Line Communications on Low Voltage Buried Cable

Fawzi Issa*, Daniel Chaffanjon*, Andre Pacaue* *Electricit6 de France (Research and Development Division), France **EcoleSup6rieure d'Electricit6 (Service Radio6lectricit6 et Electronique), France Email: [email protected], Tel: +33(0) 147655515, Fax: +33(0) 147653277
Abstract The purpose of this paper is the study of the radiations due t o the injection of power line communications (PLC) signals belonging t o the frequency range [ l M H z - 30MHzI on a low voltage buried cable. An original model has been developed and simulated with the N E C 4 code (Numerical Electromagnetic Code) t o characterize the radiated emissions. The numerical results will be validated using experimental data. Besides, a definition of an electric coupling factor is proposed, a study of the propagating waves wavelengths and the energetic exchanges are detailed too.
Keywords power line communications, radiated emissions, antenna theory, N E C 4 code, eiectromagnetic fields measurement

I. Introduction

The possible use of the electrical low voltage network as a communication medium becomes very attractive as no additive cabling is necessary. The PLC signals are assumed t o belong to a frequency range beginning at 1MH.z up to 30MH.z thus enabling the transmission of high data rates of the order . of a few Mbitsls (internet, digital movies, ...) (see e.g. [dostert]). The first developments of PLC networks in France will be located in urban areas where the transmission medium which is used more and more is the buried one. This specific choice is related to a will for networks protection against noises, storms and for obvious environmental esthetic reasons. It is the reason why we are going to focus on the radiated emissions associated with PLC signals injected on a low voltage underground cable since many studies have already been done on overhead multiconductor lines (se e.g. [damore] [helier] [naredo]). In the following section, we will present the studied cable and we will show why we have chosen a common mode propagation approach instead of a pure diierential one. Then, an original model taking into account all the propagation phenomena and the ground properties will be presented. Some numerical simulations based on the antenna theory will be done which will enable us to get some important informations about the propagating waves wavelengths above the ground, the energetic exchanges between the cable and the ground and we will finally introduce the coupling factor concept. All these simulation results will be compared to measurements done in real experimental conditions on a low voltage buried cable.

11. Characterization of the radiated emissions

The low voltage underground cable used in the French distribution grid is the HN33S33 cable depicted on figure 1with table (1) detailing the associated caption.

Fig. 1. Low voltage underground HN33S33 cable

1 Phase conductor(a1uminium) 2 Chemically reticuled polyethylene(CRP) 3 Neutral conductor(aluminium) 4 Sheath of the neutral conductor 5 Shield(stee1y ribbon) 6 Exterior CRP sheath

In normal operational conditions, a signal injection is practised between the neutral conductor and a phase conductor thus creating a pure differential establishing mode. In fact, the steely ribbon in galvanic contact with the neutral conductor can be seen as a shield which will mask the majority of the radiated emissions coming from this previous propagation mode. Moreover, the neutral conductor is connected to the ground via metallic guides which will create a circulation of P L C signals in a secondary loop integrating the ground. It is the reason why we can suppose that the majority of the radiated emissions comes from a common mode propagation. To take into account the ground, we can consider a coaxial model, the core representing the whole cable and the shield the ground. The ground will then present dielectric losses properties and the core is assumed to be a perfect conductor. The separation between the two previous materials is chemically reticuled polyethylene (CRP). The core radius corresponds to the equivalent cable radius and the shield radius has been taken in order of the penetration depth in the ground at 1MHz. Figures 2 and 3 give our coaxial model shape.
Perfect condusulr(son)

Fig. 2. Coaxial cable model into a common mode approach

~crfcslconduslor(sorc)

/
Ground (shield)

C R P (insul.tion)

Fig. 3. Coaxial cable model into a common mode approach

1 1 Simulations based o n t h e antenna theory 1. We have chosen the antenna theory as it is the most rigorous method for electromagnetic fields evaluation [burkel]. We used the NEC4 code implementing the antenna theory via the method of moments for radiating structures embedded in a lossy medium characterized by a relative permittivity and a finite conductivity [burke3]. The losses are taken into account using the Sommerfeld integrals. l ie The only assumption made in N E C is the thin wire hypothesis which assumes that al the w r s radius must be much smaller than the injected signals wavelength Xo. Hence, for a structure that can be modelled as a union of straight elementary segments, we have to spatially discretize it [burke2] with a step 6 verifying condition (2).

where cu is a weighting coefficient which can take typical values between 10 and 20. In the following simulations, we took a monochromatic signal generator delivering an excitation like Eo cos(2nfot) with an internal impedance of lOOR where Eo = 5V and fo E [lMHz...30MHzI. We added a terminal resistive load of lOOR too.

A. Simulation in a semi-infinite ground In this section, we have considered a burial depth of 80cm which is the depth used by transport and distribution utilities for underground cables practical buring. Figure 4 illustrates the simulation model and the chosen observation line (OL) is given by equation (3).
O L = {(x,y,z) E R~ I -70m 5 x 5 70m,y =Orn,z=2m) (3)

OL

2
4

OL
AUI

GROUND

Fig.4. Low voltage buried cable in a semi-infinite ground

For the given observation line defined by equation (3), we have calculated the electric and magnetic fields distributions for the three following typical frequencies 4MHz, lOMHz and 17MHz. The obtained simulations results are depicted on figure 5 (resp. 6) for the electric (resp. magnetic) field. We can clearly see on figures 5 and 6 that we have a non tranverse electromagnetic (TEM) propagation mode and we were right for not having applied some less rigorous and faster methods such as the quasistatic theory or the transmission line theory [baraton]. From a theoretical point of view, the non T E M propagation is related to the external coaxial model radius (10m) which is not neglictible compared to the injected signals wavelengths (10m at 30MHz in free space). Another interesting note is relative to the visible oscillations of the main components of the electric and magnetic fields. As there is an impedance mismatch at the terminal port, we have some standing waves phenomenon such as the distance between two local minima (or maxima) is quite equal to half the wavelength of the injected signal. For example, for 4MHz, 10MHz and 17MHz, the associated freespace wavelengths are 75m, 30m and 18m, respectively. Considering figures 5 and 6, we approximately ihd for the same frequencies, 72m, 28m and 20m for the wavelengths. These values are not exactly the same as freespace values as one part of the radiated emissions takes place in the ground which characteristics are different from the vacuum properties.

Fig. 5. Electric fields components according to OL

Fig. 6. Magnetic field components according to OL

B. Energetic exchanges By using NEC4, it is possible to get the energetic exchanges between the low voltage buried cable
and the ground. For example, we present in table (4) the power budget for the three typical frequencies 4MHz, lOMHz and 17MHz where f, IP, RP, WL and E denote the studied frequency, the injected power, the radiated power, the wire loss and the efficiency respectively.

The observation of the power budget of table (4) let us to conclude that the efficiency coefficient, which is defined as E = RPIIP, globally decreases when frequency increases. Roughtly speaking, this means that higher the frequency is and lower the radiated emissions are important for the common mode propagation for this kind of cable.

C. Coupling factor The coupling factor tc represents a local transfer function between the electric field magnitude E and these values obviously depend on the studied frequency f . The definition of K is the injected power Pi, given according to equation (5).

Figure 7 shows, for example, the evolution of the coupling factor in the frequency range [lMHz 30MHzI for three observation points located at (x, y ,z) = (-20m, Om, 2m), (x, y, z) = (Om, Om, 2m) and (x, y, z) = (20m, Om, 2m) which are located at a distance from the injection point of 30m, 50m and 70m, respectively. These results are for a common mode propagation only and not a differential mode propagation.
Coupling F a c t o r (in d B )

. . , -----------.---....,---.------~~~~~~--~~~--~~~~ :-----------:---------------:
* \,

10

15 Frequency (in

20

25

30

MHz)

Fig. 7. Calculated coupling factor

We can check that the electric coupling factor is a decreasing function of frequency specially when we are near the injection point, typically between 30m and 50m. Besides, we can note that the dynamics of the coupling factor values is spanned, for one given given frequency, between 13dB and 20dB.

IV. Experimental validation

Our experiment took place in a piece of field located in Paris suburb. The site has been chosen for not having any cable (telecomunications, electricity, water, ...) in its near neighbourhood. Hence, there was no parasit coupling with our experimental cable coming from other near cables. The cable was a HN33S33 one of lOOm long and has been buried at a 80crn depth. Only its extremities were embedded in air. To take into account a common mode propagation, we add near the two cable extremities two metallic guides of l m long. Figure 8 summarizes the general configuration of the experimentation. The measurement of the three electric components has been done by using a dipole antenna and a H F H 2 Z l rod antenna (Rohde&Schwarz) whereas this has been done for the magnetic field by using a HFH2Z2 loop antenna (Rohde&Schwarz) connected to a HP4395A (Hewlett-Packard) spectrum analyzer.

Fig. 8. Description of the measurement set-up

195

To compare the experimental data with the simulated data, we have considered the following measurement line ( M L ) defined by equation (6) and refering to figure 8.

We obtained for example the results of figure 9 for the vertical electric component for a frequency of 17MH.z and an injected power of l5dBm delivered by a Rohde&Schwarz generator.

Fig. 9. Simulated and measurement electric field

We can note that the difference existing between the simulation results and experimental data does not exceed 7dB in average. This is a clear validation of our physical coaxial model to simulate the radiations coming from a common mode propagation on the low voltage buried cable.

V. Conclusions
In this paper, some numerical simulations have been done for characterising the radiated emissions due to PLC signals. These P L C signals are assumed to belonging to a frequency range beginning at l M H z up to 30MHz. We have focused our study on a low voltage buried cable as this medium is increasingly used on the low voltage grid. These simulations enabled us to get some values of the electric coupling factor at one given frequency. Hence, knowing the value of this coupling factor and the injected power on a power line network, we can give an estimate of the electric radiated emissions related to the injection of P L C signals. Let us recall that the coupling factor results are relative to a common mode propagation only. The simulations results have been validated by experimental data. The measurement have been practised on a HN33S33 cable buried in real conditions. To refine our study, one could study the radiations of this cable for a differential mode injection to deduce an estimate of a longitudinal conversion loss ( L C L ) .

VI. Acknowledgement
We would like to gratefully thank Jerry Burke from the Lawrence Livermore Laboratory for having simulated our model, for his helpful advices and his kindness. We would like to acknowledge Marc Hblier from the Ecole Supbrieure d'Electricit6 for his helpful advices concerning the N E C code and his kindness too.

REFERENCES
[dostert] K. Dostert, Aspects of high speed powerline communications, E M C international symposium, June 2000 [damore] M. D'Amore, M.S. Sarto, Electromagnetic field radiated from broadband signal transmission on power line carrier channels, IEEE Transactions on power delivery, Vol. 12, No. 2, April 1997 [naredo] J.L. Naredo, J.L. Silva, R. Romero, P. Moreno, Application of approximated modal analysis methods for PLC systems design, IEEE Transactions on power delivery, Vol. 2, No. 1,January 1987 [helier] M. Hblier, N. Recrosio, G. Fine, Analysis of radiation characteristics of distribution line carriers with the N E C code, prod. I E E E , 1993 [burkel] G.J. Burke, A.J. Poggio, Numerical Electromagnetic Code ( N E C ) : Description theory, Lawrence Livermore Laboratory, 1981 [burkea] G.J. Burke, A.J. Poggio, Numerical Electromagnetic Code ( N E C ) : User's guide, Lawrence Livermore Laboratory, 1981 [burke3] G.J. Burke, N E C 4 : Description theory, Lawrence Livermore Laboratory, 1985 [baraton] P. Baraton, Validity domain of coupling fields to cable theories, Collection des notes internes de la D E R , E D F , 1993