Evaluation of the Interference Potential of PLC Systems

Ming Zhang and Wilfred Lauber

Communications Research Centre Canada, 3701 Carling Avenue

Ottawa, Ontario, K2H 8S2, Canada ming.zhang@crc.ca

Abstract: The purpose of this paper is to analyze the

radiation characteristics of PLC (Power Line Communication) systems by numerical modeling. This should help in the development of a measurement methodology for these systems. Based on a review of currently operating PLC systems, technical parameters are established. A typical medium-voltage (MV) overhead three-phase power line wiring configuration was chosen to characterize real PLC wiring structures with various combinations of devices such as transformers, other conductors in the vicinity, and corner angles. An analysis of the models provides an understanding of unintentional radiation arising from real PLC systems.

Keywords: Power-line communication, Numerical simulation, EMC.

1. INTRODUCTION

With technological advancements in the past few years, PLC is becoming an attractive alternative for the supply of Internet services. This new technology uses existing MV and low-voltage (LV) power distribution facilities to carry radio frequency signals and provide access to broadband services. Before this technology can be deployed on a large-scale, the potential of MV wires to cause harmful interference with other users must be resolved. Some numerical analyses of the physical mechanisms have been conducted [1-3]. However, most of the simulations did not consider transformers, and used CW signals rather than actual wideband data signals. This has given rise to debate as to the accuracy and reliability of the analyses. The main contribution of this work is to overcome these two deficiencies.

Section 2 presents the electromagnetic chara-cterization of PLC systems with computer modeling. A powerful calculating engine based on the finite difference time domain (FDTD) method was used. A broadband Gaussian excitation source models the interaction of wideband data signals with PLC systems. The algorithm, presented in Section 3, is used to scale the source level according to the signal power level within certain bandwidths.

The PLC signals launched onto MV power lines do not simply travel point to point along the wires; they also escape as undesired radiated emissions. This radiation can be studied in the context of a “wire transmitting antenna”, whose reflection coefficient and input impedance at the injection point are the two key characteristics. The input

impedance is needed to scale the calculated field strength for a given power density level (PSD). These two characteristics are calculated and presented in Section 4. Radiation patterns are calculated at certain frequencies, to examine the connection between reflection coefficients and radiation from the power line.

Section 5 presents simulation results for near field distributions. A series of observation planes is specified, consisting of horizontal planes at various heights and two sets of vertical planes: 1) cuts through the line and 2) parallel to the line with different horizontal separation distances, each side of the power line. The field strengths on these planes and lines are obtained, showing how the field distributions vary with height and horizontal distances. This reveals the general propagation behaviour along the PLC signal-energized power lines in the immediate vicinity of the structure.

This work also investigates interference issues regarding compliance with FCC limits.

In Section 6, a brief summary of the results, together with some discussion regarding the reduction of the potential interference from PLC systems, are presented.

2. CHARACTERIZING PLC THROUGH COMPUTER MODELING

A representative MV power line wiring configuration is shown in Fig.1. This configuration is similar to the Amperion system which uses WiFi to end-users. It was chosen because Amperion has a trial system operating in Canada [4]. The configuration consists of three-phase parallel lines and one neutral line. These three- phase lines are spaced in a horizontally parallel configuration 1 meter apart, and 10 meters above a good soil ground ( 15,=rε 5= mSσ ) with 10 meters thickness. The neutral line is 0.9 meter above the central phase line. All lines are modeled using 1cm x 1cm cross section copper wires. In Fig.1, the wiring structure is supported by 19 wood poles spaced 30 meters apart. The wires run straight for 360 meters, and at the 13th pole turn right at a 60 degree angle for the remaining 180 meters. The line is thus divided into two straight sections for the purpose of the analysis. Fig.2 shows the structure on the top of each pole, where one steel cross arm and three porcelain insulators are used to tie the wires on each pole. The steel cross arm is in the vicinity of the lines and its effects on the near field distribution cannot be neglected.

At both ends, the three-phase lines are terminated with either transformers or resistive loadings. In either case the termination wiring of the three-phase lines was arranged in a wye-type configuration and connected to the neutral line. This line runs vertically into the earth to a conducting object 5 meters deep in the soil, as shown in Figs. 3 and 4.

The transformers act as large inductors in power line wiring (20 Ω resistor in series with 0.3 H inductor for a 2000KVA transformer 12.5 KV to 600V [5]). Second-order parasitic impedances at high-frequency (HF) are neglected.

A number of configurations, with combinations of the location of signal injection, the number of transformers, and their locations, was simulated. This paper presents data from only two cases. Case 1 is terminated with resistive loadings at both ends. The front end of the lines is loaded with 575 Ω resistors and the back end with 50 Ω resistors. This is quite similar to models used in NTIA simulations [1]. Case 2 is not only terminated with the two end-transformers, but also loaded with two middle transformers, each located at the centers of the two sections. This is more realistic, and produces significantly different radiating behaviour.

Fig.1: MV Power Line Geometry.

Fig.2: Wiring structure on the top of each pole.

Fig. 3: Resistive loads connected with the wye- connected phase line wiring of Case 1.

Fig.4: Power transformer with the wye-connected phase line wiring of Case 2.

3. NUMERICAL SIMULATION

PLC signals injected onto power lines are typically in the frequency range 5–45 MHz, with a bandwidth of a few MHz. Source parameters used are from the Ofcom report [6]. In the numerical simulation, a Gaussian pulse centered at 15 MHz with a 20 MHz bandwidth is used as a voltage excitation source. The source was launched onto the middle line at the 3rd pole (60m from the end), and its level initially set to 1 volt with horizontal polarization along the line. The source impedance was given a real value of 50 Ω in series with the source.

With the above initial setup, numerical simulations have been carried out. Note that numerical results, such as input impedance and reflection coefficients, are independent of the applied source level, whereas other

results such as electric field strengths and radiated power are directly proportional to the source level. Source levels vary with frequency as the Gaussian distribution. With an excitation normalization feature in the software (EMPIRE [7]), this variation with frequency caused by the applied Gaussian pulse itself, can be automatically removed.

The field strength, obtained from the initial simulation, must be scaled according to the power level launched onto the lines. The scaling factor with respect to the 1-volt source level must be calculated. Real world PLC systems use various spread spectrum techniques (for example OFDM) such that the injected power on the wiring is defined in terms of power spectral density (PSD) in dBm/MHz. Numerically, PSD can be calculated by taking the average of the transmitted power from the source over a certain bandwidth. Since the impedance in PLC systems varies in a complex manner with frequency, an algorithm was developed to calculate the scaling factor. This factor can be evaluated as:

1010 log transPP FSPRBW

= ×

(1)

Where: P = Average power over the spectrum (FSP) in dBm; FSP = Occupied frequency spectrum in MHz; RBW = Resolution bandwidth in MHz;

transP = Averaged transmitted power obtained at the signal injection port, that is

2

,1

1Re( )

n

trans i transi in

VP pn Z== × =∑ (2);

n = Number of sampling points in the FSP, related to the resolution (RBW);

,i transp = Transmitted power at the ith frequency component within the spectrum; V = Averaged excitation pulse voltage over the spectrum (FSP);

,1

1Re( ) Re( ),n

in i ini

Z Zn=

= ×∑ real part of the averaged

input impedance in Ω over the spectrum;

,i inZ = Input impedance at the ith frequency component within the spectrum; PSD = P/ FSP . (3)

Given a launched PSD in Equation (3), the scaling factor, V or the required excitation pulse voltage, can be obtained from Equations (1) to (3), that is

PSD31010 10 Re( )inV RBW Z−= × × × . (4)

A Matlab post-process program based on Equation (4) was used to calculate the factor, which scaled the power level up to the PSD level of -50 dBm/Hz (or 10 dBm/MHz) [6]. In each scaling evaluation, the frequency span (FSP) was set to 3 MHz and the resolution bandwidth (RBW) to 3.5 kHz.

4. REFLECTION COEFFICIENTS AND RESULTS IN FAR FIELD

Reflection coefficients for the two Cases have been obtained at the signal injection port, and presented as Figures 5 and 6, each having a 3.5 kHz frequency resolution.

It is well known that for an antenna to be a good radiator, its reflection coefficient should be very small, i.e., < –10 dB. However, it is undesirable that PLC-energized power lines act as good radiators because of potential interference to other radio services. As is shown in these plots, the probability of power line structures acting as efficient radiators is very small. Statistically, the probability can be evaluated as the ratio of the number of data points with very small reflection coefficients, over the total number of data points. As shown in these plots, the ratios in percentage for the two cases are 0.52% and 1.22%, respectively. Information about the bandwidths and positions of the resonant points is very helpful in predicting at which frequencies and with how wide a bandwidth the system could be radiating efficiently.

Fig. 5 shows the reflection coefficient values of Case 1, with only 5 resonances above 25 MHz. The maximum bandwidth of these resonances is less than 42 kHz (or 12 points). However, this case does not include the significant inductive effect, associated with transformers, on the radiation characteristics of PLC systems, and is thus not very realistic.

In contrast, Case 2 includes transformers. The inductive reactance of transformers considerably increases the impedance discontinuities, compared to the resistive loading used in Case 1. Numerical results for the reflection coefficient display a significant resonant behaviour. As shown in Fig.6, all resonances are less than 20 kHz in bandwidth. However they are all spread across the frequency band. Other cases studied show that the resonance behaviour of a particular power line varies greatly with impedance loading (transformer etc). Some configurations have very few resonances.

Extensive simulations have been conducted at a series of resonant and non resonant frequencies to see how closely a PLC wiring configuration acts as a good or poor radiator, and its relationship with reflection coefficient values. Two radiation patterns with the wiring configuration of Case 2 are presented here as examples. In these two examples, radiation patterns at the azimuth angles of o o0 and 90φ = are calculated at two frequencies, one at a resonant frequency and the other at a non resonant

frequency, but close to this resonant frequency. It is seen from Fig. 7 that the two patterns are almost the same in shape, but the main beam levels at the resonant frequency are about 10-13 dB higher than those at the non resonant frequency.

Fig.5: Reflection Coefficients (dB) versus frequency for Case 1.

Fig.6: Reflection Coefficients (dB) versus frequency for Case 2.

Fig.7: Elevation patterns at o0φ = oand 90 and at the frequencies of 23.1039 and 23.2505 MHz, for Case 2.

5. NUMERICAL RESULTS IN NEAR FIELD AND FCC COMPLIANCE

With power lines energized with RF signals, the resulting electromagnetic fields can be categorized as either guided or radiated. The characteristics of the radiated fields have been examined in the context of a “wire antenna” in the previous section. The guided fields carry the RF signals

along the line. Multiple reflections are generated by the impedance discontinuities. Each impedance discontinuity is the leading cause of emission from the wires. In this study, electromagnetic compatibility (EMC) was examined by consideration of the near field distributions.

Examples of electric field distributions in both horizontal and vertical polarizations are presented here, Figs 8 and 10, at 15 MHz and two meters above the ground. In the distributions, the center frequency is 15 MHz with a 3 MHz bandwidth. Field strengths were calculated and plotted versus distance along the lines, for a horizontal separation ox = 10 meters on each side of the power line. This is the height and horizontal separation distance suggested by an FCC Report and Order [8]. Results at other frequencies (5, 10, 20, 25, 30 and 35 MHz) and heights, as well as with other separations, are available [9].

The field distributions indicate that at the height of 2 meters above the ground, field strength levels in the vertical polarization are much higher than those in the horizontal polarization. This is due to the fact that the primary propagation mechanism of PLC emissions below the power line height and close to the ground is an approximate guided wave [3]. As a result, the field values versus distances calculated at the 10 meter separation are presented only for vertical polarization, and displayed in Figs. 9 and 11.

As shown in the field distributions of Figs. 8 to 11, even with a broadband Gaussian RF signal launched onto the power line, each frequency exhibits an approximate standing-wave behaviour, with a one-half wavelength spacing between adjacent maxima or minima. As discussed, the standing waves and associated radiation are due to the multiple reflections of signals at impedance discontinuities. However, significant distortions in the wave pattern manifest as changes both in the spacing and field strengths around these discontinuities. It is this dramatic change in field strength that has the potential to cause substantial high level interference to other users in the immediate vicinity of the PLC lines. The analysis also shows that the dissipative soil ground has only a slight effect on the guided fields at frequency below 30 MHz.

Fig.12 shows the two vertical field distributions (x-z plane) through the power line at 15 MHz. Vertical polarization electric field strengths were calculated at the horizontal distances of ym=74.64 and ym=72.20 meters for Case 1 and Case 2, respectively. These locations were determined from the global peak values of the fields at the line height. The largest EM fields are highly concentrated around the phase and neutral lines, that is, within a few meters (exhibited as the red region in each plot). Beyond that region, a dramatic drop in the field strength can be seen. For example, the field strength two meters above the ground and 10 meters away from the center conductor is at least 40 dB smaller than that at the power line height. However, significantly strong fields in the two cases (operating at -50 dBm/Hz) can still be observed within the

range of 20-40 meters. As discussed earlier, the results of Case 2 are more realistic.

Finally, based on the numerical results of Case 2, the interference issue regarding compliance with FCC limits was examined. The FCC Report and Order sets forth measurement procedures and limits for the RF energy emitted by PLC systems. For systems operating below 30 MHz, field strengths must not exceed:

lim 10 slant( / ) 29.5 40log (30 / )E in dB V m dµ = + , (5)

A measurement height of 2 meters above the ground, gives a slant distance slantd of 12.8 meters. To match the FCC Part 15.209 radiated emission measuring conditions, 3 dB was added to the FCC limits to compensate for the difference between using the quasi-peak detector (FCC) and the peak value (simulation) [6]. Equation (5) gives an FCC limit value of 47.3 dBµV/m. Numerical results for the field strengths in the frequency range of 21-24 MHz were calculated, with a transmitted PSD of -50dBm/Hz [6]. Following FCC measurement procedures, the field strengths at distances of 0, ¼, ½, ¾, and 1 wavelength down the line from the PLC injection point were calculated at a horizontal distance of 10 meters on both sides of the line ( 10ox = ± meters). These values are presented in Fig.13, and compared with the FCC limit. Within the frequency range of 21-24 MHz, the maximum field level is 58 dBµV/m, 18 dB higher than the FCC limit.

(a) (b) Fig.8: Electric field strength (dBµV/m) at 15 MHz and at

two meters above the ground for Case 1. (a): vertical and (b): horizontal polarization.

Fig.9: Vertical polarization field strength (dBµV/m) for Case

1 at 15 MHz versus the distance along the power line calculated at a horizontal separation distance of ox = 10 meters on each side of the power line, two meters above the line.

(a) (b) Fig.10: Electric field strength (dBµV/m) at 15 MHz and at

two meters above the ground for Case 2. (a): vertical and (b): horizontal polarization.

Fig.11: Vertical polarization field strength (dBµV/m) for

Case 2 at 15 MHz versus the distance along the power line calculated at a horizontal separation distance of

ox = 10 meters on each side of the power line, two meters above the line.

Fig.12: Vertical polarization electric field strength (dBµV /m) on the x-z cross sections at 15 MHz. (a): Case 1 at ym=74.64 m and (b): Case 2 at ym=72.20 m.

(a): down the line from the left side

(b): down the line from the right side

Fig.13: Vertical polarization Electric field strengths (dBµV/m) for frequency range 21-24 MHz, calculated at two meters above the ground for Case 2. (a): Down the line from the left side 10ox = − m and (b): down the line from the right side 10ox = + m.

6. SUMMARY AND CONCLUSION

This study has examined representative MV PLC systems with various wiring configurations, excited with a broadband Gaussian signal source. Extensive numerical simulations give the interference characteristics of PLC systems with a broadband signal. Three key conclusions of the analysis are:

• Transformers with significant inductive reactance cannot be ignored in the analysis of the radiation mechanisms of PLC systems. • PLC emission is governed mainly by two parameters: signal power and the nature of impedance discontinuities on the power line. This interaction is very complex and essentially unpredictable. Possible methods to reduce the emission such as a “Notching Tone” have been suggested, whereby the PLC system might itself determine automatically which parts of the spectrum are occupied by radio signals and avoid them. This determination could be based on the reflection coefficients at the signal injection port. • Interfering emissions are typically confined to the immediate vicinity of the PLC wire.

REFERENCES

[1] NTIA, “Potential Interference from Broadband over Power Line (PBL) Systems to Federal Government Radiocommunications at 1.7 - 80 MHz, Phase 1 Study”, Volumes 1 and 2, NTIA Report 04-413, April 2004.

[2] M. Gebhardt, F. Weinman and K. Dostert, Physical and Regulatory Constraints for Communication over the Power Supply Grid, IEEE Communications Magazine, May 2003, pp. 84-90.

[3] P. S. Henry, Interference Characteristics of Broadband Power Line Communication Systems Using Aerial Medium Voltage Wires , IEEE Communications Magazine, April 2005, pp. 92-98.

[4] Amperion web: Products and Solutions at: http://www.amperion.com/products.asp?id=98.

[5] Medium Voltage Transformer Parameters provided by Touchstone Energy Cooperatives at: http://polk-burnett. apogee.net/pd/aptit.asp /.

[6] Ofcom, “Amperion PLT Measurements in Crieff”, Technical Report, Sept. 20, 2005.

[7] User’s Manual for EMPIRE, version 4.13, Oct 2005, at: http://www.empire.de/.

[8] FCC, Report and Order, “In the Matter of Amendment of Part 15 regarding new requirements and measurement guidelines for Access Broadband over Power Line Systems” Adopted October 14, 2004, Released October 28, 2004.

[9] M. Zhang and W. Lauber, “Evaluation of Interference Potential of PLC System”, CRC Technical Report Phase I, VPTWS-TM-05-02-09 (Internal Publication), Ottawa, Canada, Feb., 2005.