Challenge B: An environmentely friendly railway
FAST AND ACCURATE MEASUREMENT OF RADIATED EMISSIONS OF MOVING TRAINS
ACCORDING TO IEC 62236
E. Fedeli1, S. A. Pignari2, and G. Spadacini2
1
R.F.I S.p.A., Rome, Italy, 2Politecnico di Milano, Milan, Italy
Abstract
This work deals with the experimental characterization of the spectrum of electromagnetic radiated
emissions (RE) of moving trains. In particular, a fully-automated measurement system and an
optimized measurement procedure are designed and implemented, conforming to requirements of the
current International Standards IEC 62236-2 and EN 50121-2. The proposed system represents an
optimization of a previous version, obtained by resorting to a unique hybrid biconical/log-periodic
antenna instead of the two separate antennas (biconical and log-periodic), as suggested in the
aforementioned Standards. As a result, the measurement system is simpler, cheaper, it allows
minimization of the number of required train runs (i.e. testing time and thus costs) and, as a whole, it
leads to improved performance of the measurement process. Full compliance of the proposed
measurement procedure and system with IEC 62236 and CISPR 16 in terms of measurement
uncertainty is discussed. As a specific example, test results obtained with measurements carried out
along the high-speed railway line interconnecting Milan to Bologna, Italy, are reported.
Introduction
Measurement of radiated emissions (RE) of moving trains in the frequency range 9 kHz – 1 GHz is
one of the most complex and time-consuming tests for assessing the Electromagnetic Compatibility
(EMC) of a railway system. The implementation of the measurement procedure according to EN
50121-2 and IEC 62236-2, [1], [2], requires to trade off between the need to obtain enough
information about the emission phenomenon, and the minimization of costs and economics of time of
extensive tests.
Indeed, due to the electrically-large dimension of the railway infrastructure, and due to the fast moving
source of interference, the electromagnetic emissions radiated by the railway environment are
distributed in space and non-stationary in time. Actually, the minimum measured data set accepted by
the aforementioned Standards for the mere task of compliance verification does not provide enough
information to allow for a satisfactory physical interpretation of the complex emission phenomena
which contribute to the EMC of the system under test [3], [4]. The need for fast sampling of the
electromagnetic field, and for a detailed spectral characterization, not strictly required by the
Standards (such as the association of the measured RE with the train position, and the automatic
detection of broadband transient events) has recently led to the development of sophisticated
measurement systems characterized by high redundancy (i.e., systems involving the simultaneous
use of multiple instruments of the same type) [5].
On the other hand, it should also be noted here that even the acquisition of the minimum data set
accepted by the Standards (monitoring of three frequencies per decade) involves: a) time-consuming
measurement campaigns (requiring dedicated use of a railway line), and b) several train runs.
Nevertheless, by adopting specific measurement procedures and modern measurement-automation
techniques, it is possible to drastically reduce testing time and costs. To this end, a few years ago an
optimized measurement system was developed by resorting to a measurement chain involving three
broadband antennas (loop, biconical and log-periodic) in line with the indications provided in EN
50121 and IEC 62236 [6]. According to that solution, the antennas were connected to a spectrum
analyzer (SA) by a computer-controlled switch and, in order to minimize the number of train runs
needed to characterize the RE in the whole frequency range of interest, different SA operation modes
were used in specific sub-bands. Particularly, the entire electric-field measurement band was covered
within a train run only, This was done by exploiting two consecutive frequency sweeps in succession:
30 MHz – 300 MHz (biconical antenna) and 300 MHz – 1 GHz (log-periodic antenna) [6]. In this work,
the aforementioned measurement system has been re-designed and further optimized by exploiting a
hybrid biconical/log-periodic antenna to perform a unique sweep from 30 MHz to 1 GHz, without the
need for a switch.
Challenge B: An environmentely friendly railway
The optimized measurement chain proposed in this paper allows for cost abatement by minimizing the
train runs, and it ensures superior performance in the characterization of the time-variant character of
the railway-system RE phenomenon. Results of a preliminary experimental campaign, aimed at
surveying the RE of the Italian high-speed railway lines, demonstrates the potential of the proposed
measurement method and, at the same time, it confirms that emissions are well below the limits
foreseen by the international Standards.
Measurement system
The measurement system is sketched in Fig. 1. It is composed of two antennas: a loop antenna for
measurement of the magnetic field, and a hybrid biconical/log-periodic antenna for the measurement
of the electric field. A laptop is used to control the SA via an USB-to-GPIB interface.
The difference between the proposed measurement chain and the previous system, [6], resides in the
use of a unique hybrid antenna, which was firstly developed in [7] and is nowadays widespread in
most EMC laboratories. This antenna covers the bandwidth of both the classical biconical antenna (30
MHz – 300 MHz) and the log-periodic antenna (300 MHz – 1 GHz). As a consequence, this allows for
avoiding the use of a computer-controlled switch that in [6] was employed as a fast multiplexer to
allow the SA to manage two consecutive sweep in succession with two different antennas.
Conversely, by using the measurement system in Fig. 1, the entire frequency range 30 MHz – 1 GHz
can be analyzed with a single frequency sweep. In addition to sizable simplification of the
measurement setup and related procedure, the main advantage of this new implementation is the
elimination of: (a) the switching time; (b) the delay time related to continuous resetting of the SA
start/stop frequencies; (c) the switch insertion-loss. While the latter aspect improves the measurement
system sensitivity, the formers allow for increasing the acquisition speed and, therefore, to provide a
more detailed characterization of the RE in the time-frequency plane.
Loop Antenna
[9 KHz – 30 MHz]
Hybrid Biconical/Log-periodic Antenna
[30 MHz – 1 GHz]
Spectrum Analyzer
Computer-controlled measurement
(via GPIB bus)
Figure 1: Measurement system
Challenge B: An environmentely friendly railway
SA Operation Modes
The Standard IEC 62236-2 / EN 50121-2 specifies that RE should be characterized by analyzing at
least three frequencies per decade in the frequency interval of interest 9 kHz – 1 GHz [2, Section
5.2.1]. The basic operation mode foresees to continuously monitor RE at each single frequency of
interest, tuned on a receiver (i.e., a spectrum analyzer operating in the zero-span mode) with 50 ms
sweep time (which is the observation time) and peak detector, during the time of a complete train run
on the railway line, in front of the measurement system. In particular, in the sub-range 30 MHz – 1
GHz the test have to be performed twice, as the electric-field antenna has to operate both in the
vertical and the horizontal polarization. Note that this last requirement appears as a new statement of
the current version of EN 50121-2 [2], whereas there was no evidence of such a statement in the
2001 version of the Standard, referred in the previous work [6]. Consequently, the minimum number
of train runs required by such a non-optimized Standard test mode sums to a number between 18 and
21 (depending on specific choices involved in the design of the measurement procedure), according
to Table I.
TABLE I
Minimum number of train runs by operating the system in a non-optimized mode
Frequency [MHz]
0.009
Measured Quantity
Resolution Bandwidth
No. of antenna polarizations
Minimum no. of frequencies
Minimum no. of train runs
0.1 0.15
200 Hz
3
3
1
Magnetic Field
9 kHz
1
3
3
3
3
10
30
100 300
1000
Electric Field
120 kHz
2
3
3
3 to 6
6
In order to speed up the testing process, the Standard IEC 62236-2 / EN 50121-2 allows to use
specific operation modes of the SA, which are non-conventional in the field of EMC-compliant RE
tests. Such operation modes are:

Cyclic zero-span measurement: Several selected frequencies are analyzed during each train
run. The frequencies are continuously tuned in succession, with the SA operating in zerospan mode, and held for the sweep time (50 ms recommended) plus a time interval long
enough to acquire the reading (via computer-controlled measurement automation).

Frequency-sweep measurement: The continuous frequency-sweep technique (i.e., a
continuous sweep from a starting frequency f start to a stop frequency f stop ) is used to analyze
a specific portion of the spectrum during each train run. The number of frequencies is called
trace points and is set by the operator. In this case the sweep time is defined as the time
needed to complete the entire frequency sweep, and cannot be arbitrarily set. Indeed, the
sweep time t sweep has to fulfill the following condition:
tsweep  k
f
2
RBW
(1)
where RBW is the resolution bandwidth, f  f stop  f start is the frequency span, and k is a
coefficient depending on SA technology. Nowadays, SAs are provided with embedded
software systems that automatically set the minimum sweep time according to (1).
It is clear that, on the one hand, use of these optimized techniques leads to a reduction of the number
of train runs required to characterize the whole spectrum but, on the other hand, it makes the data
acquisition slower in term of sampling time. Indeed, the monitoring of several frequencies in
successions (instead of a single frequency) means that a significant time interval separates
subsequent readings corresponding to the same frequency. Of course, it is important to preserve
Challenge B: An environmentely friendly railway
consistency between this time interval and the train dynamics. In practice, a time interval of one
second is accepted by the Standard [2, Annex B11].
The above descrided two optimized operation modes were exploited in the automated measurement
system presented in [6]. Moreover, the concept of cyclic measurement was extended to the
frequency-sweep operation, as a third operation mode was conceived: the cyclic frequency-sweep,
which consists of two consecutive frequency sweeps in succession during a train run. In particular,
this measurement mode was used to analyze the entire electric-field measurement band within a train
run only by exploiting two consecutive sweeps in succession: 30 MHz – 300 MHz and 300 MHz – 1
GHz [6]. The need for two sweeps was justified by the use of separate antennas: a biconical antenna
for the first, and log-periodic antenna for the second sweep (in line with recommendations of the
Standard). Although a fast computer-controlled electromechanical switch was exploited to commute
the SA input port from an antenna to another, the switching time and the delay time related to
continuous resetting of the SA start/stop frequencies were not negligible and contributed to increase
the time interval, that is, to slow down the acquisition of measurement data.
The novel measurement system shown in Fig. 1, thanks to the use of a hybrid biconical/log-pediodic
antenna, can cover the whole electric-field band (30 MHz – 1 GHz) in one frequency-sweep without
using any switch.
Optimized Test Procedure
The measurement system in Fig. 1 can be used, in general, with any SA operation mode allowed by
the Standard IEC 62236-2 and EN 50121-2 as described in the previous Section. Particularly, the
operation can be optimized in order to minimize the number of train runs required for full-compliance
verification, i.e., for the acquisition of the minimum data set. This operation mode, referred in the
following to as “optimized test procedure”, foresees the following four sub-band measurements:
1) Magnetic Field, 9 kHz - 150 kHz
In this sub-band the SA is operated in cyclic zero-span mode. In particular, the specific SA employed
in our tests and the software developed for measurement automation allow for the monitoring of up to
5 frequencies within the time interval of one second.
Note that frequency-sweep operation of the SA is unsuitable in this frequency interval, since the very
low resolution bandwidth (200 Hz) would lead to unacceptable sweep times (several seconds)
according to (1).
2) Magnetic Field, 150 kHz – 30 MHz
In this sub-band the SA is operated in frequency-sweep mode. In particular, the specific SA employed
in our tests and the software developed for measurement automation allow for monitoring the band
within the time interval of one second, with 6640 trace points.
3) Electric Field, Vertical Polarization, 30 MHz – 1 GHz
In this sub-band the SA is operated in frequency-sweep mode. In particular, the used SA and the
developed automation software allow monitoring the band within the time interval of 0.35 seconds,
with 8192 trace points.
4) Electric Field, Horizontal Polarization, 30 MHz – 1 GHz
The operation mode is the same as in 3).
In conclusion, from the standpoint of economics of time and costs, the characterization of the
electromagnetic emissions radiated by a railway system according to the optimized procedure
proposed in this paper requires only 4 train runs (see Table II). If such a requirement is compared with
the 18-21 train runs needed in order to perform emission measurement according to the nonoptimized mode (see Table I), the reduction amounts to some 80%. Moreover, as one can see in
Table II, the number of sampled frequencies is by far greater than the minimum data set strictly
required by the Standard.
Challenge B: An environmentely friendly railway
TABLE II
Number of train runs with the optimized test procedure
Frequency [MHz]
0.009
Measured Quantity
Resolution Bandwidth
No. of antenna polarizations
No. of frequencies
Train runs
0.15
200 Hz
5
1
Magnetic Field
9 kHz
1
6640
1
30
1000
Electric Field
120 kHz
2
8192
2
Measurement uncertainty
A preliminary characterization of the system in terms of expanded measurement instrumentation
uncertainty (EMIU) [8] has demonstrated compliance with CISPR requirements [9]. EMIU estimation
comprises the identification of all the components of uncertainty that potentially affect the system,
These are partly related to the instrumentation, and partly related to the test site and the
measurement procedure. Particularly, by referring to the guidelines [8], the following system
components have to be accounted for via an appropriate uncertainty analysis: SA, antennas,
connections between SA and antennas through coaxial cables, measurement test site and test
repeatability. Calculation details are not reported in this paper for the sake of conciseness.
It is worth noting that, to the aim of EMIU estimation, the proposed system presents only two
differences with respect to the previous version [6] (for which full compliance with CISPR requirement
was proved): a) absence of a switch; b) hybrid biconical/log-periodic antenna. Concerning the former
difference, the absence of insertion-loss related to the switch is an unquestionable advantage as a
relevant source of uncertainty is taken out. Moreover, as regards the hybrid antenna, it can be
assumed that its performance in terms of EMIU is comparable to that of a biconical dipole in the lower
frequency range, and to the EMIU of a regular log-periodic dipole array at higher frequencies, except
for some factors such as height dependency, phase-center variation with frequency, antenna
directivity, beam pattern, possible capacitive loading (used for low-frequency improvements) which
demands for additional and specific considerations [10].
Experimental measurements
This section reports some results obtained by operating the measurement system in the optimized
mode. Tests were performed on the 25 kV – 50 Hz high-speed railway line interconnecting Milan to
Bologna, Italy. Aim of this first measurement campaign along that line was not EMC assessment, but
rather validation of the developed measurement system, and preliminary exploration of the
characteristics of the RE spectral content via repeated measurements.
As an example, Fig. 2 reports the outcome of four different tests, that is, four different train runs. The
sub-band here considered is 150 kHz – 30 MHz and the measured quantity is the magnetic-field.
Accordingly, measurement was performed with the SA operating in the frequency-sweep mode. In
particular, for each frequency, the maximum RE measured during the testing time interval is reported
in the plot. The green curve represents a preliminary measurement carried out in the absence of
trains visible to the naked eye at the place where the measurement system was placed. This was
done in order to obtain information on the background noise. This preliminary measurement is
requested by the Standards, and allows for the identification of possible sources of intentional
radiation, both from the outside world and from the railway infrastructure, which are leaved out of any
consideration related to EMC assessment. In this specific case, these sources are amplitudemodulation (AM) and short-wave (SW) radio broadcasting, as well as the 27 MHz Eurobalise
transmission system. The remaining coloured curves refer to four different test results (i.e., different
train runs). Peaks inside the highlighting circles, which are above the background noise, are due to
the train passage in front of the measurement system. The black dashed curve indicates the limit
values imposed by Standards IEC 62236-2 and EN 50121-2. No spectral lines associated with
emission radiated by the train running along the railway line exceed the limits, and the vast majority of
the spectrum results to be below the limits by tens of decibels.
In Fig. 3, the time evolution of RE (magnetic field) at the frequency of 329.24 kHz during test #1 (see
also Fig. 2) is plotted.
Challenge B: An environmentely friendly railway
80
test #1
test #2
test #3
test #4
background noise
limits 25 kV ac
magnetic field [dBA/m]
70
60
AM
50
Eurobalise
(27 MHz)
40
30
SW
20
10
0
-10 5
10
6
10
7
10
8
10
frequency [Hz]
Figure 2: Radiated emissions (magnetic field) in the frequency sub-band 150 kHz – 30 MHz. The
green line represents a measurement performed in the absence of trains visible to the naked eye on
the railway line. This measurement was carried out in order to characterize background noise. The
following sources of intentional radiation cab be observed in the plotted spectra: AM and SW radio
broadcasting, and 27 MHz Eurobalise transmission system. The remaining coloured curves refer to
four different test results (repeated measurements). Spectral peaks inside the highlighting circles are
ascribed to the train RE. The black dashed line indicates the limit values imposed by the Standards
IEC 62236-2 and EN 50121-2.
Challenge B: An environmentely friendly railway
test #1 - frequency = 329.24 kHz
50
magnetic field [dBA/m]
45
40
35
30
25
20
15
10
5
0
0
10
20
30
40
50
60
time [s]
Figure 3: Time evolution of radiated emissions (magnetic field) at the frequency of 329.24 kHz during
test #1 (see also Fig. 2).
Acknowledgements
This work was supported in part by the Italian Ministry of Education, University and Research (MIUR)
under a Programme for the development of Research of National Interest (PRIN Grant
#20089J4SM9), and in part by Italcertifer S.C.p.A, Rome, Italy (Contract # PAO8RICC01).
References
[1] IEC 62236-2, Railway applications – Electromagnetic compatibility; Part 2: Emission of the whole
railway system to the outside world, IEC, 2008.
[2] EN 50121-2, Railway applications - Electromagnetic compatibility; Part 2: Emission of the whole
railway system to the outside world, CENELEC, 2006.
[3] S. A. Pignari and D. Bellan, “Impact of the infrastructure on the electromagnetic emissions
radiated by a railway system,” in Proc. 8th World Congress on Railway Research, May 18-22,
2008, Seoul, Korea, Paper S.3.1.4.1, pp 1-7.
[4] A. Cozza and B. Démoulin, “On the modeling of electric railway lines for the assessment of
infrastructure impact in radiated emission tests of rolling stock,” IEEE Trans. Electromagn.
Compat., vol. 50, no. 3, pp. 566-576, Aug. 2008.
[5] A. J. Rowell, D. Bozec, S. A. Seller, L. M. McCormack, C. A. Marshman, and A. C. Marvin,
“Improved measurement of radiated emissions from moving rail vehicles in the frequency range 9
kHz to 1 GHz,” in Proc. 2004 Int. Symp. on Electromagn. Compat., Santa Clara, CA, USA, Aug.
9-13, 2004, pp. 19-24.
[6] S. A. Pignari, G. Spadacini, D. Bellan, and A. Gaggelli, “Measurement of rolling-stock radiated
emissions according to Standard EN 50121,” in Proc. EMC−Zurich Symposium, Singapore, Feb.
27-Mar. 3, 2006, Workshop Notes, pp. 250-255.
[7] S. J. Porter, A. C. Marvin, “A new broadband EMC antenna for emissions and immunity,” in Proc.
EMC’94 Roma Int. Symp. Electromagn. Compat., Sept, 13-16, 1994, Rome, Italy.
Challenge B: An environmentely friendly railway
[8] IEC CISPR 16, Specification for radio disturbance and immunity measuring apparatus and
methods – Part 4-2: Uncertainties, statistics and limit modeling – Uncertainty in EMC
measurements, IEC, 2003.
[9] IEC CISPR 16, Specification for radio disturbance and immunity measuring apparatus and
methods – Part 1: Radio disturbance and immunity measuring apparatus, IEC, 2003.
[10] Z. Chen, “Understanding the measurement uncertainties of the bicon/log hybrid antenna,” ITEM
1999 EMC Guide, pp. 1-6.