Australasian Universities Power Engineering Conference (AUPEC 2004) 26-29 September 2004, Brisbane, Australia
MODELING HIGH FREQUENCY SIGNAL PROPAGATION OVER LOW VOLTAGE DISTRIBUTION LINES R. Keyhani and D. Birtwhistle Queensland University of Technology School of Electrical and Electronic Systems Engineering George Street, Brisbane, QLD 4001 Tel: (617) 3864 9124 Fax: (617) 3864 1516 Email:
[email protected]
Abstract This paper examines fundamentals of signal propagation over low-voltage (LV) bare conductor power lines. Overhead lines, service cables and loads are modelled using the electromagnetic power system transients program (EMTP) and LV systems are analysed with different load types and magnitudes. It is found that high loads significantly attenuate the high frequency signals on LV lines. A novel technique is proposed to improve the high frequency signal propagation characteristics in which coupling circuits disconnect loads at high frequencies.
1.
INTRODUCTION
Recently there has been considerable interest in the use of overhead power lines as a medium for data communications. Dostert [1] has demonstrated experimentally that data transmission at rates of up to 20 Mb/s is possible for distances of 1km over European low-voltage (LV) networks that presumably comprise predominantly mainly underground cable. Hooijen [2] has made experimental examinations of the data transmission capability and types of noise present on residential power line carrier (RPLC) systems comprising similar cable networks at frequencies of up to 95 kHz. In this study it was found that there were very low input impedances to the networks that were attributed to effects of many parallel loads and particularly to capacitors in street lights, and that data transmission was limited to about 500m. Zimmermann and Dostert [3] measured propagation at frequencies of 500 kHz to 20 MHz over LV cable networks. The networks investigated appeared to comprise multiple cables with no connected loads. Abbot [4] presents an interesting review of recent developments in RPLC technology. And concludes that whilst there has been considerable research in the field there are no commercially available products that can provide megabit communications over long distances (miles). It
is suggested that the market is dominated by companies offering electronic components that allow “some form of communication over power lines” but that economic feasibility of systems available is still to be demonstrated. Pavilidouin [5] gives a review of the current status of the RPLC area concludes that “many studies are still necessary to better understand and improve the performance of power lines for high-bit-rate transmission”. All previous work on RPLC appears to have been carried out on underground cable networks which have fundamentally different characteristics to the open wire, overhead, multiply-earthed-neutral neutral distribution networks that are common in much of Australia. All research to date appears to have been conducted by communications engineers who deal with the power networks as a communications channels and there has been no attempt made to gain understanding about which aspects of the networks are responsible for attenuation of signals. In this study we present a fundamental investigation into high frequency signal propagation over LV overhead networks of the type common in Australia. The approach taken is to make use of the considerable previous work that has been done by electrical power
engineers in modelling high-frequency power line phenomena and to use existing models to undertake a systematic examination of factors contributing to signal attenuation. We also present a new solution that could remove existing limits to signal propagation and discuss potential application of the technique. 2. SIMULATION OF HIGH-FREQUENCY SIGNAL PROPAGATION ON LV OVERHEAD LINES A little previous research has been done on characterising signal propagation in overhead power lines. In this paper the characteristics of signal propagation over low-voltage MEN distribution lines have been modelled using the Alternative Transient Program (ATP) [6]. In this simulation the effects of loads and lines are studied for 4-wire, 3-phase lowvoltage overhead power lines of flat configuration of the type commonly used in Australia. Details of typical parameters of the lines are included in Table 1. Parameter Span length (m) Conductors height (m) Sag (m) Ground resistivity (Ω-m) Spans per section Conductor type Conductor resistance (mΩ/m) Pole electrode resistance (Ω) Consumer electrode resistance (Ω) Service length (m)
Value 50 15 2 100 2 ACSR 0.1237 (phase) 0.7388 (neutral) 10 10 50 m per section
Table 1. Parameters of simulated LV network The Marti model [7] of three-phase lines provides better representation during transient events and has fixed frequency responses that are similar to those provided by lumped-circuit models. The Marti model has been implemented in the Alternative Electromagnetic Transients Program and therefore this program has been used to examine the limitations to signal propagation on LV lines with a view to using the same models for later investigations of fault conditions. In the ATP simulation the effect of loads and grounding conditions on signal propagation were examined. All load are assumed to be lumped because the load size usually is much smaller than the wavelength of highest frequency in the simulations. Extensive simulation has been done and it
is found that low impedance loads attenuate significantly the communication signals. Figure 1. shows the result of signal propagation simulations for a radial low voltage distribution line. Electric loads are distributed and connected at the end of each section. The amount of impedance connected for each phase at each section of line is indicated for each plot. For purely resistive loads (unity power factor) the attenuation of signal increases with decrease of load impedance. Results of ATP simulations are shown in Figure 1. Figure 1(a) shows the line attenuation with purely resistive loads of 1 Ω (53 kW), 10 Ω (5.3 kW), 100 Ω (0.53 kW), and 1000Ω (0.053 kW) (powers are given per phase for each pole). It can be seen that there is significant increase in attenuation with frequency at the higher levels of loading and that with the modest loading imposed by 10 Ω resistance the signals rolls of at around 10 kHz. Higher loadings (low resistances) cause the signal to roll of at even lower frequencies. The attenuation is caused by increase of reactance of the line with frequency causing a voltage divider effect with the load resistances. Figure 1(a) also shows that at higher frequencies standing-wave behaviour causes peaks in the attenuation-frequency characteristics at frequencies with periods that correspond to multiples of the line propagation time. Figure 1(b) shows the attenuation-frequency characteristics of a purely inductive load. For this condition it can be seen that with loads of over about 1mH there is little reduction in attenuation for frequencies up to 100 kHz and there is considerable amplification of the signal at standing-wave frequencies. The purely inductive load is not a realistic practical condition for LV powerlines. Figure 1(c) shows the effects of placing inductance in series with a resistive load of 1000 Ω. In this case it can be seen that for the values of inductance examined there is little attenuation up to standing wave frequencies and at these frequencies there is considerable signal amplification however series inductance introduce some voltage drop at 50Hz. The very high amplification obtained at standing-wave frequencies is no doubt due to limitation of current penetration into the loads by the reactance of the series inductors causing the line to behave as an unloaded transmission line. In practice it is possible that lumped capacitance in the network will tend to reduce this amplification.
20
20
R=1000
0
10
Attenuat ion [ db]
Attenuation [db]
R=100 -20
R=10
-40
L=10mH
0
L=1mH L=0.1mH
-10
-60
-80
0.001
R=1
0.1
10
103
-20
105
0.001
0 .1
Frequency [kHz] [kHz] Frequency
10
103
105
Frequency [kHz]
1(a)
1(b)
20
20
C = 0.1µF C = 10µF C = 1000µF
15
0
10
-20
Atte nua tion [db]
Attenuation [db]
R=1000 for all curves
5
-40
L=10mH
0
-60
L=1mH L=0.1mH
-5
0.001
0.1
10 Frequency [kHz] Freq uency [kH z]
103
105
-80
0.001
0.1
10
103
105
Frequency [kHz]
1(c)
1(d)
Fig. 1. Effect of frequency and consumer loads on the attenuation of phase voltage signals on a LV overhead line (a) Resistive load (b) Pure Inductive load (c) R and L load (d) Pure Capacitive load
Figure 1(d) shows the attenuation-frequency characteristics for purely capacitive loads. In this case it can be seen that high values of shunt capacitance load causes low-frequency roll off of signal in a similar fashion to that obtained with high resistive loads. With the lower capacitance value of 0.1 µ F it can be seen that the roll-off occurs at about 100 kHz. With all capacitance loadings there are superimposed peaks that may be due to resonances between the shunt capacitances and adjacent line inductance. It can be seen that variation in loads normally occur over a daily cycle, which has a profound effect on signal propagation in PLC applications.
3.
IMPROVED SIGNAL PROPAGATION
Results of simulations showed that the low impedance capacitive and resistive loads have drastic attenuation affects on communication signals. However a small inductor in series with load can improve the signal propagation and reduce significantly attenuation. Adding a small inductor to low impedance loads has no considerable effect on 50Hz power transfer but can improve the amount of received signal at the end of line. However inductor blocks the passing of high frequency signals to electric loads. Thus the load impedance can be increased to improve the propagation characteristics of
L
Power lines
C
C n :1
ZL
200
Effective load resistance from supply side of coupler circuit Actual load resistance
the distribution system but the signal will not be received at the loads or power plugs. Ideally the coupling circuit must increase the impedance seen from high frequency source but it must not affect lowfrequency power signals. This means the coupling circuit should isolate the low impedance load from high frequency source and not affect low-frequency signals. Figure 2 shows a coupling circuit which provides two different paths for low and high frequency signals.
160
120
R=1Ω 80
R=10Ω R=100Ω R=1000Ω
40
0
0.001
103
10
0.1
105
107
Frequency [kHz]
Figure 3. Load impedance reflected into supply by new coupling circuit (Equivalent circuit as in Figure2, L = 0.1 mH , C = 1µF ) (Load values are indicated)
New coupler circuit Figure 2. Proposed Coupling Circuit
At low frequencies the series impedance of inductance impedance is very low and the voltage drop across it can be neglected. The capacitors however show very high impedance and effectively disconnect the transformer. It can be seen the coupler does not have a noticeable affect on low-frequency signals and power delivery.
1
1
0.8
0.8
L= 0.1 mH for all curves
0.6
Filter's output voltage Filter's input voltage
Filter's output voltage Filter's input voltage
At high frequencies the small series inductance has a very high impedance. The circuits and the capacitors impedances will be very small. The load will therefore be coupled to high-frequency sources attached to remote parts of power lines by the transformer. In this case the 2 impedance will be seen n times larger from the supply side.
0.4
0.2
L=10mH L=1mH
L=0.1mH
L=0.01mH
C=1µF for all curves 0.6
0.4
0.2
C=10µF C=1µF C=0.1µF C=0.01µF 0 0.001
0.1
10
Frequency [kHz]
4(a)
103
105
0
0.001
0.1
10
Freq uency [kH z]
4(b)
Figure 4. Transfer characteristics of new coupling device (Equivalent circuit as Figure 2.)
103
105
From the above discussion it can be understood that the coupler improves the signal propagation characteristics by the fact that in high frequencies the load is increased n 2 times. The effect of the transformer is not only to reduce the signal n times at the output (load side) but also to result stronger signal being received at output. Figure3. shows the ratio of impedance change for the whole frequency range. As it can be seen the for frequencies above 1MHz the ratio of transferred load impedance to load impedance itself is 100 which means power lines effectively will be disconnected from load impedance but at the same time one tenth of signal on power line can be received at loads or power plugs.
Figure 4 shows the ratio of output voltage for different values of L and C parameters. Figure 4(a) implies that the change of capacitors affects the ratio of voltage at high frequency signals while figure 4(b) indicates that the change of inductor value affects the low frequency voltage ratio. In order to avoid any voltage drop on normal power transfer the value of inductance must be selected to be as small as possible. Here in the simulation L is chosen to be 0.1mH, which makes just a 0.75V voltage drop for a 10 Ω load at power frequency. The capacitance value is required to be as small as possible to give better isolation of high frequency signals from power distribution signals but smaller values of capacitance causes more limited bandwidth.
0
40
R=1Ω
R=10Ω VL1 0 VL1
-40
Attenuation [db]
Attenuation [db]
VL2
-40
VL2
-80 VL3 -80 VL3
-120
0.001
0.1
103
10
Frequency [kH z]
-120
105
0.001
0.1
10
103
Frequency [kHz]
5(a)
105
5(b)
40
40
R=100Ω
R=1000Ω
0
0
Attenuation[db]
Attenuation [db]
VL3
VL1 -40 VL2
VL1 -40
VL2
VL3
-80
-120
-80
0.001
0.1
-120 10
Freque ncy [kHz]
5(c)
103
105
0.001
0.1
10
Frequen cy [KHz]
5(d)
Figure 5. Effect of new coupler on signal received at load V L1 : Case with new coupling device- voltage at supply side of coupler V L 2 : Case with new coupling device- voltage at load side of coupler V L3 : Case without new coupling device- voltage at load
103
105
4. EFFECTS OF PROPOSED COUPLING CIRCUIT ON SIGNAL PROPAGATION The proposed coupler was applied to the model of the power distribution system described in section 2 and simulation done with a range load resistances where couplers are applied to each load. The results of simulation for resistive loads from 1 Ω to 1000 Ω are shown in Figures 5(a) to 5(d). Each Figure shows the voltage received at the supply side of coupler VL1 , Voltage at the load when the coupler is included V L 2 and the voltage received at the load with no coupler V L 3 . As it can be seen from Figures 5(a) to 5(d) V L1 is ten times greater than the V L 2 for high frequency signals because of the transformer ratio. For R=1,5,10,100 Ω the received signal at load V L 2 is 40 db greater than the voltage at the same point when no coupler is used. It can be seen from Figures 5(a) to 5(c) that the coupler significantly improves the propagation characteristics of power distribution lines. The effect of using couplers is to increase the impedance of all loads for high frequency (communication) signals. In distribution systems without couplers most of the signal would be reflected back to the supply because of mismatch between line and load impedances while in the same system with couplers less signal will be reflected back to the high frequency source. VL1 is which is available at pole junctions are 10 times stronger than V L 2 and it can be used for reliable data integrity checks.
frequency signals but at the same time it can couple part of received signal to loads to facilitate data transmission. The coupler circuit increases the magnitude of the received signals and makes it possible for them to be received at power plugs. The result of simulations for resistive loads shows that the proposed new coupler makes the communication signal to be transmitted over power lines even in high load situations.
6.
REFERENCES
[1]
K. M. Dostert, "Power lines as high speed data transmission channels-modelling the physical limits VO - 2," presented at Spread Spectrum Techniques and Applications, 1998. Proceedings , 1998 IEEE 5th International Symposium on, 1998.
[2]
O. G. Hooijen, "A channel model for the residential power circuit used as a digital communications medium," Electromagnetic Compatibility, IEEE Transactions on, vol. 40, pp. 331-336, 1998
[3]
M. Zimmermann and K. Dostert, "A multipath model for the powerline channel," Communications, IEEE Transactions on, vol. 50, pp. 553-559, 2002.
[4]
R. E. Abbott, "High speed power line communications," presented at Power Engineering Society Summer Meeting, 2002 IEEE, 2002.
For R of 1000 Ω it can be seen that the received signal at load is less than the case when no coupler is used. The coupler makes this impedance greater and improves the degree of signal mismatch. So the received signal will be weaker in this case but it is big enough to be detected. 5.
[5] N. Pavlidou, A. J. Han Vinck, J. Yazdani, and B. Honary, "Power line communications: state of the art and future trends," Communications Magazine, IEEE, vol. 41, pp. 34-40, 2003.
CONCLUSION
Signal propagation over low voltage distribution system was modelled using ATP program. The frequency dependent parameters skin effect and ground effects were incorporated in the model. The results of simulation show that low impedance loads on distribution system can drastically attenuate the high frequency signals A new coupling circuit was proposed to improve the high frequency signal propagation over LV power reticulation lines. The coupler has no effect on low frequency signals and normal power transfer. However it effectively decouples the electric loads from high
[6]
Alternative Electromagnetic Transients Program Rule Book, Canadian-American EMTP Users Group, 1997
[7]
J. R. Marti, ”Accurate Modelling of Frequency Dependent Transmission Lines in Electromagnetic Transient Simulations”, IEEE Transactions on Power Applications and Systems, Vol. 101, No. 1, pp. 147-155, January 1982.
