Broadband Power Line Communications: The factors Influencing Wave Propagations in the Medium Voltage Lines

Size: px

Start display at page:

Download "Broadband Power Line Communications: The factors Influencing Wave Propagations in the Medium Voltage Lines"

Alberta Fields
5 years ago
Views:

1 Broadband Power Line Communications: he factors Influencing Wave Propagations in the Medium Voltage Lines Justinian Anatory, Nelson heethayi 2, Mussa Kissaka and Nerey Mvungi Faculty of Electrical and Computer Systems Eng. University of Dar es Salaam, anzania 2 Division for Electricity and Lightning Research, Uppsala University, Sweden. Abstract--this paper presents the influencing of load impedance, line length and branches on the performance of medium voltage power line communication (PLC) network. he power line network topology adopted here is similar to that of the system in anzania. Different investigation with regard to network load impedances, direct line length (from transmitter to receiver), branched line length and number of branches has been investigated. From the frequency response of the transfer function (ratio of the received and transmitted signal), it is seen that position of notches and peaks in the magnitude responses are largely affected in terms of attenuation and dispersion by the above said network parameters/configuration. he observations presented in the paper could be helpful in suitable design of the PLC systems for a better data transfer and system performance. Index erms-- Channel model, Broadband power line, Load impedance, ransfer function, ransmission lines, Communication channels. R I. INRODUCION ecently there has been a lot of interest to utilize power line infrastructure for broadband communication services. However, it has been observed that, there are a number of challenges associated with data transfer through such network. Existing power line topology (geometry and transmission voltage levels) varies from region to region and country to country. In countries like anzania, it has been observed that the medium voltage systems exhibit a potential scope to extend the broadband services to end users. he typical line length between far distribution transformer and the substation is about 4km, with around 2 (distribution) transformers distributed along the line (interconnected) leading to branched network configurations, with the connecting branch ranges between 3m to m. Moreover, the terminal loads experienced by such line configuration may not be always characteristic impedance or resistive loads. It is thus appropriate that with such medium voltage channel topology a number of case studies are to be carried out so as to provide guidelines for future optimal planning and design of communication systems. he sponsorship of SIDA/SAREC through the Faculty of Electrical and Computer Systems, University of Dar es Salaam, anzania is acknowledged. Different studies regarding the effect of load impedances, branches, etc., have been reported in Matthias et al. [], Pavlidou et al. [2], Zimmermann et al. [3], etc. Zimmermann et al. [3] and Mathias et al. [] pointed out that maximum distance for a lossless data transmission through power line is about 3m. Pavlidou et al. in [2] concludes that studies are still necessary to better understand and improve the performance of power lines for higher bit rate transmission. Researchers have investigated the variation time/frequency responses due to the influence of load impedance, line length and branches without mentioning exactly/clearly, what was the contribution of each parameter to the stochastic behavior (shown later) of channel responses. For example, the answers to the following questions should draw some conclusion for the performance improvement of medium voltage communication channel. How much the number of branches (N ob ) contribute to the signal response? How line lengths (d L ) from transmitter to the receiver and branched line length () influence the signal response? How the terminal load (infinite and low) impedances (Z inf & Z Low ) influence the signal response? Moreover in the literature, no complete systematic study was made to address all the above questions together, perhaps due to the complexities and uncertainties involved in the system. Consequently, the authors in this work are only attempting to seek some answers to the above questions. In the analyses to be presented the frequency response (or the transfer function as applicable) is calculated based on [4-6] as discussed next. II. POWER LINE CHANNEL MODEL It has been said that optimization of a transmission system is realizable only when a reasonably accurate channel model is available [7]; for investigating the power line network performance in detail. Hensen [8] proposed a simple power line model, the model was straight forward, the attenuation was increasing with frequency and do not take into consideration the multipath phenomenon. he second model was proposed by Philipps et al. [9], whose transfer function is given by (). In () out of N number of possible signal flow paths, each path delayed by timeτ i is multiplied by a complex factor ρ i (product of transmission and reflection factors). N i= j2πfτ H( f ) = i ρ ie () /7/$ IEEE. 27

2 2 he method in [9] was extended by Zimmermann et al. [3] to account for the attenuation of the signal flow and is given by (2). In (2) each path is characterized by weighting factor g (product of transmission and reflections factors) and path i length d i. he attenuation factor is modeled by the parameters a, a and k, which are obtained from measurements. Banwell et al. [] proposed a model which accounts for a multi-conductor configuration. However, the method used widely by power line researches is that due to Zimmermann et al. [3], as its easy to apply and due to model agreement with measurements. Although a popular model, it still suffers from some drawbacks, as highlighted in [5] and []. N H( f ) = g e i= i ( a + a k f ).d i e j2πf di v p In our investigation the method proposed in [4-6] is used. For a transmission line with multiple branches at a single node (e.g. node B in fig. ) the generalized transfer function can be represented by (3a). In (3a), N is the total number of branches connected at node B and terminated in any arbitrary load. Let n, m, M, H mn ( f ) and Lm, represent any branch number, any referenced (terminated) load, number of reflections (with total L number of reflections), transfer function between line n to a referenced load m, transmission factor at the referenced load m, respectively. With these the signal contribution factor α mn is given by (3b), where ρnm is the reflection factor at node B between line n to the referenced load m, γ n is the propagation constant of linen that has line length n. All terminal reflection factors PLn in general are given by (3c), except at source where ρ L = ρs is the source reflection factor [5]. Also Z s is the source impedance, Zn is the characteristic impedance of any terminal with source whilev s and ZL are source voltage and load impedance respectively based on fig.. (2).5μs. he output referenced voltage V m (f) in frequency domain is given by (4). he general time domain response is obtained by inverse Fourier transform of (4). Z Ln V m ( f ) H m ( f ) V s Z Ln Z = (4) + s Fig. : Power line network with multiple branches at a single node ABLE : PARAMEERS USED IN EAMPLE (E) LINE NUMBER Amplitude in Volts.2 [V] LINE LENGH LINE CHARACERISICS IMPEDANCE ERMINAED LOAD 2m Z =589 Z L=Z s=z 2 km Z 2=85 Z L2 =5 3 km Z 3=587 Z L3=Z 3=587 4 km Z 4=624 Z L4= 5 5 3m Z 5=57 Z L5= MODEL RESULS ime in Microsecond AP-EMP RESULS H m ( f ) = L N Lm M = n= Τ α H ( f ) n m (3a) mn mn M Ln mn M nm γ n( 2( M ) n ) α = P ρ e (3b) [us] 6 (file anatory etal26june5alh.pl4; x-v ar t) v:br Fig. 2: Simulation Results Model Result AP-EMP Results ρs n = ( source ) PLn = (3c) ρln, otherwise Consider now an example (E) of an arbitrary power line configuration with N =5 and parameters as given in table. he configuration was excited by a rectangular pulse with pulse width μs and amplitude of 2V. he pulse is shifted by For the above example the time domain response at Z 3 is given in fig. 2. he same configuration was implemented in the widely used AP-EMP software [] and the corresponding result is shown in fig. 2. It can be observed that the results are comparable. Now to derive a more generalized case applicable to any line configuration consider a power line network with 28

3 3 distributed branches as shown in fig.3, the transfer function of such network is given by (5a). In (5a) the parameters used has the same meaning as in (3a) and M is the total number of distributed nodes, d is any referenced node ( M ), H mnd (f) is the transfer function between line n to a referenced load m at a referenced node d similar to (3a). All parameters used in (5a, 5b and 5c) are similar to (3a, 3b and 3c) respectively but with reference node d. Next let us study the effects of various parameters using the above model. Amplitude in Volts MODEL RESULS.5 [V] ime in μs.3.2 H Fig. 3: Power line network with distributed branches M L N mm ( f ) = Lmdα mndhmnd( f ) d= M = n= V mm mnd Τ n m (5a) M Lnd M γ nd ( 2( M ) nd ) nmd e α = P ρ (5b) P Lnd ( f ρs = ρ Lnd, ) d = n = ( source ) otherwise (5c) Ldn = H ( f )* V mm s (6) Z Ldn + Z s Z [us] (f ile anatory etal26june7jmosi.pl4; x-v ar t) v :ZL33 Fig. 5: Simulation Results Model Result AP-EMP Results III. INFLUENCE OF LINE LENGH A. Length from transmitter to the Receiver We now consider a typical medium voltage line of anzanian power network. All the lines have the following per unit length parameters with ( Le=.9648μH/m, Ce=5.6627pF/m, R=.472/m). he configuration under study is given in fig. 6 with Z L =Z S =624. he line length AC was varied as 4km, 2km km and 5m, with length of BD constant at 3m. In the study here B is always the mid point of AC. Point D was terminated in 5. Fig. 4: Power line network with three distributed branches Consider the for example configuration fig. 4 with three distributed branches, the value of M is equal to three. he lengths were considered as ( = = = = = = =2m, =m) and load Z L2, Z L22 and Z L32 were terminated in open circuit while Z L3 and Z L33 in the line characteristic impedance. he configuration was excited by a rectangular pulse with pulse width.5μs and amplitude of 2V. he pulse is shifted by.5μs. he received response is shown in fig. 5. he same configuration was also implemented in AP-EMP the response is as in fig. 5. Note fig. 5a & 5b are comparable. Fig. 6: Power line network with a branch Fig. 7(a-d) shows frequency response of transfer function relating the load voltages at C and the sending end as given by (5) for 4km, 2km, km and 5m, respectively. From fig.7 the peak values of signal response were not attenuating Le μ D, = cosh πε D (m) is the π 2 a Ce = D cosh 2 a separation between two lines and a (mm 2 ) is the radius of conductors. 29

4 4 significantly with either frequency or line length. Any differences that are seen could be attributed to the finite losses due to the line series resistance. he position of notches in the signal response of medium voltage channel also does not depend on length from transmitter to receiver Notches Peaks Fig. 7: Simulation Results for Medium Channel of powerline link with one branch 4km 2 km km 5m. Fig.8 (a-d) shows the corresponding phase responses. It is observed that as the line length increases there are rapid changes in the phase response. he effects of branch length are studied next Fig. 8: Phase response for medium channel of powerline link with one branch 4km 2 km km 5m. B. Branched Length We now consider the configuration as given in fig. 6; i.e. the length of a line from point A to C was kept constant at 4km. he branched length was varied as (BD=5m, 3m, 6m and m) with B always at the mid of line AC. Point D was terminated in 5 as in the previous case and we repeat the same exercise as before of calculating the transfer characteristics with respect to the load at C. Figs. 9 a-d show the corresponding frequency responses for various branch line lengths. It is observed that in all cases the peaks of frequency responses was not either attenuating with frequencies or branch length similar to the earlier case. Where as, the position of the peaks and notches is case dependant unlike the previous case. he generalized expression for frequency position (f i in MHz) of i th peak in terms of branched line length ( in m) is approximately given by (8). Similarly, the positions of the notches are given by (9). f 7.7 i = ( 2 ), i =,,2... (8) i + f i 45.5 = i, i =,,2... (9) As the length of branched line increases the number of notches increase. he phase response for the case under study had similar behavior as in fig. 8a. It indicates that the length of branched transmission line still doesn t affect the phase response of medium voltage channel. Next let us study the effect of number of interconnected branches Fig. 9: Simulation Results for Medium Channel of 4km with one branch of length 5m 3m 6 m. IV. INFLUENCE OF NUMBER OF BRANCHES We consider the medium voltage channel with distributed branches as shown in fig. he number of branches was varied in the link between points A and J. he distance between points A and J was 4km, while all branches were 3m long. he number of branches was varied as 2, 5, and 5. Note that for each case the distances between the braches were equal and equally distributed between the link A and J. he terminations of all the branches were 5. Fig. a-d shows the corresponding frequency responses for different number of branches. It is observed that the positions of notches are not changed. But as the number of branches increases the attenuations of notched point tends to increase. he phase responses were comparable to previous case as shown in fig. 8a. Fig. : Power line medium voltage network with distributed branches 3

5 Fig. : Simulation results for medium voltage channel with distributed branches 2 branches 5 branches branches 5 branches V. INFLUENCE OF LOAD IMPEDANCE his study is emphasized here because, it is common that the loads at the termination of branched lines are not always line characteristic impedance or resistive, rather it could be a case dependant arbitrary load, like, low or high impedance (R type) compared to line characteristic impedance and practical load impedance (RL type) representing transformers, machines, etc. For discussions below we considered the configuration as in fig. 2. he length of line AC was kept constant and equal to 4km, while branch BE of length 3m is connected to the middle of line AB. he termination of point E was varied according to the given load impedance under investigation. Note that Z S and Z L are the characteristic impedances of the line AB. Fig. 2: Power line medium voltage network with a branch A. Resistive Load We consider the following load impedances with values 2, 2, 2, 624, 2k and 2k terminated at E. Note 624 is the characteristic impedance of the line BE. Fig. 6(ad) shows the frequency response for medium voltage channel for various termination impedance at BE. For the load impedances less than channel characteristic impedance the position of notches is unchanged with no attenuation (see figs. 3a and b). It is interesting to observe that when the load impedance lower the peaks are at db and the notches are at 4dB. As the load increase the peaks are increase and the notches decrease. As the load is characteristic impedance peaks and notches disappear. When the load impedance increase beyond the characteristic impedance the peaks and notches behave in the same way as if it were approaching lower impedances, but with a shift in their frequency position. Note the generalized expression for the frequency position of notches for the load impedance terminated in impedance less than line characteristic impedance is given by (9) while for termination impedance greater than line characteristic impedance is given by (). Similarly, the position of peak frequency for load impedance less than line characteristic impedance is given by (8), while for load impedance greater than line characteristic impedance is given by (2). he phase responses in the frequency ranges -MHz has similar features as in fig. 8a i fi = ( + 2 ), i =,,2... () Fig. 3: Results for a medium voltage channel with a branch terminated in low impedances (e) 2k (f) 2k f i 44 = i, i =,,2... (2) B. Inductive Loads he effect of inductive terminal loads is also worth investigating. For this an RL load termination at terminal E of fig. 2 was considered, where the inductance varies as.mh, mh, mh and mh, with constant resistance of 5. he frequency response has the same behavior as in fig. 3d. his indicates that the behavior of inductive load is like open circuit at high frequency as expected. he phase response had minor difference were observed compared to fig. 8a. VI. CONCLUSIONS he notches in Medium voltage channel do not depend on the line length from the position of the transmitter and receiver. he pulse distortion does not depend on the line length between transmitter and receiver. he position of notches in frequency response depends on the branched line length. he increase in branched line length tends to limit the available bandwidth in the medium voltage channel. he position of deep notches does not change with a number of distributed branches. As the number of distributed branches increases the amplitude of notched points tends to increase. he peak points in the frequency responses tend to fluctuate as number of distributed branches increases. 3

6 6 As the load impedance increases towards the line characteristic impedance the peaks attenuations tend to increase and notches tends to improve. As the termination impedance tends to an open circuit signals are less attenuated. he position of notches in frequency response tends to shift from one region to another region as the termination impedance changes. he sensitivity analysis presented here has important implication for the possible design considerations of PLC equipment. REFERENCES [] G. Mathias, M. Rapp and K. Dostert, Power Line Channel Characteristics and their Effect on Communication System Design, IEEE Communications Magazine, April 24. pp [2] N. Pavlidou, A.J. H. Vinck, J. Yazdani and B. Honary, Power Line Communications: State of the Art and Future rends, IEEE Communications Magazine, April 23. pp [3] M. Zimmermann and K. Dostert; A Multipath Model for the Powerline Channel, IEEE rans. On Communications, vol. 5, No. 4, April 22. pp [4] J. Anatory, N. heethayi, R. hottappillil, M.M. Kissaka and N.H. Mvungi, he effects of Interconnections and Branched Network in the Broadband Powerline Communications, International Gathering of Radio Science, India, 23 rd 29th October, 25. [5] J. Anatory, M.M. Kissaka and N.H. Mvungi, Channel Model for Broadband Powerline Communication, IEEE rans. On Power Delivery, Vol., January 27, pp [6] J. Anatory, M.M. Kissaka and N.H. Mvungi, Powerline Communications: he effects of Branches on the network performance, IEEE-ISPLC26, Florida, USA, March 26. pp [7] E. Biglieri and P. orino, Coding and Modulation for a Horrible Channel, IEEE Communications Magazine, May, 23. pp [8] C. Hensen and W. Schulz, ime Dependence of the Channel Characteristics of Low Voltage Power-Lines and its Effects on Hardware Implimentation, AEU Int`l. J. Electronics and Communication, vol. 54, no., Feb. 2, pp [9] H. Philipps, Modelling of Powerline Communication Channels, Proc. 3 rd Int l. Symp. Power-Line Communications and its Applications, Lancaster, UK, 999, pp []. Banwell and S. Gali, A Novel Approach to the Modeling of the Indoor Powerline Channel -Part I: Circuit Analysis and Companion Model, IEEE rans. On Power Delivery, vol. 2, no.2, April 25. pp [] H. W. Dommel, Electromagnetic transients program (EMP theory book), Bonneville Power Administration,

MODELLING OF BROADBAND POWERLINE COMMUNICATION CHANNELS

Vol.2(4) December 2 SOUTH AFRICAN INSTITUTE OF ELECTRICAL ENGINEERS 7 MODELLING OF BROADBAND POWERLINE COMMUNICATION CHANNELS C.T. Mulangu, T.J. Afullo and N.M. Ijumba School of Electrical, Electronic