, “Current flow in coaxial braided cable shields,” IEEE
Trans. Electromagn. Compat., vol. 45, no. 3, pp. 531–537, Aug. 2003.
by a cable is then obtained by summation of contributions from [22] A. H. Badr, F. A. Benson, and J. E. Sitch, “Interference between braided
all dipoles. The proposed model can be applied to open line as coaxial cables,” IEE Proc., vol. 128, no. 5, pp. 347–353, Jul. 1981.
well as to coaxial cable systems. This paper also demonstrated [23] S. Sali, F. A. Benson, and J. E. Sitch, “General crosstalk equations between
two braided coaxial cables in free space,” IEE Proc., vol. 130, no. 6,
the capacitance of end wires in the calculations of radiated pp. 306–312, Sep. 1983.
fields. The intermediate level model method generally requires [24] T. Kley, “Optimized single-braided cable shields,” IEEE Trans. Electro-
much lower computational resources than a traditional full-wave magn. Compat., vol. 35, no. 1, pp. 1–9, Feb. 1993.
[25] S. Sali, “Response of externally excited coaxial cables with wire braided
simulation. shields,” IEE Proc. Sci. Meas. Technol., vol. 141, no. 4, pp. 266–272, Jul.
1994.
[26] S. Sali, “A circuit-based approach for crosstalk between coaxial ca-
REFERENCES bles with optimum braided shields,” IEEE Trans. Electromagn. Compat.,
vol. 35, no. 2, pp. 300–311, May 1993.
[1] CISPR 16-1 Specification for Radio Disturbance and Immunity Measur-
ing Apparatus and Methods, Part 1: Radio Disturbance and Immunity
Measuring Apparatus, 1993.
[2] CISPR 16-2 Specification for Radio Disturbance and Immunity Measuring
Apparatus and Methods, Part 2: Methods of Measurement of Disturbances
and Immunity, 1996. Jin Meng was born in Henan, China, in 1977. He re-
[3] P. Wilson, G. Koepke, J. Ladbury, and C. L. Holloway, “Emission and im- ceived the B.S. degree in electrical engineering from
munity standards: Replacing field-at-a-distance measurements with total Chongqing College of Communication, Chongqing,
radiated-power measurements,” in Proc. IEEE Int. Symp. Electromagn. China, in 1999, the M.S. degree and Ph.D. degrees
Compat., Aug. 13–17, 2001, vol. 2, pp. 964–969. both in electrical engineering from the Naval Uni-
[4] P. F. Wilson, D. A. Hill, and C. L. Holloway, “On determining the max- versity of Engineering, Wuhan, China, in 2002 and
imum emissions from electrically large sources,” IEEE Trans. Electro- 2006, respectively.
magn. Compat., vol. 44, no. 1, pp. 79–86, Feb. 2002. In 2006, he joined the Research Institute of Power
[5] R. S. Shi, A. Darcherif, and J. C. Sabonnadiere, “Computation of transient Electronic Technology, Naval University of Engi-
electromagnetic fields radiated by a transmission line: An exact model,” neering, as a Lecturer, where he is currently a Profes-
IEEE Trans. Magn., vol. 31, no. 4, pp. 2423–2431, Jul. 1995. sor in electrical engineering. From October 2011 to
[6] D. A. Hill, G. Camell, K. H. Cavcey, and G. H. Koepke, “Radiated emis- September 2012, he was as a Visiting Scholar at the George Green Institute for
sions and immunity of microstrip transmission lines: Theory and rever- Electromagnetics Research, University of Nottingham, Nottingham, U.K. His
beration chamber measurements,” IEEE Trans. Electromagn. Compat., research interests include electromagnetic compatibility of power conversion,
vol. 38, no. 2, pp. 165–172, May 1996. developing models for radiated and conducted electromagnetic interference
[7] A. Cozza and B. Denoulin, “Close-form expressions for the total power (EMI) characteristics and prediction, EMI reduction techniques, and computer-
radiated by an electrically long multiconductor line,” IEEE Trans. Elec- aided circuit simulation.
tromagn. Compat., vol. 51, no. 1, pp. 119–130, Feb. 2009.
[8] C. R. Paul and D. R. Bush, “A comparison of the contribution of common-
mode and differential-mode currents in radiated emissions,” IEEE Trans.
Electromagn. Compat., vol. 31, no. 2, pp. 189–193, May 1989.
[9] D. W. P. Thomas, C. Christopoulos, and E. T. Pereira, “Calculation of ra-
diated electromagnetic fields from cables using time-domain simulation,” Yu Xian Teo was born in Kuala Lumpur, Malaysia,
IEEE Trans. Electromagn. Compat., vol. 36, no. 3, pp. 201–205, Aug. in 1987. He received the B.Eng. degree in electri-
1994. cal and electronic engineering from the University
[10] B. U. Musa, W. H. Siew, and M. D. Judd, “Computation of transient elec- of Nottingham, Nottingham, U.K., in 2009. Since
tromagnetic fields due to switching in high-voltage substations,” IEEE 2009, he has been working toward the Ph.D. de-
Trans. Power Del., vol. 25, no. 2, pp. 1154–1161, Apr. 2010. gree in electrical and electronic engineering at the
[11] Y. Kami and R. Sato, “Analysis of radiation characteristics of a finite- University of Nottingham, U.K., partially sponsored
length transmission line using a circuit-concept approach,” IEEE Trans. by the University under Tower Innovation Scholar-
Electromagn. Compat., vol. 30, no. 2, pp. 114–121, May 1988. ship.
[12] M. Leone, “Closed-form expressions for the electromagnetic radiation of His research interests include computational
microstrip signal traces,” IEEE Trans. Electromagn. Compat., vol. 49, electromagnetics, electromagnetic compatibility with
no. 2, pp. 322–328, May 2007. measurements, and transmission-line modeling method.
MENG et al.: FAST PREDICTION OF TRANSMISSION LINE RADIATED EMISSIONS 1303

David W. P. Thomas (M’94–SM’09) received the Christos Christopoulos (M’92–SM’04–F’06) was

B.Sc. degree in physics from the Imperial College of born in Patras, Greece, in 1946. He received the
Science and Technology, London, U.K., the M.Phil. Diploma degree in electrical and mechanical engi-
degree in space physics from Sheffield University, neering from the National Technical University of
Sheffield, U.K., and the Ph.D. degree in electrical en- Athens, Athens, Greece, in 1969, and the M.Sc.
gineering from Nottingham University, Nottingham, and D. Phil. degrees from the University of Sussex,
U.K., in 1981, 1987, and 1990, respectively. Brighton, U.K., in 1979 and 1974, respectively.
In 1990, he joined the Department of Electrical In 1974, he joined the Arc Research Project at
and Electronic Engineering, University of Notting- the University of Liverpool, Liverpool, U.K., and in-
ham, Nottingham, as a Lecturer, where he is cur- volved in the research on vacuum arcs and breakdown
rently a Professor of electromagnetics applications. at the Culham Laboratory, U.K. Atomic Energy Au-
His research interests include electromagnetic compatibility, electromagnetic thority, for two years. In 1976, he joined the University of Durham, Durham,
simulations, power system transients, and power system protection. U.K., as a Senior Demonstrator in electrical engineering science. In October
Prof. Thomas is a Member of the CIGRE and a Convener for CIGRE JWG 1978, he joined the Department of Electrical and Electronic Engineering, Uni-
4.207 and JWG C4.31. He is also a Vice-Chair for the IEEE EMC Technical versity of Nottingham, Nottingham, U.K., where he is currently an Emeritus
committee 7 “Low frequency EMC.” Professor of electrical engineering for Electromagnetics Research. He is the au-
thor or coauthor of more than 400 research publications, seven books, and seven
chapter contributions in books. His research interests include computational
electromagnetics, electromagnetic compatibility, signal integrity, protection and
simulation of power networks, and electrical discharges and plasmas.
Prof. Christopoulos has received the Electronics Letters Premium, the Snell
Premium, and the Measurement and Technology Premium by the Institute of En-
gineering and Technology (IET) and several conference best paper awards. He
is a Member of the IET. He was an Executive Team Chairman of the Institution
of Electrical Engineers Professional Network in Electromagnetic Compatibil-
ity and an Associate Editor of the IEEE EMC TRANSACTIONS. He was the
Chairman of Union Radio-Scientifique Internationale Commission E on the
Electromagnetic Environment and Interference. In 2010, he received the Am-
brose Fleming Medal. He is Fellow of the Royal Academy of Engineering.